Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Non-destructive detection of corrosion of epoxy coated rebar Goffin, Brigitte 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2018_february_goffin_brigitte.pdf [ 12.03MB ]
Metadata
JSON: 24-1.0362873.json
JSON-LD: 24-1.0362873-ld.json
RDF/XML (Pretty): 24-1.0362873-rdf.xml
RDF/JSON: 24-1.0362873-rdf.json
Turtle: 24-1.0362873-turtle.txt
N-Triples: 24-1.0362873-rdf-ntriples.txt
Original Record: 24-1.0362873-source.json
Full Text
24-1.0362873-fulltext.txt
Citation
24-1.0362873.ris

Full Text

 NON-DESTRUCTIVE DETECTION OF CORROSION OF  EPOXY COATED REBAR  by BRIGITTE GOFFIN   A DISSERTATION IN PARTIAL FULFILLMENT OF THE REQUIREMENTS  FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Civil Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   December 2017 ©Brigitte Goffin, 2017 Abstract   ii Abstract Deteriorating infrastructure is an acute and dangerous problem, which is often caused by the corrosion of concrete reinforcement. Marine structures and bridge decks, where sea water and de-icing salts lead to chloride ion diffusion into the concrete are particularly at risk. Epoxy coated rebar (ECR) is a popular choice for the latter structures. However, corrosion of ECR, which occurs due to coating damage, poses a challenge to non-destructive corrosion detection.  This study investigated the corrosion behaviour, accelerated corrosion and non-destructive corrosion detection of ECR.  The electrochemical corrosion behaviour of ECR in simulated concrete pore solutions was studied and compared to uncoated rebar (UCR). The polarized area of ECR was related but not proportional to the sodium ion concentration of the test solution. Furthermore, ECR was more susceptible to corrosion than UCR, particularly in the presence of NaCl and in NaHCO3 solution. A test solution of Na2CO3 and NaHCO3 led to the formation of a very fragile passive layer, that grew slowly but continuously. However, the protective layer was sensitive to even small amounts of NaCl.  Corrosion of ECR was accelerated in neutral and alkaline NaCl solutions as well as in concrete. Neutral and alkaline environments promoted coating holiday and undercoating corrosion, respectively. Part of the undercoating corrosion process was cathodic delamination, whose acceleration prior to corrosion acceleration slowed down the lateral corrosion expansion. Once corrosion had expanded across the entire surface, the subsequent corrosion rate was not affected by the initial cathodic disbondment.  Successful ECR corrosion detection was limited with the linear polarization resistance and ground penetrating radar method. However, concrete properties such as maturity, moisture and chloride content had a significant effect on the measurements. Corroded bars affected the Hall effect (HE) voltage to a lesser extent than intact rebar. Furthermore, corrosion of ECR led to higher concrete surface and lower bar temperatures during active infrared thermography (IRT) tests. The thermal results of ECR opposed those of UCR. The HE and IRT tests showed that the effects of corrosion on the thermal and magnetic behaviour during induction thermography would complement and oppose each other for UCR and ECR respectively. Lay Summary   iii Lay Summary Epoxy coated rebar (ECR) is often used in structures exposed to chloride-rich  environments. The epoxy coating is a barrier between the concrete and steel and is designed to protect the steel from corrosion.  However, damage in the coating often leads to delamination and subsequent corrosion expansion under the coating. Unfortunately, this damage becomes apparent only after significant corrosion has occurred. This study investigated the corrosion behaviour of ECR and compared it to that of uncoated rebar (UCR). ECR with coating damage was found to be more susceptible to corrosion in chloride rich environments than UCR. Furthermore, the suitability of conventional and new non-destructive testing methods for ECR corrosion detection were assessed. Conventional methods such as Ground Penetrating Radar (GPR) and Electrochemical Measurements (EM) showed limited success. Two other novel methods including Infrared Thermography and Hall Effect were relatively more successful. Preface   iv Preface This dissertation is the original work by Brigitte Goffin and all experimental work was carried out at UBC. Financial support was provided by the British Columbia Ministry of Transportation and Infrastructure, by the Natural Sciences and Engineering Research Council of Canada (NSERC) through the Sustainable Building Science Program (SBSP) and by the India-Canada Centre for Innovative Multidisciplinary Partnerships to Accelerate Community Transformation and Sustainability (IC-IMPACTS). The research work presented in chapters 3.1 and 4.1 has led to the following journal paper submission: Brigitte Goffin, Nemkumar Banthia, “Corrosion behaviour of epoxy coated and uncoated rebar in simulated concrete pore solutions” The research work presented in chapter 3.3.2.1.2, 3.3.2.2.2 and 4.3.2.2 has led to the following journal paper submission: Brigitte Goffin, Nemkumar Banthia, Noboru Yonemitsu “Effect of corrosion on the thermal behaviour of epoxy coated and uncoated rebar embedded in concrete” The research work presented in chapter 3.3.2.1.1, 3.3.2.2.1 and 4.3.2.1 has led to the following journal paper submission: Brigitte Goffin, Nemkumar Banthia “Electromagnetic detection of corrosion of epoxy coated and uncoated rebar by means of a Hall effect sensor”    Table of Contents   v Table of Contents Abstract .................................................................................................................................................. ii Lay Summary ......................................................................................................................................... iii Preface .................................................................................................................................................. iv Table of Contents ................................................................................................................................... v List of Tables ........................................................................................................................................ viii List of Figures ......................................................................................................................................... x List of Symbols ...................................................................................................................................... xix 1 Introduction ........................................................................................................................... 1 1.1 Motivation ...................................................................................................................... 1 1.2 Scope .............................................................................................................................. 4 2 Theoretical Background ......................................................................................................... 7 2.1 Corrosion ........................................................................................................................ 7 2.1.1 Thermodynamics ........................................................................................................ 8 2.1.2 Kinetics ..................................................................................................................... 16 2.2 Accelerated Corrosion .................................................................................................. 24 2.2.1 Potentiodynamic Polarization .................................................................................. 27 2.3 Corrosion in Concrete .................................................................................................. 30 2.3.1 Rebar ........................................................................................................................ 30 2.3.2 Corrosion Behaviour of Epoxy Coated Rebar ........................................................... 33 2.3.3 Effect of Chloride and Rebar Surface Conditions ..................................................... 35 2.4 Corrosion Detection of Epoxy Coated Rebar ............................................................... 36 2.4.1 Conventional Non-Destructive Testing .................................................................... 37 2.4.2 Induction Heating ..................................................................................................... 44  Table of Contents   vi 3 Experimental Procedure....................................................................................................... 53 3.1 Corrosion Behaviour ..................................................................................................... 53 3.1.1 Specimens ................................................................................................................ 53 3.1.2 Methodology ............................................................................................................ 54 3.2 Accelerated Corrosion .................................................................................................. 59 3.2.1 Concrete ................................................................................................................... 59 3.2.2 Uncoated Rebar........................................................................................................ 59 3.2.3 Epoxy Coated Rebar ................................................................................................. 60 3.3 Corrosion Detection ..................................................................................................... 65 3.3.1 Conventional Non-Destructive Testing .................................................................... 65 3.3.2 Hall-Effect and Thermography Specimens ............................................................... 71 4 Results .................................................................................................................................. 81 4.1 Corrosion Behaviour ..................................................................................................... 81 4.1.1 Polarizing Area.......................................................................................................... 81 4.1.2 Common Simulated Concrete Pore Solutions .......................................................... 83 4.1.3 Chloride .................................................................................................................... 87 4.1.4 Carbonation .............................................................................................................. 96 4.1.5 Chloride and Carbonation ...................................................................................... 104 4.2 Accelerated Corrosion ................................................................................................ 109 4.2.1 Uncoated Rebar...................................................................................................... 109 4.2.2 Epoxy Coated Rebar ............................................................................................... 113 4.3 Corrosion Detection ................................................................................................... 124 4.3.1 Conventional Non-Destructive Testing .................................................................. 124 4.3.2 Hall-Effect and Thermography ............................................................................... 165 5 Conclusions ........................................................................................................................ 212 5.1 Corrosion Behaviour ................................................................................................... 212  Table of Contents   vii 5.2 Accelerated Corrosion ................................................................................................ 213 5.3 Corrosion Detection ................................................................................................... 213 5.3.1 Conventional Non-Destructive Testing .................................................................. 213 5.3.2 Hall-Effect Sensing .................................................................................................. 214 5.3.3 Thermal Detection.................................................................................................. 214 5.4 General ....................................................................................................................... 215 6 Future Research ................................................................................................................. 216 References .......................................................................................................................................... 217   List of Tables   viii List of Tables Table 3.1: Short term (ST) and long term (LT)  testing procedure ....................................................... 56 Table 3.2: Composition of test solutions ............................................................................................. 58 Table 3.3: Concrete mix design ............................................................................................................ 59 Table 3.4: Accelerated corrosion procedure of series A, B and C of specimens later using for Hall Effect (HE) sensing and Infrared thermography (IRT) corrosion detection experiments .............................. 63 Table 3.5: Testing procedure ............................................................................................................... 67 Table 3.6: Amount of chloride per cubic meter of concrete for each slab .......................................... 68 Table 3.7: Specifications of the steel bars (SB) embedded in concrete ............................................... 71 Table 4.1: Polarized area of base solutions during LT testing .............................................................. 82 Table 4.2: Mass loss of uncoated rebar (UCR) in g ............................................................................ 110 Table 4.3: Mass loss of uncoated rebar (UCR) in %............................................................................ 111 Table 4.4: Maximum crack width ....................................................................................................... 111 Table 4.5: Legend corresponding to Table 4.6 ................................................................................... 114 Table 4.6: Distance from corrosion location to rebar’s end for the entire specimen series ............. 115 Table 4.7: Mass loss of ECR in g ......................................................................................................... 120 Table 4.8: Mass loss of ECR in % ........................................................................................................ 121 Table 4.9: Legend corresponding to the diagrams of the LPR, Half cell and resistivity measurements in chapter 4.3.1.1 ................................................................................................................................... 124 Table 4.10: Peak to peak voltages of the reflected wave in mV ........................................................ 139 Table 4.11: Delay between the arrival time of the reflected wave in relation to the direct wave in ns ............................................................................................................................................................ 139 Table 4.12: Peak amplitude of the direct wave in mV ....................................................................... 139 Table 4.13: Colour coding of GPR waves ............................................................................................ 158 Table 4.14: Vp-p of the reflected wave in mV ..................................................................................... 163 Table 4.15: Peak amplitude of the direct wave in mV ....................................................................... 163 Table 4.16: Delay in arrival time between the DW and RW in ns ...................................................... 164 Table 4.17: Median gap between the curves of bar 7 and 9 and the curves of 8 and 10 in concrete under dry conditions with a minimum spacing of 7.5 cm .................................................................. 174 Table 4.18: Relative permeability of sample constituents ................................................................. 177   List of Tables   ix Table 4.19: Median gap between the curves of corroded and intact UCR in concrete under dry conditions with a minimum spacing of 10 cm ................................................................................... 177 Table 4.20: Median gap between the curves of corroded and intact ECR in concrete under dry conditions with a minimum spacing of 10 cm ................................................................................... 181 Table 4.21: Median gap between the curves of bar 7 and 9 and the curves of 8 and 10 in concrete under saturated conditions with a minimum spacing of 7.5 cm ....................................................... 184 Table 4.22: Median gap between the curves of corroded and intact UCR in concrete under saturated conditions with a minimum spacing of 7.5 cm .................................................................................. 187 Table 4.23: Median gap between the curves of corroded and intact ECR in concrete under saturated conditions with a minimum spacing of 10 cm ................................................................................... 190 Table 4.24: Summary of minimum measurement parameters most suitable for corrosion detection ............................................................................................................................................................ 197   List of Figures   x List of Figures Figure 1.1: Flowchart of the scope of the experimental study .............................................................. 6 Figure 2.1: Schematic illustration of the corrosion process of a metal M in hydrochloric acid (adapted from (Jones 1996)) ................................................................................................................................. 7 Figure 2.2: Potential-pH diagram for iron (adapted from (Jones 1996)) ............................................. 13 Figure 2.3: Conditions of stability for water, oxygen and hydrogen .................................................... 15 Figure 2.4: Evans Diagram .................................................................................................................... 19 Figure 2.5: Effect of solution velocity, temperature and concentration on cathodic concentration polarization (adapted from (Jones 1996)) ............................................................................................ 21 Figure 2.6: Mixed potential theory for a metal M in acid solution ...................................................... 22 Figure 2.7: Effect of noble second oxidizer, Fe3+/Fe2+, with high enough exchange current density on mixed potential (adapted from (Jones 1996)) ..................................................................................... 23 Figure 2.8: Noble second oxidizer, Fe3+/Fe2+, with too small of a exchange current density to affect the mixed potential (adapted from (Jones 1996)) ..................................................................................... 24 Figure 2.9: Anodic polarization curve of an active-passive metal........................................................ 25 Figure 2.10: Anodic polarization of an active-passive metal ............................................................... 26 Figure 2.11: Simplistic schematic of polarization of the working electrode (WE) by the auxiliary electrode (AUX) with a potentiostat and a reference electrode (REF) ................................................ 28 Figure 2.12: Anodic potentiodynamic polarization plots in theory (left) and in practice (right) of an active passive metal in the passive state (adapted from (Princeton Applied Research 2017)) ........... 29 Figure 2.13: Anodic potentiodynamic polarization plots in theory (left) and in practice (right) of an active passive metal in the active state (adapted from (Princeton Applied Research 2017)) ............. 29 Figure 2.14: Potential-pH diagram for iron in water (left) (adapted from (Jones 1996)) and in chloride contaminated water (right) (adapted from (Moreno et al. 2004)) ...................................................... 31 Figure 2.15: Electrochemical corrosion cell of steel in concrete ......................................................... 31 Figure 2.16: Illustration of the corrosion process of fusion bonded epoxy coated steel (adapted from (Nguyen and Martin 2004)) .................................................................................................................. 34 Figure 2.17: Half-cell potential equipment .......................................................................................... 38 Figure 2.18: AC resistivity NDT method with 4 electrode array Wenner probe .................................. 40   List of Figures   xi Figure 2.19: Profile view of a reinforced concrete slab with corresponding 2D GPR B-scan .............. 42 Figure 2.20: Setups of UPV Transmitter T and Receiver R ................................................................... 44 Figure 2.21: Circular magnetic field (red) induced by a current (blue) flowing through a straight conductor ............................................................................................................................................. 45 Figure 2.22: Movement of holes and electrons inside a semiconductor (adapted from (Razeghi 2009)) .............................................................................................................................................................. 47 Figure 2.23: Lorentz force on holes inside a semiconductor (adapted from (Razeghi 2009)) ............. 47 Figure 2.24: Holes in the Hall electric field (adapted from (Razeghi 2009) ......................................... 48 Figure 3.1: UCR (top) and ECR (bottom) sample for the electrochemical corrosion behaviour tests . 53 Figure 3.2: Electrochemical testing setup ............................................................................................ 54 Figure 3.3: Schematic of the electrochemical testing setup ................................................................ 55 Figure 3.4: Specimens of series C ......................................................................................................... 62 Figure 3.5: Dimensions of small lab specimens in mm ........................................................................ 66 Figure 3.6: Coating damage locations of small lab specimens............................................................. 67 Figure 3.7: Large outdoor specimens including damage locations ...................................................... 69 Figure 3.8: Steel bar (SB) reinforced concrete specimens for Hall effect (HE) sensing with concrete cover cc and dimensions in mm ........................................................................................................... 72 Figure 3.9: UCR reinforced concrete specimens for Hall effect (HE) sensing with a concrete cover of 30 mm and dimensions in mm ............................................................................................................. 73 Figure 3.10: ECR with four coating holiday locations........................................................................... 73 Figure 3.11: ECR reinforced concrete specimens for Hall effect (HE) sensing with a concrete cover of 28 mm and dimensions in mm ............................................................................................................. 74 Figure 3.12: UCR reinforced concrete specimens for infrared thermography (IRT) corrosion detection with a concrete cover of 28 mm and dimensions in mm ..................................................................... 75 Figure 3.13: ECR with longitudinal hole with a diameter of 2.03 mm ................................................. 76 Figure 3.14: ECR thermal detection specimen ..................................................................................... 76 Figure 3.15: ECR reinforced concrete specimens for infrared thermography (IRT) corrosion detection with a concrete cover of 28 mm and dimensions in mm ..................................................................... 77 Figure 3.16: Setup of sensor (S) and magnet (M) at sensor-magnet spacing (SM) in relation to the rebar under investigation at a distance (R) ................................................................................................... 78 Figure 3.17: Hall effect sensing setup with concrete (left) and PVC (right) as a space holder ............ 79 Figure 3.18: Infrared thermography setup .......................................................................................... 80   List of Figures   xii Figure 4.1: Anodic PDP curves for 3 common simulated concrete pore solutions .............................. 83 Figure 4.2: Open circuit potential of ECR and UCR in 3 common simulated concrete pore solutions 83 Figure 4.3: Anodic PDP curves for 3 common simulated concrete pore solutions after 4 weeks of exposure to aerated solution ............................................................................................................... 85 Figure 4.4: Open circuit potential of ECR and UCR in 3 common simulated concrete pore solutions after 4 weeks of exposure to aerated solution .................................................................................... 85 Figure 4.5: Anodic PDP curves for 0.9M NaOH solution with varying NaCl concentration ................. 88 Figure 4.6: Open circuit potential of ECR and UCR in 0.9M NaOH solution with varying NaCl concentration ....................................................................................................................................... 88 Figure 4.7: Anodic PDP curves for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution ........................................................................................................... 91 Figure 4.8: Corrosion rates for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution (LT) ........................................................................................................ 91 Figure 4.9: OCP pre and post potentiodynamic polarization for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution ............................................................ 92 Figure 4.10: Anodic PDP curves for solutions with varying Na2CO3, NaHCO3 and NaOH concentration .............................................................................................................................................................. 97 Figure 4.11: Open circuit potential of ECR and UCR in solutions with varying Na2CO3, NaHCO3 and NaOH concentration ............................................................................................................................ 97 Figure 4.12: Anodic PDP curves for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution .................................................................................. 100 Figure 4.13: Corrosion rate for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution ........................................................................................... 100 Figure 4.14: OCP pre and post potentiodynamic polarization for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution .......................... 101 Figure 4.15: Anodic PDP curves for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration ..................................................................................................................................... 104 Figure 4.16: Open circuit potential of ECR and UCR in 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration ................................................................................................................ 105 Figure 4.17: Anodic PDP curves for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution .......................................................... 106   List of Figures   xiii Figure 4.18: Corrosion rates for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution .......................................................... 107 Figure 4.19: OCP pre and post potentiodynamic polarization for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution ............... 107 Figure 4.20: Hall effect (HE) and infrared thermography (IRT) specimens after anodic polarization for 22, 45 and 90 days ............................................................................................................................. 112 Figure 4.21: Specimens after anodic polarization in neutral NaCl solution with epoxy coating (top) and stripped of the epoxy coating (bottom) ............................................................................................. 116 Figure 4.22: Location of coating damage on specimens after anodic polarization in neutral NaCl solution with epoxy coating (left) and stripped of the epoxy coating (right) .................................... 117 Figure 4.23: Top surface of specimen A (top), B (middle) and C (bottom) with electrical connection on the right rebar end ............................................................................................................................. 118 Figure 4.24: Bottom surface of specimen A (top), B (middle) and C (bottom) with electrical connection on the right rebar end ........................................................................................................................ 119 Figure 4.25: Results of LPR measurements of specimens without induced current ......................... 125 Figure 4.26: Results of LPR measurements of specimens exposed to a current of 200 µA/cm2 for 22 days .................................................................................................................................................... 126 Figure 4.27: Results of LPR measurements of specimens exposed to a current of 200 µA/cm2 for 45 days .................................................................................................................................................... 127 Figure 4.28: Range of measured corrosion currents with respect to the corrosion type of specimens with intact electrical connection ........................................................................................................ 128 Figure 4.29: Standard deviation of measured corrosion currents with respect to the corrosion type of specimens with intact electrical connection ...................................................................................... 128 Figure 4.30: Range of measured corrosion currents with respect to the corrosion type of specimens with corroded electrical connection .................................................................................................. 129 Figure 4.31: Standard deviation of measured corrosion currents with respect to the corrosion type of specimens with corroded electrical connection ................................................................................ 129 Figure 4.32: Range of measured corrosion currents with respect to the corrosion type .................. 130 Figure 4.33: Standard deviation of measured corrosion currents with respect to the corrosion type ............................................................................................................................................................ 130 Figure 4.34: Type of corrosion and observed corrosion current ranges and standard deviation σ .. 131 Figure 4.35: Results of HC measurements of specimens without induced current ........................... 132   List of Figures   xiv Figure 4.36: Results of HC measurements of specimens exposed to a current of 200 µA/cm2 for 22 days .................................................................................................................................................... 133 Figure 4.37: Results of HC measurements of specimens exposed to a current of 200 µA/cm2 for 45 days .................................................................................................................................................... 134 Figure 4.38: Results of resistivity measurements of specimens without induced current ................ 135 Figure 4.39: Results of resistivity measurements of specimens exposed to a current of 200 µA/cm2 for 45 days ............................................................................................................................................... 135 Figure 4.40: GPR scan of specimens without induced current .......................................................... 136 Figure 4.41: GPR scan of specimens exposed to a current of 200 µA/cm2 for 22 days ..................... 137 Figure 4.42: GPR scan of specimens exposed to a current of 200 µA/cm2 for 45 days ..................... 137 Figure 4.43: A-scans taken in the centre of the specimens showing the signal amplitude in mV versus arrival time in ns ................................................................................................................................. 138 Figure 4.44: Corrosion currents in µA of slab 1,2 and 3 measured during August 2013 (3 months exposure)............................................................................................................................................ 142 Figure 4.45: Corrosion currents in µA of slab 1,2 and 3 measured during July 2015 (26 months exposure)............................................................................................................................................ 143 Figure 4.46 Corrosion currents in µA of slab 1,2 and 3 measured during October 2015 (29 months exposure)............................................................................................................................................ 144 Figure 4.47:  Effect of rebar diameter on LPR measurements of slab #2 (August 2013) ................... 145 Figure 4.48: Effect of diameter of the rebar connection point on LPR measurements of slab #3 (August 2013) .................................................................................................................................................. 146 Figure 4.49: Effect of chloride concentration on LPR measurements (August 2013) ........................ 147 Figure 4.50: Effect of temperature on LPR measurements ............................................................... 148 Figure 4.51: Effect of moisture on LPR measurements of slab 3 (January 2014) .............................. 148 Figure 4.52: Open circuit potential in mVCSE of slab 1,2 and 3 measured during November 2013 (6 months exposure) .............................................................................................................................. 150 Figure 4.53: Open circuit potential in mVCSE of slab 1,2 and 3 measured during July 2015 (26 months exposure)............................................................................................................................................ 151 Figure 4.54: Effect of diameter of the rebar connection point on HC measurements of slab #3 ..... 152 Figure 4.55: Effect of moisture conditions on HC measurements of slab #3..................................... 153 Figure 4.56: Effect of chloride concentration on HC measurements ................................................. 154   List of Figures   xv Figure 4.57: GPR scan of slab #1 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d) .............................................................................................................. 155 Figure 4.58: GPR scan of slab #2 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d) .............................................................................................................. 155 Figure 4.59: GPR scan of slab#3 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d) .............................................................................................................. 155 Figure 4.60: GPR waves of intact ECR in slab 1, 2 and 3 obtained in 2013 ........................................ 157 Figure 4.61: GPR waves of intact ECR in slab 1, 2 and 3 obtained in 2015 ........................................ 158 Figure 4.62: GPR waves of slab 2 obtained in August 2013 ............................................................... 159 Figure 4.63: Zoomed in diagrams of GPR waves of slab 2 obtained in August 2013 ......................... 160 Figure 4.64: GPR waves of slab 2 obtained in January 2015 .............................................................. 161 Figure 4.65: Zoomed in diagram of GPR waves of slab 2 obtained in January 2015 ......................... 162 Figure 4.66: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 12 Hz ............................................................................................................................................................ 166 Figure 4.67: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 19 Hz ............................................................................................................................................................ 167 Figure 4.68: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 24 Hz ............................................................................................................................................................ 167 Figure 4.69: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 31 Hz ............................................................................................................................................................ 167 Figure 4.70: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 12 Hz ............................................................................................................................................................ 168 Figure 4.71: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 19 Hz ............................................................................................................................................................ 168 Figure 4.72: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 24 Hz ............................................................................................................................................................ 169 Figure 4.73: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 31 Hz ............................................................................................................................................................ 169 Figure 4.74: Approximate location of magnet relative to rebars EM84A-1 and EM84A-2 ................ 171 Figure 4.75: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 12 Hz ............................................................................................................................................................ 171   List of Figures   xvi Figure 4.76: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 19 Hz ............................................................................................................................................................ 172 Figure 4.77: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 24 Hz ............................................................................................................................................................ 172 Figure 4.78: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 31 Hz ............................................................................................................................................................ 172 Figure 4.79: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 12 Hz ................................................................................................................................ 174 Figure 4.80: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 19 Hz ................................................................................................................................ 175 Figure 4.81: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 24 Hz ................................................................................................................................ 175 Figure 4.82: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 31 Hz ................................................................................................................................ 176 Figure 4.83: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 12 Hz ................................................................................................................ 178 Figure 4.84: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 19 Hz ................................................................................................................ 178 Figure 4.85: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 24 Hz ................................................................................................................ 179 Figure 4.86: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 28 Hz ................................................................................................................ 179 Figure 4.87: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 31 Hz ................................................................................................................ 180 Figure 4.88: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 12 Hz ................................................................................................................ 181 Figure 4.89: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 19 Hz ................................................................................................................ 182 Figure 4.90: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 24 Hz ................................................................................................................ 182 Figure 4.91: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 28 Hz ................................................................................................................ 183   List of Figures   xvii Figure 4.92: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 31 Hz ................................................................................................................ 183 Figure 4.93: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 12 Hz ................................................................................................................... 185 Figure 4.94: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 19 Hz ................................................................................................................... 185 Figure 4.95: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 24 Hz ................................................................................................................... 186 Figure 4.96: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 31 Hz ................................................................................................................... 186 Figure 4.97: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 12 Hz ............................................................................................................ 188 Figure 4.98: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 19 Hz ............................................................................................................ 188 Figure 4.99: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 24 Hz ............................................................................................................ 189 Figure 4.100: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 31 Hz .......................................................................................... 189 Figure 4.101: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 12 Hz .......................................................................................... 190 Figure 4.102: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 19 Hz .......................................................................................... 191 Figure 4.103: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 24 Hz .......................................................................................... 191 Figure 4.104: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 31 Hz .......................................................................................... 192 Figure 4.105: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 12 Hz ................................................................................................................................ 193 Figure 4.106: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 19 Hz ................................................................................................................................ 193 Figure 4.107: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 24 Hz ................................................................................................................................ 194   List of Figures   xviii Figure 4.108: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 31 Hz ................................................................................................................................ 194 Figure 4.109: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 12 Hz ................................................................................................................................................... 195 Figure 4.110: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 19 Hz ................................................................................................................................................... 196 Figure 4.111: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 24 Hz ................................................................................................................................................... 196 Figure 4.112: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 31 Hz ................................................................................................................................................... 196 Figure 4.113: Example of an IR image of sample REF-1 with location of measurement circle .......... 199 Figure 4.114: Temperature of the rebar ends in relation to the applied power ............................... 200 Figure 4.115: Concrete surface temperature in relation to the applied power ................................ 201 Figure 4.116: Concrete surface temperature in relation to the temperature of the rebar ends ...... 202 Figure 4.117: Maximum concrete surface temperature as a percentile of the maximum rebar end temperature ....................................................................................................................................... 202 Figure 4.118: Cross section of specimen after 90 days of accelerated corrosion ............................. 203 Figure 4.119: Side view of specimen after 90 days of accelerated corrosion.................................... 203 Figure 4.120: Location of concrete surface temperature measurement........................................... 204 Figure 4.121: Temperature of the rebar ends in relation to the applied power ............................... 205 Figure 4.122: Concrete surface temperature in relation to the applied power ................................ 205 Figure 4.123: Concrete surface temperature in relation to the temperature of the rebar ends ...... 206 Figure 4.124: Maximum concrete surface temperature as a percentage of the maximum rebar end temperature ....................................................................................................................................... 207 Figure 4.125: Cross section of specimens 84B-1 (left) and 84B-2 (right) ........................................... 207 Figure 4.126: Location of concrete surface temperature measurement of a reference specimen and a specimen with uneven corrosion distribution ................................................................................... 208 Figure 4.127: Time in seconds for maximum concrete surface temperature to increase by 1°C...... 208 Figure 4.128: Cross section (left) and extracted rebar (right) of specimen 40B-2 ............................. 209   List of Symbols   xix List of Symbols FLorentz  Lorentz force  EHall  Hall electric field strength µe  Mobility of electrons  µh  Mobility of holes aP  Product activity aR  Reactant activity AUX  Auxiliary electrode ?⃗?   Magnetic flux density c  Specific heat CB  Concentration in the bulk solution CSE  Copper-Saturated Copper Sulfate half cell electrode D  Density d  Diffusivity DAS  Data acquisition system E  Potential E°Cell  Standard state cell potential ECell  Cell potential Ecorr  Corrosion potential ECR  Epoxy Coated Rebar EM  Electromagnetic emf  Electromotive force Ep  Pitting potential Epp  Primary passive potential Erev  Reversible potential F  Faraday's constant (96485 C/eq) f  Surface heat loss factor FBE  Fusion Bonded Epoxy GPR  Ground Penetrating Radar   List of Symbols   xx I  Current ia  Anodic current density iapplied  Applied current density ic  Cathodic current density icorr  Corrosion current density iL  Limiting current density 𝐽   Current density of electric field k  Thermal conductivity Keq  Equilibrium constant LPR  Linear Polarization Resistance M  Atomic mass  m  mass n  number of exchanged electrons per mol NDT  Non-Destructive Testing OCP  Open Circuit Potential p  product coefficient in the electrochemical cell reaction PDP  Potentiodynamic Polarization q  Electric charge Q  Heat QR/A  Reaction/activity quotient r  reactant coefficient in the electrochemical cell reaction R  Universal gas constant (8.314 J/molK) RC  Corrosion rate REF  Reference electrode RH  Hall constant rox  Rate of oxidation RP  Penetration rate rred  Rate of reduction SCE  Saturated Calomel half cell electrode SHE  Standard Hydrogen Electrode T  Temperature t  Time   List of Symbols   xxi UCR  Uncoated Rebar UPV  Ultrasonic Pulse Velocity 𝑣𝑒⃗⃗  ⃗   Velocity of electrons 𝑣𝑒⃗⃗  ⃗   velocity of holes WE  Working electrode δ  Thickness of the diffusion layer ΔG  Gibbs free energy change ΔG°  Standard state Gibbs free energy change ΔQCD  Heat loss through conduction ΔQCV+R  Heat loss through convection and radiation η  Overpotential     1 Introduction   1 1 Introduction 1.1 Motivation Bridge decks are an essential part of our infrastructure and are used daily by millions of people (U.S. Department of Transportation: Federal Highway Administration 2017a). Thus, their structural integrity is crucial for the safety of a large number of people. However, a third of Canada’s 75,000 highway bridges were reported to be structurally or functionally deficient in 2013 (National Research Council Canada 2013). The situation in the U.S. is very similar with 21% of the bridges of the National Highway System (NHS) being deficient in December 2015 (U.S. Department of Transportation: Federal Highway Administration 2017b). One fifth of the deficient NHS bridges are structurally deficient and it is estimated that rehabilitation of these bridges will cost over 17 billion US$  (U.S. Department of Transportation: Federal Highway Administration 2017b, 2017c). It should be noted, that these numbers only include highway bridges which only represent 10% of all bridges in the U.S..  Considering that 69% of all bridges in the U.S. are made of reinforced concrete (U.S. Department of Transportation: Federal Highway Administration 2017d) and that corrosion is a major durability concern for reinforced concrete structures (Homeland Security: Science and Technology 2010; Strategic Highway Research Program - SHRP 2 2013), the early and reliable detection of reinforcing steel corrosion is essential to ensure the safety of millions of people every day.  Epoxy coated rebar (ECR) is a popular type of reinforcement for bridge decks, where the environmental corrosiveness is very high due to the use of de-icing salts. Over 80,000 bridges in the U.S. and Canada contained ECR in 2013 (Concrete Reinforcing Steel Institute - CRSI 2017b). The epoxy coating forms a barrier between the microalloyed steel and the harsh environment to prevent the corrosion of the metal alloy (Concrete Reinforcing Steel Institute - CRSI 2017a). However, even small damage in the coating of ECR can result in localized corrosion of the reinforcement (Concrete Reinforcing Steel Institute - CRSI 2017a).  The Florida Keys bridges are probably the most famous examples of concrete structures containing corroding ECR. Five bridge decks along US 1 were observed to have undergone severe corrosion in 1986 even though construction had only been completed a few years earlier in 1978 and 1983 (A. A. Sagüés, Powers, and Kessler 2001). Approximately 100,000 structures across North   1 Introduction   2 America have been built or repaired with the type of ECR used in the Florida Keys bridge decks. The corrosion process of five bridge decks in the Florida Keys was monitored over the course of 20 years and the deterioration process did not show any sign of slowing down during these two decades (A. A. Sagüés, Powers, and Kessler 2001). An annual damage increase of approximately 0.1 spall per bent has been reported. A typical spall was reported to have covered a surface area of 0.3 m2 (or 3 ft2) (A. A. Sagüés, Powers, and Kessler 2001). While these bridge decks in the Florida Keys are the most widely reported and studied cases of corroding ECR, the environmental degradation of this type of reinforcement has also been observed in bridge decks in Canada such as the Alexander bridge and Peers bridge near Hope, BC. All these reinforced concrete structures had a high chloride content in common due to the exposure to seawater of the Florida Keys bridge decks and de-icing salts of the Alexander and Peers bridge decks. Furthermore, ECR was used in the 1979 rehabilitation of the Perley Bridge spanning over the Ottawa River, connecting Ontario and Quebec (Covino et al. 2001). Even though good quality concrete had been used both during construction and rehabilitation, initial external signs of corrosion of ECR could be observed as early as seven years after deck installation. Furthermore, Covino et al. concluded that ECR delayed visible corrosion indicators by only one to four years (Covino et al. 2001). Brown (Brown 2002) studied the performance of ECR in Virginia Bridge decks and estimated a service life extension of merely five years compared to uncoated steel reinforcement. Furthermore, this study concluded that ECR was not a cost-effective corrosion protection measure for bridge decks in Virginia (Brown 2002).  The degradation of ECR does not only involve the electrochemical corrosion of steel but also the delamination of the fusion bonded epoxy (FBE) from the steel substrate. While water induced adhesion loss of the epoxy coating can sometimes be reversed by drying, Sagüés et al. (Alberto A Sagüés and Powers 1996) observed irreversable delamination of the coating. Delamination of the coating has been reported in conjunction with undercoating corrosion but also in absence of steel corrosion (Alberto A Sagüés and Powers 1996; Nguyen, Hubbard, and Pommersheim 1996; Nguyen and Martin 2004). Damage of the epoxy coating can lead to the initiation of corrosion. However, the corrosion process is not confined to the area of the coating holidays and has been reported to expand under the FBE coating (A. A. Sagüés, Powers, and Kessler 2001; Nguyen and Martin 2004). Consequently, the epoxy coating poses as a barrier between the non-destructive corrosion detection methods positioned on the concrete surface and the corrosion at the steel-epoxy interface.  Corrosion of uncoated as well as epoxy coated rebar leads to an expansion of the corrosion products inside the concrete. As a result, the concrete is exposed to tensile stresses, which lead to   1 Introduction   3 concrete cracking. These cracks accumulate and grow, which ultimately results in concrete spalling. Concrete delamination, spalling and surface cracks are signs of advanced corrosion that can be detected by by the most common non-destructive detection procedures including visual inspection, chain dragging and hammering. These detection methods can be applied to structures containing uncoated as well as epoxy coated rebar. However, these procedures are very subjective and can only detect advanced corrosion.  Early detection is only possible with more sophisticated non-destructive testing (NDT) methods. Due to the electrochemical nature of the corrosion process (Jones 1996), it appears logical to employ electrochemical NDT methods to detect corrosion. Amongst researchers, the most popular electrochemical NDT methods to detect corrosion or estimate the likelihood of corrosion are the half-cell potential, resistivity probe and linear polarization resistance. However, most electrochemical methods rely on an electrical connection between the test equipment on the concrete surface and the embedded rebar. This connection is supplied by the concrete pore solution for uncoated rebar embedded in saturated concrete. However, the FBE coating of ECR is not electrically conductive. Thus, many popular corrosion detections techniques might not be applicaple to undercoating corrosion of ECR.  This study aimed at assessing the suitability of conventional non-destructive testing (NDT) methods as corrosion detection methods for ECR embedded in concrete. The NDT methods included the half-cell potential, resistivity probe, linear polarization resistance and ground penetrating radar.  Induction thermography is a novel NDT method that involves heating the rebar through electromagnetic induction and monitoring the thermal response on the concrete surface through infrared thermography (Kobayashi and Banthia 2011). Consequently, this method is influenced by the electromagnetic and thermal properties of the specimen. Experiments separating the effect of corrosion of ECR on the magnetic and thermal propeties were conducted. The gained knowledge can be applied to induction thermography test data of ECR specimens to improve data analysis.  Furthermore, the suitability of Hall-effect sensors as a novel corrosion detection method for ECR was studied. This electromagnetic detection method takes advantage of the ferromagnetic nature of the steel reinforcement. Concrete, FBE and corrosion products are not ferromagnetic. Thus, the Hall-effect sensor readings are not expected to be affected by these materials. Consequently, the corrosion induced mass loss of the steel reinforcement is expected to be reflected in the Hall-effect measurements.   1 Introduction   4 The suitability of NDT techniques as corrosion detection methods can only be assessed on corroding specimens. However, compared to the time frame of a typical research project, natural corrosion initiation and accumulation takes a long time. Consequently, corrosion acceleration techniques are popular amongst researchers. The impressed current technique is able to reduce the duration of the corrosion initiation period from years to days and control the corrosion rate (Ahmad 2009). This technique is a popular method to accelerate the corrosion of uncoated rebar. The study at hand assesses the applicability of the impressed current technique to epoxy coated rebar and its efficiency in accelerating the corrosion process. Experiments were performed under various conditions. Furthermore, methods to enhance the delamination of the FBE coating prior to corrosion acceleration were assessed. The results of the accelerated corrosion experiments will help in designing future experimental setups aimed at creating corrosion of epoxy coated rebar. Furthermore, limited insight into the degradation mechanisms under different conditions was gained. The corrosion behaviour of ECR was studied in more detail by means of electrochemical experiments performed in simulated concrete pore solutions. A variety of solution compositions were used in order to simulate plain concrete, chloride contaminated concrete and carbonated concrete. The corrosion behaviour of ECR under these conditions was compared to that of uncoated rebar (UCR) under identical conditions. Valuable knowledge regarding the effects of environemntal conditions on the corrosion behaviour of ECR was gained from the corrosion behaviour experiments. Furthermore, insight into the effect of pore solution compositions on the polarized area was gained. The polarized area is an important quantity, that is employed in the electrochemical NDT methods such as the linear polarization resistance technique. Consequently, the gained insight improves the data analysis of these NDT methods.  1.2 Scope This study can be divided into three parts that investigated the corrosion behaviour, accelerated corrosion techniques and non-destructive corrosion detection methods of epoxy coated rebar (ECR).  The corrosion behaviour was studied by performing electrochemical experiments on epoxy coated and uncoated rebar in a variety of simulated concrete pore solutions. The test series involved open circuit potential measurements (OCP), anodic potentiodynamic polarization (PDP) and linear polarization resistance (LPR) measurements. Specimens were tested without prior exposure to the test solution in the short-term experiments. Long-term tests involved a 4-week exposure of the specimens to the test solution prior to the electrochemical tests.    1 Introduction   5 The effects of different accelerated corrosion techniques on epoxy coated rebar were studied. Experiments were performed on ECR embedded in NaCl admixed concrete, in neutral NaCl solution and in alkaline NaCl solution. Corrosion was not only promoted by the presence of NaCl but also by an impressed current. Aeration was also explored as a method to enhance corrosion of ECR in alkaline solution. Successfully corroded ECR specimens were embedded in concrete. All concrete specimens reinforced with ECR were utilized for the non-destructive corrosion detection experiments.  Conventional non-destructive testing (NDT) methods were studied for their suitability to detect corrosion of ECR. Experiments were performed on laboratory specimens and three slabs, that were located outside.  Conventional NDT techniques employed in this study were the ground penetrating radar, linear polarization resistance, Half-Cell potential and the Wenner probe method. Furthermore, Hall effect voltage and thermal measurements were performed on small laboratory specimens in order to study the electromagnetic and thermal effect of corrosion on induction thermography separately.  The scope of the experimental study is summarized in Figure 1.1, which is divided into the four colour coded categories: study focus (yellow), rebar type (light blue), experimental conditions (navy blue and red) and employed techniques (green and red). The colour red identifies the specimens used for the corrosion detection study, that had previously undergone testing in the accelerated corrosion study.      1 Introduction   6  Figure 1.1: Flowchart of the scope of the experimental study   2 Theoretical Background   7 2 Theoretical Background 2.1 Corrosion This chapter explains the origins of diagrams, quantities and reactions used in later chapters. Knowledge of the electrochemical basics of corrosion is essential for the reader to follow and understand the selection of the experimental setups and more importantly the data analysis and the relationships between electrochemical quantities utilized in this study. Corrosion is the environmental degradation of a metal or metal alloy such as iron or steel. The destructive process is the result of chemical reactions involving electron transfer between the metal and its surroundings. The majority of corrosion cells involve electronic charge transfers within a conducting aqueous medium. Consequently, most corrosion processes are of electrochemical nature. The electrochemical corrosion cell consists of four major constituents: Anode, cathode, ionic current path and electronic path (Jones 1996). Figure 2.1 illustrates schematically the processes occurring during the corrosion of a steel in concrete pore solution.   Figure 2.1: Schematic illustration of the corrosion process of a metal M in hydrochloric acid (adapted from (Jones 1996)) The steel represents the anode, where oxidation occurs. During this process electrons are liberated and the valence of the metal is increased (Jones 1996). Oxygen and water molecules in the concrete pore solution consume electrons in the cathodic reaction to from hydroxide ions. This   2 Theoretical Background   8 process is called a reduction (Jones 1996; Harnisch 2012). An electrochemical cell reaction consists of an anodic and a cathodic half-cell reaction. The anodic half-cell reaction is an oxidation and the cathodic half-cell reaction is a reduction (Jones 1996). The ionic current and electronic path connect the anode with the cathode. The electrons liberated at the anode travel through the electronic path to the cathode where they are consumed in the reduction reaction. The electronic path is located inside the metal. The ionic current path on the other hand is situated in the solution. The metal ions formed at the anode enter the surrounding solution (Jones 1996).  The corrosion process consists of thermodynamic and kinetic aspects. Thermodynamics give information about the possibility or likelihood of the occurrence of corrosion. Kinetics on the other hand determines the rate at which the environmental degradation is occurring (Harnisch 2012; Jones 1996; Malhotra and Carino 2003).   2.1.1 Thermodynamics 2.1.1.1 Gibbs Energy & Nernst Equation The principle of Gibbs free energy change ΔG is a fundamental aspect of the thermodynamics of corrosion. The change in Gibbs free energy is a measure for the tendency of an electrochemical reaction to proceed. The change of free energy becomes negative when the energy of the reaction products is lower in comparison to the energy of the reactants. A negative change (i.e. decrease) of free energy indicates a high likelihood for the reactions to occur. Thus, a more negative value of ΔG indicates a greater tendency for a reaction to occur (Jones 1996). The relationship between the free energy change and the electrochemical cell potential is shown in Eq.(1).    ∆𝐺 = − 𝑛 ∙ 𝐹 ∙ 𝐸𝐶𝑒𝑙𝑙 (1)  With ΔG = Gibbs free energy change   n = number of exchanged electrons per mol   F = Faraday’s constant (96485 C/eq)   ECell = cell potential  The electrochemical cell potential is determined by subtraction of the anodic from the cathodic half-cell potential. The two half-cell reactions can be identified by their standard potential in the   2 Theoretical Background   9 electromotive force (emf) series. The standard state is defined as the state at which the activity of both, the reactants and the products, is unity (Jones 1996). Positive and negative potentials in the emf series are noble and active respectively. The half-cell reaction with the most negative (i.e. most active) standard potential is anodic and describes the oxidation. Consequently, the half-cell reaction with the most positive (i.e. most noble) standard potential is identified as cathodic and describes the reduction (Jones 1996). In accordance with Eq.(1) the free energy change under standard condition is determined by Eq.(2)    ∆𝐺° =  − 𝑛 ∙ 𝐹 ∙ 𝐸°𝐶𝑒𝑙𝑙 (2)  With ΔG° = Standard state Gibbs free energy change   n = number of exchanged electrons per mol   F = Faraday’s constant (96485 C/eq)   E°Cell = Standard state cell potential  A positive standard state cell potential (i.e. a negative standard state free energy change) indicates a spontaneous reaction (Jones 1996). Consequently, a negative standard state cell potential (i.e. a positive standard state free energy change) indicates a non-spontaneous reaction.   Thermodynamics relate the change in free energy under non-standard conditions (i.e. the activity is not unity) to the standard state Gibbs energy change as shown in Eq.(3).    ∆𝐺 = ∆𝐺° + 𝑅 ∙ 𝑇 ∙ ln (𝑄𝑅/𝐴) (3)  With ΔG = Gibbs free energy change   ΔG° = Standard state Gibbs free energy change   R = Universal Gas Constant (8.314 J/molK)   T = Absolute Temperature (K)   QR/A = Reaction/Activity quotient    2 Theoretical Background   10 The reaction quotient QR/A in Eq.(3) is dependent on the activity of the reaction products and reactants and is determined by Eq.(4).   𝑄𝑅/𝐴 = 𝑎𝑃1𝑝1 ∙ 𝑎𝑃2𝑝2𝑎𝑅1𝑟1 ∙ 𝑎𝑅2𝑟2 (4)  With aP = product activity   aR = reactant activity   p = product coefficient in the electrochemical cell reaction   r = reactant coefficient in the electrochemical cell reaction  The activity is approximated by the molar concentration of soluble matter. For quantitative calculations of the activity of soluble species the molar concentration needs to be multiplied by an activity coefficient. In the case of gases the activity is defined by the partial pressure in atm. Liquid water and solids have an activity of unity (Jones 1996). Substituting Eq.(1) and Eq.(2) into Eq.(3) and rearranging gives Eq.(5).    𝐸𝐶𝑒𝑙𝑙 = 𝐸𝐶𝑒𝑙𝑙° −𝑅 ∙ 𝑇𝑛 ∙ 𝐹∙ ln (𝑄𝑅/𝐴) (5)  With ECell = cell potential   E°Cell = Standard state cell potential   R = Universal Gas Constant (8.314 J/molK)   T = Absolute Temperature (K)   n = number of exchanged electrons per mol   F = Faraday’s constant (96485 C/eq)   QR/A = Reaction/Activity quotient   Rewriting Eq.(5) in the log10 form gives the generalized Nernst equation as shown in Eq.(6)    𝐸𝐶𝑒𝑙𝑙 = 𝐸𝐶𝑒𝑙𝑙° − 2.303 ∙𝑅 ∙ 𝑇𝑛 ∙ 𝐹∙ log (𝑄𝑅/𝐴) (6)    2 Theoretical Background   11 The latter expresses the effect of activity on the cell potential. By substituting the appropriate values for the Universal Gas and Faraday’s constant the Nernst equation for the standard temperature of 298 K can be rewritten as Eq.(7)   𝐸𝐶𝑒𝑙𝑙 = 𝐸𝐶𝑒𝑙𝑙° −0.0592𝑛∙ log (𝑄𝑅/𝐴) (7)  In the process of the electrochemical cell reaction, reactants are consumed and products are formed. Consequently, the reactants and products concentration decreases and increases respectively until a state of equilibrium is reached. At this state the free energy change equates to zero, resulting in a cell potential of zero (Jones 1996). At equilibrium oxidation and reduction occur at the same rate and the reaction quotient equals the equilibrium constant Keq. Therefore, the Nernst equation at equilibrium is given by Eq.(8)   0 =  𝐸𝐶𝑒𝑙𝑙° −  2.303 ∙𝑅 ∙ 𝑇𝑛 ∙ 𝐹∙ log (𝐾𝑒𝑞) (8)  Under standard condition Eq.(8) can be rewritten as Eq.(9)   0 =  𝐸𝐻+ 𝐻2⁄° −0.0592𝑛∙ log (𝐾𝑒𝑞) (9)  2.1.1.2 pH The reduction of hydrogen is shown in Eq.(10)   2𝐻+ + 2𝑒− = 𝐻2 (10)  Applying the Nernst equation at standard temperature (Eq.(8)) to this reaction gives Eq.(11)   𝐸𝐻+ 𝐻2⁄ = 𝐸𝐻+ 𝐻2⁄° −0.05922∙ log (𝑎𝐻2𝑎𝐻+2 ) (11)    2 Theoretical Background   12 The activity of gases is defined by their partial pressure in atm. Furthermore, when assuming the standard state the activity of a gas, such as hydrogen (H2), is equal to unity. According to the emf series the standard state potential of the hydrogen reduction reaction is zero (Jones 1996). Consequently Eq.(11) can be rewritten as Eq.(12) under standard conditions.   𝐸𝐻+ 𝐻2⁄ = −0.05922∙ log (1𝑎𝐻+2 ) (12)  Rearranging gives Eq.(13)   𝐸𝐻+ 𝐻2⁄ = 0.0592 ∙ log (𝑎𝐻+) (13)  The negative decimal logarithm of the hydrogen ion activity defines the pH. Therefore Eq.(13) can be rewritten as Eq.(14)   𝐸𝐻+ 𝐻2⁄ = − 0.0592 ∙ pH (14)  2.1.1.3 Pourbaix Diagrams A central figure in thermodynamics is the Pourbaix diagram, that relates the corrosion potential of a given metal to the pH of the solution (Jones 1996; Harnisch 2012). The Pourbaix atlas consists of a collection of potential – pH diagrams that are specific to water. The potentials used to construct a Pourbaix diagram are always reported versus the saturated hydrogen electrode (SHE). Figure 2.2 shows such a diagram for iron, which is the main constituent of reinforcing steel.    2 Theoretical Background   13  Figure 2.2: Potential-pH diagram for iron (adapted from (Jones 1996)) 2.1.1.3.1 Equilibria 2.1.1.3.1.1 Chemical & electrochemical The solid lines in the diagram (Figure 2.2) represent electrochemical or chemical equilibrium. Both types of equilibrium have the activity dependence in common (Jones 1996). Horizontal lines show that the reactions are only dependent on the potential but not the pH. Charge transfer occurs during the reactions. However, hydrogen ions or water molecules are not involved in the reactions as shown by the lack of pH dependence (Jones 1996). Consequently, the horizontal lines represent electrochemical equilibrium. Vertical lines show that the reactions are only dependent on the pH but not the potential. The oxidation state of the involved elements does not change and charge transfer is not required. Consequently, the vertical solid lines represent pure chemical equilibrium (Jones 1996). Diagonal lines show that the reactions are dependent on both the potential and the pH. Both hydrogen ions and/or water molecules as well as charge transfer are involved in the reactions. Consequently, the diagonal lines represent a mixed chemical and electrochemical equilibrium (Jones 1996).  2.1.1.3.1.2 Water-Gas The dashed lines labeled (a) and (b) in Figure 2.2 indicate water-gas equilibrium.    2 Theoretical Background   14 Line (a) represents the hydrogen gas - water equilibrium. At potentials below the aforementioned equilibrium hydrogen gas (H2) is the stable phase. Similarly, at potentials greater than the reversible potential water (H+ or H2O) is the stable phase (Jones 1996). The hydrogen evolution reactions are shown in Eq.(15) and Eq.(16) for acids and bases respectively.   2𝐻+  +  2𝑒−  =  𝐻2(𝑔) (15)   2𝐻2𝑂 +  2𝑒−  =  𝐻2(𝑔) +  2𝑂𝐻− (16)  Below line (a) water is reduced to hydrogen gas and above line (a) hydrogen gas is oxidized to water (Jones 1996). Line (b) represents the oxygen – water equilibrium. At potentials more active than the reversible potential, water (O2- or H2O) is the stable phase (Jones 1996). Conversely, at potentials above the equilibrium oxygen (O2) is the stable phase. Eq.(17) and Eq.(18) show the oxygen reduction reactions for acids and bases respectively.   𝑂2 +  4𝐻+  +  4𝑒−  =  2𝐻2𝑂 (17)   𝑂2 +  2𝐻2𝑂 +  4𝑒−  =  4𝑂𝐻− (18)  Below line (b) oxygen is reduced to water and above line (b) water is oxidized to oxygen. Figure 2.3 visualizes the stable phases in the potential – pH diagram.   2 Theoretical Background   15  Figure 2.3: Conditions of stability for water, oxygen and hydrogen   2.1.1.3.2 Aeration Aerated water is saturated with dissolved air. Consequently, if a metal is immersed in aerated water, metal oxidation occurs simultaneously with oxygen reduction to form a corrosion cell (Jones 1996). In the presence of aerated water all reactions with a reversible potential below that of line (b) in Figure 2.2 are oxidized. Deaerated water does not have any dissolved oxygen. Hence, a corrosion cell of a metal immersed in deaerated water consists of metal oxidation and hydrogen reduction. However, the aforementioned cell can only form if the reversible potential of the metal is below line (a) in Figure 2.2.  It can be observed that the range of reversible potentials to form a corrosion cell is much larger for aerated water in comparison to deaerated water. Consequently, aeration increases the oxidizing power of water (Jones 1996). 2.1.1.3.3 Corrosion Tendencies Three major areas, that describe the corrosion tendency of a metal, can be identified in the potential-pH diagram (Figure 2.2), namely:   I Immunity    2 Theoretical Background   16 II Active corrosion  III Passivity  A metal is considered immune when the metallic state is the most stable state. Corrosion of the metal does not occur in the immune area of the Pourbaix diagram. Active corrosion is only possible in area II (Harnisch 2012; Jones 1996). The metal’s cation is the most stable element in this area of the potential-pH diagram. Passivity occurs when a layer forms on the metal surface that limits active corrosion of the substrate. This surface modification is commonly referred to as the passive layer (Harnisch 2012; Jones 1996).  2.1.2 Kinetics 2.1.2.1 Corrosion Rate Corrosion is the result of electrochemical reactions at the metal - electrolyte interface. During the aforementioned reactions, charge transfers occur at the interface, where electrons travel through the electronic path from the anode to the cathode. The rate of the electron flow is a measure for the rate of the electrochemical reaction and hence for the rate of corrosion. The resulting metal loss is proportional to the aforementioned current flow (Jones 1996).  Faraday’s law describes the relationship between mass loss and current flow as shown in Eq.(19)   𝑚 = 𝐼 ∙ 𝑡 ∙ 𝑀𝑛 ∙ 𝐹 (19) With m = mass loss (g)   I = current (A = C/s)   t = time (s)   M = atomic weight (g/mol)   n = number of exchanged electrons per mol   F = Faraday’s constant (96485 C/eq)  The corrosion rate describes the mass loss over time per unit area. Hence, the corrosion rate is determined by dividing Eq.(19) by the area and time, resulting in Eq.(20)   𝑅𝐶 = 𝑖𝐶𝑜𝑟𝑟 ∙𝑀𝑛 ∙ 𝐹 (20)   2 Theoretical Background   17  Eq.(20) shows the proportionality between the corrosion rate RC and the corrosion current density icorr and hence the current flow (Jones 1996). Consequently, the current flow is a measure for the corrosion rate as stated earlier. The penetration rate RP is determined by the division of the corrosion rate by the density D of the oxidizing metal or alloy (Jones 1996). The resulting formula for the depth of penetration into the anode over time is shown in Eq.(21)   𝑅𝑃 = 𝑖𝐶𝑜𝑟𝑟 ∙𝑀𝐷 ∙ 𝑛 ∙ 𝐹 (21)  2.1.2.2 Equilibrium The state of equilibrium is defined as the state at which the rate of oxidation rox and the rate of reduction rred are equal (Jones 1996). Hence, Eq.(20) can be rewritten as Eq.(22)   𝑟𝑜𝑥 = 𝑟𝑟𝑒𝑑 = 𝑖𝑜 ∙𝑀𝑛 ∙ 𝐹 (22)    The exchange current density io needs to be determined experimentally because it is a surface dependant parameter (Jones 1996). It is proportional to the reversible rate of reaction at equilibrium. Similarly, the corresponding electrochemical half-cell potential at the aforementioned state is called the reversible potential. At potentials higher than the reversible potential oxidation occurs and at potentials lower than the reversible potential reduction occurs (Jones 1996).  2.1.2.3 Polarization The process of changing the half-cell potential to a value E other than the reversible potential Erev is called polarization (Jones 1996). The aforementioned change in potential is called the overpotential 𝜂 and it is determined by Eq.(23)   𝜂 = 𝐸 − 𝐸𝑟𝑒𝑣  (23)  The overpotential can be either positive or negative, indicating anodic and cathodic polarization respectively (Jones 1996). There are two types of polarization mechanisms that can occur in an   2 Theoretical Background   18 electrochemical cell, namely activation and concentration polarization. These two mechanisms can occur at the same time (Jones 1996).  2.1.2.3.1 Activation Polarization The following four processes occur in the order listed at the metal – electrolyte interface during electrochemical corrosion of a metal (Jones 1996). 1. Adsorption 2. Electron transfer 3. Combination 4. Desorption This sequence of reactions controls the activation polarization and the slowest process controls the rate of reaction (Jones 1996). Consequently, this step in the reaction sequence is called the rate-determining step (RDS). This polarization mechanism can only control the rate of reaction, when the concentration of products and reactants at the metal – electrolyte interface equates to the concentrations in the bulk electrolyte (Jones 1996).  As shown in Eq.(20), the current density is a measure for the rate of reaction. Similarly, the potential is a measure for the driving force for the reaction. Hence, the Butler - Volmer equation relates the rate of reaction to the driving force for the reaction (Jones 1996) as shown in Eq.(24)   𝑖 = 𝑖𝑜 ∙ 𝑒[𝛽∙𝑛∙𝐹∙𝜂𝑅∙𝑇 ] − 𝑖𝑜 ∙ 𝑒[−(1−𝛽)∙𝑛∙𝐹∙𝜂𝑅∙𝑇 ] (24)     = 𝑖𝑜 ∙ 𝑒[𝛽∙𝑛∙𝐹∙𝜂𝑅∙𝑇 ] ∙ (1 − 𝑒−[𝑛∙𝐹∙𝜂𝑅∙𝑇 ]) (25)  In case of anodic polarization with an overpotential larger than 50 mV the term in round parenthesis in Eq.(25) becomes negligible and the Butler – Volmer equation can be simplified to the term in Eq.(26)    𝑖 = 𝑖𝑜 ∙ 𝑒[𝛽∙𝑛∙𝐹∙𝜂𝑅∙𝑇 ] (26)    2 Theoretical Background   19 The overpotential can be determined by rearranging Eq.(26) and substituting the Euler logarithm for the decimal logarithm.    𝜂 𝑎𝑐𝑡 =  2.3 ∙𝑅 ∙ 𝑇𝛽 ∙ 𝑛 ∙ 𝐹∙ 𝑙𝑜𝑔 (𝑖𝑖𝑜) (27)  The term in front of the logarithm is defined as the incline of the curve between the reversible potential and the overpotential and is often called ba or bc for the anodic and cathodic branch respectively (Jones 1996). Substituting this notation into Eq.(27) gives Eq.(28) and Eq.(29)   𝜂𝑎𝑐𝑡,𝑎  =  𝑏𝑎 ∙ 𝑙𝑜𝑔 (𝑖𝑖𝑜) (28)     𝜂𝑎𝑐𝑡,𝑐  =  𝑏𝑐 ∙ 𝑙𝑜𝑔 (𝑖𝑖𝑜) (29)  Eq.(28) and Eq.(29) are known as the Tafel equations. In a potential versus logarithm of the current density plot (Figure 2.4) these equations are straight lines, that meet at the reversible potential and the logarithm of the exchange current density (Jones 1996). Such a plot visualizes the polarization kinetics under activation control and is known as the Evans diagram.  Figure 2.4: Evans Diagram    2 Theoretical Background   20  2.1.2.3.2 Concentration Polarization Concentration polarization is caused by limitations of the mass transfer, that result in concentration changes at the metal – electrolyte interface. These variations can reduce the rate of reaction below that of activation polarization, in which case concentration polarization becomes rate controlling (Jones 1996). This type of polarization can become particularly important for cathodic reactions.  Similarly to the Eq.(27), the overpotential caused by concentration changes is determined by Eq.(30)   𝜂𝑐𝑜𝑛,𝑐 =  2.3 ∙𝑅 ∙ 𝑇𝑛 ∙ 𝐹∙ 𝑙𝑜𝑔 (1 −𝑖𝑐𝑜𝑟𝑟𝑖𝐿) (30)  The limiting current density iL is caused by the mass transport limitations. Hence, it is dependent on the diffusivity d of the reactants and the thickness of the diffusion layer δ as well as the concentration CB in the bulk solution (Jones 1996). It is determined by Eq.(31)   𝑖𝐿 = 𝑑 ∙ 𝑛 ∙ 𝐹 ∙ 𝐶𝐵𝛿 (31)  The limiting current density is also influenced by the solution conditions. It increases with increasing velocity, temperature and concentration (Jones 1996). The effect of the solution conditions on the concentration polarization is visualized in Figure 2.5.   2 Theoretical Background   21  Figure 2.5: Effect of solution velocity, temperature and concentration on cathodic concentration polarization (adapted from (Jones 1996))  2.1.2.3.3 Total Polarization The total overpotential is determined by the sum of the activation and concentration overpotential (Jones 1996). The resulting total anodic and cathodic overpotentials are shown in Eq.(32) and Eq.(33) respectively.   𝜂𝑡𝑜𝑡𝑎𝑙,𝑎  =  𝑏𝑎 ∙ 𝑙𝑜𝑔 (𝑖𝑖𝑜) (32)     𝜂𝑡𝑜𝑡𝑎𝑙,𝑐  =  𝑏𝑐 ∙ 𝑙𝑜𝑔 (𝑖𝑖𝑜) + 2.3 ∙𝑅 ∙ 𝑇𝑛 ∙ 𝐹∙ 𝑙𝑜𝑔 (1 −𝑖𝑐𝑜𝑟𝑟𝑖𝐿) (33)  It can be observed that the cathodic overpotential consists of an activation and a concentration controlled part. The anodic overpotential is only activation controlled (Jones 1996). The reduction process is concentration controlled in the case of high cathodic polarization. The reduction process only becomes activation controlled at low cathodic polarization.   2 Theoretical Background   22 2.1.2.4 Mixed Potential Theory Even though the anodic and cathodic reactions are not the same during the corrosion process, they share an important parameter. Due to the principle of charge conservation the sum of all anodic currents is equal to the sum of all cathodic currents. In other words, the oxidation and reduction currents are the same and are commonly referred to as the corrosion current (Jones 1996). In case the oxidation and reduction occur on equal surface area, the anodic and cathodic current densities are identical. This current density is commonly referred to as the corrosion current density icorr (Jones 1996). In a potential – current density semilog plot, that shows the Evans diagrams of the cathode and anode, the corrosion current density corresponds to the point of intersection of the anodic oxidation curve and the cathodic reduction curve (Jones 1996). Such a plot is shown in Figure 2.6.  Figure 2.6: Mixed potential theory for a metal M in acid solution The potential at this point is called the corrosion potential Ecorr. It is also commonly referred to as the open circuit potential (OCP) or the free potential. It can be observed that the corrosion potential is usually far away from the reversible potential of the anode and cathode (Jones 1996). Consequently, the electrodes are polarized and hence not at equilibrium during the corrosion process. The corrosion potential is a mixed potential because it is influenced by the reversible potential of the anode and the cathode and represents a combination of the anodic and cathodic voltages (Jones 1996).    2 Theoretical Background   23 The corrosion potential and current density depend on thermodynamics in the form of the reversible potentials as well as kinetics in the form of the exchange current densities and the Tafel slopes.  2.1.2.4.1 Effect of Current Density Both the reversible potentials as well as the exchange current densities influence the two corrosion parameters. However, in most cases the effect of the kinetic parameters outweighs the effect of the thermodynamic driving force on the corrosion rate (Jones 1996). 2.1.2.4.2 Effect of Oxidizer Addition of a very strong oxidizer with a reversible potential that is much nobler than that of the remaining corrosion cell constituents can have a significant effect on the corrosion behaviour of the cell (Jones 1996). In the case where the oxidizer is very noble and has a sufficiently large exchange current density, the oxidizer can increase both the corrosion potential as well as the corrosion rate of the electrochemical cell (Jones 1996). This is shown in Figure 2.7 with Fe2+ as an oxidizer.  Figure 2.7: Effect of noble second oxidizer, Fe3+/Fe2+, with high enough exchange current density on mixed potential (adapted from (Jones 1996)) The case of a noble oxidizer Fe2+ with an exchange current density too low to affect the corrosion rate is shown in Figure 2.8.   2 Theoretical Background   24  Figure 2.8: Noble second oxidizer, Fe3+/Fe2+, with too small of a exchange current density to affect the mixed potential (adapted from (Jones 1996)) 2.1.2.4.3 Effect of Concentration Polarization In the case of high cathodic polarization, the reduction process is concentration controlled. As mentioned earlier the limiting current density is highly dependent on the solution conditions and increases with increasing temperature, concentration and velocity. Consequently, the corrosion current density and hence the corrosion rate also increases with an increased velocity (and/or temperature, concentration) (Jones 1996).  An increase of the limiting current density only leads to an increase of the corrosion current density as long as the reduction process is concentration controlled. Once the limiting current density exceeds the anodic current density, the reduction process becomes activation controlled (Jones 1996). Consequently, the solution conditions no longer have an effect on the corrosion current density and hence corrosion rate. 2.2 Accelerated Corrosion Active-passive metals such as steel, are metals, that display active or passive corrosion behaviour dependent on the electrochemical potential (Jones 1996). Figure 2.9 shows the anodic polarization curve of an active passive metal.    2 Theoretical Background   25  Figure 2.9: Anodic polarization curve of an active-passive metal Three regions can be identified in Figure 2.9. The active region between I and II, the passive region between III and IV and the transpassive region between IV and V. In the active region, the metal undergoes active corrosion until the primary passive potential Epp is reached at II. The Epp indicates the formation and stabilization of a protective passive layer on the surface of the metal (Jones 1996). However, the protective layer is very unstable between II and III and small environmental disturbances such as a change in pH or temperature, can lead to active metal dissolution.  Between III and IV, the passive layer is stable and greatly reduces the anodic current density and thus corrosion rate (Jones 1996). Further increase of the potential past IV, leads to transpassive behaviour. In the transpassive region, pitting corrosion occurs, leading to a sudden increase in current density (Jones 1996).   Anodic polarization is a very popular way to accelerate the corrosion of rebar (Oy 2003; Ahmad 2009; Austin, Lyons, and Ing 2004; Caré and Raharinaivo 2007; Alonso et al. 1998; Saricimen et al. 1997).  The electrochemical potential between a reference electrode and the metal to be corroded is raised from Ecorr by application of a current density iapplied through a counter electrode as shown in Figure 2.10.    2 Theoretical Background   26  Figure 2.10: Anodic polarization of an active-passive metal The applied current density iapplied can be determined by the difference between the anodic ia and cathodic ic current densities in Figure 2.10 as shown in Eq. (35) (Jones 1996)   𝑖𝑎𝑝𝑝𝑙𝑖𝑒𝑑 = 𝑖𝑎 − 𝑖𝑐 (34)  With iapplied = applied current density ia = anodic current density   ic = cathodic current density  During the anodic polarization electrons are drawn out of the metal, which results in more positive potentials (Jones 1996). Thus, the potential can be raised from the passive region (between III and IV in Figure 2.10) to the transpassive region (between IV and V in Figure 2.10). Consequently, the anodic and cathodic current density are increased and decreased respectively by the anodic polarization (Figure 2.10). Thus, the oxidation or corrosion rate of the metal is increased (Jones 1996).  Two types of anodic polarization can be used to accelerate the corrosion. Galvanostatic polarization applies a constant current (Oy 2003; Alonso et al. 1998; Caré and Raharinaivo 2007), whereas potentiostatic polarization applies a current needed a constant potential to the corroding metal (Rengaswamy, Vedhalakshmi, and Balakrishnan 1995; Saricimen et al. 1997). The advantage of   2 Theoretical Background   27 the galvanostatic polarization lies in the constant applied current and thus corrosion rate of the metal undergoing accelerated corrosion. Both the corrosion rate as well as the mass loss can be predicted with the constant applied current. Furthermore, the galvanostatic polarization requires a DC source and DC current calibrator, which are significantly less expensive than a potentiostat needed for the potentiostatic polarization. However, in order to choose the appropriate applied current, one needs to know the polarized area. Knowledge of this area is not required for the potentiostatic polarization. 2.2.1 Potentiodynamic Polarization Similar to the potentiostatic polarization, the potential rather then the current is controlled during the potentiodynamic polarization (PDP). For simplicity, these types of polarization are often described as impressed or applied potential methods (Saricimen et al. 1997; Rengaswamy, Vedhalakshmi, and Balakrishnan 1995). However, it is practically impossible to apply a potential. Therefore, a polarizing current is applied by an auxiliary electrode (AUX) to control the potential between the working electrode (WE) and the reference electrode (REF). The applied current is automatically adjusted by the potentiostat to keep the potential between the WE and REF at a constant prescribed voltage increase during the potentiodynamic polarization (Jones 1996). Figure 2.11 shows a very simplified schematic of the experimental setup and visualizes the polarization of the WE by application of a current through the AUX and the potential measurement between the REF and the WE. The direct current is supplied to the AUX by the potentiostat, which also measures the potential between the REF and the WE.    2 Theoretical Background   28  Figure 2.11: Simplistic schematic of polarization of the working electrode (WE) by the auxiliary electrode (AUX) with a potentiostat and a reference electrode (REF) It should be noted, that the REF is usually not placed directly in the solution as shown in Figure 2.11 for simplicity but rather into a Luggin probe with a salt bridge. The potential of the REF does not change during the polarization and only very little current if any passes through the potential measuring circuit (Jones 1996).  A complete anodic potentiodynamic polarization (PDP) curve of an active-passive alloy is schematically shown in Figure 2.9. However, if the cathodic curve intersects the anodic curve in the passive region such as in Figure 2.12 (left), the resulting anodic PDP curve will only consist of the part of the curve above the Ecorr (or OCP) (Princeton Applied Research 2017). This is shown in Figure 2.12 (right). An example of this type of behaviour is reinforcing steel in a very alkaline solution such as 0.9M NaOH in which the steel is protected by a passive layer.     2 Theoretical Background   29  Figure 2.12: Anodic potentiodynamic polarization plots in theory (left) and in practice (right) of an active passive metal in the passive state (adapted from (Princeton Applied Research 2017)) If the cathodic curve intersects the anodic in the active region as shown in Figure 2.13 (left), all three regions (active, passive, transpassive) can be identified from the resulting anodic PDP curve as shown in Figure 2.13 (right).  Figure 2.13: Anodic potentiodynamic polarization plots in theory (left) and in practice (right) of an active passive metal in the active state (adapted from (Princeton Applied Research 2017))     2 Theoretical Background   30 2.3 Corrosion in Concrete Steel is produced from natural iron oxides. During the smelting process the energy level of the metal is increased. Depending on the environmental conditions the metal will always strive to reach the energy level of the oxides it is made of (Harnisch 2012). The process of the metal or alloy returning to the aforementioned energy level is called corrosion (Jones 1996; Harnisch 2012).  2.3.1 Rebar  Concrete is a highly alkaline composite material with a pH of the pore solution exceeding 12. Hence, a passive layer consisting of Fe3O4 and γ-Fe2O3 is formed on the surface of the reinforcing bar (Harnisch 2012). The oxide layer prevents the formation of an electronic and ionic current path and therefore protects the steel from corroding. If the concrete successfully prevents water and air from reaching the surface of the rebar the passive layer can remain indefinitely. However, the passive layer is not very stable and can be damaged or even destroyed under more aggressive circumstances such as exposure to chloride adsorption, carbonation of the concrete or mechanical impact (Jones 1996; Harnisch 2012). Consequently, active corrosion can take place.  Chloride ions impair the passivity of the reinforcing steel and thus lead to its susceptibility to environmental degradation. The chloride ions also form an acid solution with the ferrous corrosion products. Consequently the pH of the pore solution experiences a further decrease which further enhances the corrosion of the rebar (Jones 1996) Carbonation of concrete occurs when carbon dioxide CO2 enters the concrete and reacts with the saturated Calcium Hydroxide Ca(OH)2 in the pore water. The resulting Calcium Carbonate CaCO3 causes a reduction of the pore-water pH and thus enables corrosion of the rebar (Jones 1996) Figure 2.14 shows the effect of the pH of the electrolyte on the ability of iron to corrode in a chloride solution. A reduction in the pH of the pore solution is equivalent to a shift to the left in the diagram where active (area II) corrosion is possible.   2 Theoretical Background   31  Figure 2.14: Potential-pH diagram for iron in water (left) (adapted from (Jones 1996)) and in chloride contaminated water (right) (adapted from (Moreno et al. 2004)) Furthermore, the dotted lines represent higher Cl- concentrations of the solution. Thus, the passive region of the diagram decreases and the area, in which corrosion occurs, increases with increasing Cl- concentration.  Figure 2.15 shows the electrochemical corrosion cell of steel in concrete. The electrolyte represents the concrete pore solution surrounding the corroding reinforcing bar.  Figure 2.15: Electrochemical corrosion cell of steel in concrete    2 Theoretical Background   32  Oxygen from the atmosphere diffuses into the concrete and forms hydroxide ions with the water molecules of the pore solution as shown in reaction (35).    𝑂2 + 2𝐻2𝑂 + 4𝑒− → 4𝑂𝐻− (35)  This reducing reaction consumes electrons. These required electrons are obtained from the oxidizing anodic reaction of the iron as shown in reaction (36).    𝐹𝑒 → 𝐹𝑒2+ + 2𝑒− (36)  The liberated electrons travel from the anode through the rebar to the metal electrolyte interface where the cathodic reaction (35) takes place (Jones 1996; Harnisch 2012; Malhotra and Carino 2003; Schneider Bautabellen Fuer Ingenieure 2006)  The iron ions formed in the anodic reaction enter the electrolyte to form iron(II) hydroxide (or ferrous hydroxide) with the reduction products (Harnisch 2012; Schneider Bautabellen Fuer Ingenieure 2006)   𝐹𝑒2+ + 2𝑂𝐻− → 𝐹𝑒(𝑂𝐻)2 (37)  A further oxidation of the iron ions with the help of oxygen is also possible (Harnisch 2012).    4𝐹𝑒2+ + 𝑂2 → 4𝐹𝑒3+ + 2𝑂2− (38)  Corrosion of the reinforcing steel can only occur when water and oxygen are present in the concrete and the passive layer is damaged. Hence, corrosion does not occur in very dry concrete because of the lack of water. Furthermore, the rebar in continuously saturated concrete does not corrode either because of the lack of oxygen.    2 Theoretical Background   33 2.3.2 Corrosion Behaviour of Epoxy Coated Rebar 2.3.2.1 Degradation of Epoxy The epoxy coating works as a barrier to protect the steel from environmental predators such as water and chloride ions. However, even though the diffusion coefficient of water and chloride ions is significantly reduced by the fusion bonded epoxy (FBE) coating the transport of harmful substances to the metal surface is not eliminated. The diffusion coefficient of water is reported to lie between 2.44x10-9 and 3x10-9 cm2/s (Legghe et al. 2009; Powers 2009). The diffusion of chloride ions occurs at a significantly lower rate with a coefficient of 4.67x10-12 cm2/s (Legghe et al. 2009). Tensile stress increases the diffusion coefficient and consequently accelerates the diffusion process (Legghe et al. 2009).  The diffusion of water through the FBE layer leads to adhesion loss between the epoxy coating and the steel substrate (Legghe et al. 2009; Powers 2009; Nguyen and Martin 1996, 2004). The loss of adhesion is mostly recovered upon drying. However, in the case of complete adhesion loss recovery is not possible (Legghe et al. 2009; Powers 2009; Nguyen and Martin 1996, 2004). The adhesion between the metal and polymer consists of chemical and physical links, that are sensitive to water. Some groups in the epoxy resin are able to form hydrogen bonds with the steel. However, these groups need to compete with the diffused water molecules (Legghe et al. 2009). Furthermore, water can hydrolyze chemical bonds in the epoxy network. The damage caused by hydrolysis cannot be recovered by water desorption (Powers 2009).The discussed degradation mechanisms of the epoxy can occur regardless of the presence of coating imperfections.  According to A775 − 07b “Standard Specification for Epoxy-Coated Steel Reinforcing Bars” only 3 holidays per meter of ECR are permissible and the repaired surface area cannot exceed 1% per foot of ECR (ASTM International 2014a). However, after shipping and fabrication, which can induce further damage and even disbondment, ECR is often kept in storage yards. During storage, deterioration, damage and disbondment can be initiated or promoted by thermal cycles, ultraviolet exposure, improper handling and sea spray in coastal areas (A. A. Sagüés, Powers, and Kessler 2001). Further damage to the coating can occur during the casting process, particularly during rebar cage building and positioning in formwork as well as concrete pouring and vibration (A. A. Sagüés, Powers, and Kessler 2001). Consequently, even if all stages between the manufacturing of ECR and the final structure are performed according to the norm (ASTM International 2013, 2014a, 2015), a small amount of coating damage is likely to be present.    2 Theoretical Background   34 Once the ECR is embedded in concrete, the pore solution interacts with the epoxy coating. This interaction occurs even in absence of chloride ions and the concrete pore solution penetrates between the coating and the metal substrate, in locations where disbondment has already occurred. Delamination is aggravated in the presence of chloride ions (A. A. Sagüés, Powers, and Kessler 2001) and has been observed in widespread areas of bridges older than four years even when chloride concentrations were very low (Alberto A and Powers 1996). Furthermore, this adhesion loss cannot be reversed by drying (Alberto A and Powers 1996).  2.3.2.2 Degradation of Steel The corrosion process of ECR falls in the category of crevice corrosion. The environmental degradation of the steel substrate consists of three mechanisms, namely adhesion loss, anodic blistering and cathodic delamination (Nguyen and Martin 1996, 2004). The corrosion process of ECR is illustrated in Figure 2.16.  Figure 2.16: Illustration of the corrosion process of fusion bonded epoxy coated steel (adapted from (Nguyen and Martin 2004)) Adhesion loss of the coating is caused by water diffusion through the FBE layer. Consequently, water accumulates along the entire metal-polymer interface as shown in Figure 2.16 (Legghe et al. 2009; Powers 2009; Nguyen and Martin 1996, 2004).  Chloride ions migrate from the coating holiday along the substrate-coating interface. As a result, water and chloride are present under the coating next to the damage. The iron of the steel substrate reacts with the water and chloride in the anodic reaction that releases electrons. Accumulation of the   2 Theoretical Background   35 corrosion products caused by the aforementioned reaction leads to the formation of anodic blisters (Nguyen and Martin 1996, 2004; A. Sagüés et al. 2008). The concentration of chloride ions inside the blister is reported to be four to six times that of the bulk solution (Nguyen and Martin 1996, 2004). Furthermore, the pH inside the anodic blister is very low and Nguyen and Martin reported values between 3.5 and 5 (Nguyen and Martin 1996, 2004; A. Sagüés et al. 2008). Increased exposure time leads to an increase in the size of the blisters and the blister growth is accelerated at elevated temperatures (Nguyen and Martin 1996, 2004). Sodium ions are transported along the metal-epoxy interface from the coating damage to the cathodic area (Nguyen and Martin 1996, 2004). In addition to the Na+ ions, water and oxygen molecules, that diffused through the FBE coating, as well as electrons, released by the anodic oxidation, are present in this area. Consequently, sodium hydroxide is formed in the cathodic reduction, that increases the pH of the solution under the FBE coating. Nguyen and Martin observed a pH between 10 and 11 in the cathodically delaminated region (Nguyen and Martin 1996, 2004). The cathodic delamination expands around the anodic blister away from the coating damage (Nguyen and Martin 1996, 2004; A. Sagüés et al. 2008). Furthermore, cathodic sites have also been observed at the coating damage (A. Sagüés et al. 2008). In the case of electrically connected ECR by contact points of coating damage, as well as low resisitivity concrete, long range corrosion macrocell formation has been observed. The latter results in undesirable anode-to-cathode ratios (A. A. Sagüés, Perez-Duran, and Powers 1991), that increase the corrosion rate. Furthermore, the presence of oxygen is not necessary for the corrosion of ECR. Thus, corrosion can even occur in saturated concrete (Lau, Sagüés, and Powers 2010).  It should be noted that exposure to a neutral NaCl environment is reported to cause corrosion of the steel at the location of the damage (Nguyen and Martin 1996, 2004). Consequently, no anodic blisters form. However, cathodic delamination still occurs (Nguyen and Martin 1996, 2004).   2.3.3 Effect of Chloride and Rebar Surface Conditions While the presence of chloride ions in concrete is well known to promote the corrosion of reinforcing steel, the halogen also affects the cathodic disbondment of ECR (Nguyen, Hubbard, and Pommersheim 1996; Nguyen and Martin 2004; L. Li and Sagues 2001; Alberto A and Powers 1996) Even though Sagüés et al. (Alberto A and Powers 1996) observed disbondment of the epoxy coating in absence of chloride ions, chloride ions have been reported to aggravate the delamination of the coating from its substrate (A. A. Sagüés, Powers, and Kessler 2001). Thus, chloride accelerates the disbondment but is not necessary for the delamination initiation.    2 Theoretical Background   36 The surface condition of ECR and UCR differ greatly. Both rebar types are made of hot rolled microalloyed steel, which is heated to 900 °C (i.e. 1650F) before shaping the alloy into the desired geometry and dimensions. During this heating process the surface of the metal oxidizes, resulting in an oxide scale, that gives the steel an anthracite appearance. This oxide layer consists mainly of hematite, magnetite and wustite (Iordanova et al. 2000; Sun et al. 2004). This oxide layer covers the surface of the uncoated rebars and protects the underlying alloy. However, the epoxy coated rebars do not have such a protective layer, because it is mechanically removed before the application of the epoxy coating (ASTM International 2014a, 2013). Thus, once the epoxy coating is damaged, the surface conditions of the two rebar types are not identical.  The black oxide layer has been reported to increase the critical chloride concentration (Mohammed and Hamada 2006; Mammoliti et al. 1996), above which breakdown of the passive layer and active corrosion occurs. However, damage of the epoxy coating exposes a significantly smaller area of the steel of ECR compared to the entire surface of UCR. A decrease in the specimen size has been found to increase the critical chloride content (Angst et al. 2011). Thus, the lack of a black oxide layer for ECR decreases the critical chloride content and the smaller surface area of ECR increases the critical chloride content compared to UCR.  2.4 Corrosion Detection of Epoxy Coated Rebar A variety of non-destructive testing (NDT) methods is available to evaluate the deterioration of reinforced concrete structures. NDT methods that assess the extent of deterioration due to reinforcing steel corrosion can be distinguished into three categories. These include methods that measure general degradation of the concrete itself, methods that assess the extent of delamination of the concrete cover over the reinforcing steel and techniques that detect the probability or rate of corrosion (Strategic Highway Research Program - SHRP 2 2013). Forms of concrete degradation that are indicators for the environmental corrosiveness or the extent of corrosion include increased chloride content and microcracking of the concrete. Furthermore, delamination is a consequence of corrosion of the reinforcement.  The environmental corrosiveness can be an indicator for the likelihood of corrosion of uncoated rebar (UCR) (Malhotra and Carino 2003). However, under identical environmental conditions the corrosion behaviour of ECR greatly varies depending on the damage level of the epoxy coating (Sturgeon et al. 2010). Hence, the environmental corrosiveness includes only very limited information about the corrosion of ECR.    2 Theoretical Background   37 Corrosion needs to be fairly extensive to lead to delamination or cracking of concrete. However, early detection is desirable for safety and economic reasons. Consequently, NDT methods for corrosion detection are the most desirable for the condition assessment of bridge decks. Well known NDT techniques to detect corrosion of reinforcing bars are the Half Cell and Resistivity Probe (Strategic Highway Research Program - SHRP 2 2013). Recently, Ground Penetrating Radar has been used in research studies to identify corrosion of reinforcing bars embedded in concrete (Kabir and Zaki 2011)(Dinh 2014). The linear polarization resistance (LPR) technique is an electrochemical method to detect corrosion, that is mainly popular in research studies (Millard et al. 2001)  2.4.1 Conventional Non-Destructive Testing 2.4.1.1 Half Cell The electrical potential between a reference electrode and the steel reinforcement is measured with the half-cell method. This thermodynamic information indicates the probability of corrosion (Malhotra and Carino 2003; Jones 1996).  Figure 2.17 shows the NDT equipment of the half-cell method. A half-cell reference electrode (REF) is connected to a data acquisition system, which in turn is connected to the rebar under investigation. The REF is positioned on the concrete surface over top of the rebar. Direct contact between the REF solution and the pore water of the concrete is created by a sponge at the bottom end of the half-cell electrode (Malhotra and Carino 2003). The data acquisition system then measures the electrochemical potential between the REF and the rebar (Malhotra and Carino 2003). This open circuit potential (OCP) is a mixed potential because it is influenced by the reversible potential of both the anode and the cathode (chapter 2.1.2.4, Figure 2.6). Thus, the OCP represents a combination of the reversible anodic and cathodic potentials. At this potential the sum of all cathodic currents equals the sum of all anodic currents (Jones 1996). The OCP is an important thermodynamic parameter, that gives information about the likelihood of corrosion.  (Malhotra and Carino 2003; Jones 1996). It should be emphasized that it is not an indicator for the rate of corrosion. A lower (i.e. more negative) measured OCP indicates a higher probability of corrosion taking place. Thus, a lower or more negative OCP is commonly referred to as a more active OCP. Conversely, a higher, more positive OCP is referred to as a nobler OCP. However, no conclusions can be drawn to the speed of the corrosion process. The copper-saturated copper sulfate (CSE) (Figure 2.17) and saturated calomel (SCE) half-cell electrodes are common choices for the reference electrode (Malhotra and Carino 2003). The   2 Theoretical Background   38 advantage of the CSE lies in its rugged and simple design. The SCE is commonly chosen for corrosion experiments due to its steady potential that can easily be controlled by maintaining saturation of the KCl solution inside the SCE (Jones 1996). All potentials below -350 mVCSE and above -200 mVCSE are an indicator of active and passive corrosion behaviour, respectively. Free potentials between these two limits indicate uncertain corrosion behaviour (Malhotra and Carino 2003).  Figure 2.17: Half-cell potential equipment Saricimen et al. (Saricimen et al. 1997) monitored the free potential of ECR in NaCl contaminated concrete over a course of 200 days and observed a drop in the free potential, that coincided with corrosion initiation. The latter drop occurred up to 44 days sooner for ECR with mechanical coating damage compared to intact ECR. However, the corrosion of uncoated rebar initiated at approximately the same time as the intact ECR. Furthermore, UV damaged ECR led to the least active potentials and smallest potential drop. Furthermore, the intact ECR specimens reached potentials as low as -600 mVSCE but the specimens with mechanically damaged epoxy coating only reached potentials as low as -440 mVSCE (Saricimen et al. 1997).   2.4.1.2 Linear Polarization Resistance The linear polarization resistance (LPR) method forces the reinforcing steel to leave its state of equilibrium. The consequent change in potential is measured over a fixed time interval leading to the evaluation of the corrosion rate (Jones 1996; Malhotra and Carino 2003).   2 Theoretical Background   39 The general set-up of the linear polarization resistance method incorporates three different electrodes, namely the reference (REF), auxiliary (AUX) and working electrode (Malhotra and Carino 2003), that are all connected to a potentiostat. The reinforcing steel represents the working electrode. The REF is a half-cell electrode such as a copper-copper sulfate (CSE) or saturated calomel (SCE) electrode (Jones 1996). Furthermore, the AUX is a ring electrode that is place around the reference electrode. The REF, surrounded by the AUX, is placed on the concrete surface and a potentiostatic pulse is applied to the concrete through the AUX. Consequently, a change in the electrochemical potential of the rebar occurs, that is measured by the potentiostat over a short period of time. With the acquired information the corrosion current density and the corrosion rate can be evaluated (Malhotra and Carino 2003).   2.4.1.3 Resistivity Methods There are two types of concrete resistivity measurements, that differ in the type of applied current i.e. AC and DC methods. The Wenner probe belongs to the AC techniques and the iCor is a DC method. 2.4.1.3.1 Wenner Probe The resistivity equipment using a Wenner probe is pictured in Figure 2.18. The probe consists of an array of four electrodes that are positioned on the concrete surface. The concrete functions as a conducting medium with the pore solution representing a conducting electrolyte.    2 Theoretical Background   40  Figure 2.18: AC resistivity NDT method with 4 electrode array Wenner probe An alternating current is applied to the concrete through the outer electrodes that are connected to an AC current source (Malhotra and Carino 2003) located inside the data acquisition system (DAS) in Figure 2.18. Consequently the reinforcing steel is polarized and a potential drop is measured by the DAS between the inner two electrodes. With Ohm’s law the electrical resistivity can now be determined as the ratio of the potential drop to the induced current (Malhotra and Carino 2003). The magnitude of the detected potential change is affected by the distribution of the conductivity. Therefore, the resistivity is highly dependent on the properties of the pore structure of the concrete since the pore solution functions as a conducting electrolyte (Malhotra and Carino 2003) 2.4.1.3.2  iCor The iCor is a new piece of equipment that also consists of a 4 electrode array connected to a DAS. Similar to the Wenner probe, the two outer electrodes of the iCor array apply a current and a potential drop between the inner two electrodes is measured by the DAS. However, the applied current is a direct current. Unlike the Wenner probe, the iCor does not measure the voltage response at a single frequency but rather over a frequency sweep (Ghods, Alizadeh, and Salehi 2014). A frequency sweep of the applied current from high to low frequencies is utilized and the resulting voltages are recorded (Ghods, Alizadeh, and Salehi 2014). In the case of corroding rebar the measured voltage is low and independent of the frequency. In the case of non-corroding rebar however, the voltage response decreases with increasing frequency (Ghods, Alizadeh, and Salehi 2014).    2 Theoretical Background   41 However, low frequency measurements are not only noise sensitive but also very time consuming (Ghods, Alizadeh, and Salehi 2014). Consequently, the low frequency response of the reinforcement is determined by application of a narrow DC current pulses (Ghods, Alizadeh, and Salehi 2014). The aforementioned pulse is applied for a short period of time and the voltage response is recorded with a very high sampling rate. Subsequently, the low frequency impedance is determined through signal processing (Ghods, Alizadeh, and Salehi 2014).  2.4.1.4  Ground Penetrating Radar The ground penetrating radar (GPR) technique is an electromagnetic (EM) method, that has been gaining in popularity as a corrosion detection method in the last two decades (Hong, Lai, and Helmerich 2012; Lai, Kind, and Wiggenhauser 2010; Kabir and Zaki 2011; East and Conference 2007; Eisenmann et al. 2013; Strategic Highway Research Program - SHRP 2 2013). The equipment consists of a transmitter and a receiver antenna. The transmitter applies a high frequency electromagnetic wave to the object under investigation. When the wave hits an anomaly, such as a rebar inside of concrete, a portion of the wave is reflected and detected by the receiver. The reflection of the EM wave is dependent on the electrical properties of the anomaly and the difference to the surrounding medium. Electrical conductivity and relative dielectric permittivity have a great influence on the reflected signal (Malhotra and Carino 2003). Reinforcing bars are a very good reflector and Figure 2.19 shows that a rebar can be identified as a hyperbola in the GPR B-scan.     2 Theoretical Background   42  Figure 2.19: Profile view of a reinforced concrete slab with corresponding 2D GPR B-scan  The clarity of the hyperbolas is dependent on the difference in dielectric constant or relative electrical permittivity at the concrete-rebar interface. The latter constant is a measure of a material’s ability to store electrical energy. Metals such as microalloyed steel have an infinitely large dielectric constant, whereas concrete has a low constant between 8 and 10 (Strategic Highway Research Program - SHRP 2 2013).  Thus, a large part of the incident wave is reflected at the rebar-concrete interface, which results in a clear B-scan as shown in Figure 2.19. Unlike microalloyed steel, corrosion products such as iron oxide have a very low electrical conductivity and thus dielectric constant. Consequently, the difference in electrical permittivity at the concrete-rust interface is smaller compared to the concrete-steel interface. Thus, a smaller part of the incident wave is reflected back to the receiver by a corroded rebar compared to an intact rebar, which results in reduced clarity of the hyperbola’s in the B-scan (Strategic Highway Research Program - SHRP 2 2013). The B-scan is made up of the amplitude and time of the EM wave and is integrated into commercially available software and is thus easy to attain even without expertise of the technology. Thus, it is a popular tool to visualize the GPR data (Strategic Highway Research Program - SHRP 2 2013; Eisenmann et al. 2013; Kabir and Zaki 2011; East and Conference 2007). However, the analysis of the B-scan is very subjective and the quantitative analysis of measures such as the amplitude, travel time, frequency spectrum and signal energy is not yet integrated into all commercially available software (Eisenmann et al. 2013). While the A-scan, which consists of the plot of the signal voltage in relation to its arrival time, is made available in some software, the quantitative data the plots are made of is not. Thus, many researches have developed their own data analysis tools to quantitatively measure the effects of corrosion of reinforcing bars on the GPR signal.  During the early stages of corrosion of rebar embedded in concrete, corrosion products diffuse into the concrete. Thus, the corrosion products are located closer to the concrete surface than the   2 Theoretical Background   43 rebar. The EM waves are then reflected at the concrete-rust interface rather than at the concrete-rebar interface (Lai, Kind, and Wiggenhauser 2010). Due to the reduced travel distance of the waves, the signal travel time decreases and the signal energy as well as the frequency increases (Hong, Lai, and Helmerich 2012; Lai, Kind, and Wiggenhauser 2010). Further corrosion activity, leads to the formation of cracks, that increase in number, width and length with increasing amount of corrosion. Consequently, additional interfaces, such as air-rust and air-concrete, are created, at which the EM waves are reflected (Lai, Kind, and Wiggenhauser 2010). Consequently, the initial corrosion effects are reversed and the travel time increases, whereas the amplitude and frequency are reduced (Lai, Kind, and Wiggenhauser 2010; Eisenmann et al. 2013). The effects of chloride ions on the GPR signal have been investigated by Hong et al. (Hong, Lai, and Helmerich 2012), who concluded that the increased conductivity due to the addition of chloride to concrete led to a reduction in the real part and an increase in the imaginary part of the complex dielectric constant. The reduction of the real part led to stronger direct waves and delayed reflected waves. The enlarged imaginary part resulted in decreased signal energy of the reflected wave with a reduction of both the amplitude and centre frequency (Hong, Lai, and Helmerich 2012). Consequently, chloride contamination and corrosion of reinforcing bars initially have opposing effects on the GPR signal. However, once a larger degree of corrosion is reached, the impact of the corrosion on the GPR signal reverses and thus becomes similar to the impact of chloride ions on the GPR signal. Unlike chloride and advanced corrosion, moisture reduces the amplitude of the direct wave. However, all three aspects attenuate the reflected signal. Eisenmann et al. analyzed the peak to peak voltage of the A-scan of uncoated as well as epoxy coated rebar with varying degrees of corrosion (Eisenmann et al. 2013). A reduction in the voltage was observed for the uncoated bar whose diameter had been reduced by 50 % due to corrosion. However, the epoxy coated bar with the same degree of corrosion showed an increase in the peak to peak voltage (Eisenmann et al. 2013).  2.4.1.5 Ultrasonic Pulse Velocity (UPV) The ultrasonic pulse velocity method is based on the principle of stress wave propagation and relates the wave velocity to the properties of the object under investigation. A dynamic load causes three types of stress waves in a solid elastic medium, namely compressional, shear and surface waves. The stress wave with the highest velocity is the compressional wave. This type of wave is also known as the P-wave or longitudinal wave (Malhotra and Carino 2003; Watanabe et al. 2014). The velocity of the stress wave greatly depends on the elastic properties of the medium it is travelling through.    2 Theoretical Background   44 The discussed NDT method consists of a transmitting transducer T, that applies a stress wave to the object under investigation and a receiving transducer R that detects the pulse at a distance L away from the transmitter (Malhotra and Carino 2003).  Three different test set-ups are shown in Figure 2.20. The indirect set-up is the most practical measurement configuration because access from only one side of the object under investigation is required.   Figure 2.20: Setups of UPV Transmitter T and Receiver R Both transducers are connected to a data acquisition system that measures the transit time of the stress wave. The transit time is defined as the time between pulse application by the transmitter and pulse detection by the receiver. The quotient of the distance between the transducers and the transit time is equal to the P-wave velocity (Malhotra and Carino 2003). When using the indirect set-up a reduction in the distance between the transducers results in the stress wave travelling closer to the surface (Malhotra and Carino 2003).  Watanabe et al. have shown that small amounts of corrosion induced cracking lead to a reduction of the measured velocity (Watanabe et al. 2014). However, a further increase in corrosion does not change the velocity significantly. Hence, early detection of corrosion is possible but the amount of corrosion cannot be related to the measured wave velocity (Watanabe et al. 2014). Furthermore, research has only been performed on specimens with a concrete cover of 3 cm (Watanabe et al. 2014).  2.4.2 Induction Heating Active infrared thermography involves the heating of an object and monitoring its thermal behaviour.  Heating of the object under investigation can be accomplished through solar radiation or a heater such as an induction heater (Strategic Highway Research Program - SHRP 2 2013; Kobayashi and Banthia 2011). Induction heating is advantageous because only magnetic objects are being heated and do not need to be in direct contact with the heater (Kobayashi and Banthia 2011).   2 Theoretical Background   45 Consequently, in the case of reinforced concrete only the reinforcing bars are being heated and not the surrounding concrete (Kobayashi and Banthia 2011).  Infrared (IR) cameras measure and record the thermal radiation, that is emitted by an object. Regions of different temperatures can be identified which relate to different thermal properties of the object under investigation. Thermal conductivity, specific heat capacity and density are the material properties that influence the heat distribution and heat flow the most. Corrosion products have a lower density and consequently lower thermal conductivity in comparison to microalloyed steel. Consequently, it is expected that corroded rebar does not cool down as quickly as sound rebar (Kobayashi and Banthia 2011). However, not only the thermal properties of corrosion products differ from those of steel but also the magnetic properties. Thus, in order to fully understand the thermal response of a reinforced concrete specimen to induction heating, one needs to separate the magnetic and thermal behaviour.   2.4.2.1 Hall-Effect Sensing 2.4.2.1.1 Induction The basic principles of electromagnetism dictate that if a current is flowing through a straight conductor, a magnetic field is created around the conductor with its orientation being determined by the direction of the current. Consequently, reversing the direction of the current flowing through the conductor, reverses the orientation of the magnetic field surrounding the conductor. This is shown in Figure 2.21, where the green cylinder represents the conductor, the blue arrow represents the direction of the electric current and the red circular arrows represent the direction of the magnetic flux.   Figure 2.21: Circular magnetic field (red) induced by a current (blue) flowing through a straight conductor Conversely, Faraday’s Law of electromagnetic induction states that if a conductor is exposed to a changing magnetic field, an electromotive force (emf), is induced in the conductor. The emf is an induced voltage, whose magnitude depends on the rate of the changing magnetic flux. A faster rate leads to a higher voltage. If the orientation of the magnetic field is continuously reversed at a constant   2 Theoretical Background   46 rate, the induced voltage is continuously reversed between positive and negative polarity. In other word, an AC output voltage is generated inside the conductor. In the case of the conductor being a closed circuit, a current will flow through the conductor as a result of the induced voltage. Based on the basic principles of electromagnetism, this current creates a magnetic field around the conductor (Figure 2.21). This self-induced electromotive force opposes the change that caused it. Furthermore, a faster rate of change leads to a larger electromotive force. This phenomenon of an opposing self-induced emf or voltage is commonly known as Lenz’s law.  In a nutshell, if there is a relative motion between a static magnetic field and a conductor, according to Faraday’s law, a voltage is induced in the conductor, which in turn causes a circulating current to flow within the conductor. These circulating currents are known as Eddy currents. According to Lenz’s law, each Eddy current induces a magnetic field around itself and consequently induces an emf voltage. The direction of this self-induced emf opposes the change that caused it. Consequently, eddy currents act like a negative force and increase the resistance of the conductor to the induced current. This leads to resistive heating and power loss.  2.4.2.1.2 Hall-Effect When a semiconductor, carrying an electric current, is placed in a magnetic field, the charge carriers in the semiconductor are subject to a force. This is the basic principle of the Hall effect as well as the Lorentz force.  Holes (+q) and electrons (-q) are charge carriers inside a semiconductor, that move based on their charge in and against the direction of an applied electrical current (Razeghi 2009). If a constant current passes through a semiconductor in the x-direction, creating a uniform electric field with current density 𝐽 , the charge carriers move along the x axis as shown in Figure 2.22.   2 Theoretical Background   47  Figure 2.22: Movement of holes and electrons inside a semiconductor (adapted from (Razeghi 2009)) Next, the semiconductor is placed in a magnetic field with flux density ?⃗? , whose orientation (z-direction) is perpendicular to the direction of the electrical current (x-direction). The charge carriers are now subject to the Lorentz force 𝐹𝐿𝑜𝑟𝑒𝑛𝑡𝑧⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  , whose orientation in the y-direction is perpendicular to both 𝐽  and ?⃗?  as shown in Figure 2.23. The Lorentz force influences the direction of movement of the charge carriers (Razeghi 2009).   Figure 2.23: Lorentz force on holes inside a semiconductor (adapted from (Razeghi 2009)) In an infinite semiconductor, the charge carriers would circulate around an axis parallel to the direction of the magnetic flux density (z-direction). In a finite semiconductor, such as the one shown in Figure 2.24, positively and negatively charge carriers, that are subject to the Lorentz force, accumulate on opposing sides of the semiconductor (Razeghi 2009). This creates a charge gradient in the y-direction, perpendicular to the direction of the electrical current as well as magnetic flux density   2 Theoretical Background   48 and parallel to the direction of the Lorentz force (Figure 2.23). The electric field resulting from the charge gradient is called the Hall electric field (Razeghi 2009). The force resulting from the Hall electric field runs in the opposite direction to the Lorentz force.   Figure 2.24: Holes in the Hall electric field (adapted from (Razeghi 2009) Semiconductors, whose charge carriers are predominantly holes (positive charge) are called p-type semiconductors and the resulting Lorentz force is shown in Eq. (39), where q denotes the electric charge and 𝑣ℎ⃗⃗⃗⃗  denotes the velocity of the holes in the x-direction (Razeghi 2009). Similarly, semiconductors whose charge carriers are predominantly electrons (negative charge) are called n-type semiconductors and the resulting Lorentz force is shown in Eq. (40), and 𝑣𝑒⃗⃗  ⃗ denotes the velocity of the electrons in the x-direction.  𝐹𝐿𝑜𝑟𝑒𝑛𝑡𝑧⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  =  𝑞 ∙ 𝑣ℎ⃗⃗⃗⃗ × ?⃗?  (39)  𝐹𝐿𝑜𝑟𝑒𝑛𝑡𝑧⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  =  − 𝑞 ∙ 𝑣𝑒⃗⃗  ⃗× ?⃗?  (40)  At equilibrium, the Hall electric field force and the Lorentz force balance each other out (Razeghi 2009). The resulting Hall electric field strength is shown in Eq. (41) and Eq. (42) for the p- and n-type semiconductors respectively.    𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ =  − 𝑣ℎ⃗⃗⃗⃗ × ?⃗?  (41)  𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ =  − 𝑣𝑒⃗⃗  ⃗× ?⃗?  (42) In Eq. (41) and Eq. (42) the electric field is oriented in the y-direction, the velocity is oriented in the x direction and the magnetic flux is oriented in the z-direction (Razeghi 2009). Substituting the drift velocity of the holes and electrons into the Hall electric field strength leads to Eq. (43) and Eq.   2 Theoretical Background   49 (44), where p and n denote the concentrations of the positive (i.e. holes) and negative (i.e. electrons) charge carriers in the semiconductor.  𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ =  𝐽 𝑝 ∙ 𝑞× ?⃗?  (43)  𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ =  − 𝐽 𝑛 ∙ 𝑞× ?⃗?  (44) In Eq. (43) and Eq. (44) ,𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  is an experimentally measured quantity and 𝐽  and ?⃗?  are experimental parameters. Furthermore, p and n are material properties of the semiconductor (Razeghi 2009). Extracting the material properties of the semiconductor from Eq. (43) and Eq. (44) gives the Hall constant shown in Eq. (45) and Eq. (46) for p- and n-type semiconductor respectively.  𝑅𝐻 = 1𝑝 ∙ 𝑞 (45)  𝑅𝐻 = − 1𝑛 ∙ 𝑞 (46) Consequently, the equation for 𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  can be rewritten as Eq. (47)  𝐸𝐻𝑎𝑙𝑙⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗  =  𝑅𝐻 ∙ (𝐽  × ?⃗? ) (47) For semiconductors, that contain an equal amount of both holes and electrons, the Hall constant RH can be determined with Eq. (48), where µh and µe are the hole and electron mobility respectively (Razeghi 2009).   𝑅𝐻 = 1𝑞∙𝑝 ∙ 𝜇ℎ2 − 𝑛 ∙ 𝜇𝑒2(𝑝 ∙ 𝜇ℎ + 𝑛 ∙ 𝜇𝑒)2 (48)  2.4.2.1.3 Hall Effect Sensors In the case of Hall effect sensors, the sensor is made of a semiconductor. Consequently, the Hall constant RH is a material property of the Hall effect sensor. The uniform magnetic field 𝐽  is applied to the sensor through a power supply such as a standard battery. Essentially, RH and 𝐽  are experimental constants and the only variable in Eq. (47) is the applied magnetic flux ?⃗? . Thus, changes in the magnetic field can be detected by a change in the measured Hall effect voltage.  The presence of ferromagnetic materials such as microalloyed steel cause changes in an existing magnetic field. Para- and diamagnetic materials on the other hand do not distort the magnetic field. Corrosion products consist primarily of iron oxides, which are paramagnetic and thus do not affect the magnetic field. Concrete and air are also paramagnetic and water is diamagnetic. Consequently,   2 Theoretical Background   50 mass loss of the rebar embedded in concrete due to corrosion would result in a smaller change of an applied magnetic field than an intact rebar would.  2.4.2.2 Thermal Detection 2.4.2.2.1 Thermal Properties The thermal properties of carbon steel, corrosion products, fusion bonded epoxy and concrete are very different and together create a complex thermal system. Changing the amount of one of these thermal system constituents affects the thermal response of the entire system. This effect is particularly strong if the amount of the material with the highest or lowest thermal conductivity is changed. Steel has the highest and corrosion products have the lowest thermal conductivity. Thus, the corrosion process, during which steel is consumed and corrosion products are produced, is expected to have a significant effect on the thermal response of the whole system.  Kobayashi and Banthia (Kobayashi and Banthia 2011) cited thermal conductivities of 2.7 W/mK for concrete and 0.07 W/mK for corrosion products from Oshita et al. (Oshita et al. 2009). Carbon steel reinforcing bars have a significantly larger thermal conductivity of 80.4 W/mK (CMC Commercial Metals 2015). Fusion bonded epoxy coatings have a thermal conductivity of approximately 0.3 W/mK (3M 2012, 2013; Axalta 2014). 2.4.2.2.2 Heating Process Heating a rebar from the inside can be achieved by running a DC current through a high resistance heating alloy wire such as nichrome wire inserted into the rebar. The heat Q required to elevate the temperature of an object by ΔT is determined by its mass m and specific heat c (Rama Corporation 2017; Incropera et al. 2007) as shown in Eq. (49).   𝑄 = 𝑚 ∙ 𝑐 ∙ ∆𝑇 (49) Thus, the total heat Q*, necessary to heat a piece of steel by means of a Nichrome wire inserted into the steel, is determined by Eq. (50) (Rama Corporation 2017), where the heat of the steel QS as well as the heat of the Nichrome wire QNC are determined by Eq. (49).  𝑄∗ = 𝑄𝑆 + 𝑄𝑁𝐶 (50) However, Eq. (50) does not take heat losses due to conduction, convection or radiation into account. In many cases, such as rebar embedded in concrete, conduction is the primary heat loss phenomenon. The amount of heat lost ΔQCD through conduction during an exposure time te is determined for the one dimensional case by Eq. (51) (Rama Corporation 2017; Incropera et al. 2007), where k is the thermal conductivity, d is the thickness of the material and A the surface area of the   2 Theoretical Background   51 interface, where the conduction occurs. In the case of rebar embedded in concrete, A would the entire surface area of the rebar, that is in contact with the concrete.   ∆𝑄𝐶𝐷 = 𝑘 ∙  𝐴𝑑∙ ∆𝑇 ∙ 𝑡𝑒 (51) The combined heat loss ΔQCV+R due to convection and radiation can be calculated with Eq. (52) and the surface heat loss factor f (Rama Corporation 2017). This heat loss factor combines the convection heat transfer coefficient, the emissivity and the Stefan Boltzman (Incropera et al. 2007) constant into one value. In the case of a piece of rebar embedded in concrete with exposed rebar ends, ΔQCV+R only needs to be determined for the exposed rebar ends, of which the surface area represents the area A in Eq. (52).  ∆𝑄𝐶𝑉+𝑅 = 𝐴 ∙ 𝑓 ∙ 𝑡𝑒 (52) The total heat loss can simply be determined by the sum of the heat losses from Eq. (51) and Eq. (52). To determine the required power P, one needs to employ Eq. (53), in which ts and te are the start-up and the exposure time respectively (Rama Corporation 2017).   𝑃 =∑𝑄𝑡𝑠+ 23∙∑∆𝑄𝑡𝑒 (53) The power applied to the system of the DC power supply can be determined by Ohm’s law as shown in Eq. (54), where R is the resistance and I is the applied current (Rama Corporation 2017; Incropera et al. 2007).  𝑃 = 𝑅 ∙ 𝐼2 (54) Setting the required and applied power equal gives Eq. (55) with the start-up time ts as the only unknown.   𝑅 ∙ 𝐼2 = 𝑄𝑁𝐶 + 𝑄𝑆𝑡𝑠+ 23∙∆𝑄𝐸 + ∆𝑄𝐶𝑃 + ∆𝑄𝐶𝑡𝑒 (55) 2.4.2.2.3 Infrared Thermography Light and infrared (IR) radiation are electromagnetic (EM) waves. IR waves are located between microwaves and visible light in the EM wave spectrum. The wave lengths range from 780 nm to 1 mm and the frequencies range from 5 × 1011 Hz to 5 × 1014 Hz (Vollmer and Moellmann 2010). The speed of propagation is defined as the product of the frequency and wavelength.  Thermal radiation is a type of IR radiation and its wave lengths range from 780 nm to approximately 14 µm (Vollmer and Moellmann 2010). This thermal IR wave spectrum is divided into three spectral ranges called the long wave (7 to 14 µm), mid wave (3 to 5 µm) and short wave (0.9 to   2 Theoretical Background   52 0.7 µm) region. Most IR cameras are specifically made for one of these ranges (Vollmer and Moellmann 2010).  Every body emits thermal radiation in the form of EM waves at temperatures larger than 0 K (-273.15 °C). The amount and distribution of the thermal radiation is a function of the wavelength, which in turn is dependent on the temperature as well as material properties. Solids are opaque to IR radiation and in extension thermal radiation (Vollmer and Moellmann 2010). Hence, the thermal emission, recorded by IR cameras, only refers to the surface of the solid. The emissivity is a material property that is highly dependent on the surface conditions of the object under investigation. It is equal to the ratio of the actual amount of thermal radiation emitted to that of a blackbody under identical conditions (Vollmer and Moellmann 2010). A blackbody is a perfect emitter of thermal radiation with an ideal surface. Consequently, emissivity can have values between zero and one.  Max Planck derived a concept relating the spectral excitance (or total radiant power in a wavelength interval) of a blackbody to its given temperature (Vollmer and Moellmann 2010). For any given wavelength, radiance increases with an increase in temperature. Furthermore, the wavelength depends on the temperature, as an increase in temperature decreases the wavelength. Hence, short wavelengths indicate high temperatures and long wavelengths indicate low temperatures (Vollmer and Moellmann 2010). A practical example of this relation between wavelength and temperature is a simple hot plate, that turns red once it reaches a high enough temperature. The high temperature leads to wavelengths so short, that they fall into the visible instead of the IR spectrum.  The Stefan Boltzman Law builds on the concept of spectral excitance and dictates that its integral gives the total excitance (also referred to as emissive power). The latter depends solely on the temperature for blackbodies (Vollmer and Moellmann 2010).  The thermal radiation, that is oriented into the direction of an IR camera, is focused on the detector of the IR camera and measured quantitatively. The IR camera measures the total radiant power within a certain spectral interval. The temperature of the object is then determined from the measured total radiant power and the emissivity.     3 Experimental Procedure   53 3 Experimental Procedure The research project was divided into three parts, that investigated the corrosion behaviour, accelerated corrosion and the corrosion detection of epoxy coated rebar.  3.1 Corrosion Behaviour 3.1.1 Specimens Both epoxy coated (ECR) as well as uncoated (UCR) rebar were used in this project. The ECR was sourced from Harris Rebar in Richmond, BC and met ASTM A775. All specimens were made of 5 cm long pieces of 10M rebar and Figure 3.1 shows a UCR and ECR sample. The specimens were cleaned in reagent alcohol in an ultrasonic bath for 5 minutes and subsequently dried at room temperature. Once the specimens were dry, a copper wire was connected to each specimen with conductive silver epoxy. After a 24 hour curing period both ends were covered with a two part epoxy repair material designed for fusion bonded epoxy (3M™ Scotchkote™ 413/215 Patch Compound).   Figure 3.1: UCR (top) and ECR (bottom) sample for the electrochemical corrosion behaviour tests Consequently, only the curved surface area of the uncoated rebar shown in Figure 3.1 was exposed to the solution during the electrochemical corrosion behaviour testing. Assuming a nominal perimeter of 35.5 mm the exposed area was 17.75 cm2. The coating of the epoxy coated rebar was damaged in six locations by drilling a hole into the epoxy to create simulated holidays in the coating (Figure 3.1). Each damage exposed an area of the microalloyed steel of 1 mm2 resulting in a total area   3 Experimental Procedure   54 per specimen of 6 mm2. Consequently, the exposed area of an ECR specimen was equal to 0.33 % of the exposed area of a UCR specimen.  3.1.2 Methodology 3.1.2.1  Electrochemical Testing A Versastat 4 potentiostat, connected to a silver/silver chloride reference electrode (REF), a graphite counter electrode (or auxiliary electrode AUX) and the test specimen as the working electrode (WE), was used for all electrochemical measurements (Figure 3.2). The composition of the test solutions varied to simulate the effect of environmental parameters on the corrosion behaviour.   Figure 3.2: Electrochemical testing setup The experimental setup is shown in Figure 3.2 and Figure 3.3 illustrates the role of the three electrodes in the electrical circuit. All potential measurements are performed between the REF and the WE and signify the difference in potential ΔE between the REF and WE. The potential EREF of the REF is known. Thus, the potential EWE of the WE is determined by the sum of EREF  and the measured ΔE.   3 Experimental Procedure   55 Depending on the type of electrochemical test, a current is applied to the system through the AUX (also referred to as counter electrode). Application of a current through the AUX leads to a change in the potential of the WE. However, the potential of the REF remains constant. Thus, any change in the potential measurements results from a change in the potential of the WE.    Figure 3.3: Schematic of the electrochemical testing setup Short term and long term tests were performed. For the short term (ST) tests the specimens were not submerged in the solution prior to testing. For the long term (LT) tests the specimens were submerged in the testing solution for four weeks prior to testing. During these four weeks the solution was aerated and kept at 35°C (Nguyen and Martin 2004). Submerging of the specimens in an aerated solution at an elevated temperature for four weeks during the long term testing was expected to lead to adhesion loss of the coating (Nguyen and Martin 2004) and corrosion at or near the coating damage. The testing procedure of the ST and LT series are shown in Table 3.1.       3 Experimental Procedure   56 Table 3.1: Short term (ST) and long term (LT)  testing procedure Short-term (ST) Long-term (LT) Process Notes Process Notes  Incubator 4 weeks at 35°C OCP 4 hours LPR -20 mV to +20 mV vs. OCP at 0.6 V/h OCP 4 hours OCP 4 hours PDP 0 V to +2 V vs. OCP at 0.6 V/h PDP 0 V to +2 V vs. OCP at 0.6 V/h OCP 4 hours OCP 4 hours 3.1.2.1.1 Open Circuit Potential The open circuit potential (OCP) was measured between the working electrode and the reference electrode. The open circuit potential measurement was performed over a period of 4 hours. The aforementioned time period was chosen to allow for the potential to stabilize.  3.1.2.1.2 Potentiodynamic Polarization The potentiodynamic polarization (PDP) was performed immediately after the OCP measurement. During the potentiodynamic polarization the potential of the corrosion cell was increased from the OCP by application of a current through the counter electrode (chapter 2.2.1). The specimens were anodically polarized to a potential of +2 V versus OCP with a scan rate of 0.6 V/h (=0.1667 mV/s) (ASTM International 2014c). The specimens were only anodically polarized because prior cathodic polarization could have altered the entire polarization curve. The PDP results and data analysis consist of a potential to current density plot. The current density was determined by normalizing the current to the polarized area. The polarized area was assumed to be 17.75 cm2 for UCR and 6 mm2 for ECR specimens in the ST testing. These assumptions were based on the entire lateral surface area of UCR and the area of the coating holidays of ECR respectively.  Other researchers assumed the same surface area for UCR and ECR specimens of the same size (Sturgeon et al. 2010; Erdogdu 1992; Saricimen et al. 1997). However, the epoxy coating was not electrically conductive and polarization of the entire ECR surface was unlikely. Consequently, the current density and the corrosion rate would have been greatly underestimated. Nonetheless, during the LT testing, water uptake of the coating, water induced adhesion loss and cathodic disbondment (Nguyen and Martin 1996, 2004; Nguyen, Hubbard, and Pommersheim 1996; Alberto A   3 Experimental Procedure   57 and Powers 1996) occurred. Thus, the polarized area of the ECR specimens of the LT test series were increased. Consequently, the polarized area for ECR specimens during LT measurements was estimated experimentally as described in chapter 4.1.1..  3.1.2.1.3 Linear Polarization Resistance The linear polarization resistance (LPR) method was only applied to the long-term (LT) specimens. The LPR measurements were not the focus of this study and were used as supporting information. During the LPR test the potential dropped to -20 mV below the OCP and was raised to +20 mV above the OCP at a scan rate of 0.6 V/h (=0.1667 mV/s) (ASTM International 2014b) and the applied current was recorded. Subsequently, the LPR was determined by Eq. (56) from the change in potential ΔE and the change in applied current ΔI normalized to the polarized area A. Implementing the result from equation Eq.(56) into the Stern Geary equation led to the corrosion current density as shown in Eq. (57). The constant B was estimated as 26 mV and 52 mV for active and passive behaviour respectively (Moreno et al. 2004; Andrade et al. 1986).  𝐿𝑃𝑅 = 𝑖𝐶𝑜𝑟𝑟 ∙∆E∆I/A (56)  𝑖𝐶𝑜𝑟𝑟 = 𝐵𝐿𝑃𝑅 (57) The obtained corrosion current density iCorr is in the unit of µA/cm2. In order to obtain the corrosion rates in the unit of µm/year, iCorr is multiplied by the conversion factor of 11.6 (Alonso et al. 1998; Jones 1996). This conversion factor is based on the simplifying assumption of corrosion of iron (Jones 1996).  3.1.2.1.4  Solutions The corrosion behaviour was investigated for carbonated as well as un-carbonated conditions. Furthermore, the effect of chloride ions on the corrosion behaviour of the reinforcement was analyzed. For the best case scenario of no carbonation and no chloride contamination, the corrosion behaviour in three commonly used simulated concrete pore solutions (SCPS) was compared (Moreno et al. 2004; Nakayama and Obuchi 2003; Poursaee 2010; Shi et al. 2011; Tan, Wijesinghe, and Blackwood 2014). The compositions of the solutions are shown in Table 3.2 under the SCPS category. Various authors proposed that a high concentration of bicarbonate ions enhances the stability of a passive film on the metal surface and thus protects the steel from corroding (Moreno et al. 2004; Tan, Wijesinghe, and Blackwood 2014). A low concentration of bicarbonate ions however, is believed   3 Experimental Procedure   58 to accelerate the corrosion of steel due to the pH reduction and the resulting breakdown of the passive layer (Moreno et al. 2004). The corrosion behaviour of reinforcing bars in solutions of varying carbonate and bicarbonate concentrations was investigated. The solution compositions are shown under the Carbonation category in Table 3.2. The compositions of the carbonate-bicarbonate solutions were adapted from Moreno et al. (Moreno et al. 2004). Additionally, solutions of only carbonate and only bicarbonate were used to separate the effect of the two substances.    The effect of chloride ions on the degradation process under carbonated as well as un-carbonated conditions was tested. The chloride concentration of the SCPSs was increased by addition of sodium chloride. Five concentrations of 0M, 0.001M, 0.01M, 0.1M and 1M were used. Sodium chloride was added to the sodium hydroxide and a weakly concentrated carbonate-bicarbonate base solution as shown in Table 3.2.  Calcium has a low solubility compared to sodium (Lenntech BV 2017). As a result, calcium -hydroxide, -carbonate and -bicarbonate, which can be found in concrete, were replaced by their sodium counterparts in the Carbonation and Chloride category (Moreno et al. 2004). Table 3.2: Composition of test solutions Category Base solution pH NaCl concentration SCPS 0.9M NaOH 13.5 0 Sat. Ca(OH)2 12.5 0 0.03M Ca(OH)2 + 0.3M KOH + 0.1M NaOH 13 0 Carbonation 0.03M NaHCO3 + 0.01M Na2CO3 0.01M NaHCO3 + 0.0033M Na2CO3 0.01M NaHCO3 0.01M Na2CO3 9 9 8 11 0 0 0 0 Chloride 0.01M NaHCO3 + 0.0033M Na2CO3 0.9M NaOH 9 13.5 0M, 0.001M, 0.01M, 0.1M, 1M 0M, 0.001M, 0.01M, 0.1M, 1M  In order to avoid carbonation of the solutions during the LT tests, the test solutions were replaced by a fresh solution on a weekly basis and the pH was monitored daily. The electrochemical tests were performed in a fresh solution (Nguyen and Martin 2004).    3 Experimental Procedure   59 3.2 Accelerated Corrosion 3.2.1 Concrete All reinforced concrete specimens, except for the large outdoor specimens (chapter 3.3.1.2), were cast with the mix design shown in Table 3.3. A high water to cement ratio of 0.6 was chosen to ensure workability in the case of admixed NaCl. Furthermore, a high w/c increases the permeability and thus corrosion. The maximum aggregate size was 10 mm. Table 3.3: Concrete mix design  3.2.2 Uncoated Rebar Rebar was embedded in NaCl admixed concrete with a NaCl to cement ratio of 3 weight percent. The concrete mix design is described in chapter 3.2.1 and the specimen dimensions can be found in chapter 3.3.2.1.1.2 and 3.3.2.1.2.1. The specimens were cured for one week before the accelerated corrosion process was started. An induced current accelerated corrosion technique was applied to the samples. A stainless steel mesh was used as the cathode and placed in a Rubbermaid® container. Three specimens were then places face down on top of the stainless steel mesh. The bottom of the Rubbermaid® container was filled with a 3 % NaCl solution until the solution level was just below the rebar. The mesh was connected to the negative lead of a power supply with the three specimens connected in parallel to the positive lead.  The maximum naturally occurring corrosion current density is reported to be 100 µA/cm2 (Austin, Lyons, and Ing 2004; Alonso et al. 1998; Caré and Raharinaivo 2007). However, it has been shown that application of a current density of 100 µA/cm2 and 200 µA/cm2 lead to almost identical strain versus mass loss curves (El Maaddawy and Soudki 2003). Consequently, the applied current density in this study was 200 µA/cm2 based on the surface area of the rebars. Based on the literature (Austin, Lyons, and Ing 2004; El Maaddawy and Soudki 2003) it was estimated that a current density of 200 µA/cm2 in combination with a concrete cover of 5.5 cm would lead to a 5% mass loss after 22 days, a 10% mass loss after 45 days and a 20% mass loss after 90 days   3 Experimental Procedure   60 of UCR in chloride contaminated concrete. Consequently, the duration of the current application was chosen as 22, 45 and 90 days. The designation of the specimen was chosen according to the total duration of the applied current. This accelerated corrosion method was applied to the UCR specimens of the Hall effect sensing and the thermal detection test series.  After the Hall effect and thermal experiments were completed, the rebars were extracted from the concrete by splitting the concrete specimens in half. The extracted rebars were then pickled (Rengaswamy, Vedhalakshmi, and Balakrishnan 1995) and mechanically cleaned with a wire brush to remove corrosion products and small concrete pieces. Subsequently, the rebars were submerged in the rust remover POR-15® Metal Prep and placed in an ultrasonic bath for 30 minutes. Afterwards, the rebars were sandblasted at a low pressure of 85 psi. The sandblasted specimens were then weighed. All rust removal techniques were applied to the corroded as well as the reference specimens to allow for accurate mass loss calculations (ASTM International 2016).  3.2.3 Epoxy Coated Rebar 3.2.3.1 In Concrete The concrete mix design and specimen dimensions can be found in 3.2.1 and 3.3.1.1.1 respectively. In order to accelerate corrosion of ECR 5% sodium chloride was added to the concrete mix design. In addition to the admixed NaCl, corrosion was enhanced through galvanostatic polarization. Each specimen was submerged in a container filled with 3% sodium chloride solution. A graphite plate that worked as a cathode was placed in the solution on top of the specimen. A DC power source was connected to the graphite plate and the two rebars. The current flowed from the power source to the graphite plate and back to the power source through the rebars. A constant current was applied to the system with a current density of 200 µA/cm2 based on the area of the coating damage. The current was applied for a total duration of 0, 22 and 45 days. The specimens, that had not been polarized served as the reference.  For each polarization duration 3 replica specimens were used, resulting in a total of 9 concrete specimens.  In order to determine how much the specimens had corroded, the rebars were extracted from the concrete and visually inspected for corrosion products. Subsequently, the ECR were soaked in acetone to allow the epoxy coating to soften. The coating was then peeled off by hand and light wire brushing. Once the coating was fully removed, the rebars were again visually inspected for corrosion products and the locations of corrosion sites were documented.    3 Experimental Procedure   61 3.2.3.2 In Neutral NaCl Solution The specimens in this test series consisted of 20 cm long pieces of 10M ECR. One small damage of the epoxy coating of each specimen, measuring 3 by 2 mm, was created by cutting into the coating and subsequently peeling off the coating. This coating holiday was located 6.5 cm away from the rebar end. A copper wire was connected to the end further away from the damage (i.e. 13.5 cm away from damage). The wire was connected by inserting it into a small hole in the rebar. The hole was then filled by conductive silver epoxy. Both ends including the electrical connection were then sealed with 3M™ Scotchkote™ 413/215, followed by a layer of silicon tape. After each sealant had been cured for at least 24 hours, the specimens were submerged in a 3% NaCl solution at room temperature. The copper wires as well as two graphite counter electrodes were connected to a TENMA® 72-6694 current calibrator and the current was adjusted to a current density of 200 µA/cm2 with respect to the coating damage. 3.2.3.3 In Alkaline NaCl Solution Three test series, which each consisted of a delamination (stage 1) and a corrosion phase (stage 2), were performed. The three series differed in the techniques used to accelerate the delamination as well as the corrosion and are referred to as series A, B and C. All three procedures are summarized in Table 3.4. It should e noted, that all solutions were replaced with a fresh solution once a week to avoid carbonation and changes in concentration due to evaporation. Furthermore, the specimens were later used for Hall effect (HE) sensing and infrared thermography (IRT) corrosion detection experiments. Therefore, the bar preparations and dimensions can be found in chapter 3.3.2.1.1.3 for HE specimens and in chapter 3.3.2.1.2.2 for IRT specimens. However, one set up specimens is shown in Figure 3.4.   3 Experimental Procedure   62  Figure 3.4: Specimens of series C The delamination stage of series A and C involved galvanostatic cathodic polarization, whereas series B was subject to thermal cycling during this first phase. The delamination phase of all three series lasted 21 days.  The anode in the galvanostatic setup of series A and C consisted of two graphite blocks, that were placed along opposite sides in a thermally stable plastic box. The two blocks were connected in parallel to the positive lead of a current calibrator. The ECRs were placed inside the box with one end elevated on top of one graphite rod. The elevated end was connected in parallel to the negative lead of the current calibrator. The box was then filled with highly alkaline 0.9M NaOH solution, that contained 3% NaCl, to a level that left the elevated rebar end dry. The latter simply served as a protection measure for the electrical connection of the rebars to the current calibrator. The entire box was kept in an oven set to 50°C and the current calibrator supplied a current density of 162 µA/cm2 relative to the entire rebar surface area. The latter current density is equivalent to a current density of 40.68 mA/cm2 relative to the entire exposed area of the ECR (i.e. the area of the coating holidays).  The specimens of series B were placed in a thermally stable plastic box. The box was filled with highly alkaline 0.9M NaOH solution, that contained 3% NaCl, to a level, that covered the rebar. Thermal cycling was performed by alternating between placing the box of series B in a fridge at 6°C and an oven set to 50°C. Each cycle lasts 24 hours.    3 Experimental Procedure   63 Table 3.4: Accelerated corrosion procedure of series A, B and C of specimens later using for Hall Effect (HE) sensing and Infrared thermography (IRT) corrosion detection experiments  A B CName GSCP Thermal cycles GSCPDuration 21 d 21 d = 21 · 1 d 21 dTemperature 50 °C 50 °C & 6 °C 50 °CApplied current density -162 µA/cm2 - -162 µA/cm2Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaClName GSAP GSAP AeratedDuration 45 d 45 d 45 dTemperature 50 °C 50 °C 50 °CApplied current density 8 µA/cm2 8 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaClName GSAP GSAP AeratedDuration 17 d 17 d 45 dTemperature 50 °C 50 °C 50 °CApplied current density 8 µA/cm2 8 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaClName GSAP GSAP AeratedDuration 40 d 40 d 40 dTemperature 50 °C 50 °C 50 °CApplied current density 12 µA/cm2 12 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl2 HE and 2 IRT specimens removed for testing2 HE and 2 IRT specimens removed for testingVisual inspection did not show signs of corrosionName GSAP GSAP AeratedDuration 21 d  21 d 21 dTemperature 50 °C 50 °C 50 °CApplied current density 12 µA/cm2 12 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaClName GSAP GSAP AeratedDuration 13 d 13 d 13 dTemperature 50 °C 50 °C 50 °CApplied current density 79 µA/cm2 79 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaClName GSAP GSAP AeratedDuration 10 d 10 d 10 dTemperature 50 °C 50 °C 50 °CApplied current density 108 µA/cm2 108 µA/cm2 -Solution 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl 0.9M NaOH + 3% NaCl3 HE and 2 IRT specimens removed for testing3 HE and 3 IRT specimens removed for testingVisual inspection did not show signs of corrosion2 Corrosion2 Corrosion2 Corrosion1 DelaminationStageSeriesPurpose2 Corrosion2 Corrosion2 CorrosionOne specimen of each series was removed for assess the degree of corrosion  3 Experimental Procedure   64 The accelerated corrosion phase consisted of galvanostatic anodic polarization for series A and B and aeration for series C. All specimens were kept in the same oven with the temperature set to 50°C throughout the entire accelerated corrosion phase.  The same setup as the galvanostatic cathodic polarization setup described above was used for the anodic polarization of specimens A and B. However, the ECRs were connected to the positive lead and the graphite was connected to the negative lead of the current calibrator.  The applied current density was initially set to 0.8 µA/cm2 relative to the entire surface area, which is equivalent to 200 µA/cm2 relative to the exposed surface area of the coating damage.  After 45 days of the accelerated phase, one ECR of each series was taken out of the solution. The coating was removed by soaking the rebars in acetone and subsequently peeling off the epoxy coating. The stripped specimens were visually inspected for corrosion sites.  The remaining rebars stayed in the accelerated corrosion phase for another 17 days before the current density of specimens A and B was increased to 12 µA/cm2 with respect to the entire surface area, which equals 3 mA/cm2 with respect to the exposed surface area. Furthermore, the graphite cathodes were replaced by copper mesh cathodes at the same time the current density was increased. The copper mesh ensured a more even distribution of the current. The accelerated corrosion conditions of series C remained the same. After 40 days in these conditions, 2 rebars of each series designated for the Hall effect sensing and 2 rebars of each series designated for the thermal testing were removed from the accelerated corrosion chamber and are referred to as 40A and 40B in the remainder of this document.  After an additional 21 days, the current density applied to specimens A and B was increased to 79 µA/cm2 with respect to the entire surface area, which is equivalent to 19.84 mA/cm2 with respect to the exposed surface area. After 13 days, the current density was increased to 108 µA/cm2 with respect to the entire surface area, which equates to 27.12 mA/cm2 with respect to the exposed surface area. Specimens A and B were exposed to the aforementioned final current density for 10 days before the accelerated corrosion phase terminated and are referred to as 84A and 84B in the remainder of this document. The conditions of the accelerated corrosion phase of series C did not change and the total length of the accelerated corrosion phase of all three series was the same.  It should be noted, that all solutions were replaced with a fresh solution once a week to avoid carbonation of the solution.    3 Experimental Procedure   65 The specimens, that had successfully been corroded were embedded in concrete and used for Hall effect sensing and thermal detection experiments. After completion of the NDT experiments, the rebars were extracted by splitting the concrete specimens in half. The extracted bars were then soaked in acetone to allow the epoxy coating to soften. Subsequently, the coating was mechanically removed by peeling and light wire brushing. The stripped bars were then pickled (Rengaswamy, Vedhalakshmi, and Balakrishnan 1995) followed by light wire brushing to remove corrosion products and concrete residue. Afterwards, the bars were submerged in POR-15® Metal Prep rust removing liquid and placed in an ultrasonic bath for 30 minutes. Subsequently, the bars were sandblasted at a low pressure of 85 psi. The weight of the sandblasted specimens was then determined gravimetrically.  All rust removal techniques were applied to the corroded as well as the reference specimens to allow for accurate mass loss calculations (ASTM International 2016).  3.3 Corrosion Detection The mix design of all small lab specimens, Hall effect and thermal specimens was the same and is shown in chapter 3.2.1. 3.3.1 Conventional Non-Destructive Testing 3.3.1.1  Small Lab Specimens 3.3.1.1.1  Specimens The concrete specimens were 10 cm tall with a width of 40 and a length of 38 cm as shown in Figure 3.4. Each specimen included two 10M epoxy coated rebars, that ran parallel to each other with a distance of 10 cm. The rebars were 39 cm long and were sticking out of the concrete by 1 cm on one side. Figure 3.5 schematically shows a concrete specimen. A total of 9 concrete specimens were cast.    3 Experimental Procedure   66  Figure 3.5: Dimensions of small lab specimens in mm Before the ECR was embedded in concrete, the epoxy coating was damaged with a file in 12 locations. The damage locations along two bars are shown in Figure 3.6. Half of the holidays were located on the top rebar surface and the other half on the bottom surface. The locations of the holidays are shown by a blue arrow or a red circle in Figure 3.6 depending on the surface. Taking all 12 intentional coating defects into account, the holidays were approximately evenly spaced along the rebar.       3 Experimental Procedure   67  Figure 3.6: Coating damage locations of small lab specimens A small hole was drilled into the end of the rebar, that was sticking out of the concrete and a wire was attached with silver conductive epoxy. All exposed rebar surfaces as well as the rebar-wire joint were coated with 3M™ Scotchkote™ 413/215 to avoid edge effects. After the concrete had been cured for 7 days, the corrosion of the embedded ECRs was accelerated according to method described in chapter 3.2.3.1.  3.3.1.1.2 Testing Procedure Testing started 24 hours after termination of the accelerated corrosion procedure and took four days in total. Table 3.5 shows the testing schedule including a 48 hour drying period.  Table 3.5: Testing procedure Day Activity 1 Linear Polarization Resistance, Half Cell, Wenner probe  2 Drying 3 Drying 4 Ground Penetrating Radar  3.3.1.2 Large Outdoor Specimens 3.3.1.2.1 Specimens Three slabs of different chloride content were cast on May 16 2013. The concrete was supplied by Lafarge. A general use 35 MPa concrete mix design with entrained air and a maximum slump of 75 mm was used. The amount of chloride admixed to the concrete of each slab is shown in Table 3.6.   3 Experimental Procedure   68 Table 3.6: Amount of chloride per cubic meter of concrete for each slab Slab Admixed NaCl Admixed Cl- [-] [kg/m3] [kg/m3] #1 0 0 #2 8 4.8 #3 11 6.6  All slabs were reinforced with epoxy coated 10M and 20M rebars. The epoxy coating of the reinforcing bars in slab #2 and #3 was damaged with a file. Pre-existing damage of the coating was not repaired. The locations of all coating holidays were recorded and the file holidays are shown as a large X and the accidental holidays as a small x in Figure 3.7. The 20M bars are shown in blue and 10M bars are shown in black and were located over top of the 20M bars. The concrete cover was 65 cm for all three slabs. The slabs were covered with a plastic sheet and kept wet for five weeks. Subsequently, the slabs were sprayed with water every other day before the first day of testing August 16 2013. After the initial round of testing had been completed, water ponding cycles started on September 6 2013. Each cycle consisted of 7 days of water ponding, followed by 7 days without water ponding. The slabs were covered by a roof and thus did not experience any natural precipitation.    3 Experimental Procedure   69  Figure 3.7: Large outdoor specimens including damage locations   3 Experimental Procedure   70 3.3.1.2.2 Testing Procedure Testing of the three slabs took two weeks in total due to the size of the specimens. Each test series started after a 7 day period of water ponding. In order to take advantage of the high moisture content due to the water ponding, the electrochemical NDT methods were performed first. The electrochemical techniques consisted of the Linear Polarization Resistance (LPR) method as a measure for the corrosion rate and the Half-cell method. After completion of the electrochemical testing, the slabs were allowed to dry for 7 days before the Ground Penetrating Radar (GPR) was used as a measure for the amount of corrosion. Testing of the large outdoor specimens was performed in August 2013, November 2013, December 2013, January 2014, January 2015, July 2015 and October 2015.  3.3.1.3 NDT Methods The Wenner probe with the Proceq Canin+ data acquisition unit was used for the resistiity measurements. Resistivity measurements were obtained by placing the Wenner probe on the concrete surface, directly over top of the rebar. The 4 electrode array was oriented parallel to the embedded rebar.  The Half Cell equipment consisted of a copper-saturated copper sulfate (CSE) half cell electrode and the Proceq Canin+ data acquisition unit. The open circuit potentials were obtained directly over top of the rebar and were analyzed in the voltage against the CSE. The linear polarization resistance equipment consisted of the Gamry Reference 600 potentiostat, a copper-saturated copper sulfate (CSE) half cell electrode and a stainless steel ring as the auxiliary electrode. A scan rate of 0.6 V/h (0.1667 mV/s) (ASTM International 2014b) was employed to polarize the rebar. Measurements were taken directly over top of the rebar.  The GSSI SIR® 3000 unit with the 1.6 GHz antenna was employed for the Ground Penetrating radar measurements. The movement of the antenna was perpendicular to the rebar orientation and the software RADAN 6.6 was used for the data analysis.    3 Experimental Procedure   71 3.3.2 Hall-Effect and Thermography Specimens 3.3.2.1 Samples 3.3.2.1.1 Hall-Effect Sensing 3.3.2.1.1.1 Steel Bars Microalloyed steel bars (SB) with diameters of 10.2 mm, 9.6 mm, 8.7 mm and 7.5 mm were initially tested to establish the feasibility of Hall effect sensing as a corrosion detection method. The bars were fabricated by reducing the diameter of regular 10M rebar. The resulting diameter and mass loss relative to the original 10M rebar simulated the effects of corrosion and are shown in Table 3.7. The SB were embedded in concrete with the mix design shown in chapter 3.2.1. The dimensions of the SB reinforced concrete specimens are shown in Figure 3.8 in mm. Irregularities of the concrete moulds resulted in concrete covers between 2.7 cm and 3.0 cm as shown in Table 3.7.  Table 3.7: Specifications of the steel bars (SB) embedded in concrete Designation Diameter Mass Loss Concrete cover [-] [mm] [%] [cm] 10 10.2 18.23 3.00 9 9.6 28.00 2.85 8 8.7 41.37 3.00 7 7.5 55.46 2.70    3 Experimental Procedure   72  Figure 3.8: Steel bar (SB) reinforced concrete specimens for Hall effect (HE) sensing with concrete cover cc and dimensions in mm   3.3.2.1.1.2 Uncoated Rebar (UCR) Regular 10M rebar was used for all testing involving UCR. The length of the rebar used for testing without concrete was 20 cm. The mass loss of the corroded bar, that was tested without concrete was 8.94%. The rebar in the reinforced concrete samples was 25 cm long with a concrete cover of 3 cm. The dimensions of the concrete samples were 22 cm by 12.5 cm by 12.5 cm and are shown in Figure 3.9 in mm. These specimens had undergone accelerated corrosion in accordance with chapter 3.2.2 after the concrete had been cured for one week. Furthermore, unreinforced concrete specimens of the same dimensions and mix design, with and without admixed NaCl, were also cast.   3 Experimental Procedure   73  Figure 3.9: UCR reinforced concrete specimens for Hall effect (HE) sensing with a concrete cover of 30 mm and dimensions in mm  3.3.2.1.1.3 Epoxy Coated Rebar (ECR) HE sensing samples were made with 20 cm long, 10M ECR. Four holes were drilled into the coating, which were just deep enough to expose circular steel surface areas with a 3 mm diameter each (ASTM International 2014a). The holes were located between rebar ribs and had a spacing of five ribs between them as shown in Figure 3.10.  Figure 3.10: ECR with four coating holiday locations The bars had undergone accelerated corrosion according to chapter 3.2.3.3 before testing. After this initial round of testing, the corroded as well as the reference ECR were embedded in concrete with a concrete cover of 2.8 cm. The dimensions of the concrete specimens were 18 cm by 12 cm by 10 cm and are shown in Figure 3.11 in mm. Reinforced as well as unreinforced specimens with the aforementioned dimension were cast. NaCl was admixed to half of the unreinforced specimens with a NaCl to cement ratio of 3 % by mass.    3 Experimental Procedure   74  Figure 3.11: ECR reinforced concrete specimens for Hall effect (HE) sensing with a concrete cover of 28 mm and dimensions in mm  3.3.2.1.2 Thermal Detection 3.3.2.1.2.1 Uncoated Rebar (UCR) The reinforced concrete specimens were 14 cm long and had a cross sectional area of 12 cm by 9.5 cm. Each specimen contained a 15 cm long 10M rebar with a concrete cover of 2.8 cm as shown in Figure 3.12.    3 Experimental Procedure   75  Figure 3.12: UCR reinforced concrete specimens for infrared thermography (IRT) corrosion detection with a concrete cover of 28 mm and dimensions in mm A hole with a 3.1 mm diameter had been drilled through the entire length of the rebar. After the UCR had been embedded in concrete and cured for one week, the hole was cleaned and sealed with silicon. The accelerated corrosion process was then started according to chapter 3.2.2. Once the accelerated corrosion process had been completed, the samples were submerged in water for 48 hours to saturate the concrete. Subsequently, the silicon was removed from the rebar ends and the hole was cleaned again. This was followed by drying the samples in an oven set to 50 °C until a daily moisture loss of no more than 0.1% had been reached.    Once the specimens had been dried, three twisted 25 AWG Nickel Chromium wires were inserted into the longitudinal hole in the rebar. The resistance heating wires were composed of 80% Nickel and 20% Chromium and each wire had a diameter of 0.45 mm and a resistivity of 6.9 Ω/m. The hole was then filled with the high temperature chemical set cement Omegabond 700 (Omega Engineering Inc. 2017). The latter cement served as an electrical insulator with a high thermal conductivity and resistance up to 827°C.  3.3.2.1.2.2 Epoxy Coated Rebar (ECR) For the thermal detection specimens 20 cm long pieces of 10M ECR were used. Four holes, deep enough to expose four circular steel surface areas with a 3 mm diameter each, had been drilled into   3 Experimental Procedure   76 the epoxy coating (ASTM International 2014a). The holes had a 5 rib spacing and were positioned between the rebar ribs (damage size and location shown in Figure 3.10).  A longitudinal hole was drilled into both rebar ends as shown in Figure 3.13. The two holes met in the centre of the rebar, resulting in a longitudinal hole throughout the entire length of the rebar. A drill bit with a diameter of 2.03 mm had been used for both holes.   Figure 3.13: ECR with longitudinal hole with a diameter of 2.03 mm Both ends of the bars were sealed with silicon to allow for easy removal of the sealant. Once the silicon had been cured for 24 hours, a layer of 3M electrical tape was applied to avoid etching of the underlying silicon in the alkaline solution (Zubel et al. 2001; Xia et al. 2001). The latter tape was covered with 3M™ Scotchkote™ 413/215 to avoid dissolution of the tape’s glue. The final layer consisted of two finger cots that were sealed around the rebar ends by a zip tie and an elastic to avoid cathodic disbondment of the 3M™ Scotchkote™ 413/215. The sealed rebars then underwent the accelerated corrosion procedure described in chapter 3.2.3.3. Figure 3.14 shows on sample after all sealants had been applied.  Figure 3.14: ECR thermal detection specimen Once corroded, the sealants were removed from the bar ends and the ECRs were embedded in concrete with a cover of 2.8 cm. The dimensions of the reinforced concrete specimens were 18 cm by 12 cm by 10 cm, leading to 1 cm of rebar sticking out on both ends as shown in Figure 3.15.    3 Experimental Procedure   77  Figure 3.15: ECR reinforced concrete specimens for infrared thermography (IRT) corrosion detection with a concrete cover of 28 mm and dimensions in mm The concrete was cured for seven days, after which the holes were once again sealed with electrical tape. The specimens were submerged in water for 48 hours to allow for saturation of the concrete. Subsequently, the holes were opened and cleaned before the specimens were placed in an oven set to 50 °C. Once the daily moisture loss did not exceed 0.07% anymore, the specimens were removed from the oven.  Three twisted 25 AWG Nickel Chromium wires were inserted into the longitudinal hole of each rebar. The holes were then filled with electrically insulating high temperature conductive cement Omegabond 700 (Omega Engineering Inc. 2017).  3.3.2.2 Testing Procedure 3.3.2.2.1 Hall-Effect Sensing The Hall effect (HE) voltage was measured by means of a linear Hall Effect Sensor (Melexis MLX90242). A standard 9 V alkaline battery was connected to the HE sensor as a power supply.  The varying magnetic field was supplied by a rare earth magnet, that was attached to an adjustable speed motor by an aluminum rod. The speed of the motor determined the rotational speed   3 Experimental Procedure   78 of the magnet and its rotational frequency. All specimens were tested at 12 Hz, 19 Hz, 24 Hz and 31Hz. An additional frequency of 28 Hz was used for ECR and UCR specimens embedded in concrete under dry conditions.  The Hall effect sensor was connected to a signal conditioner, which applied a gain of 100 and a 50 Hz filter to the signal. The signal conditioner was connected to a data acquisition system, that recorded the voltage over time.   Figure 3.16: Setup of sensor (S) and magnet (M) at sensor-magnet spacing (SM) in relation to the rebar under investigation at a distance (R) Figure 3.16 schematically shows the positioning of the HE sensor S and the magnet M relative to the rebar being tested. The magnet rotated around its own axis and stayed in the same location relative to the rebar during the entire experiment. The sensor on the other hand was moved away from the magnet and parallel to the direction of the rebar during the experiment. Thus, the distance SM in Figure 3.16 increased during an experiment with an increment of 2.5 cm. Data was collected at each sensor magnet spacing (SM).  The distance R from the magnet and the sensor to the rebar did not change during an experiment. A 2.5 cm wide PVC block was placed between the sensor-magnet assembly and the rebar to ensure a constant distance R throughout an experiment. Once the specimens had been embedded in concrete, the concrete cover determined the distance R. The two test setups are shown in Figure 3.17. Reinforced concrete samples were tested under saturated and dry conditions.     3 Experimental Procedure   79  Figure 3.17: Hall effect sensing setup with concrete (left) and PVC (right) as a space holder The data acquisition unit recorded the peak to peak voltage Vp-p of the Hall effect (HE) sensor every 0.5 seconds. The median voltage of a 10 second time frame was used for the data analysis. Furthermore, the median Vp-p of an unreinforced concrete sample under the same measurement conditions was subtracted from the voltage of the specimen under investigation. Consequently, the change in HE voltage cause by the rebar was obtained.   3.3.2.2.2 Thermal Detection The twisted Nickel Chromium wire was connected to the switching mode power supply PSC-6616 and a constant current of 25.4 A was applied. Due to the low resistivity of the Nickel Chromium wire, the twisted wire heated up and acted as a heating element when the current of 25.4 A was applied. The power supply was manually turned off once the rebar ends had reached a temperature of approximately 93.5 °C.  The thermal response was recorded with the FLIR E50x infrared camera. The camera was positioned on a tripod and the lens was 60 cm away from the sample surface. The IR camera was connected to a laptop running the IR software FLIR Tools+. The software recorded an IR video of each sample for one hour. The measurement setup is shown in Figure 3.18.   3 Experimental Procedure   80  Figure 3.18: Infrared thermography setup   4 Results   81 4 Results 4.1 Corrosion Behaviour The corrosion behaviour of epoxy coated rebar (ECR) and uncoated rebar (UCR) samples (chapter 3.1.1) in simulated concrete pore solutions was studied in a series of electrochemical tests including open circuit potential (OCP), potentiodynamic polarization (PDP) and linear polarization (LPR) measurements. Short term (ST) measurements that did not involve any exposure to the test solution prior to testing were performed. Furthermore, long term (LT) experiments were conducted, which consisted of a 4 week exposure of the specimen to an aerated test solution at elevated temperatures prior to testing (chapter 3.1.2).  4.1.1 Polarized Area The original exposed area of the ECR and UCR specimens was 0.06 cm2 and 17.75 cm2 respectively. The polarized area of the short term (ST) specimens was assumed to be equal to the original exposed area due to the electrically insulating nature of the epoxy coating.  However, the long term (LT) specimens were exposed to the test solutions for four weeks prior to testing. During this four-week exposure water uptake of the coating, water induced adhesion loss and cathodic disbondment are likely to have occurred on the ECR specimens (Nguyen and Martin 1996, 2004; Nguyen, Hubbard, and Pommersheim 1996; Alberto A and Powers 1996). Consequently, the polarized area increased for the ECR samples between the ST and the LT testing. However, the polarized area could not be easily measured. Thus, the polarized area of the LT ECR specimens was estimated based on the assumption that the corrosion rates of UCR and ECR were the same in relatively non-corrosive solutions with an alkaline pH and no sodium chloride. Due to the lack of chloride ions and the alkaline pH, corrosion did not initiate on either sample. Thus, the observed corrosion rates were believed to have been the same for ECR and UCR. Consequently, the polarized area of ECR during the LT testing was determined by setting the measured corrosion rates of ECR and UCR equal and solving for the polarized area of ECR. The resulting polarized area of ECR during the LT testing is shown in Table 4.1 for each chloride free alkaline base solution. The factor by which the original exposed area of ECR (0.06cm2) had to be multiplied to result in the polarized area of the LT measurements is shown as the coefficient in Table 4.1.  4 Results   82 Table 4.1: Polarized area of base solutions during LT testing  Table 4.1 shows the largest increase in polarized area for the 0.9M NaOH solution and the smallest for the saturated calcium hydroxide solution. While the polarized area was not proportional to the sodium ion concentration, the two were related. Hence, the polarized area of the saturated calcium hydroxide solution did not increase between the ST and LT testing due to the lack of sodium ions. The sodium hydroxide solution on the other hand had by far the largest amount of sodium ions and consequently polarized area. Furthermore, the two carbonate-bicarbonate solutions resulted in the same pH but the polarized area increased with an increase in sodium ions. Studies by Nguyen and Martin showed the promoting effect of sodium ions on the cathodic delamination (Nguyen and Martin 2004). However, the mechanisms of water uptake of the coating and water induced adhesion loss do not involve sodium ions. Thus, the observed increase in the polarized area is mainly contributed to cathodic delamination, which was enhanced by the presence of sodium ions. This led to the largest increase of the polarized area by a factor of 74 for the 0.9M NaOH solution and no change for the sodium free saturated Ca(OH)2 solution.  However, the 0.01M NaHCO3 solution and the 0.03M Ca(OH)2 + 0.3M KOH + 0.1M NaOH solution resulted in the same polarized area despite a Na+ concentration of 0.01M and 0.1M respectively. Thus, factors other than the sodium ion concentration affected the polarized area. The exact causes of the increase in polarized area are unclear and cannot be determined from the experiments performed within the scope of this study. Chemical analysis of the steel-epoxy interface as well as the precise extend of water uptake of the coating, water induced adhesion loss as well as cathodic delamination are required in order to determine the exact causes of the increase in polarized area.     4 Results   83 4.1.2 Common Simulated Concrete Pore Solutions 4.1.2.1 Short Term The potentiodynamic polarization (PDP) curves and the open circuit potentials (OCP) for three common simulated concrete pore solutions are shown in Figure 4.1 and Figure 4.2 respectively. The solid lines represent the specimens without any coating and the dashed lines show the results of the epoxy coated rebar specimens. One can observe that the two types of specimens resulted in very similar PDP curves under identical conditions. The specimens without any coating tended to produce PDP curves with a steeper incline, indicating a more passive behaviour. Furthermore, at a given potential the transpassive current density of ECR was greater than UCR in all three solutions.  Figure 4.1: Anodic PDP curves for 3 common simulated concrete pore solutions  Figure 4.2: Open circuit potential of ECR and UCR in 3 common simulated concrete pore solutions   4 Results   84 The solution consisting of calcium-, potassium- and sodium hydroxide was considered to represent the composition of concrete pore solution the most accurately out of the three SCPSs (Poursaee 2010).  The measured open circuit potential (OCP) was the most positive and therefore least active for the saturated calcium hydroxide solution. Conversely, the most active OCP was observed for the sodium hydroxide solution. All solutions resulted in similar OCPs for the two sample types with a tendency for a more positive potential for the ECR samples.  The saturated calcium hydroxide solution produced the lowest current densities for both ECR and UCR. In the passive region, current densities for the other two solutions were up to 10 times higher. The significantly lower current densities in the passive region suggested the presence of a more stable passive layer on both sample types in the saturated calcium hydroxide solution. The 0.9M sodium hydroxide solution resulted in higher current densities for the ECR specimens and almost identical current densities for the UCR specimens compared to the three component solution with the same sample type. However, the increase in current densities from the three component solution to the NaOH solution for ECR samples was small and not continuously observed over the entire passive region. The incline of the curve in the passive region was the steepest for the sodium hydroxide solution, which is another indicator for increased stability of the passive layer. The aforementioned observation was particularly pronounced for the ECR samples. Nonetheless, the UCR samples led to a significantly steeper incline compared to the ECR samples regardless of the test solution.  The potentiodynamic polarization of the UCR samples in the calcium hydroxide solution led to current fluctuations throughout most of the passive region. The unprocessed data showed a similar effect for the ECR samples under identical conditions too. In the sodium hydroxide and the three component solution, these current fluctuations were only present for ECR specimens. The extent of the fluctuations decreased from the saturated calcium hydroxide solution to the three component solution, where instabilities were only observed close to the OCP. A further decrease in the amount and range of current fluctuations was observed for the sodium hydroxide solution. However, the current fluctuations were particularly pronounced for specimens polarized in solutions containing calcium hydroxide even though the surface preparation was identical for all specimens of all solutions.  Stable pitting corrosion did not occur until the pitting potential had been reached. The pitting potential was indicated by a sudden large increase in the current density. Consequently, the PDP curve  4 Results   85 ran horizontal at the pitting potential. Figure 4.1 shows a pitting potential of approximately 0.7 V regardless of the SCPS and the sample type.  The PDP curves obtained for the saturated calcium hydroxide solution were not very stable. The aforementioned observation was particularly pronounced at potentials close to the OCP.  4.1.2.2 Long Term Figure 4.3 shows the PDP curves and Figure 4.4  shows the OCP obtained for three commonly used SCPSs after four weeks of exposure to an aerated SCPS.   Figure 4.3: Anodic PDP curves for 3 common simulated concrete pore solutions after 4 weeks of exposure to aerated solution  Figure 4.4: Open circuit potential of ECR and UCR in 3 common simulated concrete pore solutions after 4 weeks of exposure to aerated solution Both solutions containing sodium hydroxide led to almost identical OCP measurements. However, the measured potential for the ECR samples tended to be more negative and thus more  4 Results   86 active in comparison to the UCR specimens in the same solution. The saturated calcium hydroxide solution led to the similar OCP for the ECR samples. However, the OCP of the UCR sample in the aforementioned medium is significantly more negative and thus more active compared to the solutions containing sodium hydroxide.  The UCR specimens led to a steeper slope in the passive region for the saturated calcium hydroxide solution in comparison to the three component solution indicating a higher stability of the passive layer. However, the opposite is given by the measured current densities, which were initially larger for the calcium hydroxide solution, indicating a less stable passive layer.  Current fluctuations were only observed for ECR specimens in saturated calcium hydroxide solution and to a much lesser extent in the three component solution. The resulting current fluctuations were particularly prominent close to the OCP.  The critical pitting potential that marks the onset of the transpassive corrosion behaviour decreased from the saturated calcium hydroxide to the sodium hydroxide solution. This trend was observed for both sample types. The pitting potential in the Ca(OH)2 solution was approximately 0.9 V and 0.7 V in the NaOH solution.  4.1.2.3 Discussion Comparison between the ST and LT measurements showed that the OCP was the most stable over time for the sodium hydroxide solution and decreased significantly over time for the saturated calcium hydroxide solution. In the case of the three component solution, the OCP only decreased for the ECR samples with increasing exposure time but did not change significantly for the UCR specimens. Nonetheless, the three component solution led to a similar OCP after the four week exposure to the one obtained with the sodium hydroxide solution for both specimen types.   The current densities tended to increase with increasing exposure time regardless of the specimen type. Furthermore, this observation was made for all three SCPSs used in this study. The increase in current density was more pronounced for the ECR samples. However, it should be noted that the saturated calcium hydroxide solution led to a significant increase in current density over time for the UCR samples. The increased exposure time resulted in more stable PDP curves, which was indicated by the lack of current fluctuations in comparison to the ST testing. Nonetheless, ECR specimens showed a substantial amount of current fluctuations in saturated calcium hydroxide solution even during the long term testing. Pitting is unlikely to have caused the current instabilities due to the lack of chloride  4 Results   87 ions and the high pH of the solutions. Regardless of the exact cause of the current fluctuations, they were reduced by a decreasing Ca(OH)2 and increasing NaOH content.  The critical pitting potential tended to increase with exposure time. However, the increase was rather small and not consistently observed. The calcium hydroxide led to more variation in the test results and to more instability of the PDP curves. Furthermore, the overall shape of the PDP curves obtained with the sodium hydroxide solution was very similar to the one corresponding to the three component solution. Additionally, the exposure length had a very similar effect on both sample types in the sodium hydroxide and the three component solution. Consequently, the 0.9M NaOH solution was chosen as the base for the solution representing non-carbonated conditions. This solution led to very similar results compared to the three component solution with less instabilities due to the lack of calcium hydroxide.  4.1.3  Chloride 4.1.3.1 Short Term The PDP curves and OCP for 0.9M NaOH solutions with varying NaCl levels are shown in Figure 4.5 and Figure 4.6 respectively. With the exception of the highest NaCl concentration of 1M (green) all curves were very similar in shape and magnitude up until a potential of 0.75 V was reached. Above the aforementioned voltage the current densities were significantly higher for the epoxy coated rebar (ECR) samples.   4 Results   88   Figure 4.5: Anodic PDP curves for 0.9M NaOH solution with varying NaCl concentration  Figure 4.6: Open circuit potential of ECR and UCR in 0.9M NaOH solution with varying NaCl concentration The OCP tended to be more positive for the ECR samples in comparison to the UCR samples regardless of the salt level. The OCP did not change significantly with the addition of small amounts of sodium chloride. Nonetheless, both types of specimens showed a much more negative OCP at a 1M NaCl concentration. The reduction in OCP indicates a more aggressive environment that enhanced corrosion activity.  The current densities of the passive region were very similar in magnitude for all samples tested in solutions of less than 1M NaCl content. The highest sodium chloride concentration however greatly increased the current densities at potentials below 0.85 V. The current densities increased by up to two orders of magnitude for the UCR samples and more than three orders of magnitude for the ECR samples. This large upsurge indicated the onset of stable pitting corrosion.   4 Results   89 The slope in the passive region was noticeably steeper for the UCR samples compared to the ECR samples. This indicates a more stable passive layer for the uncoated bars. The 1M NaCl curves showed a steeper slope for the UCR specimens compared to the ECR specimens too. However, both sample types were actively corroding in the aforementioned solution but at different rates. The ECR samples were corroding at a higher rate than the UCR specimens. At a potential of 0.2 V the slope of the PDP curve of the ECR specimen in the solution containing 1M NaCl suddenly increased. The potential increased with an almost steady current density. This indicated diffusion control of the corrosion process. In other words, the rate of the corrosion process was limited by the diffusion of ions to the corrosion site. This could have been caused by the buildup of corrosion products on the sample surface. Consequently, the ions had to diffuse through that layer of corrosion products. Furthermore, crevice corrosion was very likely to be occurring on the ECR samples in the 1M NaCl solution. Diffusion control is very common for this type of corrosion because the ions need to either diffuse through the epoxy coating or migrate via the steel-epoxy interface to the corrosion site under the coating.  Passive behaviour was observed for all specimens, that were tested in a solution of 0M to 0.1M NaCl concentration. However, current fluctuations close to the OCP could be observed for ECR in solutions with 0.001M, 0.01M and 0.1M NaCl. Current instabilities can be caused by the repeated formation and breakdown of the passive layer (Princeton Applied Research 2017), which can occur in solutions with a high hydroxide to chloride ratio. However, current fluctuations could also be observed for UCR in the chloride free solution, in which a breakdown of the passive layer was unlikely due to the high pH and the lack of chloride ions. A high degree of surface roughness of the tested sample can cause current instabilities. In order to better simulate field conditions, the samples in this study were not polished before testing. Thus, ECR surface conditions differed from those of UCR and the roughness of the sample surfaces was relatively high. Furthermore, current fluctuations can be caused by equipment noise. This is usually the case at particularly low currents, such as the ones observed in the passive region.  The pitting potential was approximately 0.75 V for both sample types. Furthermore, pitting potential did not significantly change with increasing NaCl content.   The corrosion behaviour was not very sensitive to salt concentration changes at a concentration of 0.1M NaCl and below. Nonetheless, low NaCl concentrations caused minor increases in the current densities for both sample types.  4 Results   90 Large amounts of NaCl however led to a significantly more active OCP for UCR and ECR. Furthermore, the observed current densities were much higher as well as the corrosion rate and the shape of the PDP curve indicated a much more active corrosion behaviour. 4.1.3.2 Long Term The anodic PDP curves and corrosion rates for 0.9M NaOH solutions with varying NaCl concentrations after four weeks of exposure to the aerated solution are shown in Figure 4.7 and Figure 4.8 respectively. Figure 4.9 shows the OCP measured before and after the PDP.  At low NaCl concentrations of 0.001M and 0.01M a weak trend of more positive OCPs and lower current densities was observed compared to the salt free solutions. Furthermore, the corrosion rates did not increase with the addition of up to 0.01 M NaCl regardless of the rebar type.  The current densities of ECR in chloride free and 0.001 M NaCl solution resulted in higher current densities in comparison to UCR. However, 0.01 M NaCl led to almost identical current densities for the two sample types below the pitting potential Ep and the corrosion rates were very similar for the two samples types. Even though the current densities as well as the corrosion rates were very low in solutions up to 0.01M NaCl, the shape of the curve did not indicate passive behaviour for either sample type.  Regardless of the NaCl concentration, the pitting potential was approximately 0.75 V and 0.7 V for ECR and UCR respectively. Above these potentials, the transpassive region of the polarization curve started and corrosion pits stabilized. The sudden increase of the current density at the pitting potential was approximately one order of magnitude larger for the UCR specimens compared to the ECR samples.   Furthermore, current fluctuations at potentials above the OCP were observed for the UCR samples at sodium chloride concentrations of 0.001 M and 0.01 M. ECR samples only showed a small amount of current fluctuations very close to the OCP at 0.001 M NaCl. These current instabilities were likely caused by the differences in the surface conditions between the UCR and ECR samples and equipment noise. Consequently, the corrosion rates shown in Figure 4.8 fluctuated at salt concentrations up to 0.01 M and showed no clear trend.   4 Results   91   Figure 4.7: Anodic PDP curves for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution  Figure 4.8: Corrosion rates for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution (LT)  4 Results   92  Figure 4.9: OCP pre and post potentiodynamic polarization for 0.9M NaOH solution with varying NaCl concentration after 4 weeks of exposure to aerated solution One can observe a similar OCP for ECR and UCR samples in solutions with the two highest concentrations of NaCl (0.1M and 1M) in Figure 4.7. At the latter NaCl levels the OCP was significantly lower and thus more active in comparison to the sodium chloride free solutions (orange). Furthermore, Figure 4.9 shows that salt levels of 0.1M and 1M led to lower OCP measurements after the PDP compared to the initial OCP. Conversely, NaCl concentrations up to 0.01M led to an increase in OCP from pre- to post-polarization.  The recorded current densities of UCR were significantly higher for NaCl concentrations of 0.1M and 1M compared to the salt free solution, indicating active corrosion behaviour. ECR on the other hand did not show a significant increase of the current densities at 0.1 M NaCl and 1 M NaCl led to a much smaller shift towards higher current densities compared to UCR in Figure 4.7. Nonetheless, the corrosion rates increased for both sample types as shown in Figure 4.8. Moreover, the corrosion rate in the 1 M NaCl solution of ECR of 28 µm/cm2 even surpassed the rate of UCR at 20 µm/cm2.  The observed current densities at 0.1 M and 1 M NaCl were higher for the UCR specimens in comparison to the ECR specimens. However, both sample types resulted in an initially shallow slope of the curves, that quickly became a steep incline. The steep slope should not be mistaken for passive behaviour. Particularly at sodium chloride levels of 1M the presence of a passive layer was highly unlikely. The steep slope was likely the result of diffusion control. Since the specimens had been exposed to an aerated version of the testing solution for 4 weeks prior to testing, corrosion products had already been present at the beginning of the electrochemical testing. Consequently, the diffusion  4 Results   93 rate of ions through the corrosion layer controlled the current density even at potentials close to the OCP.  In the case of ECR specimens, current fluctuations were only observed for NaCl levels of 0.1M. However, UCR samples did not show any current fluctuations at this sodium chloride concentration. These instabilities of the currents were likely the result of inhomogeneity of the diffusion layer, that governed the dissolution process.  Low amounts of sodium chloride reduced the current densities and made the OCP more positive. The aforementioned NaCl levels did not increase the corrosion rate and thus did not have a significant effect on either type of sample. Large amounts of sodium chloride on the other hand increased the current densities and made the OCP more negative. The aforementioned NaCl levels had a corrosion promoting effect on both types of samples, which led to increasing corrosion rates with increasing NaCl concentrations. Higher current densities  and similar corrosion rates were observed for UCR compared to ECR samples at NaCl concentrations of at least 0.1 M.  4.1.3.3 Discussion 4.1.3.3.1 Exposure time Salt levels at and below 0.01M did not promote corrosion and even led to reduced current densities and positive OCP after four weeks of exposure to the aerated solutions. Furthermore, the corrosion rates did not increase at these NaCl levels. Salt levels up to 0.1 M did not have a significant effect on the short term testing.  The range of NaCl concentrations, that did not affect the corrosion behaviour negatively (i.e. promoting), was smaller for the long term testing. Indicators of the corrosion promoting behaviour are a more negative OCP, increased current density and increased corrosion rate as well as the shape of the potentiodynamic polarization curve. The aforementioned indicators were observed for concentrations of 0.1M NaCl only during the long term testing and 1M NaCl during both short and long term testing. Thus, the passive layer formed in the 0.9 M NaOH solution could only withstand a salt level of 0.1 M NaCl temporarily and broke down after prolonged exposure.  4.1.3.3.2 Uncoated Rebar The recorded current densities of the UCR samples decreased for NaCl levels up to 0.01 M and increased for 0.1M from the short to long term test results. Furthermore, the OCP increased from before to after the PDP for salt levels up to 0.01 M. The opposite trend was observed for NaCl  4 Results   94 concentrations of at least 0.1M. This is explained by the instability of the passive layer. At low NaCl levels such as 0M, 0.001M and 0.01M the passive layer was present. The layer grew and stabilized during the LT exposure to the solution, leading to reduced current densities and increased OCP (Abd El Haleem et al. 2010).  Equipment noise at low passive current densities and the high degree of surface roughness resulted in current fluctuations above the OCP in solutions with NaCl levels up to 0.01M. Once the potential had been raised far enough into the passive region, higher but more stable currents could be observed. Furthermore, the increased post-polarization OCP at low NaCl concentrations indicated, that the passive layer was quickly restored after the PDP (W. Li et al. 2014; Han et al. 2009). A high NaCl level of 0.1M created a very aggressive environment, which the passive layer could not withstand for four weeks. Consequently, the current densities increased from the short term to the long term testing. Furthermore, the reduced post-polarization OCP indicated, an increase in corrosion activity due to the anodic polarization. At a salt concentration of 1M, the [Cl-]/[OH-] ion concentration ratio was high at 1.11 and thus the passive layer could not form. Consequently, the current densities were high and very similar for the short and long term testing. However, due to the buildup of corrosion products on the sample surface during the four week exposure, the corrosion process was diffusion controlled in the long term testing resulting in a steeper incline of the PDP curve in comparison to the short term curve.  4.1.3.3.3 Epoxy Coated Rebar The recorded current densities of the ECR samples marginally increased from the short term to the long term testing for NaCl concentrations up to 0.01 M and resulted in a shallower incline.  However, the current densities did not change over time for 0.1 M NaCl, despite a reduction of the OCP by 0.35 V, and decreased for 1 M NaCl. Furthermore, the current densities increased by approximately two orders of magnitude at the pitting potential during the LT testing and by four orders of magnitude during the ST testing. These changes were likely caused by the difference in the normalizing area used to determine the current densities for the two exposure times. The ST current densities were determined by the size of the coating damage, 0.06 cm2, because significant water uptake, water induced adhesion loss and cathodic disbondment were not expected to have occurred yet. The LT current densities were normalized by 4.44 cm2, to account for water uptake, water induced adhesion loss and cathodic disbondment of the epoxy coating. Thus, the current density measurements of the LT testing were the result of corrosion processes or lack thereof occurring at the coating holidays (0.06 cm2) as well as under the coating (4.38 cm2). If the damaged locations were  4 Results   95 heavily corroding, the resulting currents would be high. If simultaneously the area under the coating was passivated, the resulting currents would be very low. The current measurements of the entire sample would then be very low because the passivated area under the coating (4.38 cm2) makes up 99% of the total area used to normalize the currents (4.44 cm2) in the LT tests. Thus, limited information can be obtained of the corrosion behaviour of the samples from the current densities of the LT testing. The ST tests however, only measure the corrosion behaviour at the coating damage and thus provide a more accurate presentation of the corrosion behaviour of ECR.  The NaCl concentration of 0.001 M and 0.1 M led to current fluctuations regardless of the exposure time, whereas lower salt levels of 0.01 M only resulted in current fluctuations during the ST testing. Irregularities in the surface roughness between the specimens was likely the cause for inconsistencies in the observation of current fluctuations. Furthermore, equipment noise contributed to the current instabilities. The polarized area of the LT testing was 74 times larger than the area of the ST measurements. Thus, the absolute currents of the LT measurements were significantly larger than the ST currents. Consequently, the effect of equipment noise on the current instabilities was more pronounced for the ST measurements.   The OCP of samples in solutions up to 0.01M NaCl increased from pre to post polarization, whereas the opposite was observed for 0.1M and 1M NaCl. The reduction of the OCP indicated an increase in corrosion activity (W. Li et al. 2014; Han et al. 2009) and the lack of a passive layer for NaCl levels of 0.1M and 1M. Passivity occurred at salt levels below 0.1 M and diffusion control governed the corrosion process at salt levels above 0.1 M, which is indicated by the steep slope of the 1 M NaCl curve in Figure 4.7. Furthermore, the corrosion rates increased with the addition of 0.1 M and 1 M NaCl. 4.1.3.3.4 Sample Type The passive current densities obtained during the short and long term testing were very similar for the ECR and UCR specimens. However, due to the uncertainties of the LT current densities of ECR mentioned in the previous section, conclusions regarding the differences of the LT current densities of ECR and UCR could not be drawn.  Despite the similarities of the passive current densities of the two sample types during the ST measurements, ECR led to much larger active current densities combined with a shallower slope at a NaCl concentration of 1 M compared to the UCR samples.   4 Results   96 During the short term testing both types of specimens were protected by the passive layer, which formed on the steel surface due to the high pH of the sodium hydroxide solution. As a result, the recorded current densities at NaCl levels below 1 M were very low and the curves’ incline very steep.  However, the passive layer could not form in the solution containing 1 M NaCl due to the oxidizing effect of the chloride ions. Consequently, the current densities increased for both sample types. Nonetheless, the uncoated rebar was still protected by the oxide layer on the steel surface (Mohammed and Hamada 2006; Mammoliti et al. 1996). Consequently, the increase in current density caused by the loss of the passive layer was reduced for the uncoated rebar in comparison to the epoxy coated rebar. ECR samples did not have any protection layer left once the epoxy coating and the passive layer were removed and could corrode much faster. The corrosion rates of both rebar types increased with the addition of at least 0.1 M NaCl during the LT testing and the corresponding PDP curves even showed signs of diffusion controlled corrosion. The corrosion rates were very similar for the two types except for the rate at 1 M NaCl. At the latter salt concentration, ECR led to higher corrosion rate in comparison to UCR.   4.1.4 Carbonation 4.1.4.1 Short Term The PDP curves and OCP shown in Figure 4.10 and Figure 4.11 respectively correspond to specimens polarized in solutions containing varying levels of sodium carbonate and bicarbonate. The green graphs represent the reference solution of 0.9M NaOH.   4 Results   97  Figure 4.10: Anodic PDP curves for solutions with varying Na2CO3, NaHCO3 and NaOH concentration  Figure 4.11: Open circuit potential of ECR and UCR in solutions with varying Na2CO3, NaHCO3 and NaOH concentration One can observe, that the 0.01M Na2CO3 solution (light blue) as well the reference solution (green) resulted in similar PDP curves for ECR samples compared to UCR samples. The current densities obtained in the carbonate solution (light blue) were significantly lower compared to the reference solution. Similarly, to the reference solution, the carbonate solution led to lower current densities for the UCR specimens in comparison to the ECR samples. Additionally, the slope of the continuous 0.1M Na2CO3 curve was noticeably steeper than the dashed curve of the same kind. The lower current densities and steeper slope for the UCR samples indicate a more stable passive layer in comparison to the ECR specimens, which was observed for not only the 0.1M Na2CO3 solution but also the reference solution.   4 Results   98 The OCP tended to be more positive for the 0.01M Na2CO3 solution compared to the 0.9M NaOH solution of the same specimen type. Thus, both types of samples showed a more passive behaviour in the 0.01M Na2CO3 solution compared to the 0.9M NaOH solution. Furthermore, the critical pitting potential in the carbonate solution was raised in comparison to the sodium hydroxide solution.   The solution with a NaHCO3 concentration of 0.01M led to a PDP curve, whose shape indicates a much more active behaviour in comparison to the reference solution. Active behaviour was observed up until a potential of approximately +0.1 V had been reached. Passive behaviour was observed above the aforementioned potential. The transpassive region started at a voltage of approximately +1.2 V. These regions of the PDP curve, indicating the corrosion behaviour, were observed for both types of specimens. However, the ECR samples led to significantly higher current densities in comparison to the UCR specimens. This is indicated by a shift to the right of the PDP curve in Figure 4.10. Furthermore, one can observe, that the anodic PDP curve of the ECR sample in the bicarbonate solution started at a potential more negative compared to the UCR sample under identical conditions. Thus the OCP of the ECR sample was more negative than the OCP of the UCR sample in the 0.01M NaHCO3 solution.  The shape of the PDP curve indicated more active behaviour, the current densities were higher and the OCPs were more negative for the 0.01M NaHCO3 solution compared to the reference solution for both types of specimens. Thus, both samples showed a more active behaviour in the 0.01M NaHCO3 solution compared to the 0.9M NaOH solution.   The low concentrated carbonate-bicarbonate solution with a concentration of 0.01M NaHCO3 and 0.0033M Na2CO3 resulted in curves very similar to the bicarbonate results. The graphs showed a slight shift to the left compared to the bicarbonate curves. This indicates the weak passivating effect of the sodium carbonate. The accelerating effect of the sodium bicarbonate was very strong, as indicated by the large shift to the right of the 0.01M NaHCO3 and 0.0033M Na2CO3 curves in comparison to the 0.9M NaOH curves.   The orange curves, that represent the results of highly concentrated carbonate-bicarbonate solution, varied based on the specimen type.  The shape of the curve representing the UCR sample indicates passive behaviour. The current densities were reduced and the OCP more positive in comparison to the light blue carbonate curve.  4 Results   99 The passivating effect in the 0.03M NaHCO3 and 0.01M Na2CO3 solution originated from the presence of carbonate, whose effect on the corrosion behaviour was stronger in comparison to the corrosion promoting effect of the bicarbonate for UCR samples. The shape of the dashed orange curve, which shows the results of the ECR samples, is typical for active behaviour of an active-passive material. The passivating effect of the sodium carbonate was stronger in comparison to the 0.01M NaHCO3 and 0.0033M Na2CO3 solution, which was indicated by a shift to the left (lower current densities) and more positive OCP. However, the accelerating effect of the sodium carbonate was still very strong as indicated by the active shape, a more negative OCP and a shift to the right in comparison to the 0.01M Na2CO3 solution (light blue-dashed). Thus, the 0.03M NaHCO3 and 0.01M Na2CO3 solution had a passivating effect on the UCR samples and an accelerating effect on the ECR samples.   Carbonate passivated the steel at a concentration of 0.1M. However, bicarbonate promoted the corrosion of the steel at the aforementioned concentration. When combining the two substances in one solution with a ratio of three to one, the resulting effect depended on the specimen type as well as the solution concentration. In the case of a weak solution of 0.01M NaHCO3 and 0.0033M Na2CO3 the accelerating effect of the bicarbonate dominated for both types of samples. In the case of a stronger solution of 0.03M NaHCO3 and 0.01M Na2CO3 the accelerating effect of the bicarbonate dominated only for the ECR sample. Under identical conditions the UCR sample was passivated due to the dominance of the carbonate. 4.1.4.2 Long Term Figure 4.12 and Figure 4.13 respectively show the anodic PDP curves and corrosion rates of specimens, that were submerged in solutions with varying levels of sodium carbonate and sodium bicarbonate for four weeks prior to testing. The green curves represent specimens, that underwent  4 Results   100 the same conditions in the reference solution of 0.9M NaOH. Furthermore, the OCP measured before and after the polarization is shown in Figure 4.14.    Figure 4.12: Anodic PDP curves for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution    Figure 4.13: Corrosion rate for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution   4 Results   101  Figure 4.14: OCP pre and post potentiodynamic polarization for solutions with varying Na2CO3, NaHCO3 and NaOH concentration after 4 weeks of exposure to aerated solution Both types of specimens showed a more passive behaviour in the 0.01M Na2CO3 as well as in the 0.01M NaHCO3 solution in comparison to the reference solution. Both bicarbonate and carbonate decreased the current densities as well as the corrosion rate and led to a more positive OCP in comparison to the 0.9M NaOH. The carbonate led to slightly lower current densities and corrosion rates than the bicarbonate. Similar to the 0.9M NaOH solution, the 0.01M Na2CO3 as well as the 0.01M NaHCO3 solution resulted in higher current densities for the ECR specimens compared to the UCR specimens despite identical corrosion rates (Figure 4.13) (due to the choice of the normalizing area of ECR).  Furthermore, the bicarbonate led to more positive potentials and thus OCP compared to the carbonate. However, both Na2CO3 as well as NaHCO3 resulted in significant reductions of the OCP after the polarization as shown in Figure 4.14. The drop between the pre and post polarization OCP was between 0.27 V and 0.48 V for both rebar types in the solutions with varying levels of sodium carbonate and sodium bicarbonate. The only exception was ECR in 0.01M NaHCO3 solution, which led to a drop of less than 0.04 V. Sodium hydroxide on the other hand resulted in an increase of at least 0.23 V of the OCP.    Both substances led to a steeper incline of the PDP curve and thus a more passive behaviour in comparison to the reference solution. For both sample types the pitting potential increased from the 0.9M NaOH to the 0.01M Na2CO3 and further to the 0.01M NaHCO3 solution. The pitting potential was higher for the ECR samples in all three cases.   4 Results   102 Thus, both carbonate and bicarbonate had a passivating effect on the steel. The effects were very similar even though carbonate reduced the current density and corrosion rate slightly more than bicarbonate did and the bicarbonate led to more positive OCPs in comparison to the carbonate.   Considering that carbonate and bicarbonate had a very similar passivating effect on the corrosion behaviour, it is only logical that 0.01M NaHCO3 and 0.0033M Na2CO3 also passivated the steel. The PDP curve of the weakly concentrated carbonate-bicarbonate solution was very similar to the 0.01M Na2CO3 as well as the 0.01M NaHCO3 curve. Furthermore, the corrosion rates were in between the rates of the carbonate and of the bicarbonate solutions as shown in Figure 4.13. The total concentration of carbonate and bicarbonate ions was higher in the 0.01M NaHCO3 and 0.0033M Na2CO3 solution, which led to even lower current densities and a larger drop in OCP after the polarization. A significant amount of current instabilities was observed, which was much more pronounced for the UCR compared to the ECR specimens.   Similarly, the solution containing 0.03M NaHCO3 and 0.01M Na2CO3 led to the smallest observed current densities and corrosion rates because it had the highest concentration of carbonate and bicarbonate ions. Furthermore, the drop in post polarization OCP was maximized with this solution and resulted in an OCP reduction of 0.48 V and 0.42 V for ECR and UCR respectively. One can observe that the difference in current densities between the UCR and ECR samples was the smallest for the strongly concentrated carbonate-bicarbonate solution. The solution containing 0.01M NaHCO3 + 0.0033M Na2CO3 resulted in current densities 10 times larger for the ECR compared to the UCR specimens. Both sample types showed large amounts of current fluctuations in the highly concentrated solution. However, the current fluctuations of the ECR samples were steadier compared to the UCR specimens. In other words, the range of the fluctuations varied much more for the UCR samples.  Carbonate and bicarbonate had a passivating effect on both types of samples. The current densities as well as the corrosion rates were reduced and the OCP became more positive. However, both carbonate and bicarbonate resulted in a significant drop of the post polarization OCP. A higher concentration of carbonate or bicarbonate led to a larger decrease in current density and thus corrosion activity. Significant amounts of current fluctuations could be observed at low current densities, indicating equipment noise as a likely cause. However, current instabilities were notably  4 Results   103 more pronounce for UCR specimens despite the larger polarized area and thus larger absolute currents of UCR compared to ECR. Thus, the sample surface conditions likely contributed to the current fluctuations.  4.1.4.3 Discussion Sodium carbonate had a passivating effect on the corrosion behaviour regardless of the exposure time. This was indicated by positive OCPs, low current densities, a steep incline of the PDP curves and low corrosion rates. Sodium carbonate solution had a very high pH around 11, which promoted the formation of a passive layer even after a short period of time. The passive layer on the surface of the steel protected the underlying metal from corrosion. This protective layer formed quickly and was very effective in reducing the current densities and even led to values below those of sodium hydroxide solution with a pH of 13.  The corrosion promoting effect of sodium bicarbonate diminished with increasing exposure time and even passivated the steel after four weeks of exposure to the aerated solution. ECR samples led to higher current densities for both short term and long term testing compared to the UCR samples.  Sodium bicarbonate solution had a pH around 8, which is the lower end of the alkaline spectrum. Consequently, the passive layer took much longer to form and stabilize compared to the sodium carbonate and the sodium hydroxide solution. Due to the low pH, the corrosion was accelerated during the short term testing in comparison to the sodium hydroxide solution. However, eventually a weak passive layer formed, which reduced the current densities significantly.   As a result of the aforementioned effect of carbonate and bicarbonate, the carbonate-bicarbonate solutions passivated the steel regardless of the concentration in the long term testing. The opposing short term effects of carbonate and bicarbonate led to the corrosion behaviour being highly dependent on the carbonate and bicarbonate concentrations as well as on the sample type. Furthermore, UCR was observed to passivate more readily in the carbonate-bicarbonate solutions in comparison to ECR.   Even though sodium hydroxide solution had a high pH around 13.5, the observed current densities were higher in comparison to the sodium carbonate and the sodium bicarbonate solution  4 Results   104 under long term testing. Nonetheless, the high pH led to an almost instantaneous passivation, whose protective layer could quickly reform after it had been broken down by polarization.   4.1.5 Chloride and Carbonation 4.1.5.1 Short Term The PDP curves and OCP shown in Figure 4.15 and Figure 4.16 represent solutions containing 0.01M NaHCO3 and 0.0033M Na2CO3 and varying levels of NaCl between 0M and 1M.  The recorded current densities increased with increasing NaCl content, which resulted in a shift to the right of the graphs. The addition of NaCl did not only lead to an increase in current density but also a more negative OCP. Both observations were made for both types of samples and indicated a rise in corrosion activity.  Figure 4.15: Anodic PDP curves for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration   4 Results   105  Figure 4.16: Open circuit potential of ECR and UCR in 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration Even though the effect of the chloride under carbonated conditions was identical for the ECR and UCR samples, the observed active current densities were consistently 17 times larger for the ECR samples. This indicated higher corrosion activity of the ECR samples throughout the testing series. Furthermore, the initial slope of the curve was shallower and the OCP tended to be more negative for the ECR specimens, which is another indicator of higher corrosion activity.  The shape of the curves corresponding to the solution without any chloride was very typical for an active passive allow, clearly showing an active, passive and transpassive region. However, even small amounts of chloride changed the shape of the PDP curve significantly. Instead of a passive and transpassive region, the current density in chloride contaminated solutions remained almost constant once a potential of approximately 0.5 V had been reached. This shape indicated a diffusion controlled corrosion behaviour, whose rate was limited by the diffusion of the chloride ions to the corrosion site.   Figure 4.15 shows that the increase in current density was proportional to the increase in NaCl concentration, even at low salt levels. This differed from the observation made for the 0.9M NaOH solution in Figure 4.5. In the sodium hydroxide solution only high NaCl levels of 1M increased the current densities and decreased the OCP significantly. Furthermore, the corrosion promoting effect of NaCl on the corrosion behaviour was much more significant for the ECR specimens in comparison to the UCR specimens in the sodium hydroxide solution. However, Figure 4.15 showed that NaCl had the same effect on the corrosion behaviour regardless of the specimen type.    4 Results   106 4.1.5.2 Long Term Figure 4.17 and Figure 4.18 show the anodic PDP curves and corrosion rates obtained from samples that were submerged in an aerated solution of 0.01M NaHCO3 and 0.0033M Na2CO3 with varying levels of NaCl for four weeks.    Figure 4.17: Anodic PDP curves for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution   4 Results   107  Figure 4.18: Corrosion rates for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution   Figure 4.19: OCP pre and post potentiodynamic polarization for 0.01M NaHCO3 + 0.0033M Na2CO3 solution with varying NaCl concentration after 4 weeks of exposure to aerated solution One can observe a significant drop in OCP with increasing amount of NaCl. Furthermore, the recorded current densities increased with increasing NaCl content, indicating an increase in corrosion rate. The latter was verified by the LPR measurements, that showed a steady increase of the corrosion rate with increasing chloride content even at low concentrations of 0.001 M. At low potentials the current densities tended to be higher for the ECR samples compared to the UCR samples. However, the difference in current densities between the two sample types diminished in the transpassive region. The OCP was very similar for the two sample types with a tendency for a more negative OCP for the ECR samples. Furthermore, Figure 4.19 shows that the OCP decreased from pre to post  4 Results   108 polarization for both samples regardless of the NaCl level. However, the largest drops in OCP were observed for both sample types in the salt free solution and UCR in 0.001M NaCl solution. The latter three cases showed passive current densities in Figure 4.17 and corrosion rates in Figure 4.18.  The corrosion rate of ECR was more sensitive to the addition of NaCl, which led to a corrosion rate of 232 µm/year compared to the UCR value of 9 µm/year. This is in line with the findings of the PDP, which resulted in significantly higher current densities for ECR than UCR near the OCP. Further addition of the oxidizing salt led to similar deterioration speeds as well as current densities of the two sample types. Nonetheless, the addition of 1 M NaCl led to double the corrosion rate for ECR compared to UCR despite similar current densities.   The shape of the curves indicated that diffusion control of the corrosion process became more dominant with an increasing amount of sodium chloride. The diffusion control was more pronounced for the ECR specimens, which was indicated by a steeper slope of the dashed curves compared to the continuous counterparts. Furthermore, the dashed curves were not as smooth and had many small current fluctuations. The more dominant diffusion control and the current fluctuations of the ECR samples are explained with crevice corrosion initiating under the epoxy coating. This type of corrosion is very dependent on the diffusion speed because the ions need to diffuse to the corrosion site under the coating. This happens either through the coating or through the steel-epoxy interface.   Even small amounts of NaCl had a significant effect on the corrosion behaviour under carbonated conditions as shown in Figure 4.17. This differs from the observations made under very alkaline conditions as shown in Figure 4.7. Adding sodium chloride to the 0.9M sodium hydroxide solutions only led to an increase in corrosion activity at high NaCl concentrations of 0.1M and particularly 1M. However, the effect of the NaCl on the corrosion behaviour was the same for the ECR and UCR for both types of base solutions. Furthermore, diffusion control was particularly pronounced in the case of the ECR samples regardless of the base solution. This was due to the fact that crevice corrosion occurred on the ECR sample but not on the UCR specimens.  4.1.5.3 Discussion The current densities decreased with increasing exposure time. This trend was observed for both types of samples. However, this trend was much stronger for the ECR samples whose current densities were up to 100 times higher for the short term testing compared to the long term testing.   4 Results   109 Furthermore, the OCP became more positive with increasing exposure time. This trend diminished with increasing NaCl content.  Diffusion control was the current limiting phenomenon in all solutions containing sodium chloride regardless of the concentration and the exposure time. However, the dominance of the diffusion control was the most pronounced after four weeks for the ECR samples. This was likely due to the buildup of corrosion products and more importantly the initiation of crevice corrosion under the epoxy coating.  The corrosion rates, obtained during the LT testing, increased with increasing NaCl level even at very low NaCl concentrations of 0.001 M regardless of the sample type. However, ECR was more sensitive to the addition of NaCl, which resulted in significantly higher corrosion rates at NaCl levels of 0.001 M and 1 M.   The corrosion activity increased with the addition of even small amounts of sodium chloride. However, under very alkaline conditions, the corrosion promoting effect of NaCl was only observed when large amounts of NaCl such as 1M were present. This observation was made for both exposure times and both specimen types in 0.9 M NaOH solution.  Less NaCl was needed to destroy the passive layer under carbonated conditions in comparison to the sodium hydroxide solution. The stability of the passive layer was higher under the more alkaline conditions, despite the fact that the observed current densities and corrosion rates of the base solution were higher in comparison to the carbonated base solution.  4.2 Accelerated Corrosion 4.2.1 Uncoated Rebar The empirical and measured mass loss of the uncoated rebar (UCR) (chapters 3.3.2.1.1.2 and 3.3.2.1.2.1), that had been polarized for 22, 45 and 90 days (chapter 3.2.2) are shown in Table 4.2 and Table 4.3 as absolute values and percentiles of the mass of the reference specimens respectively. The empirical values were determined by Faraday’s law, using Eq.(19) from chapter 2.1.2.1 (page 16), which is repeated below.  𝑚 = 𝐼 ∙ 𝑡 ∙ 𝑀𝑛 ∙ 𝐹 (19) The molar mass M and number of equivalents n were based on the simplifying assumption of corrosion of iron. The current I was determined by the product of the applied current density of 200  4 Results   110 µA/cm2 (chapter 3.2.2) and the embedded surface area of the UCR. The time t equated to the duration of the accelerated corrosion process of 22, 45 and 90 days. The applied current density of the Hall effect and thermography specimens was 200 µA/cm2 (chapter 3.2.2). However, the UCR of the thermography specimens was 15 cm long, of which 14 cm were embedded in concrete (chapter 3.3.2.1.2.1), and the rebar of the Hall effect samples was 25 cm long, of which 22 cm were embedded in concrete (chapter 3.3.2.1.1.2). Thus, the absolute mass loss shown in Table 4.2 was notably lower for the thermography specimens compared to the Hall effect samples. The same procedure for corrosion acceleration had been performed on both sample types. However, due to the differences in surface area, polarization was performed on two separate electrical circuits.  All measured values of the 22 day exposure specimens (referred to as “22”) exceeded the empirically determined mass loss by up to 1.8 % or 3.34 g. However, while a similar difference between the empirical and gravimetric value was observed for the 45 HE bars, the measured weight loss of the 45 IRT specimens was below that determined by Faraday’s law. With the exception of IRT45-2, the increase in weight loss of the IRT specimens was very small between 22 and 45 days of galvanostatic polarization.  Table 4.2: Mass loss of uncoated rebar (UCR) in g  Furthermore, only a small increase in measured weight loss occurred between a polarization duration of 45 and 90 days for the HE bars even though the mass loss had been expected to double. The mass loss of the 90 IRT specimens was approximately twice as large as the mass loss after 45 days of accelerated corrosion. The latter observation was in line with the empirical values. However, the Empirical Measured Empirical Measured[g] [g] [g] [g]1 13.10 7.402 10.44 7.473 9.53 7.601 19.79 8.852 19.28 10.643 21.84 8.611 21.91 16.532 26.52 18.953 24.05 16.6445 19.97 11.9890 39.94 23.97Hall effect Thermography22 9.76Specimen5.86 4 Results   111 empirical values were significantly larger than the measured values due to the low gravimetric mass loss after 45 days.  Table 4.3: Mass loss of uncoated rebar (UCR) in %  The final percentile mass loss of the two specimen types was very similar after 90 days of galvanostatic corrosion acceleration. However, due to the reduced actual corrosion rate between 45 and 90 days for the HE bars and between 22 and 45 days for the IRT bars, the final measured mass loss was significantly lower than the empirical mass loss.  Table 4.4 shows the maximum crack width measured on the top of each specimen. The top surface of six specimens is shown in Figure 4.20. The crack width increased with increasing polarization duration for both specimen types. However, while the largest increase occurred between a polarization duration of 22 and 45 days for the HE specimen (except for HE90-2), the largest increase in crack width of the IRT specimens was observed between 45 and 90 days.  Table 4.4: Maximum crack width  Empirical Measured Empirical Measured1 7.1% 7.0%2 5.7% 7.1%3 5.2% 7.2%1 10.8% 8.3%2 10.4% 9.9%3 11.8% 8.1%1 11.9% 15.5%2 14.2% 17.8%3 13.0% 15.6%Hall effect Thermography224590Specimen5.2%10.7%21.4%5.5%11.3%22.6%Hall Effect Thermography[mm] [mm]1 3.00 1.502 2.50 1.503 1.50 2.001 3.50 2.002 3.00 3.003 3.00 2.501 3.50 3.002 4.50 4.503 3.00 4.00459022Specimen 4 Results   112  7 Figure 4.20: Hall effect (HE) and infrared thermography (IRT) specimens after anodic polarization for 22, 45 and 90 days Soudki et al. suggested that the accumulation of corrosion products could hinder the diffusion of hydroxide or Fe2+ ions or a combination thereof (El Maaddawy and Soudki 2003). Consequently, the corrosion rate would decrease leading to a lower gravimetric than empirical mass loss at high corrosion degrees. However, this is unlikely to have occurred because the specimens had been galvanostatically polarized. During the corrosion acceleration, a constant current was applied. If the resistance increased due to the accumulation of corrosion products, the power supply would have increased the voltage to maintain a constant current.  Andrade et al. suggested heat generation as the primary cause for a less efficient current application (Alonso et al. 1998). Conversely, a higher gravimetric than empirical mass loss is explained  4 Results   113 by acidification and non-corroding inclusions of the metal (Alonso et al. 1998). The induced corrosion process can lead to acidification, which in turn causes a simultaneous additional corrosion process to take place. Non-corroding metal inclusions do not take part in the anodic dissolution process, but fall out once the surrounding metal has oxidized.  The exact causes of the inconsistent mass loss are unclear. Heat generation, acidification, non-corroding inclusions as well as human error could have had an impact on the actual corrosion rate. To identify the specific origins of the higher and lower than empirically predicted mass loss additional experiments would need to be performed. These would need to include pH measurements to assess the level of acidification, thermal measurements to assess the heat generation as well as identification of the corrosion products and steel substrate inclusions.  4.2.2 Epoxy Coated Rebar 4.2.2.1 In Concrete Concrete specimens (chapter 3.3.1.1.1), each reinforced with two ECR, had been polarized for 0, 22 and 45 days (chapter 3.2.3.1). The coating of the ECR had intentionally been damaged prior to casting of the concrete specimens.  Corrosion at the coating damage was observed during the initial visual inspection of the rebar. The corrosion locations at the coating holidays measured from the rebar’s end without electrical connection are marked by an X in Table 4.6. Furthermore, rebars, that did not show any sign of corrosion at any of the coating defects can be identified by a grey background in Table 4.6. Even though some crevice corrosion at coating holidays was observed, the epoxy coating did not show any sign of accumulation of corrosion products under the coating (for example in the form of blisters).  Once the epoxy coating had been removed, the full extent of the corrosion was uncovered.  Corrosion at the electrical connection was observed for all but two polarized specimens. The presence and lack of corrosion of the electrical connection can be identified by the red and green font in Table 4.6. Rebar R-1 and R-7 had been polarized for 22 days and 45 days respectively. These two specimens were the only polarized rebars, that did not corrode at the electrical connection. None of the reference specimens, showed signs of corrosion at the electrical connection.  Undercoating corrosion was observed on almost all specimens including the unpolarised references samples. However, the extent of the undercoating corrosion varied greatly and is shown in brown in Table 4.6. A legend explaining how to identify different types of corrosion in Table 4.6 is shown in Table 4.5.  4 Results   114 Table 4.5: Legend corresponding to Table 4.6 Legend Type of corrosion X Corrosion at coating holiday  No corrosion at any coating holiday Red Corrosion at electrical connection point Green No corrosion at electrical connection point  Undercoating corrosion X Undercoating corrosion and corrosion at coating holiday  4 Results   115  Table 4.6: Distance from corrosion location to rebar’s end for the entire specimen series 35[days] [-]R-13 top X XbottomR-14 topbottomR-15 topbottomR-16 top `bottomR-17 top XbottomR-18 topbottomR-1 top XbottomR-4 topbottomR-2 top X X X XbottomR-5 top X X XbottomR-3 topbottomR-6 topbottomR-7 topbottomR-12 top X XbottomR-8 top XbottomR-9 top X XbottomR-10 top X XbottomR-11 top X Xbottom45Induced current durationDistance from rebar's end to corrosion location3428 29 30 31 32 3322 23 24 25 26 27Surface17 18 19 20Rebar12 13 14 15 160[cm]220 1 2 3 94 5 6 7 8 2110 11 4 Results   116  Even though the reference specimens had not been polarized, corrosion could be observed on all six bars. However, except for R-17, the unpolarised specimens either corroded under the coating or at a coating damage but not both. Furthermore, while the extent of corrosion was very small, corrosion had initiated in multiple locations on each bar.  The polarized specimens resulted in a much larger extent of the corrosion. However, even though a larger amount of surface area of the rebar had been corroding, the corrosion depth was still minimal. Thus, neither mass nor diameter loss measurements could be used to quantify the amount of corrosion. The extent of the corrosion of the polarized specimens varied greatly and were not proportionate to the duration of the applied current. Nonetheless, most specimens, that had been polarized for 45 days resulted in a larger amount of corrosion compared to the 22 day specimens. A combination of corrosion at the coating holidays and undercoating corrosion was observed for most polarized specimens.  4.2.2.2 In NaCl Solution Epoxy coated rebar (ECR) with one intentional coating damage measuring 3 mm by 2 mm was submerged in a neutral 3% NaCl solution and polarized with a current density of 200 µA/cm2 for 45 days (chapter 3.2.3.2).  Figure 4.21 and Figure 4.22 show two specimens after having undergone the accelerated corrosion procedure in neutral NaCl solution. Both specimens were epoxy coated during the accelerated corrosion experiment. The epoxy coating was only removed after polarization had stopped and the specimens were taken out of the test solution. The red circles show the location of the coating damage.   Figure 4.21: Specimens after anodic polarization in neutral NaCl solution with epoxy coating (top) and stripped of the epoxy coating (bottom)   4 Results   117  Figure 4.22: Location of coating damage on specimens after anodic polarization in neutral NaCl solution with epoxy coating (left) and stripped of the epoxy coating (right) An accumulation of corrosion products was observed at the coating damage of each specimen. Furthermore, crevice corrosion along the edges of the coating holiday could be observed. Once the epoxy coating had been removed, the entire surface of the steel rebar was revealed but no additional corrosion could be observed. The corrosion had not expanded under the coating beyond 2 mm, even though the specimens had not been confined in concrete. These results were in line with the findings of Nguyen and Martin (Nguyen and Martin 1996, 2004), who reported anodic sites at the coating damage rather than under the coating for ECR in a neutral environment in the presence of NaCl.  The blue rectangle in Figure 4.21 encloses an area of the stripped ECR with a slightly brighter surface colour compared to the rebar surface to the left and right of the rectangle. Nguyen and Martin observed a white metallic appearance of surface areas, which had experienced cathodic disbondment in saturated calcium hydroxide solution with 3.5% NaCl (Nguyen and Martin 2004). The reported pH of the surface with the white metallic appearance was between 10 and 11. Furthermore, cathodic delamination in conjunction with anodic sites at the coating damage of specimens in neutral NaCl solution have been reported by Nguyen et al. (Nguyen, Hubbard, and Pommersheim 1996).  Consequently, the brighter appearance of the surface enclosed by the blue rectangle in Figure 4.21 was likely the result of cathodic disbondment of the epoxy coating.   4.2.2.3 In Alkaline Solution Epoxy coated rebar (ECR) with four intentional coating holidays (chapter 3.3.2.1.1.3 and 3.3.2.1.2.2) were submerged in an alkaline solution containing 3% NaCl. The specimens were exposed to one of three test series, that each consisted of a delamination stage (stage 1), followed by a corrosion stage (stage 2) (chapter 3.2.3.3). Galvanostatic cathodic polarization was employed during stage 1 of series A and C and thermal cycling was used to enhance coating delamination of series C.  The subsequent corrosion stage consisted of galvanostatic anodic polarization for series A and B and of solution aeration for series C.   4 Results   118 4.2.2.3.1 After 45 Days of Stage 2 Figure 4.23 and Figure 4.24 show the rebars, that had been exposed to stage 2 of the accelerated corrosion procedure for 45 days. No corrosion was observed for specimen C, that had been cathodically polarized followed by exposure to the aerated solution. Both specimens A and B resulted in small amounts of corrosion. Specimen A had been cathodically polarized while rebar B had been exposed to thermal cycles during stage 1 of the accelerated corrosion procedure. Both A and B were cathodically polarized during stage 2.   Figure 4.23: Top surface of specimen A (top), B (middle) and C (bottom) with electrical connection on the right rebar end While specimen A corroded at all four coating holidays, only one of the coating defects of specimen B showed corrosion products. However, specimen B showed more widespread corrosion under the coating. The differences in the extent of undercoating corrosion could be observed on both surfaces, but were more pronounced on the bottom surface shown in Figure 4.24. Undercoating corrosion of specimen A could only be observed towards the right rebar end, close to the electrical connection (Figure 4.23, Figure 4.24). Specimen B on the other hand showed corrosion products under the coating along the entire rebar.   4 Results   119  Figure 4.24: Bottom surface of specimen A (top), B (middle) and C (bottom) with electrical connection on the right rebar end Comparison of specimen A and C showed the effects of anodic polarization versus aeration as means to accelerate the corrosion of ECR. Figure 4.23 and Figure 4.24 clearly show, that only anodic polarization led to the formation of corrosion products.  The differences in the extent of corrosion of samples A and B, highlight the effect of cathodic polarization compared to thermal cycles during stage 1. One possible explanation of the more extensive undercoating corrosion of sample B is, that thermal cycling was more effective in causing disbondment of the coating compared to cathodic polarization. Another explanation is, that the rebar surface under the coating of sample A was protected by a passive film caused by a high pH layer, that had formed due to cathodic disbondment during stage 1. During the cathodic disbondment process, sodium ions migrate along the steel epoxy interface and react with the water and oxygen molecules, that diffuse through the epoxy coating. Sodium hydroxide is formed in this cathodic reaction and accumulates under at the coating-substrate interface. Consequently, the pH on the steel surface under the coating is increased (Nguyen and Martin 2004). Thus, corrosion could only take place at the coating holidays, where the availability of chloride ions was higher compared to the coating-steel interface. Even though cathodic disbondment is believed to have occurred on both specimen A and B, this phenomenon was likely enhanced on sample A due to the cathodic polarization of the sample. Mechanically speaking, this would mean that option B might not have been any more effective in causing disbondment. Nonetheless, thermal cycling was more effective in promoting subsequent undercoating corrosion compared to cathodic polarization.  4.2.2.3.2 After 40 and 84 Days of Increased Current of Stage 2 The absolute and percentile mass loss of series A and B ECR specimens is shown in Table 4.7 and Table 4.8 respectively. The measured mass loss of all but two specimens surpassed the empirical mass loss determined by Faraday’s law. This gap between empirical and gravimetric mass loss was  4 Results   120 particularly pronounced for series B. Series B had been exposed to thermal cycles and series A to cathodic polarization during the delamination stage. Both specimen series had been anodically polarized in order to accelerated the corrosion. Furthermore, both series had been kept in an oven set to 50 °C during the corrosion phase. However, series A, B and C had been located on the middle, bottom and top rack in the oven respectively and significantly more evaporation of the solution of series B could be observed compared to series A and C. Thus, a temperature gradient caused by the location of the heating elements likely led to higher temperatures of the solution of series B compared to series A and C.   Table 4.7: Mass loss of ECR in g  Consequently, the average mass loss of both series was compared. 40A and 40B specimens had lost on average 1.7% and 2.4% respectively. Furthermore, 84A and 84B specimens had lost on average 5.4% and 6% respectively. Thus, the difference in mass loss between the two series did not increase with increasing exposure to the accelerated corrosion stage. Consequently, the temperature gradient in the oven had a minor effect on the corrosion rate.  Empirical Measured Empirical Measured[g] [g] [g] [g]1 0.86 2.882 1.56 4.751 2.42 5.202 2.87 3.531 8.04 12.192 7.13 6.493 6.401 11.08 7.402 10.40 7.453 8.48 7.9784B 5.92 5.925.925.9284A40B 1.731.731.731.73Hall effect ThermographySpecimen40A 4 Results   121  Table 4.8: Mass loss of ECR in %  Therefore, thermal cycling as the mechanism of the delamination stage caused the larger amount of corrosion of series B compared to series A, which had undergone cathodic polarization during the first stage. Corrosion had already expanded underneath the coating along the entire rebar for the 40A and 40B specimens. No further lateral expansion of the corrosion could occur between the 40 and 84 specimens. The delamination stage affected the initial stages of corrosion. The less effective cathodic polarization slowed down the initial lateral expansion of corrosion under the coating of series A compared to series B. This led to a larger mass loss for series B. However, once the corrosion of both series had expanded along the entire rebar surface leading to cracking of the epoxy coating, the corrosion progressed at a similar rate for series A and B. Thus, the initial difference in mass loss did not increase over time. Table 4.8 shows a greater variability of the mass loss of series A compared to series B. Furthermore, only the measured mass loss of series A varied notably more than the gravimetric mass loss of UCR shown in Table 4.3 on page 111. Thus, the increased variability likely originated from the cathodic polarization stage of series A. The origins of the inconsistencies area unclear and could have been caused by human error. Additional experiments are needed to determine the origin of the increased variability.  4.2.2.4 Discussion All specimens, that had been embedded in concrete prior to the accelerated corrosion process (chapter 4.2.2.1) resulted in the initiation and lateral expansion of corrosion under the coating. Empirical Measured Empirical Measured[%] [%] [%] [%]1 0.6% 2.0%2 1.1% 3.3%1 1.6% 3.5%2 1.9% 2.4%1 5.4% 8.3%2 4.8% 4.4%3 - 4.3%1 7.5% 5.0%2 7.0% 5.1%3 5.8% 5.4%SpecimenHall effect Thermography1.2%1.2%3.9%3.9%1.2%1.2%3.9%3.9%40A40B84A84B 4 Results   122 However, corrosion was only observed at coating defects of the ECR specimens, that had not been embedded in concrete but instead had been submerged in a neutral 3% NaCl solution during the anodic polarization (chapter 4.2.2.2).  The same current density (relative to the area of the coating damage) had been applied to the ECR in neutral NaCl solution and the embedded specimens. Furthermore, the room temperature of both sets of specimens had been similar and hydration temperature increase had been minimal during the anodic polarization of the specimen at a concrete age of 7 days (Zou et al. 2012). Despite these similarities of the exposure conditions, the embedded ECRs were more prone to undercoating corrosion than the ECR in neutral NaCl solution. The confinement of ECR in solution was reduced compared to the embedded ECR. Furthermore, the coating water uptake was likely similar or increased (due to the lack of confinement) for the ECR in neutral NaCl solution compared to ECR embedded in concrete. A reduction in the confinement and increased water uptake of the coating would increase the likelihood of water induced adhesion loss. In the presence of coating defects, the adhesion loss would be expected to make it easier for ions to enter the epoxy-steel interface at the coating damage and migrate along the interface. However, ECR embedded in concrete corroded under the epoxy coating despite the increased confinement compared to ECR in neutral NaCl solution. The surrounding pH was lower for the ECR in solution compared to the alkaline pH of concrete. Thus, the reduced pH of the 3% NaCl solution compared to the concrete likely caused the corrosion to only initiate and progress at the coating damage (Nguyen, Hubbard, and Pommersheim 1996). The high pH and sodium content of the concrete promoted cathodic disbondment of the epoxy coating combined with the initiation of anodic sites under the coating rather than at the coating damage (Nguyen and Martin 1996, 2004). Furthermore, the alkaline pH built a passive layer on the steel surface at the coating damage, thus protecting it from corrosion initiation at the coating holiday. The latter was not the case for all embedded rebar as the passivating effect of the high pH competed with the corrosion promoting effect of the added NaCl. Cathodic disbondment can occur even in neutral electrolytes such as a 3% NaCl solution (Nguyen and Martin 2004). However, the low pH of the neutral solution did not protect the steel at the coating damage. Thus, corrosion initiated and expanded at the coating damage rather than under the epoxy coating, where the steel was protected by a high pH and reduced amounts of chloride ions.   4 Results   123 The initial current density of the corrosion stage of series A and B in alkaline solution (chapter 4.2.2.3) was 200 µA/cm2 relative to the area of the coating damage and thus identical to the embedded specimens and ECR in neutral NaCl solution.  The specimens in neutral NaCl solution led to a similar corrosion pattern as specimen A (cathodic polarization followed by anodic polarization). Both specimen types corroded only at the coating damage even though cathodic disbondment is believed to have been more extensive for specimen A compared to ECR in neutral solution. Thus, a high concentration of chloride ions at the coating damage combined with a high pH at the coating steel interface resulted in the protection of the steel surface under the coating and degradation of the steel at the coating damage.  Specimen B had been exposed to thermal cycles prior to anodic polarization. This sample resulted in predominantly undercoating corrosion compared to corrosion at the coating damage of samples in neutral NaCl solution. These observations are in line with the findings of Nguyen and Martin (Nguyen and Martin 2004; Nguyen, Hubbard, and Pommersheim 1996). Both sample B and the specimens in neutral NaCl solution are believed to have experienced water induced adhesion loss and cathodic disbondment (Nguyen and Martin 2004). However, the solution, that sample B was exposed to was very alkaline and thus protected the steel at the coating defects. Hence, anodic sites initiated under the coating as opposed to at the coating holiday locations of sample B. The larger pH gradient between the coating damage and the cathodically delaminated sites of the specimen in neutral NaCl solution (Nguyen, Hubbard, and Pommersheim 1996) resulted in corrosion at the coating holiday as opposed to under the coating.  Aeration of specimen C did not result in any corrosion. The protection of the steel due to a high pH dominated over the corrosion promoting chloride ions and aeration of the solution. Similar to sample A, the cathodic polarization of specimen C promoted the cathodic delamination, which increased the pH at the coating steel interface.   The largest mass loss was observed with method B, which consisted of thermal cycling followed by anodic polarization at 50°C. However, method A, which employed cathodic polarization in the delamination phase, led to the same corrosion rate once corrosion had expanded across the entire rebar surface. Thus, the delamination stage affected the lateral expansion and thus the initial corrosion rate. Consequently, the mass loss of specimens 40A was lower compared to the 40B bars. However, once corrosion had initiated on the entire rebar surface and the lateral expansion  4 Results   124 stagnated, corrosion progressed at a similar rate for samples A and B. Thus, the difference in mass loss between 84A and 84B was almost identical to the difference between 40A and 40B.   4.3 Corrosion Detection 4.3.1 Conventional Non-Destructive Testing 4.3.1.1 Small Lab Specimens The corrosion of epoxy coated rebar (ECR) embedded in concrete (chapter 3.3.1.1.1) had been accelerated by galvanostatic polarization (chapter 3.2.3.1). Subsequently, the linear polarization resistance, half cell, Wenner probe and the ground penetrating radar method were employed to detect the corrosion non-destructively (chapter 3.3.1.1.2).  The reinforcing bars had corroded at the exposed steel surface at the coating damage or underneath the coating. The corrosion at the coating damage is referred to as exposed or visible corrosion and the corrosion under the epoxy coating is referred to as hidden or undercoating corrosion in this chapter.  Specimens, that did not shown any signs of exposed corrosion are shown as dashed lines and visibly corroding rebars are shown as continuous lines in the following diagrams. Furthermore, the specimens, that experienced undercoating corrosion can be identified by a triangular marker and specimens with a negligible amount of hidden corrosion have a round marker. A legend corresponding to the diagrams in this chapter is shown in Table 4.9. Furthermore, the locations of the exposed and hidden corrosion are shown in Table 4.6 on page 115 in chapter 4.2.2.1.  Table 4.9: Legend corresponding to the diagrams of the LPR, Half cell and resistivity measurements in chapter 4.3.1.1 Legend Type of corrosion ― Exposed corrosion --- No exposed corrosion Δ Undercoating corrosion o No undercoating corrosion  4.3.1.1.1 Linear Polarization Resistance Figure 4.25, Figure 4.26 and Figure 4.27 show the measured corrosion currents in µA for specimens that had been galvanostatically polarized for 0, 22 and 45 days respectively.   4 Results   125 Figure 4.25 shows the measured corrosion currents for the specimen, that had not been polarized. Even though all specimens shown in Figure 4.25 had experienced some degree of undercoating corrosion, the amount of undercoating corrosion was considered negligible. Figure 4.25 shows that the two rebars with exposed corrosion, R-13 and R-17, resulted in higher currents compared to the specimens without corrosion at the coating damage.   Figure 4.25: Results of LPR measurements of specimens without induced current Furthermore, the corrosion currents of R-13 and R-17 varied more, but the locations of higher currents did not identify the corrosion locations. Table 4.6 (on page 115) shows that the exposed corrosion of R-17 was located 16.2 cm away from the rebar’s end. However, Figure 4.25 shows that the detected current peaked at 12 cm and 27 cm. Specimen R-13 showed signs of corrosion at 23.5 cm and 32.8 cm and the current peaks were observed at 7 cm and 32 cm. Hence, only one of the three corrosion locations matched a location of a peak current.  Figure 4.26 shows the corrosion rates of the specimens that had undergone galvanostatic polarization for 22 days. Specimens R-3, R-4 and R-6 did not show any sign of exposed corrosion. However, all specimens were corroding under the epoxy coating with the exception of R-1.   4 Results   126  Figure 4.26: Results of LPR measurements of specimens exposed to a current of 200 µA/cm2 for 22 days Figure 4.26 shows that the corrosion currents observed for the rebars without any exposed corrosion were well below 15 µA. Conversely, the corrosion rates observed for the rebars with visible corrosion were well above 15 µA. Along rebar R-5, currents above 30 µA were observed for exposed corrosion locations. The measured currents below 30 µA corresponded to locations without exposed corrosion. However, the trend of higher rates corresponding to exposed corrosion locations was not observed for R-2 and R-1. Furthermore, the measured currents for exposed corrosion specimens varied more (higher standard deviation) along a rebar in comparison to rebars without exposed corrosion. The corrosion currents of bars that had only undergone hidden corrosion (R-3, R-4, R-6) were below 15 µA and the currents did not fluctuate significantly along these bars.   Figure 4.27 shows the corrosion currents obtained for the specimens that had been galvanostatically polarized for 45 days. Specimen R-7 did not show any sign of corrosion at the coating holidays and the amount of undercoating corrosion of this bar was negligible. Consequently, R-7 resulted in the lowest corrosion currents of the specimen series shown in Figure 4.27. Furthermore, the corrosion current of R-7 did not vary significantly along the bar length.   4 Results   127  Figure 4.27: Results of LPR measurements of specimens exposed to a current of 200 µA/cm2 for 45 days Specimens R-8 through R-12 had all been corroding at coating defects as well as under the coating. Consequently, they all showed higher values and variability in comparison to R-7. The peaks of the currents of specimens R-8 and R-12 coincided with the locations of a combination of exposed and undercoating corrosion. The remaining locations of only undercoating corrosion did not result in current peaks. Furthermore, the current peaks of R-11 were obtained at locations of only undercoating corrosion and exposed corrosion of this sample did not result in an increase of the measured current. Specimens R-9 and R-10 had been corroding under the coating as well as at the coating damage. However, no combination of the two corrosion types led to a peak of the measured corrosion current. All corroding specimens shown in Figure 4.27 led to more fluctuations of the measured currents in comparison to specimen R-7.   Current peaks did not reliably identify corrosion locations regardless of the type of corrosion. Furthermore, even though a significant increase in corrosion currents was observed between the unpolarised (Figure 4.25) and polarized specimens (Figure 4.26 and Figure 4.27), an increase in the duration of the induced current did not increase the corrosion current further.  This was due to the nature of the LPR method, which is a measure for the corrosion rate rather than the amount of corrosion. The corrosion rate did not change between the specimens polarized for 22 and 45 days because the same current density had been applied. Only the amount of corrosion had changed due to the difference in polarization duration.  Upon inspection of the rebars, corrosion at the electrical connection of all polarized specimens was discovered. Consequently, the corrosion currents of specimens with a corroding and intact  4 Results   128 electrical connection were analyzed separately. Figure 4.28 and Figure 4.29 show the range and standard deviation of the measured corrosion currents with respect to the type of corrosion for specimens with intact electrical connection. Specimens with intact electrical connection include all reference specimens that had not been polarized, specimen R-1 that had been polarized for 22 days and specimen R-7 that had been polarized for 45 days. None of the specimens with an intact electrical connection had experienced significant amounts of undercoating corrosion.   Figure 4.28: Range of measured corrosion currents with respect to the corrosion type of specimens with intact electrical connection No corrosion was observed for specimens with corrosion currents below 3.3 µA and corrosion at the coating damage was observed for specimens with corrosion currents above 4.8 µA. Furthermore, the standard deviation of the currents measured along one rebar were observed to be above 0.8 µA for visibly corroding and below 0.5 µA for non corroding specimens.   Figure 4.29: Standard deviation of measured corrosion currents with respect to the corrosion type of specimens with intact electrical connection Thus, both the magnitude of the corrosion currents as well as the standard deviation correctly identified bars with corrosion at the coating damage for specimens with intact electrical connection.  The corrosion current ranges and the standard deviation with respect to the corrosion type, of specimens with corroding electrical connection, are shown in Figure 4.30 and Figure 4.31.   4 Results   129  Figure 4.30: Range of measured corrosion currents with respect to the corrosion type of specimens with corroded electrical connection The lowest corrosion current and lowest standard deviation of current were observed for bars without corrosion. Conversely, the highest corrosion current and highest standard deviation of current were observed for bars with both type of corrosion. Corrosion currents up to 6.9 µA and standard deviations no larger than 0.2 µA identified a lack of corrosion. Currents above 20.2 µA were only observed for specimens with a combination of exposed and hidden corrosion. Corrosion currents between 7.8 µA and 11.9 µA identified samples with only undercoating corrosion. However, corrosion currents between 11.9 µA and 20.2 µA identified the presence of corrosion but not the corrosion type. Nonetheless, Figure 4.31 shows that the corrosion type could be identified by the standard deviation of the measured corrosion currents along one rebar. Samples with only undercoating corrosion led to standard deviations between 0.2 µA and 0.4 µA, whereas exposed corrosion led to a standard deviation of 0.7 µA. The simultaneous occurrence of both corrosion types led to the highest standard deviation of at least 0.7 µA.    Figure 4.31: Standard deviation of measured corrosion currents with respect to the corrosion type of specimens with corroded electrical connection Consequently, the standard deviation of the measured corrosion currents was a better indicator for the type of corrosion for samples with a corroding electrical connection. Nonetheless, the magnitude of the measured corrosion current did correctly identify corroding rebar. The corrosion current ranges and standard deviations regardless of the condition of the electrical connection are shown in Figure 4.32 and Figure 4.33 respectively.     4 Results   130  Figure 4.32: Range of measured corrosion currents with respect to the corrosion type Undercoating corrosion was correctly identified by the magnitude of the corrosion current regardless of the presence of exposed corrosion. However, corrosion only at damage was not correctly identified by the magnitude of the corrosion current since the presence of only visible corrosion led to currents as low as 4.8 µA, which was well below the maximum current of intact specimens of 6.9 µA. However, visible corrosion led to standard deviations of at least 0.7 µA and a lack of corrosion resulted in a maximum standard deviation of 0.5 µA. The standard deviations of samples with only undercoating corrosion were between 0.2 µA and 0.4 µA, which falls within the range of the intact specimens.   Figure 4.33: Standard deviation of measured corrosion currents with respect to the corrosion type Thus, the magnitude of the corrosion current correctly identified undercoating corrosion. The standard deviation of the corrosion currents on the other hand correctly identified exposed corrosion. Initial inspection of the corrosion current identified all specimens, that had undergone undercoating corrosion and some specimens, that had only suffered from exposed corrosion by corrosion currents above 6.9 µA. Furthermore, corrosion currents below 4.8 identified rebars without any corrosion. The standard deviation of corrosion currents between 4.8 µA and 6.9 µA identified specimens with only exposed corrosion by values of at least 0.7 µA and specimens without corrosion by values up to 0.5 µA.   Figure 4.34 shows a conservative summary of the discussed results. No corrosion was observed at measured corrosion currents between 0 µA and approximately 5 µA. Corrosion was possible when a corrosion current between 5 µA and 8 µA was detected and likely if the standard deviation along the bar was above 0.7. Corrosion currents between 8 µA and 12 µA were observed for ECR with either  4 Results   131 undercoating corrosion or corrosion at the coating defect but not both. Measured corrosion currents above 12 µA indicated that simultaneous corrosion at the coating damage and under the coating was possible.   Figure 4.34: Type of corrosion and observed corrosion current ranges and standard deviation σ Corrosion of the rebars could be identified by a combination of the magnitude and the standard deviation of the measured corrosion currents. The standard deviation was independent of the state of corrosion of the electrical connection. Nonetheless, the exact location of corrosion could not be identified by the LPR method.  4.3.1.1.2 Half Cell Figure 4.35, Figure 4.36 and Figure 4.37 show the measured half-cell potential against a copper-copper(II) sulfate reference electrode (CSE) in mVCSE for specimens, that had been galvanostatically polarized for 0, 22 and 45 days respectively.    Figure 4.35 shows that the specimens without any corrosion had a tendency of a more positive half-cell potential, indicating a nobler and less active behaviour. However, the aforementioned trend was very weak.  4 Results   132  Figure 4.35: Results of HC measurements of specimens without induced current The highest variations in potentials were observed for specimens R-18 and R-15, neither of which showed any corrosion. The potentials corresponding to specimens R-13 and R-14 varied the least despite the fact that R-13 showed corrosion products at a distance of 23.5 cm and 32.8 cm from the rebar’s end. The latter locations of R-13 did not result in more negative potentials compared to the remainder of the sample. Corrosion of the specimen R-17 was observed 16.2 cm from the rebar’s end, which coincided with a dip in the R-17 curve in Figure 4.35. However, a second dip at 30 cm did not correspond to a corrosion location.   Figure 4.36 shows that all measured potentials were within a range of -550 mVCSE to -700 mVCSE regardless of the conditions of the rebars. Specimen R-1 was the only rebar, that had only corroded at coating holidays. Bars R-3, R-4 and R-6 had only corroded under the coating and R-2 and R-5 had corroded at the coating damage as well as under the coating.  One could observe very consistent measurements for the specimens with both corrosion types, R-2 and R-5. Furthermore, the largest variations were observed for specimens R-6 and R-3, both of which were only corroding under the coating.   4 Results   133  Figure 4.36: Results of HC measurements of specimens exposed to a current of 200 µA/cm2 for 22 days Corrosion of specimen R-1 was observed at a distance of 20.6 cm from the rebar’s end. Contrary to expectations, one could see a peak in the R-1 curve at that location, indicating a nobler potential compared to the remainder of the specimen.   Figure 4.37 shows that specimen R-7 led to the noblest potentials. R-7 was the only specimen that did not show any signs of corrosion after 45 days of galvanostatic polarization. Specimens R-8 through R-12 had corroded both under the coating as well as at the coating damage.  The R-7 curve showed a relatively large variation with a measurement range from -580 mVCSE to -635 mVCSE.   4 Results   134  Figure 4.37: Results of HC measurements of specimens exposed to a current of 200 µA/cm2 for 45 days Specimen R-10 showed slightly more positive potentials near the exposed corrosion location whereas R-12 showed slightly more negative results around the visibly corroding spots. However, specimens R-11, R-9 and R-8 did not follow either of the aforementioned trends.   The half-cell method did not detect localized corrosion of epoxy coated rebar reliably. No consistent trend of the measurements could be observed to identify the presence of corrosion at the coating damage or undercoating corrosion. However, except for specimen R-16, all measured potentials were similar to the ones observed by Saricimen et al. (Saricimen et al. 1997) for actively corroding ECR without any coating damage. The potentials for R-16 in Figure 4.35 were in line with the actively corroding potential of ECR with 3% coating damage reported by Saricimen et al. (Saricimen et al. 1997). Furthermore, all observed potentials were well below -350 mVCSE, the potential below which the likelihood for active corrosion is high (Malhotra and Carino 2003).   4.3.1.1.3 Wenner Probe Figure 4.38 and Figure 4.39 show the measured resistivity in kΩcm for specimens, that had been galvanostatically polarized for 0 and 45 days respectively. The magnitude of the measured values indicated, that the measurement depth had not extended to the depth of the reinforcing bars. Thus, the resistivity values shown in Figure 4.38 and Figure 4.39 were those of the concrete surface layer. No trend could be observed when comparing the resistivity of the corroding and intact specimens shown in Figure 4.38. None of these specimens were exposed to a current prior to the resistivity test.   4 Results   135  Figure 4.38: Results of resistivity measurements of specimens without induced current Neither the magnitude nor the standard deviation of the measured resistivity was correlated to the presence of corrosion.   The only specimen in Figure 4.39, that did not show any sign of corrosion, was R-7. The resistivity of the aforementioned specimen tended to be slightly higher in comparison to the corroding specimens. However, the difference between the corroding specimens and R-7 was not significant.   Figure 4.39: Results of resistivity measurements of specimens exposed to a current of 200 µA/cm2 for 45 days Furthermore, the variation of the measured resistivity along one specimen was not related to the presence of corrosion.    4 Results   136 The comparison of Figure 4.38 and Figure 4.39 shows that the measured resistivity was higher for the specimens that had ben galvanostatically polarized for 45 days prior to testing.  4.3.1.1.4 Ground Penetrating Radar Figure 4.40, Figure 4.41 and Figure 4.42 show the GPR images of the specimens, that had been galvanostatically polarized for 0, 22 and 45 days respectively. These images were taken by rolling the antenna over the centre of the specimens perpendicular to the rebar. Consequently, the images reflect the conditions of the rebars between 15 cm and 25 cm in Table 4.6 on page 115. The arrows point at the location of the rebar. Two rebars right next to each other mean that they were located in the same concrete specimen. Three concrete specimens are shown in each image. The darker and lighter grey indicate lower and higher signal voltages. Red and Blue indicate the most negative and positive signal voltages respectively.   Figure 4.40 shows the clearest image for R-17 and the blurriest image for R-14. None of the specimens without induced current showed significant amounts of undercoating corrosion but a small amount of undercoating corrosion was observed for all specimens. Furthermore, rebars R-13 and R-17 had corroded at coating damage locations as shown in Table 4.6 on page 115.    Figure 4.40: GPR scan of specimens without induced current Even though small differences between the GPR image of rebars R-13 through R-18 could be identified, they were not caused by varying amounts of corrosion. All rebar images in Figure 4.40 were blurry regardless of the specimen.   The two arrows on the left of Figure 4.41 show the location of R-3 and R-6 in the same concrete specimen. The two rebars had a very similar GPR image even though only R-6 showed signs of corrosion in the area under investigation.    4 Results   137  Figure 4.41: GPR scan of specimens exposed to a current of 200 µA/cm2 for 22 days Rebar R-5 showed a significantly clearer image than R-2 located in the same concrete specimen. Even though both rebars were corroding in the area under investigation, R-2 showed a significantly higher amount of corrosion compared to R-5. However, both specimen R-1 and R-4 led to the blurriest images in Figure 4.41 even though, these rebars had only shown small amounts of corrosion in the area under investigation.  The images of all rebars shown in Figure 4.42 were very clear. However, small differences could still be observed. The GPR scans of rebars R-12 and R-7 are shown on the left of Figure 4.42. R-7 did not show significant amounts of corrosion and led to a very clear GPR image. R-12 on the other hand showed significant amounts of corrosion in the area under investigation. However, the corrosion was mostly located on the underside of the rebar. R-12 led to an image, that was blurrier than the one of R-7.   Figure 4.42: GPR scan of specimens exposed to a current of 200 µA/cm2 for 45 days Rebars R-11 and R-10 were located in the same concrete specimen. R-11 led to a very clear image and R-10 to a blurrier image than R-11. R-11 showed signs of corrosion in the area under investigation whereas R-10 did not corrode in this location.    4 Results   138 Even though localized corrosion could not consistently be detected with the GPR B-scan, one could see differences between the B-scans obtained from specimens, that had been galvanostatically polarized for 0, 22 and 45 days. The images became clearer with increasing duration of the galvanostatic polarization.   Figure 4.43 shows the signal amplitude in relation to the arrival time for rebar R-2, R-10, R-11 and R-13 in yellow, blue, orange and black respectively. R-13 (black) was located in a reference slab and was not corroding in the location of the scan. R-2 (yellow) had been polarized for 22 days and corrosion products were observed in the scan location. Rebars R-10 (blue) and R-11 (orange) were located in the same concrete specimen, that had been polarized for 45 days. However, only R-11 was corroding in the scan location.   Figure 4.43: A-scans taken in the centre of the specimens showing the signal amplitude in mV versus arrival time in ns The first peak of the curves resulted from the direct wave (DW). The peak amplitude of the DW increased with increasing duration of the polarization. The first minima were caused by the air-concrete interface of the specimen surfaces. Furthermore, the third maxima (and second above 5000 mV) was the result of a reflected wave (RW) cause by the embedded rebars. The peak value of the reflected wave was the largest for R-11, the corroding bar, that had been polarized for 45 days. Furthermore, the lowest amplitude of the RW was observed for the reference bar R-13. The peak to  4 Results   139 peak voltage Vp-p of the RW is determined by the difference between the lowest and highest voltage of the RW and is shown in Table 4.10.   An increase in Vp-p with increasing polarization duration was observed. Furthermore, the corroding R-11 led to a higher voltage compared to R-10, which had not corroded at the scan location.  Table 4.10: Peak to peak voltages of the reflected wave in mV Corrosion Polarization Rebar Vp-p No 0 days R-13 5884 Yes 22 days R-2 9038 No 45 days R-10 10385 Yes 45 days R-11 12038  The delay between the arrival time of the DW and the RW is shown in Table 4.11. The longest delay was observed for the reference bar R-13. Furthermore, the corroding bar, that had been polarized for 45 days (R-11) resulted in the shortest delay.   Table 4.11: Delay between the arrival time of the reflected wave in relation to the direct wave in ns Corrosion Polarization Rebar t(RW-DW) No 0 days R-13 1.48 Yes 22 days R-2 1.44 No 45 days R-10 1.44 Yes 45 days R-11 1.17  The peak amplitudes of the direct wave of the previously discussed specimens are shown in Table 4.12. One can observe an increase in peak amplitude with increasing polarization duration. Furthermore, a small increase in amplitude with the presence of corrosion was observed between the R-10 and R-11 bars, that were located inside the same specimens.  Table 4.12: Peak amplitude of the direct wave in mV Corrosion Polarization Rebar DW No 0 days R-13 13010 Yes 22 days R-2 13929 No 45 days R-10 14541 Yes 45 days R-11 14732   4 Results   140 4.3.1.1.5 Discussion None of the NDT methods could detect the exact locations of corrosion. The amounts of corrosion were likely too small for exact localization. However, corroding rebars were successfully identified by the LPR technique. A combination of the magnitude and the standard deviation of the corrosion current had been used in the data analysis.  A correlation between the presence of corrosion and the half-cell potential could not be established and the concrete resistivity values measured with the Wenner probe increased with increasing duration of the polarization. Higher resistivity values of the polarized specimens might indicate that the chloride was drawn more towards the rebar due to the galvanostatic polarization of the latter. Consequently, less chloride was present in the concrete above the rebar, leading to increased resistivity values of the polarized specimens.  The GPR B-scan images became clearer the longer the specimens had been polarized. Furthermore, higher amplitudes of the reflected wave and a tendency for a shorter delay between the reflected and direct wave could be observed for increasing polarization time and with the presence of corrosion. These effects on the GPR signal could indicate the early stages of corrosion (Lai, Kind, and Wiggenhauser 2010), ECR corrosion (Eisenmann et al. 2013) or a reduction in chloride contamination (Hong, Lai, and Helmerich 2012) or a combination thereof. However, due to the extremely low amount of corrosion and the concrete resistivity readings, a reduction in chloride contamination with increasing polarization duration is suspected to have had the largest effect on the GPR signal. Nonetheless, an increase in DW amplitudes with increasing polarization duration was observed. An increase in the DW amplitude has been reported as a result of an increasing chloride content, reduced moisture content or advanced corrosion (Hong, Lai, and Helmerich 2012). Given that the previously discussed properties of the reflected wave indicated a reduction in chloride content with increasing polarization duration, the increase in DW amplitude was likely caused by a reduction in the moisture content caused by the reduction in NaCl content. Sodium chloride binds the moisture in concrete and reduces the water vapor transmission (Griffin and Henry 1964). The drying time of all specimens was identical. Hence, the moisture content at the time of testing was lower for the specimens with lower NaCl content.  4.3.1.2 Large Outdoor Specimens Non-destructive testing (NDT) was performed on the large scale specimens (chapter 3.3.1.2.1) in August 2013, November 2013, December 2013, January 2014, January 2015, July 2015 and October  4 Results   141 2015. The NDT methods used include the LPR as a measure for the corrosion rate, the HC as a measure for the corrosion likelihood and the GPR as a measure for the amount of corrosion. 4.3.1.2.1 Linear Polarization Resistance The results of the linear polarization resistance measurements were analyzed in the unit of µA. In order to convert the values to mm per year the measurements need to be divided by the polarized area in cm2 and multiplied by 0.0116. The latter conversion factor is based on the simplifying assumption of the corrosion of iron. The challenge of interpreting LPR measurements is the unknown polarized area of the metal.  Figure 4.44 through Figure 4.51 show colour coded LPR maps of slab 1, 2 and 3 and the column on the very right explains the colour coding of the maps. Italic, underlined values indicate locations, where the epoxy coating had been visibly damaged during handling of the rebar. Boxed values indicate locations, where the coating had purposely been damaged with a file. Bold values indicate an increased corrosion current compared to the remainder of the tested bar coinciding with the location of a coating damage. Slab 2 and 3 contained 8 kg/m3 and 11 kg/m3 of admixed NaCl respectively (4.8 kg/m3 and 6.6 kg/ m3 of admixed Cl-). Slab 1 served as the chloride free reference slab.  4.3.1.2.1.1 Localized Corrosion Detection The corrosion currents measured during August 2013, July 2015 and October 2015 are shown in Figure 4.44, Figure 4.45 and Figure 4.46 respectively. The values obtained in 2013 were notably larger than the currents measured in 2015. This was observed for all three slabs regardless of the chloride concentration. Furthermore, the declining trend continued between July and October 2015. The reference values of slab 1 were on average 7 and 10 times larger in August 2013 compared to July and October 2015 respectively. However, the corrosion currents of slabs 2 and 3 in Figure 4.44 were on average 28 and 22 times larger than the currents obtained 2 years later (Figure 4.45) and 40 and 30 times larger than the currents obtained in October 2015 (Figure 4.46). Thus, the chloride concentration had a significant impact on the initial high corrosion rates but was not the sole cause. Furthermore, the decline in corrosion currents between July and October 2015 was likely due to varying environmental temperatures.   4 Results   142  Figure 4.44: Corrosion currents in µA of slab 1,2 and 3 measured during August 2013 (3 months exposure) Regardless of the overall magnitude of the measured corrosion currents, localized corrosion could not be detected reliably. In the salt-free slab 1, 33.3% of coating holidays coincided with an increase in measured corrosion current in Figure 4.44. The chance of detection of coating damage was higher for slab 2 with 54.5% and lower for slab 3 with 22.2%. Between August 2013 (Figure 4.44) and July 2015 (Figure 4.45) the chance of coating damage detection increased slightly to 37.5% for slab 1 but decreased significantly for slab 2 to 18.1% and slightly for slab 3 to 17.6%. The probability of detection of the coating damage by an increased corrosion current was 27.3% and 11.1% for slab 2 and 3 respectively. However, none of the coating defects of the reference slab were detected by increased corrosion currents.  Thus, the probability of detection of coating damage decreased over time for the chloride contaminated slabs. This was likely caused by a reduction in corrosion activity over time. NaCl had been admixed and was thus present at the ECR-concrete interface from the time the slabs were cast. Thus, corrosion activity was believed to have been high from the beginning on. While repeated water ponding and draining might have reduced the chloride content in the surface layer of the concrete, the chloride contamination at the rebar level is not likely to have changed significantly over time. However, the high initial corrosion activity might have led to an accumulation of corrosion products at the coating defects. Consequently, the corrosion products blocked access to the anodic reaction sites and reduced the corrosion activity.  Even though the magnitude of the current decreased for the chloride free slab 1 between August 2013 and July 2015, the probability of coating damage detection remained almost constant (33.3% to 37.5%). Slab 1 did not have any admixed NaCl, thus the corrosion activity of slab 1 did not change significantly over time. Consequently, the probability of coating damage detection did not change between August 2013 and July 2015.  M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 2.09 2.37 3.25 3.31 3.63 3.87 3.24 M20 25.25 18.88 25.47 17.43 14.94 23.28 16.14 M20 6.03 8.72 10.24 12.06 15.61 9.22 11.56 M20 0.01.41 1.17 0.35 2.61 12.93 18.07 7.94 8.20 10.97 11.34 5.69 9.40 4.0M20 2.56 2.14 4.34 4.60 4.33 3.81 3.54 M20 13.80 21.59 14.54 11.83 12.46 8.29 11.24 M20 11.50 8.49 19.78 10.50 24.09 32.19 43.73 M20 8.01.70 0.98 2.80 13.03 17.27 16.80 10.48 8.50 15.56 15.86 10.43 12.0M20 2.73 5.18 6.40 5.27 6.60 5.23 6.59 M20 13.47 24.42 26.15 33.86 34.95 7.38 17.76 M20 23.13 25.51 31.05 31.82 31.80 16.46 9.54 M20 16.02.14 1.47 2.42 15.03 17.46 7.52 9.80 9.74 5.44 11.64 9.65 20.0M20 1.73 2.07 2.47 4.96 2.78 5.90 5.85 M20 37.89 26.26 23.08 28.97 14.58 29.78 17.34 M20 12.81 27.81 22.56 43.34 27.09 43.93 37.47 M20 24.02.28 1.32 3.43 11.52 5.55 11.01 10.44 8.46 25.27 8.40 17.82 28.0M20 8.78 5.27 3.80 4.02 3.56 6.42 5.22 M20 38.63 17.50 16.46 13.84 33.09 14.64 28.71 M20 9.91 9.52 12.10 23.10 9.62 20.10 10.49 M20 32.02.00 0.96 3.20 9.33 15.94 14.11 13.35 9.08 13.63 11.24 16.89 36.0M20 0.78 0.43 1.16 1.56 2.02 2.11 1.91 M20 30.40 36.53 38.70 41.68 20.91 27.70 48.33 M20 12.44 13.12 12.91 23.70 14.49 15.36 18.67 M20 40.01.77 0.93 1.91 18.99 11.92 17.98 18.50 15.16 14.88 7.43 14.30 44.0M20 0.43 0.80 0.51 0.82 0.75 1.03 0.93 M20 42.77 11.80 30.62 46.95 17.59 13.69 53.22 M20 12.23 13.49 13.09 20.02 12.13 12.03 11.53 M20 48.01.00 0.94 0.22 15.59 24.05 34.32 28.73 10.02 11.48 6.26 8.08 52.0M20 1.35 1.47 1.53 1.61 1.77 1.70 1.74 M20 42.62 41.41 17.82 23.83 25.40 15.07 12.41 M20 20.96 15.45 8.01 7.66 9.19 8.74 7.42 M20 56.0[µA] [µA]August 2013[µA]Slab #1 Slab #2 Slab #3Connected to 20M at intersection 4 Results   143  Figure 4.45: Corrosion currents in µA of slab 1,2 and 3 measured during July 2015 (26 months exposure) The lack of detection of any coating damage of slab 1 in October 2015 was likely due to reduced environmental temperatures. Higher temperatures promoted the corrosion during the summer months and lower temperatures in the fall slowed down the environmental degradation process of the rebars. The already low corrosion rates of slab 1 (due to the lack of NaCl and the high pH of concrete) reduced further between July and October 2015, thus making localized corrosion detection impossible. However, the corrosion currents of slab 2 and 3 were overall observed to have been higher due to the admixture of NaCl. Thus, a reduction in the environmental temperatures did in fact reduce the probability of localized corrosion detection, but there was still enough corrosion activity to result in a small chance of detection. Nonetheless, it should be noted, that the successfully detected locations of August 2013 were not the same as the ones correctly identified in 2015. The latter observation was made for all three slabs.  M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 0.32 0.31 0.24 0.22 0.36 0.21 M20 1.65 1.09 1.32 0.97 0.96 0.96 0.90 M20 1.07 0.98 1.02 0.99 1.10 1.02 0.93 M20 0.00.29 1.79 0.30 0.65 0.38 0.21 0.67 0.38 0.2M20 0.37 0.39 0.46 0.67 0.71 0.33 M20 0.46 1.51 1.43 0.39 0.44 0.47 0.46 M20 0.71 0.49 0.45 0.42 0.43 0.44 0.50 M20 0.50.75 0.43 0.57 0.25 0.85 0.28 0.64 0.42 0.57 0.40 0.7M20 0.34 0.27 1.18 0.42 0.21 0.29 M20 1.29 0.88 0.77 1.18 1.26 1.08 0.90 M20 1.06 1.03 0.91 1.13 1.02 0.95 0.99 M20 0.90.24 0.42 0.34 0.29 0.26 0.31 0.57 0.55 0.26 0.66 0.34 1.1M20 1.36 0.28 0.38 0.40 0.60 0.33 M20 0.72 0.72 1.22 0.69 0.71 0.72 0.73 M20 0.96 0.97 0.95 0.78 0.71 0.76 0.69 M20 1.40.26 0.44 0.33 0.41 0.25 0.22 0.65 0.43 0.25 0.75 0.37 1.6M20 0.27 0.65 0.67 0.71 1.12 0.73 M20 1.49 1.22 1.37 1.32 1.72 1.66 1.68 M20 1.88 2.40 1.61 1.77 1.51 1.62 1.74 M20 1.80.26 0.24 0.08 0.26 0.26 0.00 0.63 0.51 0.26 1.13 2.0M20 0.45 0.80 0.51 0.47 0.79 0.52 M20 1.98 1.41 1.50 1.21 1.56 1.59 1.76 M20 1.40 1.33 1.68 1.78 1.88 1.75 1.79 M20 2.30.27 0.58 0.07 0.24 0.37 0.00 0.61 0.67 0.37 0.91 0.46 2.5M20 0.09 0.27 0.31 0.14 0.25 0.24 M20 2.18 1.86 3.07 2.53 2.77 2.80 3.08 M20 1.84 2.23 2.12 2.22 2.08 2.24 2.27 M20 2.70.22 0.45 0.09 0.23 0.33 0.40 0.66 0.49 0.33 0.64 0.56 2.9M20 0.28 0.46 0.47 0.43 0.66 0.44 M20 0.78 0.68 0.89 1.05 0.83 0.75 0.93 M20 1.23 1.06 1.07 1.02 0.94 1.06 1.08 M20 3.2July 2015Connected to 20M at intersectionSlab #1 Slab #2 Slab #3[µA] [µA] [µA] 4 Results   144  Figure 4.46 Corrosion currents in µA of slab 1,2 and 3 measured during October 2015 (29 months exposure) The probability of successful coating damage/corrosion detection and its complement could be determined from the corrosion currents shown in Figure 4.44, Figure 4.45 and Figure 4.46. However,  the probability of erroneously identified damage locations by high corrosion currents with intact epoxy coating could not be calculated because the intact coating would need to be verified by rebar extraction. Once the coating and underlying rebar conditions are verified, more comprehensive statistical calculations should be performed in order to determine the reliability of detection of coating damage and corrosion by the LPR method.  4.3.1.2.1.2 Influencing Factors Figure 4.47 shows the measured currents of slab 2. The map on the left and right show the measurements performed on 10M and 20M rebar respectively. For each measurement the equipment was directly connected to the rebar under investigation. The measurements were performed during August 2013.  M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 0.14 0.15 0.15 0.22 0.16 0.12 M20 0.90 0.77 0.78 0.78 0.79 0.84 0.75 M20 0.90 0.93 0.84 0.84 0.80 0.93 0.99 M20 0.00.30 0.31 0.17 0.42 0.33 0.24 0.44 0.36 0.2M20 0.23 0.24 0.27 0.29 0.23 0.22 M20 0.31 0.30 0.32 0.33 0.30 0.29 0.41 M20 0.43 0.39 0.39 0.38 0.42 0.50 0.39 M20 0.50.40 0.27 0.06 0.23 0.30 0.16 0.46 0.24 0.20 0.38 0.33 0.7M20 0.38 0.38 0.37 0.36 0.37 0.36 M20 0.65 0.46 0.40 0.42 0.45 0.49 0.70 M20 0.64 0.70 0.73 0.76 0.69 0.75 0.70 M20 0.90.32 0.21 0.04 0.29 0.24 0.15 0.44 0.31 0.23 0.39 0.30 1.1M20 0.25 0.21 0.26 0.28 0.24 0.29 M20 0.50 0.49 0.53 0.55 0.51 0.59 0.64 M20 0.66 0.50 0.50 0.54 0.52 0.49 0.51 M20 1.40.42 0.32 0.05 0.27 0.30 0.23 0.52 0.30 0.32 0.38 0.24 1.6M20 0.66 0.68 0.75 0.71 0.62 0.78 M20 1.00 1.05 0.96 1.02 0.95 1.30 1.48 M20 1.00 0.97 0.84 1.11 0.92 0.90 1.65 M20 1.80.46 0.18 0.08 0.28 0.27 0.17 0.52 0.26 0.30 0.38 0.27 2.0M20 0.40 0.50 0.46 0.54 0.61 0.56 M20 1.19 1.13 1.04 1.14 1.03 1.36 1.54 M20 1.51 1.39 1.41 1.35 1.41 1.44 1.41 M20 2.30.47 0.29 0.09 0.28 0.27 0.16 0.43 0.25 0.36 0.36 0.29 2.5M20 0.34 0.17 0.16 0.19 0.17 0.15 M20 1.81 1.64 1.70 1.76 2.45 2.83 2.59 M20 1.75 1.64 1.57 1.73 1.75 1.69 1.91 M20 2.70.32 0.24 0.17 0.48 0.29 0.30 0.34 0.24 2.9M20 M20 0.70 0.95 0.67 0.76 0.77 0.88 0.78 M20 0.70 0.66 0.76 0.62 0.76 1.13 0.86 M20 3.2Slab #3[µA] [µA] [µA]October 2015Connected to 20M at intersectionSlab #1 Slab #2 4 Results   145   Figure 4.47:  Effect of rebar diameter on LPR measurements of slab #2 (August 2013) Lower currents could be observed for the 10M rebar in comparison to the 20M rebar. The median corrosion current of the 10M and 20M rebar was 15 µA and 23 µA respectively. The corrosion rate is proportional to the corrosion current normalized to the polarized area. Consequently, the larger corrosion currents of the 20M rebar were likely caused by a larger polarized area of the 20M rebar compared to the 10M rebar. However, the epoxy coating was not conductive which was confirmed by multimeter readings prior to casting. If only the part of the rebar with the damaged coating was polarized, the diameter of the underlying rebar would not be expected to influence the current reading. Consequently, the increased currents with increasing rebar diameter were likely due to water uptake of the epoxy or adhesion loss of the coating from its substrate or a combination thereof. However, the factor by which the corrosion currents of 20M were larger than 10M was not consistent and in some cases 20M rebar even resulted in lower corrosion currents (second 20M from the top).   Figure 4.48 shows the measured corrosion currents for slab 3. These measurements were performed during August 2013. All measurements were performed along the 20M rebars that were crossed by the 10M rebars in four locations. The values on the left and right map were obtained by connecting the LPR equipment to a 10M and a 20M rebar respectively.  10M 10M 10M 10M 10M 10M 10M 10M20M 5.31 11.13 11.17 12.59 20M 25.25 18.88 25.47 17.43 14.94 23.28 16.14 012.93 18.07 7.94 8.20 420M 16.17 16.98 16.65 10.10 20M 13.80 21.59 14.54 11.83 12.46 8.29 11.24 813.03 17.27 16.80 10.48 1220M 18.64 18.75 21.83 11.88 20M 13.47 24.42 26.15 33.86 34.95 7.38 17.76 1615.03 17.46 7.52 9.80 2020M 11.82 7.69 16.11 7.92 20M 37.89 26.26 23.08 28.97 14.58 29.78 17.34 2411.52 5.55 11.01 10.44 2820M 5.07 12.04 16.28 12.67 20M 38.63 17.50 16.46 13.84 33.09 14.64 28.71 329.33 15.94 14.11 13.35 3620M 15.36 18.47 17.87 22.56 20M 30.40 36.53 38.70 41.68 20.91 27.70 48.33 4018.99 11.92 17.98 18.50 4420M 21.72 18.73 23.96 23.96 20M 42.77 11.80 30.62 46.95 17.59 13.69 53.22 4815.59 24.05 34.32 28.73 5220M 18.72 17.10 15.67 12.70 20M 42.62 41.41 17.82 23.83 25.40 15.07 12.41 56Slab #2Connected to 10M Connected to 20M [µA] [µA] 4 Results   146  Figure 4.48: Effect of diameter of the rebar connection point on LPR measurements of slab #3 (August 2013) Figure 4.48 shows that significantly lower values were obtained when the equipment was connected to a 10M rebar compared to a 20M rebar. Connection to the 20M rebar delivered corrosion currents, that were on average twice as large as the ones obtained with the 10M connection. However, the increase in measured corrosion current was not consistent across the slab. In two isolated cases, the 20M currents were 10.5 and 7 times those of the 10M measurements. Otherwise the measurements with the 20M connections were up to four times larger than those of the 10M connection. Nonetheless, both rebar connections led to the detection of the same amount of coating damage locations. Furthermore, most of the successfully detected locations had been detected during both sets of measurements. Overall, regions of low and high corrosion currents were similar for the two data sets. However, the 20M connections led to higher standard deviations of the measured corrosion current, which could be an indicator of corrosion activity.  Consequently, interpretation of measurements and identification of corroded areas could be challenging. Furthermore, this led to the conclusion that the LPR equipment should always be connected to the rebar under investigation. However, this is not practical in the field because every rebar being investigated would need to be exposed.   Figure 4.49 shows the currents measured during August 2013 for slab 1, 2 and 3.  As expected the slab without any chloride contamination (slab 1) shows significantly lower currents in comparison to the slabs with admixed NaCl. However, even though slab 3 had a higher chloride concentration 10M 10M 10M 10M 10M 10M 10M 10M20M 2.38 11.61 2.65 11.34 7.06 4.39 9.33 20M 6.03 8.72 10.24 12.06 15.61 9.22 11.56 0420M 5.48 4.34 10.05 6.60 12.41 17.58 15.58 20M 11.50 8.49 19.78 10.50 24.09 32.19 43.73 81220M 11.76 7.55 2.95 13.21 10.71 5.71 6.67 20M 23.13 25.51 31.05 31.82 31.80 16.46 9.54 162020M 6.81 15.01 8.56 6.12 13.34 16.96 9.64 20M 12.81 27.81 22.56 43.34 27.09 43.93 37.47 242820M 2.94 4.09 9.11 8.66 7.32 12.70 11.23 20M 9.91 9.52 12.10 23.10 9.62 20.10 10.49 323620M 5.66 5.98 14.32 6.37 11.41 10.19 16.12 20M 12.44 13.12 12.91 23.70 14.49 15.36 18.67 404420M 12.57 10.42 9.08 16.73 15.64 12.09 6.36 20M 12.23 13.49 13.09 20.02 12.13 12.03 11.53 485220M 12.36 11.04 4.76 2.75 2.99 2.11 2.89 20M 20.96 15.45 8.01 7.66 9.19 8.74 7.42 56Slab #3Connected to 10M Connected to 20M [µA] [µA] 4 Results   147 compared to slab 2, it did not lead to higher currents. In fact the values obtained for slab 2 tended to be larger compared to slab 3. Overall the values measured on both contaminated slabs were very similar.   Figure 4.49: Effect of chloride concentration on LPR measurements (August 2013) Even though the locations of the damaged coating could not reliably be identified from the LPR measurements, the slabs containing chloride showed much higher readings. This poses the question whether the LPR measurements are an indicator for the corrosiveness of the medium or the actual corrosion rate of the rebar. Figure 4.50 shows the corrosion current of slab 3, measured in July and October 2015 respectively. The difference in environmental temperature between the testing dates was 6.4 °C. The higher temperature during July resulted in higher corrosion current measurements compared to the October measurements.   M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 2.09 2.37 3.25 3.31 3.63 3.87 3.24 M20 25.25 18.88 25.47 17.43 14.94 23.28 16.14 M20 6.03 8.72 10.24 12.06 15.61 9.22 11.56 M20 0.01.41 1.17 0.35 2.61 12.93 18.07 7.94 8.20 10.97 11.34 5.69 9.40 4.0M20 2.56 2.14 4.34 4.60 4.33 3.81 3.54 M20 13.80 21.59 14.54 11.83 12.46 8.29 11.24 M20 11.50 8.49 19.78 10.50 24.09 32.19 43.73 M20 8.01.70 0.98 2.80 13.03 17.27 16.80 10.48 8.50 15.56 15.86 10.43 12.0M20 2.73 5.18 6.40 5.27 6.60 5.23 6.59 M20 13.47 24.42 26.15 33.86 34.95 7.38 17.76 M20 23.13 25.51 31.05 31.82 31.80 16.46 9.54 M20 16.02.14 1.47 2.42 15.03 17.46 7.52 9.80 9.74 5.44 11.64 9.65 20.0M20 1.73 2.07 2.47 4.96 2.78 5.90 5.85 M20 37.89 26.26 23.08 28.97 14.58 29.78 17.34 M20 12.81 27.81 22.56 43.34 27.09 43.93 37.47 M20 24.02.28 1.32 3.43 11.52 5.55 11.01 10.44 8.46 25.27 8.40 17.82 28.0M20 8.78 5.27 3.80 4.02 3.56 6.42 5.22 M20 38.63 17.50 16.46 13.84 33.09 14.64 28.71 M20 9.91 9.52 12.10 23.10 9.62 20.10 10.49 M20 32.02.00 0.96 3.20 9.33 15.94 14.11 13.35 9.08 13.63 11.24 16.89 36.0M20 0.78 0.43 1.16 1.56 2.02 2.11 1.91 M20 30.40 36.53 38.70 41.68 20.91 27.70 48.33 M20 12.44 13.12 12.91 23.70 14.49 15.36 18.67 M20 40.01.77 0.93 1.91 18.99 11.92 17.98 18.50 15.16 14.88 7.43 14.30 44.0M20 0.43 0.80 0.51 0.82 0.75 1.03 0.93 M20 42.77 11.80 30.62 46.95 17.59 13.69 53.22 M20 12.23 13.49 13.09 20.02 12.13 12.03 11.53 M20 48.01.00 0.94 0.22 15.59 24.05 34.32 28.73 10.02 11.48 6.26 8.08 52.0M20 1.35 1.47 1.53 1.61 1.77 1.70 1.74 M20 42.62 41.41 17.82 23.83 25.40 15.07 12.41 M20 20.96 15.45 8.01 7.66 9.19 8.74 7.42 M20 56.0[µA] [µA]August 2013[µA]Slab #1 Slab #2 Slab #3Connected to 20M at intersection 4 Results   148  Figure 4.50: Effect of temperature on LPR measurements However, the difference was not significant and the same areas of high and low corrosion currents could be identified in both data sets.  Figure 4.51 shows the measured currents for slab 3 under wet and dry conditions. The measurements were performed in January 2014.    Figure 4.51: Effect of moisture on LPR measurements of slab 3 (January 2014) All measured values were very low under wet as well as dry conditions. However, the values measured under dry conditions were consistently higher in comparison to wet conditions. This was unexpected because a high moisture content of the concrete generally leads to a high corrosion rate M10 M10 M10 M10 M10 M10 M10 M10M20 1.07 0.98 1.02 0.99 1.10 1.02 0.93 M20 0.90 0.93 0.84 0.84 0.80 0.93 0.99 0.00.38 0.21 0.67 0.38 0.33 0.24 0.44 0.36 0.2M20 0.71 0.49 0.45 0.42 0.43 0.44 0.50 M20 0.43 0.39 0.39 0.38 0.42 0.50 0.39 0.40.42 0.57 0.40 0.24 0.20 0.38 0.33 0.5M20 1.06 1.03 0.91 1.13 1.02 0.95 0.99 M20 0.64 0.70 0.73 0.76 0.69 0.75 0.70 0.70.55 0.26 0.66 0.34 0.31 0.23 0.39 0.30 0.9M20 0.96 0.97 0.95 0.78 0.71 0.76 0.69 M20 0.66 0.50 0.50 0.54 0.52 0.49 0.51 1.10.43 0.25 0.75 0.37 0.30 0.32 0.38 0.24 1.2M20 1.88 2.40 1.61 1.77 1.51 1.62 1.74 M20 1.00 0.97 0.84 1.11 0.92 0.90 1.65 1.40.51 0.26 1.13 0.00 0.26 0.30 0.38 0.27 1.6M20 1.40 1.33 1.68 1.78 1.88 1.75 1.79 M20 1.51 1.39 1.41 1.35 1.41 1.44 1.41 1.80.67 0.37 0.91 0.46 0.25 0.36 0.36 0.29 1.9M20 1.84 2.23 2.12 2.22 2.08 2.24 2.27 M20 1.75 1.64 1.57 1.73 1.75 1.69 1.91 2.10.49 0.33 0.64 0.56 0.29 0.30 0.34 0.24 2.3M20 1.23 1.06 1.07 1.02 0.94 1.06 1.08 M20 0.70 0.66 0.76 0.62 0.76 1.13 0.86 2.5October 2015 16.4 °C[µA]Connected to 20M at intersectionSlab #3[µA]July 2015 22.8 °C10M 10M 10M 10M20M 2.825 3.616 3.064 20M 1.572 1.461 1.026 0.01.575 3.434 0.604 0.581 0.520M 1.862 2.109 5.059 20M 1.398 1.039 1.02.335 3.353 0.680 0.466 1.520M 3.747 5.317 4.299 20M 1.712 1.770 2.054 2.01.461 2.886 0.585 0.283 2.520M 4.041 4.590 3.084 20M 1.659 1.696 1.519 3.01.437 2.681 0.614 0.291 3.520M 1.784 3.559 2.840 20M 2.416 2.637 2.416 4.02.118 1.554 0.752 0.308 4.520M 3.909 2.059 2.013 20M 1.739 1.737 1.659 5.01.880 2.330 0.669 0.300 5.520M 3.303 4.873 6.647 20M 3.419 3.502 3.878 6.01.970 0.903 0.754 0.287 6.520M 3.909 2.825 2.124 20M 3.515 3.747 3.940 7.0[µA] [µA]Dry Wet 4 Results   149 and thus large measured currents. A possible explanation could be that the IR compensation applied by the LPR equipment was not very accurate. However, turning the IR compensation off led to a scatter of measurement points rather than a curve. Consequently, corrosion currents could not be obtained from measurements without IR compensation.    4.3.1.2.2 Half Cell Colour coded Half-Cell potential maps of slabs 1, 2 and 3 are shown in Figure 4.52 through Figure 4.56. The potential measurements are shown in mV with respect to the saturated copper electrode (CSE) and the colour coding is explained in the column on the far right next to each map.  The locations, where the epoxy coating had been damaged during handling of the rebar, can be identified by the underlined and italic font. Boxed values indicate location, at which the epoxy coating had ben purposely damaged with a file. Slab 2 and slab 3 had 8 kg/m3 and 11 kg/m3 of NaCl admixed to the concrete respectively (4.8 kg/m3 and 6.6 kg/m3 of admixed Cl-). Furthermore, starting September 2013 slab 2 and 3 had been exposed to 7 day water ponding periods on a bi-weekly basis (chapter 3.3.1.2.1). Slab 1 served as the reference without any admixed sodium chloride or water ponding. 4.3.1.2.2.1 Localized Corrosion Detection Figure 4.52 and Figure 4.53 show the potential measurements from November 2013 and July 2015 respectively. Both the largest and lowest potentials were observed for the salt free slab 1 regardless of the testing day. Furthermore, both maps show very similar potentials along each rebar and localized corrosion or coating damage did not lead to a localized change in the open circuit potential.   4 Results   150  Figure 4.52: Open circuit potential in mVCSE of slab 1,2 and 3 measured during November 2013 (6 months exposure) No consistent trend could be observed between the measurements obtained in November 2013 and July 2015. Dependent on the rebar being investigated, the values either increased or decreased over time. Both trends could be observed within each slab. Furthermore, both chloride contaminated slabs resulted in very similar potential readings and a smaller potential range in comparison to slab 1. M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 -245 -230 -240 -225 -225 -205 -250 M20 -280 -295 -295 -300 -305 -300 -305 M20 -325 -315 -315 -300 -310 -305 -290 M20 -100-265 -340 -130 -475 -265 -270 -350 -345 -255 -240 -200 -170 -140M20 -170 -160 -165 -160 -160 -155 -180 M20 -290 -295 -300 -295 -305 -300 -300 M20 -315 -315 -315 -310 -295 -260 -270 M20 -180-255 -340 -150 -485 -275 -280 -355 -360 -260 -230 -200 -170 -220M20 -285 -280 -285 -290 -290 -270 -285 M20 -230 -230 -235 -220 -225 -230 -235 M20 -245 -240 -230 -230 -225 -245 -245 M20 -260-260 -345 -150 -480 -275 -280 -340 -345 -250 -225 -200 -210 -300M20 -635 -615 -615 -635 -610 -605 -610 M20 -230 -240 -240 -235 -230 -235 -235 M20 -245 -240 -225 -215 -240 -240 -245 M20 -340-270 -340 -135 -465 -285 -280 -350 -355 -250 -230 -195 -200 -380M20 -630 -610 -620 -615 -620 -610 -615 M20 -320 -320 -320 -320 -315 -315 -310 M20 -325 -320 -315 -310 -305 -315 -320 M20 -420-270 -345 -145 -475 -290 -285 -355 -345 -245 -235 -200 -185 -460M20 -340 -335 -300 -300 -300 -285 -300 M20 -325 -320 -325 -320 -315 -315 -285 M20 -325 -315 -315 -315 -315 -320 -280 M20 -500-265 -325 -130 -470 -285 -290 -345 -330 -220 -220 -195 -170 -540M20 -485 -470 -450 -440 -450 -430 -455 M20 -355 -355 -355 -350 -340 -325 -310 M20 -345 -325 -300 -300 -315 -340 -345 M20 -580-280 -330 -110 -480 -285 -280 -310 -325 -240 -215 -195 -190 -620M20 -355 -350 -345 -350 -305 -320 -330 M20 -340 -345 -350 -355 -355 -355 -345 M20 -660November 2013[mV vs. CSE] [mV vs. CSE]Slab #2 Slab #3Connected to 20M at intersectionSlab #1[mV vs. CSE] 4 Results   151  Figure 4.53: Open circuit potential in mVCSE of slab 1,2 and 3 measured during July 2015 (26 months exposure) 4.3.1.2.2.2 Influencing factors Figure 4.54 shows the measured potentials for slab 3. The map on the left and right was obtained by connecting the HC equipment to 10M and 20M rebars respectively. All measurements were performed during August 2013.  The values in the left map are noticeably lower compared to the map on the right. Measurements that were performed with the equipment connected to a 10M rebar led to potentials that were up to 100 mV more positive in comparison to the same measurement with a 20M rebar.  M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10 M10M20 -85 -85 -90 -100 -100 -85 -115 M20 -230 -225 -225 -230 -235 -225 -230 M20 -260 -260 -265 -260 -255 -255 -245 M20 -50-190 -170 -200 -130 -250 -225 -200 -270 -235 -210 -250 -310 -80M20 -165 -175 -175 -185 -180 -165 -170 M20 -210 -210 -220 -225 -230 -215 -225 M20 -250 -255 -255 -240 -240 -215 -215 M20 -110-200 -175 -205 -155 -235 -215 -205 -275 -240 -205 -230 -305 -140M20 -130 -135 -140 -135 -150 -145 -155 M20 -175 -185 -185 -185 -190 -185 -200 M20 -220 -210 -200 -200 -190 -215 -210 M20 -170-200 -170 -220 -160 -235 -230 -195 -265 -230 -190 -230 -335 -200M20 -210 -205 -205 -205 -215 -215 -215 M20 -180 -195 -205 -200 -200 -205 -215 M20 -220 -220 -205 -200 -210 -215 -215 M20 -230-195 -160 -200 -160 -250 -235 -195 -265 -235 -190 -225 -325 -260M20 -455 -455 -450 -455 -450 -455 -465 M20 -165 -180 -190 -185 -190 -195 -200 M20 -215 -210 -205 -195 -195 -190 -195 M20 -290-190 -165 -220 -140 -260 -235 -210 -285 -215 -190 -230 -315 -320M20 -375 -370 -370 -375 -370 -370 -370 M20 -315 -320 -320 -315 -320 -320 -305 M20 -315 -315 -310 -315 -310 -315 -300 M20 -350-205 -180 -225 -150 -255 -245 -215 -265 -205 -180 -225 -315 -380M20 -155 -160 -160 -160 -160 -165 -170 M20 -350 -350 -365 -360 -365 -355 -345 M20 -360 -360 -340 -340 -340 -340 -340 M20 -410-205 -195 -260 -170 -255 -235 -205 -270 -215 -185 -220 -310 -440M20 -315 -320 -315 -310 -305 -305 -310 M20 -205 -200 -195 -200 -185 -195 -200 M20 -180 -185 -190 -190 -180 -195 -205 M20 -470July 2015Connected to 20M at intersectionSlab #1 Slab #2 Slab #3[mV vs. CSE] [mV vs. CSE] [mV vs. CSE] 4 Results   152  Figure 4.54: Effect of diameter of the rebar connection point on HC measurements of slab #3 This meant that the diameter of the rebar the equipment was connected to affected the magnitude of the measured potential. However, the difference in potentials between measurements was similar for both maps. Therefore, low potential areas in the left map were also identified as relatively low potential areas in the right map. This was also the case for high potential areas. Thus, consistency of the diameter of the rebar the half cell equipment is connected to, is required in order to be able to identify areas of low or high potential.  Figure 4.55 shows the potential maps for slab 3 under wet and dry conditions. For each measurement the equipment was connected to the rebar under investigation. At rebar intersections, the equipment was connected to a 20M rebar.  10M 10M 10M 10M 10M 10M 10M 10M-305 -295 -240 -210 -195 -200 -260 -240 -260 -380 -370 -330 -285 -275 -290 -345 -315 -335 -10020M -280 -255 -225 -260 -185 -230 -190 -165 -220 20M -365 -340 -320 -340 -280 -325 -285 -260 -295 20M -120-270 -225 -245 -245 -225 -200 -130 -165 -155 -350 -310 -325 -320 -315 -280 -215 -245 -235 -14020M -305 -250 -260 -235 -235 -170 -130 -135 -150 20M -380 -330 -335 -310 -310 -255 -230 -245 -245 20M -160-330 -300 -255 -230 -205 -215 -170 -155 -195 -395 -370 -335 -310 -285 -290 -265 -240 -285 -18020M -255 -250 -235 -180 -190 -170 -250 -260 -335 20M -325 -320 -300 -260 -265 -255 -330 -335 -405 20M -200-280 -220 -225 -205 -180 -220 -230 -295 -325 -345 -310 -300 -285 -255 -300 -315 -365 -390 -22020M -260 -255 -245 -200 -175 -240 -245 -275 -330 20M -345 -325 -315 -275 -255 -325 -325 -350 -410 20M -240-290 -285 -305 -220 -175 -225 -260 -290 -350 -350 -350 -370 -305 -245 -300 -335 -355 -415 -26020M -300 -280 -300 -255 -225 -245 -255 -300 -340 20M -375 -355 -360 -335 -300 -320 -325 -375 -405 20M -280-275 -280 -230 -265 -265 -275 -270 -220 -315 -335 -335 -310 -330 -335 -345 -340 -270 -370 -30020M -275 -270 -235 -255 -260 -265 -305 -145 -270 20M -345 -345 -300 -335 -335 -340 -370 -230 -355 20M -320-245 -210 -230 -240 -245 -275 -260 -195 -315 -315 -275 -305 -320 -315 -340 -325 -265 -365 -34020M -235 -240 -245 -170 -180 -170 -255 -305 -310 20M -315 -310 -300 -235 -250 -245 -320 -375 -370 20M -360-270 -265 -260 -195 -230 -260 -280 -285 -205 -340 -330 -330 -270 -305 -335 -345 -345 -280 -38020M -280 -265 -275 -290 -280 -290 -315 -280 -265 20M -345 -320 -345 -355 -345 -355 -375 -340 -325 20M -400-275 -285 -280 -305 -295 -295 -315 -285 -310 -345 -345 -335 -365 -360 -360 -375 -340 -370 -420[mV vs. CSE] [mV vs. CSE]Slab #3Connected to one rebar cycle10M 20M 4 Results   153   Figure 4.55: Effect of moisture conditions on HC measurements of slab #3 It can be observed that the potentials along one rebar were very similar. This became particularly apparent under wet conditions. The measured values were very similar for both moisture conditions and no particular trend could be observed.  Figure 4.56 shows the potential maps for slab 2 and 3. For each measurement the equipment was connected to the rebar under investigation. At rebar intersections, the equipment was connected to a 20M rebar.  10M 10M 10M 10M 10M 10M 10M 10M-265 -235 -200 -210 -280 -200 -205 -230 -10020M -325 -325 -315 -315 -300 -310 -305 -290 -300 20M -310 -295 -255 -305 -220 -275 -240 -210 -250 20M -120-255 -240 -200 -170 -215 -235 -190 -125 -14020M -325 -315 -315 -315 -310 -295 -260 -270 -275 20M -340 -285 -285 -275 -250 -210 -165 -190 -180 20M -160-260 -230 -200 -170 -265 -210 -185 -145 -18020M -250 -245 -240 -230 -230 -225 -245 -245 -250 20M -255 -270 -245 -205 -220 -195 -250 -280 -345 20M -200-250 -225 -200 -210 -210 -185 -205 -280 -22020M -245 -245 -240 -225 -215 -240 -240 -245 -255 20M -260 -270 -250 -210 -180 -265 -250 -285 -335 20M -240-250 -230 -195 -200 -260 -190 -210 -270 -26020M -325 -325 -320 -315 -310 -305 -315 -320 -320 20M -345 -330 -315 -290 -255 -280 -265 -335 -375 20M -280-245 -235 -200 -185 -240 -230 -255 -160 -30020M -320 -325 -315 -315 -315 -315 -320 -280 -295 20M -330 -325 -280 -320 -315 -325 -350 -210 -280 20M -320-220 -220 -195 -170 -185 -220 -250 -165 -34020M -345 -345 -325 -300 -300 -315 -340 -345 -340 20M -305 -315 -305 -235 -225 -250 -310 -365 -350 20M -360-240 -215 -195 -190 -225 -215 -235 -270 -38020M -345 -340 -345 -350 -355 -355 -355 -345 -340 20M -325 -340 -375 -350 -345 -355 -340 -315 -345 20M -400-245 -245 -200 -195 -245 -280 -265 -265 -420[mV vs. CSE] [mV vs. CSE]DryWetConnected to 20M at intersectionsSlab #3 4 Results   154  Figure 4.56: Effect of chloride concentration on HC measurements The 10M rebars tended to have more positive potentials in the slab with the higher chloride content (slab 3). However, the 20M rebars tended to have more negative potentials in slab 3. Overall the differences in potential were small with some exceptions, where a difference of 100 mV was recorded. Locations of coating damage could not be identified reliably by these potential maps. 4.3.1.2.3 Ground Penetrating Radar The Ground Penetrating Radar (GPR) equipment was used as a measure for the amount of corrosion. Due to restrictions in the available software, only the B-scan and A-scan could be analyzed. The B-scan offered fast analysis of large amounts of data, whereas the A-scan offered more quantitative analysis of small amounts of data.  Figure 4.57, Figure 4.58 and Figure 4.59 show GPR B-scans of slab 1, 2 and 3 respectively. Scan a, b, c and d were obtained on August 16th 2013, November 1st 2013, January 13th 2015 and October 20th 2015 respectively.  10M 10M 10M 10M 10M 10M 10M 10M-240 -210 -270 -280 -265 -235 -200 -210 -10020M -310 -210 -275 -265 -275 -300 -275 -280 -320 20M -325 -325 -315 -315 -300 -310 -305 -290 -300 20M -120-230 -230 -275 -235 -255 -240 -200 -170 -14020M -325 -265 -255 -285 -240 -290 -280 -270 -300 20M -325 -315 -315 -315 -310 -295 -260 -270 -275 20M -160-295 -275 -295 -300 -260 -230 -200 -170 -18020M -270 -260 -255 -270 -215 -220 -225 -250 -290 20M -250 -245 -240 -230 -230 -225 -245 -245 -250 20M -200-255 -260 -240 -270 -250 -225 -200 -210 -22020M -245 -215 -275 -275 -255 -240 -265 -285 -280 20M -245 -245 -240 -225 -215 -240 -240 -245 -255 20M -240-245 -275 -295 -290 -250 -230 -195 -200 -26020M -300 -295 -315 -315 -315 -290 -295 -305 -280 20M -325 -325 -320 -315 -310 -305 -315 -320 -320 20M -280-310 -280 -305 -285 -245 -235 -200 -185 -30020M -320 -340 -325 -325 -305 -305 -290 -180 -290 20M -320 -325 -315 -315 -315 -315 -320 -280 -295 20M -320-310 -315 -260 -230 -220 -220 -195 -170 -34020M -355 -355 -340 -325 -325 -320 -270 -245 -300 20M -345 -345 -325 -300 -300 -315 -340 -345 -340 20M -360-280 -260 -195 -215 -240 -215 -195 -190 -38020M -355 -345 -325 -270 -350 -240 -290 -285 -345 20M -345 -340 -345 -350 -355 -355 -355 -345 -340 20M -400-295 -275 -230 -295 -245 -245 -200 -195 -420[mV vs. CSE] [mV vs. CSE]Slab #2Connected to 20M at intersectionsSlab #3 4 Results   155   Figure 4.57: GPR scan of slab #1 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d)  Figure 4.58: GPR scan of slab #2 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d)  Figure 4.59: GPR scan of slab#3 on August 16, 2013 (a), November 1, 2013 (b) and January 13, 2015 (c) and October 20, 2015 (d)  4 Results   156 While the scans for the chloride free slab 1 did not notably change over time except for the scan d, the images of slab 2 and 3 increased in clarity over time. This increase in clarity was particularly pronounced for slab 2. Increased sharpness or clarity of the B-scan signifies large differences in dielectric properties, which is enhanced under dry conditions. Considering, that slab 1 only led to a notably sharper B-scan during the last measurement, the latter was likely caused by a reduced concrete moisture content compared to earlier measurements. Consequently, the enhanced clarity of scan d shown in Figure 4.58 and Figure 4.59 were likely due to a lower moisture content of slabs 2 and 3 on October 20, 2015 compared to earlier testing days.  Nonetheless, the sharpness of the B-scans improved over time for both chloride contaminated slabs between August 2013 and January 2015 even though the scans of the chloride free slabs did not change notably during that time frame. Thus, the difference in dielectric properties at the reflection interfaces increased over time for slabs 2 and 3 but remained the same for slab 1.  The reduction of the chloride content above the rebar over time would explain the improving sharpness of the B-scan. A reduction of the chloride content leads to a reduction of the conductivity of concrete and thus a larger difference in dielectric properties between the steel and the surrounding concrete. However, while a reduction of the chloride content of the surface layer might have been caused by the water ponding and draining cycles, the chloride contamination at the rebar depth was unlikely to have notably changed over time.   Even though the initial stages of rebar corrosion have been reported to have led to clearer B-scans and advanced corrosion to blurrier images, the latter observations had been made for uncoated rebar (Strategic Highway Research Program - SHRP 2 2013; Eisenmann et al. 2013). Advanced corrosion of ECR has been reported to have led to effects on the GPR signal opposing those of UCR of the same degree of corrosion (Eisenmann et al. 2013). The consequences of corrosion of ECR on the reinforced concrete can differ significantly from those of UCR. The epoxy coating can lead to reduced amounts of corrosion products diffusing into the surrounding concrete, reduced concrete cracking as well as increased density of the corrosion products. A reduction in the amount and size of concrete cracks would lead to clearer B-scans, while the reduction of the amount of corrosion products diffusing into the concrete would have the opposite effect (Lai, Kind, and Wiggenhauser 2010). The impact of increased density of the corrosion products on the GPR signal is unclear.  Even though the B-scan is created from quantitative data such as wave amplitude, arrival time and frequency, the analysis and interpretation of the B-scan is very subjective. Thus, further analysis  4 Results   157 in the form of the A-scan, which relates the signal voltage to the arrival time, was performed. The A-scan represents a vertical slice of the B-scan. Figure 4.60 and Figure 4.61 show the A-scan for a 20M rebar, that was not suspected to be corroding, in slab 1, 2 and 3 in yellow, green and blue respectively. The black curve represents a scan taken at a location of slab 1 without any rebar. The scans in Figure 4.60 and Figure 4.61 were taken in November 2013 and October 2015 respectively.  Figure 4.60: GPR waves of intact ECR in slab 1, 2 and 3 obtained in 2013 The second negative peak and the third positive peak show the wave reflected by the rebar (RW). Thus, these peaks could not be observed for the black curves. Small extrema of the black curve were the result of the inhomogeneity of the concrete. The difference in voltage between the two peaks of the RW is referred to as the peak to peak voltage Vp-p. Figure 4.60 and Figure 4.61 show a significantly larger Vp-p for slab 1 in comparison to the chloride contaminated slabs. The smallest peak to peak voltage was observed for slab 2 in 2013 as well as in 2015.   4 Results   158  Figure 4.61: GPR waves of intact ECR in slab 1, 2 and 3 obtained in 2015 Furthermore, the very first peak of the curves signified the direct wave (DW). One could observe a larger amplitude of the DW for the chloride free slab and similar amplitudes for the chloride contaminated slabs. The difference in arrival time between the RW and the DW was the shortest for slab 1 and the largest for slab 2.  This observation was made with the data obtained in 2013 as well as 2015. Based on the LPR measurements shown in Figure 4.44, five locations on slab 2 with intact and damaged coating as well as low and high corrosion current readings were selected. The resulting A-scans are shown in Figure 4.62 and the colour coding of the curves is explained in Table 4.13. The black curves represent a location on slab 2 without any rebar. Furthermore, a zoomed in version of the diagram in Figure 4.62 is shown in Figure 4.63.  Table 4.13: Colour coding of GPR waves Colour Coating condition Corrosion current  Intact Low  Damage Low  Damage High  Intact High  4 Results   159   Figure 4.62: GPR waves of slab 2 obtained in August 2013 The largest DW amplitudes were observed for the locations, that had resulted in low corrosion current readings. Furthermore, an intact coating led to the largest RW amplitudes and thus Vp-p. The longest time difference between the arrival of the DW and RW was observed for the bar location with coating damage and high corrosion currents.   4 Results   160  Figure 4.63: Zoomed in diagrams of GPR waves of slab 2 obtained in August 2013  The A-scans of the same locations obtained in January 2015 are shown in Figure 4.64 and a zoomed in version of the same diagram in Figure 4.65. None of the locations had resulted in a spike in corrosion current in 2015. All scan locations with ECR resulted in DW amplitudes larger than the scan over the rebar free section. This was in line with the August 2013 observations of an increasing DW amplitude with decreasing corrosion current. Furthermore, the corrosion currents obtained in 2015 (Figure 4.45, Figure 4.46) were significantly smaller compared to the 2013 measurements (Figure 4.44) and the presence of ECR led to a larger increase in DW amplitude in 2015 compared to 2013.   4 Results   161  Figure 4.64: GPR waves of slab 2 obtained in January 2015 Similar to the 2013 observations, the locations with intact epoxy coating led to the largest RW amplitudes. The longest arrival time difference between the DW and RW were observed for the location with coating damage, that had led to a spike in corrosion current in 2013.   4 Results   162  Figure 4.65: Zoomed in diagram of GPR waves of slab 2 obtained in January 2015  The peak to peak voltages of various locations on slab 1, 2 and 3 for August 2013 and January 2015 are shown in Table 4.14. The voltages were obtained by the difference between the negative and positive peak of the RW and subtraction of the equivalent rebar free value. The latter subtraction served the purpose of minimizing the influence of the concrete on the measurements.  The chloride free slab led to the largest peak to peak voltages both in 2013 and 2015 and slab 2 to the lowest Vp-p. Furthermore, the voltage increased from 2013 to 2015 for all three slabs. The damaged coating in slab 2 and high corrosion currents in slab 3 coincided with lower Vp-p in 2013. This trend of slab 2 was also observed in 2015 but only the location with coating damage and high corrosion current in slab 3 led to a reduced Vp-p in 2015. The latter observation of slab 3 is likely linked to a reduction in corrosion current, that was observed between 2013 and 2015. Thus, a trend of a lower Vp-p with higher corrosion currents could not be observed in 2015 due to the lack of high currents. The combination of coating damage and a high current coincided with a reduced Vp-p for all slabs regardless of the testing date.   4 Results   163 Table 4.14: Vp-p of the reflected wave in mV Coating condition Corrosion current in 2013 2013 2015 S1 S2 S3 S1 S2 S3 Intact Low 6721 2574 4302 12392 5941 6373 Intact High - 2455 2920 - 6275 6529 Damaged High 6069 1703 3808 11451 4804 1882 Damaged Low - 1188 4500 - 3569 6490  Consequently, a high corrosion current as well as damage of the epoxy coating were linked to a reduction in the Vp-p of the RW. However, the presence of chloride as well as concrete maturity had a large impact on the peak to peak voltage despite attempts to minimize the influence of concrete.  The amplitude of the DW is shown in Table 4.15 for various locations on slab 1, 2 and 3. The measurements had been performed in August 2013 and January 2015. A large increase in the DW amplitude was observed for slab 1 between 2013 and 2015. However, even though the values increased over time for slab 2 and 3, the difference between the 2015 and 2013 measurements of slab 2 and 3 was much smaller compared to slab 1. The locations of high corrosion current of slab 2 coincided with lower DW amplitudes in 2013 as well as 2015. However, this observation could only be made for slab 2. Table 4.15: Peak amplitude of the direct wave in mV Coating condition Corrosion current in 2013 2013 2015 S1 S2 S3 S1 S2 S3 Intact Low 7564 9050 8053 10784 9196 8941 Intact High - 8119 8270 -  8922 9020 Damaged High 7993 8455 7914 10549 9039 9333 Damaged Low -  8911 8191 -  9608 8863  The delay in arrival time between the DW and the RW is shown in Figure 4.16. The largest delays were observed for slab 2 regardless of the testing day. Furthermore, the location with both coating damage and a high corrosion current led to the longest delay of the RW. Furthermore, the delays increased over time for all measurement locations of slab 2 but decreased for slab 1 and 3.     4 Results   164 Table 4.16: Delay in arrival time between the DW and RW in ns Coating condition Corrosion current in 2013 2013 2015 S1 S2 S3 S1 S2 S3 Intact Low 1.68 1.78 1.98 1.66 2.09 1.91 Intact High - 1.92 1.85 -  2.04 1.81 Damaged High 1.81 1.99 2.00 1.75 2.13 2.00 Damaged Low -  1.95 1.88 -  2.08 1.81  4.3.1.2.4 Discussion Localized corrosion or coating damage could not be detected with the Half-Cell method. Furthermore, no consistent trend of the measured potentials could be observed. The largest range of potentials were observed for the salt free slab 1, which is similar to the findings of Saricimen et al. (Saricimen et al. 1997), who observed the largest drop in open circuit potential for ECR with intact epoxy coating once corrosion activity had initiated. Furthermore, the potential range of the reference slab was similar to the potential range for ECR with intact coating observed by Saricimen et al. (Saricimen et al. 1997). However, it should be noted that the magnitude of the measured potential increased with increasing diameter of the rebar the equipment was connected to. Compared to the rebar diameter, the moisture content and chloride concentration had a negligible effect on the readings.  Up to 54.5% of coating defects could be detected in chloride contaminated concrete by high corrosion currents using LPR methods and up to 37.5% in salt free concrete. However, chloride contamination as well as a reduction in moisture content led to an overall increase of the current readings regardless of the presence of coating holidays. Chloride contamination and moisture reduction increase and decrease the conductivity of concrete respectively. Thus, the origin of the aforementioned observations is unclear. The diameter of the rebar the potentiostat was connected to as well as the size of the rebar under investigation increased the corrosion currents. The environmental temperature had a comparably small effect on the measurements.  Locations of high corrosion current readings as well as locations of coating holidays resulted in a reduction of the peak to peak voltage of the RW. Furthermore, the longest delays in arrival time between the RW and DW were observed for locations of both coating damage and high corrosion currents. Wiggenhauser et al. (Lai, Kind, and Wiggenhauser 2010) reported similar observations for advanced corrosion of uncoated rebar. Chloride contamination greatly reduced the amplitude and delayed the arrival time of the RW. These observations are consistent with the findings of Helmerich  4 Results   165 et al. (Hong, Lai, and Helmerich 2012). Furthermore, increasing concrete maturity led to higher amplitudes of the RW and DW regardless of the chloride concentration.  4.3.2 Hall-Effect and Thermography Induction thermography is a potential non-destructive corrosion detection method that consists of heating the embedded rebar through induction and monitoring the thermal response from the concrete surface (chapter 2.4.2). However, this NDT method is affected by both the thermal and magnetic properties of the rebar under investigation. In order to study the effect of corrosion on the magnetic and thermal properties separately, Hall effect and thermal experiments were performed.  4.3.2.1 Hall-Effect Samples The two objectives of this study were to assess the suitability of Hall effect sensors as a non-destructive corrosion detection method as well as determining the effect of corrosion on electromagnetic induction.  Hall effect measurements were performed on microalloyed steel bars (SB), uncoated rebar (UCR) and epoxy coated rebar (ECR) (chapter 3.3.2.1.1). A rotating rare earth magnet positioned near the end of the bar under investigation was employed as the source of the magnetic flux. A Hall effect sensor was then moved along the bar away from the magnet to record the Hall voltage at multiple locations (chapter 3.3.2.2.1).  Microalloyed steel is ferromagnetic and the change in the Hall effect resulting from the presence of the steel bars (i.e. SB, UCR, ECR) is shown on the ordinate in Figure 4.66 through Figure 4.104. The distance between sensor and magnet is shown on the abscissa in Figure 4.66 through Figure 4.104. 4.3.2.1.1 PVC 4.3.2.1.1.1 PVC - Steel Bars Figure 4.66 through Figure 4.69 show the change in the Hall effect (HE) voltage at four different frequencies for the steel bar (SB) specimens. The space holder, that ensured a consistent distance of 2.5 cm between the bar under investigation and the sensor-magnet setup, consisted of a 2.5 cm deep PVC block.  No relation between the bar diameter and the change in HE voltage could be observed for small sensor magnet spacings of 2.5 cm and 5 cm. Thus, the change in HE voltage was independent of the bar diameter at spacings up to 5 cm.   4 Results   166 For larger sensor magnet spacings of 7.5 cm, 10 cm and 12.5 cm, a trend for a larger change in the HE voltage with an increase in bar diameter could be observed for all frequencies. At any given frequency, the bar with the largest diameter (10 mm) led to the largest change in the HE voltage. Conversely, the bar with the smallest diameter (7 mm) led to the smallest change in HE voltage. However, the observed reduction in HE voltage was not always proportionate to the reduction in diameter.  The largest change in HE voltage was observed for a frequency of 19 Hz (Figure 4.67) regardless of the bar diameter. However, the largest difference between the 7 mm and 10 mm curves was observed for 12 Hz, followed by 31 Hz. At the latter frequencies, the two largest bars (9 mm and 10 mm) led to very similar results. The trend for a larger change in HE voltage with a larger bar diameter could only be observed at the two largest sensor magnet spacings of 10 cm and 12.5 cm at 31 Hz.    Figure 4.66: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 12 Hz   4 Results   167  Figure 4.67: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 19 Hz  Figure 4.68: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 24 Hz  Figure 4.69: Change in HE voltage along SB with PVC as a space holder at a magnet rotation of 31 Hz  4 Results   168 4.3.2.1.1.2 PVC - Uncoated Rebar The effect of an intact as well as a corroded piece of UCR on the HE voltage are shown in Figure 4.70 through Figure 4.73. The mass loss of the corroded UCR was 8.94%.  The intact UCR led to larger changes in HE voltage in comparison to the corroded UCR. This observation could be made for sensor spacings of 5 cm, 7.5 cm, 10 cm and 12.5 cm regardless of the frequency (with one exception at 24Hz and 5 cm spacing). The largest difference between the curve corresponding to the corroded and intact piece of UCR was observed for 31 Hz, followed by 19 Hz. This suggested that 19 Hz and 31 Hz were more suitable to detect corrosion in comparison to 12 Hz 24 Hz.   Figure 4.70: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 12 Hz  Figure 4.71: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 19 Hz   4 Results   169  Figure 4.72: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 24 Hz  Figure 4.73: Change in HE voltage along UCR with PVC as a space holder at a magnet rotation of 31 Hz 4.3.2.1.1.3 PVC - Epoxy Coated Rebar To study the effect of corroded ECR on the Hall voltage, ECR had undergone the accelerated corrosion procedures described in chapter 3.2.3.3. Prior to corrosion acceleration by anodic polarization, specimens A had been exposed to cathodic polarization whereas specimens B had been exposed to thermal cycling. After an initial low anodic current application, the current was incrementally increased. The total duration of the increased current was 84 days for both specimens A and B. The reference specimens consisted of intact pieces of ECR and had not undergone any corrosion acceleration procedures.    Figure 4.75 through Figure 4.78 show the change in HE voltage caused by the presence of ECR. The green curves correspond to intact pieces of ECR and the red and blue curves to corroded ECR after 84 days of accelerated corrosion. Curves of the same line type (i.e. dashed or continuous) were  4 Results   170 tested on the same day. Only one representative curve of each set of replicas is shown in Figure 4.75 through Figure 4.78 except for the red curves, where both A84-1 and A84-2 are shown. It should be noted, that the results of specimens that had undergone 40 days of accelerated corrosion were inconclusive and are not shown in this section. Furthermore, 3 replicas of each specimen type were used in this study.  A large gap between the two red curves was observed for low frequencies. This gap decreased with an increase in frequency until almost identical results were obtained at 31 Hz as shown in Figure 4.78. Visual inspection of the bars showed more uniform corrosion for A84-1 compared to A84-2 (Figure 4.74). A84-2 showed no signs of corrosion towards the bar end where the magnet was positioned (left in Figure 4.74), whereas A84-1 showed clear signs of corrosion along the entire bar. Consequently, A84-1 led to HE values below that of A84-2.  The large gap between the two red curves in Figure 4.75 shows, that the condition of the rebar directly below the magnet influenced the HE measurement significantly. This influence decreased with increasing frequency, resulting in a decreasing gap between the red curves with increasing frequency (Figure 4.75 through Figure 4.78).  All B84 specimens (i.e. B84-1, B84-2 and B84-3) showed signs of uniform corrosion along the entire bar and resulted in very similar curves. Hence, only one representative curve is shown per frequency in Figure 4.75 through Figure 4.78.  The two green curves representing the reference specimen without corrosion were obtained from the same specimen but on different days. The differences between these two curves were the result of environmental influences. Due to the significant impact of environmental influences, one should only compare results obtained on the same time.     All uniformly corroded specimens led to HE values below the reference measurements at all magnet sensor spacings larger than 2.5 cm (for readings taken on the same day). A84-2 led to values below the reference voltage at concrete spacings larger than 5 cm. One could observe that the gap between the A84-1 curve and its reference (continuous green) was larger than the gap between the B84-2 curve and its reference (dashed green). This suggested that the A84-1 bar experienced a higher corrosion mass loss compared to the B84-2 bar. This was particularly pronounced at 12 Hz and 31 Hz, where A84-1 showed a median difference to the reference values of 0.07 V and B84-2 resulted in a median difference of 0.03 V.   4 Results   171    Figure 4.74: Approximate location of magnet relative to rebars EM84A-1 and EM84A-2   Figure 4.75: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 12 Hz    4 Results   172  Figure 4.76: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 19 Hz  Figure 4.77: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 24 Hz  Figure 4.78: Change in HE voltage along ECR with PVC as a space holder at a magnet rotation of 31 Hz  4 Results   173 4.3.2.1.2 Concrete 4.3.2.1.2.1 Dry Concrete – Steel Bars Figure 4.79 through Figure 4.82 show the effect of the steel bars embedded in concrete on the HE voltage measurements under dry conditions. Measurements were performed at magnet rotation speeds of 12 Hz, 19 Hz, 24 Hz and 31 Hz.  Due to irregularities of the concrete moulds, the four specimens did not have identical concrete covers. The concrete covers of bar 10 and 8 were 3 cm and bars 9 and 7 had concrete covers of 2.85 cm and 2.7 cm respectively (refer to Table 3.7 on page 71). The concrete cover determined the distance between the rebar and the sensor-magnet assembly and thus impacted the HE voltage reading. Consequently, the results of bars with similar or identical concrete cover were compared to identify the effect of the bar diameter on the HE measurement.  Bars 10 and 8 had identical concrete covers of 3 cm and all measurements with a sensor magnet spacing of at least 5 cm resulted in increased HE voltages with increased bar diameter regardless of the frequency. The gap between the curves of bar 10 (yellow) and bar 8 (orange) is shown in Table 4.17 for all frequencies. The latter was maximized at 24 Hz, where the median gap between these curves was 115 mV.  The concrete covers of bar 9 and 7 varied by 0.15 cm and were 2.85 cm and 2.7 cm respectively. Reduced concrete covers as well as increased bar diameters increase the impact the steel has on the HE voltage. Specimen 7 had a smaller bar diameter as well as a smaller concrete cover compared to specimen 9. Thus, the difference in HE voltage of bar 9 and 7 is reduced due to the contradicting effects of the differences in bar diameter and concrete cover.  Figure 4.79 shows higher HE voltages for specimen 7 compared to specimen 9 at a frequency of 12 Hz. This shows that the effect of the concrete cover dominated over the impact of the bar diameter at 12 Hz. At larger frequencies of 19 Hz, 24 Hz and 31 Hz the impact of the bar diameter outweighed the effect of the concrete cover at sensor magnet spacings of at least 10 cm. As a result, specimen 9 led to higher HE voltages compared to specimen 7. In the case of 24 Hz, specimen 7 resulted in HE voltages below those of specimen 9 at sensor magnet spacings as low as 5 cm. Furthermore, the median gap between the curves of specimen 7 and 9 was maximized at 24 Hz with 43 mV as shown in Table 4.17.   4 Results   174 Table 4.17: Median gap between the curves of bar 7 and 9 and the curves of 8 and 10 in concrete under dry conditions with a minimum spacing of 7.5 cm f 7-9 8-10 [Hz] [mV] [mV] 12 -38 57 19 25 53 24 43 115 31 31 80    Figure 4.79: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 12 Hz   4 Results   175  Figure 4.80: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 19 Hz   Figure 4.81: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 24 Hz   4 Results   176  Figure 4.82: Change in HE voltage along SB embedded in concrete under dry conditions at a magnet rotation of 31 Hz  4.3.2.1.2.2 Dry Concrete – Uncoated Rebar The change in Hall effect voltages along the UCR embedded in concrete with a cover of 2.5 cm are shown in Figure 4.83 through Figure 4.87.  Measurements were performed with rotation frequencies of the magnet of 12 Hz, 19 Hz, 24 Hz, 28 Hz and 31 Hz. The corroded specimens had undergone the accelerated corrosion process for 0, 22, 45 and 90 days. Only one representative curve per corrosion degree is shown in the diagrams. The Hall effect measurements obtained for sensor-magnet spacings below 10 cm were inconclusive and did not relate to the degree of corrosion. At sensor magnet spacings of at least 10 cm, the reference specimens consistently resulted in higher Hall effect voltages in comparison to the corroded specimens. However, the measured voltages did not decrease with increasing degrees of corrosion except for rotational frequencies of the magnet of 28 Hz and 31 Hz (Figure 4.86, Figure 4.87). At the latter frequencies, the measured Hall effect voltages did generally decrease (with a few outliers) with increasing amount of corrosion but not proportionately. This lack of proportionality was likely due to the changes in the concrete surrounding the rebar resulting from the corrosion process. Cracks in the concrete extending from the rebar to the surface as well as the accumulation and dispersion of corrosion products throughout the concrete were the result of the corrosion of rebar embedded in concrete. Hall effect measurements tended to result in lower voltages when performed directly over top of a crack in the concrete and in higher voltages when performed directly over top of corrosion products on the concrete surface.  Hence, concrete cracks and the accumulation of  4 Results   177 corrosion products had opposing effects on the Hall effect measurements. The relative magnetic permeability of the main constituents of the reinforced concrete samples are shown in Table 4.18. The relative magnetic permeabilities in Table 4.18 decrease from top to bottom, placing concrete between iron oxide and air. Larger permeability values result in larger HE voltages, which is why the presence of carbon steel increases the HE sensor readings. Consequently, corrosion products, which primarily consist of iron oxide, increase the HE voltage compared to concrete. Concrete cracks are filled with air or water, which have lower relative magnetic permeabilities in comparison to concrete. Hence, cracks in the concrete led to lower HE voltages compared to concrete without cracks. Furthermore, a localized higher degree of corrosion and thus mass loss might have occurred at the crack locations and consequently led to reduced HE voltage readings.  Table 4.18: Relative permeability of sample constituents Material Magnetic category Relative magnetic permeability Carbon steel Ferromagnetic 100 (Nave 2017) FeO Paramagnetic 1.0072 (Nave 2017) Concrete 1.002 (NGU Geological Survey of Norway 1998) Air 1.00000037 (Cullity and Graham 2008) Water Diamagnetic 0.9999909 (Nave 2017)  At sensor magnet spacings of at least 10 cm, the median gap between the graphs of a corroded specimen to the reference was maximized at 24 Hz regardless of the degree of corrosion as shown in Table 4.19.   Table 4.19: Median gap between the curves of corroded and intact UCR in concrete under dry conditions with a minimum spacing of 10 cm f 22 45 90 [Hz] [mV] [mV] [mV] 12 31 26 32 19 41 21 29 24 53 45 64 28 17 29 33 31 27 23 44  4 Results   178  Figure 4.83: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 12 Hz   Figure 4.84: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 19 Hz   4 Results   179  Figure 4.85: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 24 Hz   Figure 4.86: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 28 Hz   4 Results   180  Figure 4.87: Change in HE voltage along UCR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 31 Hz 4.3.2.1.2.3 Dry Concrete – Epoxy Coated Rebar The effect of ECR embedded in concrete on the HE voltage is shown in Figure 4.88 through Figure 4.92 for rotational frequencies of the magnet of 12 Hz, 19 Hz, 24 Hz, 28 Hz and 31 Hz. The concrete samples had been dried prior to testing to minimize the effect of moisture on the HE measurements. Sensor magnet spacings below 7.5 cm led to inconclusive results regardless of the frequency. Furthermore, sensor magnet spacings of at least 10 cm led to slightly higher HE voltages for the reference specimens compared to corroded specimens except for the lowest applied frequency of 12 Hz. At the latter frequency, the reference only resulted in the highest HE voltage at sensor magnet spacings of at least 12.5 cm. However, even at distances above 12.5 cm, the corroded samples led to HE voltages very close to the reference.  Figure 4.90 through Figure 4.92 show that at high frequencies, the dashed reference curve consistently runs above the continuous curves once a peak voltage has been reached. At magnet rotations of 19 Hz and 24 Hz the peak values were reached at a distance of 5 cm. At 31 Hz a local peak was reached at 7.5 cm. Consequently, the reference specimens resulted in the highest change in HE voltage at sensor magnet spacings of at least 7.5 cm for 24 Hz and 28 Hz and of at least 10 cm for 31 Hz.  At a distance between the magnet and the HE sensor between 10 cm and 15 cm, the median gap between the reference curve and the curves of the corroded samples was maximized at a magnet rotation of 19 Hz as shown in Table 4.20. Overall, corrosion resulted in lower HE voltages compared to the uncorroded reference at large sensor magnet spacings. The minimum spacing needed to identify the corroded samples depended  4 Results   181 on the rotational frequency of the magnet. However, the reduction of the recorded HE voltage was not proportional to the amount of corrosion.  Table 4.20: Median gap between the curves of corroded and intact ECR in concrete under dry conditions with a minimum spacing of 10 cm f 40A 40B 84A 84B [Hz] [mV] [mV] [mV] [mV] 12 31 17 18 27 19 26 24 27 24 24 26 28 21 18 28 12 11 30 16 31 7 14 13 29    Figure 4.88: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 12 Hz   4 Results   182  Figure 4.89: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 19 Hz   Figure 4.90: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 24 Hz   4 Results   183  Figure 4.91: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 28 Hz   Figure 4.92: Change in HE voltage along ECR with a concrete cover of 2.5 cm under dry conditions at a magnet rotation of 31 Hz 4.3.2.1.2.4 Wet Concrete – Steel Bars Figure 4.93 through Figure 4.96 show the effect of the steel bars embedded in saturated concrete on the HE voltage measurements. The specimens containing bar 10 and 8 had identical concrete covers of 3 cm. Specimen 10 consistently led to higher HE voltages compared to specimen 8 under saturated conditions. This observation was made regardless of the frequency of the magnet rotation as shown in Figure 4.93 through Figure 4.96. The median gap between the two curves was maximized with 79 mV at the lowest frequency of 12 Hz but closely followed by 31 Hz with a median gap of 77 mV.   4 Results   184 Specimens 9 and 7 had concrete covers of 2.85 cm and 2.7 cm respectively. Consequently, the HE voltage of specimen 9 was not consistently larger than that of specimen 7. At a frequency of 12 Hz, specimen 7 led to lower HE measurements compared to sample 9 at sensor-magnet spacings of at least 5 cm as shown in Figure 4.93. The impact of the smaller bar diameter of specimen 7 outweighed the effect of the larger concrete cover on the Hall effect measurements at 12 Hz. The latter observation could also be made for a frequency of 19 Hz with the exception of the 10 cm spacing. At higher frequencies of 24 Hz and 31 Hz, specimen 9 only led to larger HE voltages than sample 7 when a minimum spacing of 15 cm and 10 cm was used respectively. At smaller distances between the sensor and magnet, the effect of the concrete cover dominated over the bar diameter.  The median gap between the curves of specimen 9 and 7 was maximized at 12 Hz with 48 mV as shown in Table 4.21.  Table 4.21: Median gap between the curves of bar 7 and 9 and the curves of 8 and 10 in concrete under saturated conditions with a minimum spacing of 7.5 cm f 7-9 8-10 [Hz] [mV] [mV] 12 48 79 19 34 49 24 -24 29 31 -17 77   4 Results   185   Figure 4.93: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 12 Hz   Figure 4.94: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 19 Hz   4 Results   186  Figure 4.95: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 24 Hz   Figure 4.96: Change in HE voltage along SB embedded in concrete under saturated conditions at a magnet rotation of 31 Hz 4.3.2.1.2.5 Wet Concrete – Uncoated Rebar Figure 4.97 through Figure 4.100 show the effect of UCR embedded in saturated concrete on the HE voltage measurements. It should be noted that the specimens, that had undergone 90 days of accelerated corrosion, were tested 3 days after the measurements of the REF, 22, and 45 samples had been performed.  Specimen 90 led to anomalous results for magnet rotation speeds of 12 Hz and 24 Hz, which was likely due to human error and environmental influences on the day of testing. The latter specimens led to HE voltages below the reference measurements at sensor magnet spacings of at least 7.5 cm  4 Results   187 at a frequency of 19 Hz as shown in Figure 4.98. A frequency of 31 Hz only led to HE voltages of the 90 samples below those of the reference, when a minimum sensor magnet spacing of 15 cm had been used.  The HE voltage measurements obtained at 12 Hz were inconclusive regardless of the degree of corrosion as shown in Figure 4.97. At 19 Hz and 24 Hz, specimens 22 and 45 led to voltage measurements below the reference once a minimum sensor magnet distance of 7.5 cm had been reached. However, the gap to the reference curve was not related to the degree of corrosion. Figure 4.100 shows HE voltages of the 22 and 45 samples below the reference at sensor magnet spacings of at least 15 cm at a magnet rotation of 31 Hz. The minimum spacing at the latter frequency was comparatively large but the gap to the reference curve did in fact increase with increasing amounts of corrosion for specimens 22 and 45. The median gap to the reference curve was maximized at 19 Hz regardless of the degree of corrosion as shown in Table 4.22. The median gap did not increase with increasing degree of corrosion. However, the 90 specimens did result in a significantly larger gap compared to the 22 and 45 samples.  Overall, 19 Hz led to the most promising results with a very low minimum distance between the sensor and magnet of 5 cm and the largest gap to the reference regardless of the degree of corrosion.  Table 4.22: Median gap between the curves of corroded and intact UCR in concrete under saturated conditions with a minimum spacing of 7.5 cm f 22 45 90 [Hz] [mV] [mV] [mV] 12 -10 -4 -132 19 38 27 53 24 14 14 3 31 18 26 23  4 Results   188  Figure 4.97: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 12 Hz  Figure 4.98: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 19 Hz   4 Results   189  Figure 4.99: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 24 Hz  Figure 4.100: Change in HE voltage along UCR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 31 Hz 4.3.2.1.2.6 Wet concrete – Epoxy Coated Rebar The effect of ECR embedded in saturated concrete on the HE voltage measurements are shown in Figure 4.101 through Figure 4.104. It should be noted, that the results of specimens that had undergone 40 days of accelerated corrosion were inconclusive and are not shown in this section. The curves of corroded bars 84 A and 84 B consistently fell below the reference curve for sensor magnet spacings of at least 10 cm regardless of the frequency.  Figure 4.102 and Figure 4.104 show that the corroded bars led to lower HE voltages compared to the reference specimens even at sensor magnet spacings below 10 cm at a magnet rotation of 19 Hz and 31 Hz. The minimum spacing at the latter frequencies was 7.5 cm and 5 cm respectively. Nonetheless, the median gap to the reference  4 Results   190 was maximized at 12 Hz for samples B and 24 Hz for samples A. However, Table 4.23 shows that 12 Hz and 24 Hz led to very similar median gaps for sample A.  Even though 19 Hz and 31 Hz led to lower minimum spacings, the median gap to the reference was maximized at 12 Hz. Table 4.23: Median gap between the curves of corroded and intact ECR in concrete under saturated conditions with a minimum spacing of 10 cm f A B [Hz] [mV] [mV] 12 32 37 19 8 26 24 34 10 31 15 24    Figure 4.101: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 12 Hz   4 Results   191  Figure 4.102: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 19 Hz   Figure 4.103: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturated conditions at a magnet rotation of 24 Hz   4 Results   192  Figure 4.104: Change in HE voltage along ECR with a concrete cover of 2.5 cm under saturate conditions at a magnet rotation of 31 Hz 4.3.2.1.2.7 Effect of NaCl The effect of NaCl on the Hall voltage was studied by two unreinforced concrete specimens, that had 0% and 3% of NaCl by mass of cement admixed to the concrete. The Hall voltage measurements were performed on both specimens at four different magnet rotational frequencies.  The change in the HE voltage caused by the presence of 3% NaCl per weight of cement is shown in Figure 4.105 through Figure 4.108 for magnet rotation frequencies of 12 Hz, 19 Hz, 24 Hz and 31 Hz. The measurements were performed under dry (grey) and saturated (blue) conditions. The curves were obtained by subtracting the measurement value of a specimen with 0% NaCl from the equivalent measurement of an identical specimen with 3% NaCl. Figure 4.105 through Figure 4.108 show large fluctuations at sensor magnet spacings below 10 cm and comparatively steady values at larger spacings. Due to the lack of consistency in the voltage fluctuations at spacings below 10 cm, NaCl was likely not the sole cause of the changes in HE voltage. Environmental influences such as temperature variations, accuracy and steadiness of sensor positioning as well as equipment accuracy likely contributed to the HE voltage fluctuations.  At sensor magnet spacings between 10 cm and 20 cm the effect of NaCl on the HE voltage was generally found to have been below 0.02 V. The effect of NaCl on the measurements was minimized at 24 Hz, where almost no change in the HE voltage could be observed. This low impact of NaCl on the voltage measurements was particularly pronounced under dry conditions as shown in Figure 4.107. The effect of NaCl fluctuated more with increasing sensor magnet spacings under saturated conditions in comparison to dry conditions. The largest impact of NaCl on the HE voltage was observed under saturated conditions at 19 Hz (Figure 4.106). At the latter frequency NaCl consistently increased  4 Results   193 the HE voltage under saturated conditions and reduced it under dry conditions. No other frequency led to a clear positive or negative impact of the NaCl on the HE voltage.    Figure 4.105: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 12 Hz   Figure 4.106: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 19 Hz   4 Results   194   Figure 4.107: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 24 Hz   Figure 4.108: Difference in HE voltage between concrete containing 0% NaCl and 3% NaCl at a magnet rotation of 31 Hz 4.3.2.1.2.8 Effect of Water The HE voltage change due to the concrete moisture content is shown in Figure 4.109 through Figure 4.112 for magnet rotations of 12 Hz, 19 Hz, 24 Hz and 31 Hz. The curves were obtained by subtracting the measurements under dry conditions from the equivalent measurement of the same specimen under saturated conditions. The specimens did not contain any reinforcement.  Figure 4.109 through Figure 4.112 show that the 0% NaCl curves (grey) stabilized at a sensor magnet spacing of 10 cm regardless of the frequency. In the presence of 3% NaCl, stabilization of the  4 Results   195 impact of water on the HE voltage started at 7.5 cm except for a magnet rotation of 19 Hz. Furthermore, one could observe larger fluctuations of the 0% NaCl curve (grey) in comparison to the 3% NaCl curve (blue) regardless of the frequency. Hence, the impact of the moisture content on the HE voltage measurements was more consistent in the presence of sodium chloride. The fluctuations at small spacings were likely due to other influencing factors such as environmental temperature and manual sensor positioning.  The effect of water on the HE voltage measurements was most pronounced at 12 Hz regardless of the NaCl content. The curves stabilized at voltage changes of -0.26 V. Conversely, the impact of moisture on the HE voltages was minimized at 24 Hz. At the latter frequency, almost no change in the measurement was detected with the addition of water, particularly in the presence of 3% NaCl. The measurements performed at 31 Hz were only impacted by moisture in the absence of NaCl (grey curve Figure 4.112). At magnet rotations of 19 Hz, the addition of water increased the HE voltages in the presence of NaCl and decreased the measurement values in the absence of NaCl.    Figure 4.109: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 12 Hz   4 Results   196   Figure 4.110: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 19 Hz  Figure 4.111: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 24 Hz  Figure 4.112: Difference in HE voltage between saturated and dry concrete at a magnet rotation of 31 Hz  4 Results   197 4.3.2.1.3 Summary A minimum spacing between the sensor and the magnet was necessary to obtain sensor readings, that were affected by small differences in rebar mass. The minimum spacing as well as the most suitable rotational frequency of the magnet, was dependent on the bar type as well as the material of the space holder. Table 4.24 summarizes the minimum sensor-magnet spacing and the magnet rotational frequencies that were most suitable for corrosion detection for each combination of the space holder and the bar type.  Table 4.24: Summary of minimum measurement parameters most suitable for corrosion detection Space holder Steel bars Uncoated rebar Epoxy coated rebar min Spacing Frequency min Spacing Frequency min Spacing Frequency [-] [cm] [Hz] [cm] [Hz] [cm] [Hz] PVC 5 19; 31 5 19;31 5 12; 31 Dry concrete 5 24 10 24 10 19 Wet concrete 5 12 7.5 19 10 12; 19  The minimum spacing was not affected by the bar type when a space holder of PVC was used. Furthermore, all bar types led to good results at 31 Hz with a minimum spacing of 5 cm in the case of PVC as a space holder. An additional frequency of either 19 Hz or 12 Hz depending on the bar type led to promising readings too. Embedding the bars in concrete led to larger minimum spacings for UCR and ECR. However, the latter was not affected by the space holder type in the case of the SB. The PVC space holder was 2.5 cm thick and the concrete covers varied between 2.7 cm and 3 cm for SB and were 3 cm and 2.8 cm for UCR and ECR respectively. The increase in distance between the bar and the sensor-magnet setup likely increased the minimum spacing as well as the moisture and NaCl content of the concrete. Salt and water had a significant effect on the HE voltage measurements at sensor magnet spacings up to 7.5 cm. The effects of NaCl and water were most pronounced at 19 Hz and 12 Hz respectively. The effect of the presence of the bars on the HE voltage was larger under dry conditions and less sensitive to changes of the frequency. Consequently, the minimum amount of corrosion of ECR, that affected the HE readings was larger under saturated conditions. Furthermore, a similar effect was observed for the PVC space holder.   4 Results   198 Table 4.24 shows a reduction in frequency from the dry to the wet concrete regardless of the bar type. Furthermore, the minimum spacing did not change with the moisture content except for UCR, where it reduced slightly.   Corrosion could be detected with the HE sensing technique for UCR as well as ECR embedded in concrete. However, the technique could not always reliably identify the amount of corrosion. It was sensitive to the concrete cover as well as the moisture and NaCl content at small sensor magnet spacings. Nonetheless, it was shown that corrosion did affect the HE voltage readings and thus the induction heating process. Corroded bars had a reduced effect on the HE and would thus likely lead to a reduced temperature if heated by induction.  4.3.2.2 Thermal Samples The objective of this study was to assess the impact of corrosion on the thermal response to induction heating independently from the effect of corrosion on to the electromagnetic induction heating process. Consequently, instead of induction heating, a high resistance heating wire inserted into the rebar was employed. A direct current of 25.4 A was applied to the heating wire until the rebar ends had reached a temperature of 93.5°C. The power supply was the turned off manually. The temperature of the concrete and the rebar was monitored with an infrared thermography camera. The experimental procedure is described in chapter 3.3.2.2.2 and a description of the UCR specimens can be found in chapter 3.3.2.1.2.1 and the ECR specimens in chapter 3.3.2.1.2.2. 4.3.2.2.1 Uncoated Rebar The peak temperature change ΔT of the concrete surface with respect to its initial temperature was determined by the average temperature of a small area of approximately 380 pixels. The area was chosen as the coolest location directly over top of the rebar as shown in Figure 4.113.   4 Results   199  Figure 4.113: Example of an IR image of sample REF-1 with location of measurement circle The coolest, rather than the centre location was chosen because of the asymmetric temperature rise across the specimen.  The latter was due to differences in the exposed lengths of the rebars on both sides of the specimens. Approximately 11 mm of the rebar was sticking out of the concrete on the right side of the specimen and only 0.1 mm was sticking out on the left side of the specimen. Thus, the temperature rise was not symmetric across the specimen and in order to avoid edge effects, the coolest location on the concrete surface was chosen.  The heating process was manually terminated by shutting down the power supply once the infrared camera readings showed a maximum temperature of the rebar ends of 93.5°C (chapter 3.3.2.2.2). This manual operation as well as small environmental temperature changes between testing dates introduced inconsistencies in the maximum applied power and temperature change of the specimens. Figure 4.114 shows the max temperature rise of the rebar ends in relation to the applied power. The results show that the corroded bars reached a higher temperature at a given applied power. The bars with the highest degree of corrosion (90) reached the highest temperatures and the reference bars the lowest temperature. Nonetheless, the peak bar temperature was not directly proportional to the degree of corrosion.   4 Results   200  Figure 4.114: Temperature of the rebar ends in relation to the applied power These observations suggest, that more power was needed to heat up the intact rebar compared to the corroded bars. Thus, the corroded bars were more insulated compared to the reference UCR. A higher degree of insulation of the corroded bars resulted in an overall reduction of the thermal conductivity between the steel surface and the concrete surface facing the IR camera. The latter signified an increased resistance to the transfer of heat away from the rebar. Consequently, the heat built up inside the UCR faster and led to a reduction in the required power of the corroded bars compared to the intact UCR.    The maximum temperature rise on the concrete surface in relation to the applied power is shown in Figure 4.115. The dashed linear trendline was formed by the measurements of the reference samples. Overall, the concrete surface temperature was approximately linearly related to the applied power. However, focusing on the area between 5 and 7 kJ and the reference trendline, one can observe a trend of a smaller rise in concrete surface temperature at a given applied power for the corroded bars.   4 Results   201  Figure 4.115: Concrete surface temperature in relation to the applied power This suggests a higher degree of insulation of the corroded bars compared to the reference specimens. Consequently, the thermal conductivity was reduced and the heat could not easily be transferred to the concrete surface. Thus, the ratio of the thermal conductivity of the entire concrete specimen (including corrosion products, cracks etc.) surrounding the rebar and that of the air surrounding the rebar ends was reduced. A reduction in this ratio of thermal conductivities led to a reduction in the portion of the heat being transferred through the specimen and thus to the concrete surface of the more insulated corroded specimens. Conversely, the portion of the heat being transferred through the air surrounding the rebar ends increased for the aforementioned specimens.   Figure 4.116 shows the maximum rise in concrete surface temperature in relation to the peak temperature rise of the rebar ends. Both measurements were obtained with the IR camera. A larger temperature rise of the rebar ends of the corroded specimens could be observed compared to the reference specimens.     4 Results   202 Figure 4.116: Concrete surface temperature in relation to the temperature of the rebar ends This was particularly pronounced for the 90 specimens. Furthermore, the increase in concrete surface temperature was slightly lower for the corroded specimens. The concrete surface temperature rise as a percentage of the peak rebar temperature is shown in Figure 4.117. A trend of a decreasing percentage with increasing degree of corrosion could be observed. The rebar ends were hotter but the concrete surface temperature lower with increasing amount corrosion.   Figure 4.117: Maximum concrete surface temperature as a percentile of the maximum rebar end temperature These observations were in line with the previously described findings of an increased amount of insulation around the corroded bars in comparison to the intact UCR. Moreover, Figure 4.116 and Figure 4.117 show a trend of increased insulation with an increased degree of corrosion.   Both cracks and corrosion products in the concrete serve as an insulator to the rebar in comparison to sound concrete. The latter has a higher thermal conductivity than air filled cracks and rust. Thus, the rebar was insulated at the ends by the surrounding air and along the curved surface facing the specimen surface by the cracks and rust. Hence, the least insulated rebar surface was the back side of the rebar (i.e. the curves surface facing the back of the specimen) where less cracks and rust accumulated due to the positioning of the anode during the accelerated corrosion and the increased concrete cover. Figure 4.118 and Figure 4.119 show the distribution of the corrosion products and cracks respectively. Thus, most of the heat was transferred into the specimen rather than to the concrete surface.   4 Results   203  Figure 4.118: Cross section of specimen after 90 days of accelerated corrosion  Figure 4.119: Side view of specimen after 90 days of accelerated corrosion Nonetheless, a higher degree of corrosion led to more, longer and wider cracks as well as more corrosion products. Thus, a higher degree of corrosion led to an overall reduction of the thermal conductivity of the specimen as a whole (including concrete, cracks, rust etc.) and hence an increased amount of insulation. Consequently, the ratio of the thermal conductivity of the specimen as a whole, to that of the air surrounding the rebar ends, decreased with an increasing degree of corrosion. The latter led to a smaller portion of heat being transferred through the specimen to the concrete surface and a larger portion being transferred between the rebar ends and the surrounding air. Thus, a higher degree of corrosion resulted in lower concrete surface temperatures as well as higher rebar end temperatures at a given applied power. Conversely, less power was needed for the corroded rebar to reach a given temperature compared to the intact rebar, due to the increased amount of insulation around the corroded bars. The latter throttled the heat transfer away from the rebar and thus heat built up inside the rebar faster.    4 Results   204 However, the observed trends are likely not strong enough to reliably detect corrosion of UCR by thermal properties alone.    4.3.2.2.2 Epoxy Coated Rebar The concrete surface temperature used for the data analysis was measured directly over top of the rebar with the IR camera. The average temperature of a circle of approximately 380 pixels positioned at the coolest location was used as shown in Figure 4.120. Due to the symmetry of the specimens, the latter heated up approximately symmetrically and the circle was located near the centre of the specimen (Figure 4.120 left).   Figure 4.120: Location of concrete surface temperature measurement The rebars of specimens 84B-1 and 84B-2 had only shown signs of corrosion on one half of the rebar. This was due to evaporation of the solution used during the accelerated corrosion process. As a result, the elevated end of the rebar had dried out leading to uneven corrosion along the bars. Consequently, the specimens’ thermal response to the heating process was not symmetric as shown in Figure 4.120 (right). The latter image already shows, that corrosion did in fact have an impact on the thermal behaviour of the ECR reinforced concrete specimens. Specimens 84B-1 and 84B-2 are shown in a dull orange and other specimens of the same corrosion age (i.e. 84A and 84B-3) are shown in bright orange in the following diagrams.     The maximum rise of the rebar temperature in relation to the applied power is shown in Figure 4.121. The rebar ends were expected to have been the hottest parts of the rebar.   4 Results   205  Figure 4.121: Temperature of the rebar ends in relation to the applied power The corroded bars as well as the stripped bar (i.e. ECR, which was stripped off its epoxy coating) led to smaller rebar end temperature rises even though more power was applied to those specimens. This trend was observed for all degrees of corrosion but it was not proportional to the degree of corrosion. The latter observation suggests, that the intact ECR was more insulated in comparison to the corroded and stripped bars. Thus, heat could not easily escape from the rebar and quickly built up inside the rebar. Corroded and stripped bars were less insulated and transferred heat into the concrete during the heating process. Thus, more power was needed to reach the same ΔT of the rebar.   Figure 4.122 shows the rise in concrete surface temperature in relation to the applied power. Overall, it can be observed, that an increase in power led to higher surface temperatures.   Figure 4.122: Concrete surface temperature in relation to the applied power Nonetheless, the concrete surface temperature did increase with increasing degree of corrosion at a given amount of applied power. This was particularly clear at an applied power of approximately  4 Results   206 6.4 kJ. Furthermore, the specimen with the stripped ECR resulted in similar concrete ΔT to applied power ratio to the highly corroded specimens. These observations suggested a reduction in insulation of the rebar with increasing degree of corrosion as well as a lack of coating. Consequently, heat was easily and thus quickly transferred from the rebar to the concrete. Hence, less heat escaped through the rebar ends, that were insulated by the surrounding air with the lowest thermal conductivity of the testing system of 0.02534 W/(m K) (Dixon 2007) at room temperature. This resulted in a larger amount of heat being transferred to the concrete, which led to higher concrete surface temperatures.   Figure 4.123 shows the maximum temperature rise of the concrete surface in relation to the peak temperature rise of the rebar ends.   Figure 4.123: Concrete surface temperature in relation to the temperature of the rebar ends It can be observed that the corroded bars led to higher rises in concrete surface temperatures with smaller increases in rebar end temperatures. A trend of a larger concrete to smaller rebar temperature rise with increasing amount of corrosion was observed. Furthermore, the specimen with the stripped bar led to similar results to the corroded bars.  Figure 4.124 shows the peak concrete surface temperature rise as a percentage of the maximum temperature increase of the rebar ends.   4 Results   207  Figure 4.124: Maximum concrete surface temperature as a percentage of the maximum rebar end temperature A trend of an increasing percentage with increasing degree of corrosion can be observed. However, the stripped ECR resulted in the same percentage as the bars with a high degree of corrosion. These observations are in line with the previously described findings of a reduction in thermal insulation of the rebar for the corroded as well as the stripped specimens.   Specimens 84B-1 and 84B-2 had only corroded towards one end of the rebar and are shown in Figure 4.125 on the left and right respectively. Very concentrated corrosion was observed for 84B-1, that led to relatively short cracks in the epoxy coating. 84B-2 on the other hand showed spalling of the epoxy coating due to the expansion of the corrosion products underneath.   Figure 4.125: Cross section of specimens 84B-1 (left) and 84B-2 (right) Thus, the results of these two specimens were not always in line with the observed trends of Figure 4.121 through Figure 4.124. Consequently, the maximum rise in concrete surface temperature at the hottest spot directly above the rebar was measured and analyzed. The average temperature of a circle of approximately 380 pixels on the concrete at each rebar end was considered as shown in Figure 4.126.   4 Results   208  Figure 4.126: Location of concrete surface temperature measurement of a reference specimen and a specimen with uneven corrosion distribution Additionally, the time it took for that circle to reach its maximum temperature after the power supply had been turned off was determined. The rate at which the hottest spots of the concrete surface had heated up could then be calculated. Figure 4.127 shows the inverse of the rate at which these two hotspots heated up in seconds per degree Celsius.   Figure 4.127: Time in seconds for maximum concrete surface temperature to increase by 1°C The stripped rebar led to much shorter heating times of 81 s/°C and 85 s/°C and thus faster heating rates compared to the intact ECR of 125 s/°C. This supports the observation of a reduction in insulation with the loss of the epoxy coating. Furthermore, the corroded specimens led to heating times between those of the stripped and intact ECR except for 40B-2 and 84B-2. Figure 4.128 shows specimen 40B-2 after it had been split on the left and the extracted rebar surface, that had faced the IR camera on the right. This specimen led to heating times even longer than the intact ECR as shown in Figure 4.127.  4 Results   209  Figure 4.128: Cross section (left) and extracted rebar (right) of specimen 40B-2 It can be observed that rebar 40B-2 showed signs of corrosion products, that had expanded under the coating. However, the coating had only split open in a few locations and the coating cracks were very short. Thus, the coating had only suffered minor damage and hence the impact on its insulation power was minor too. Consequently, the time for the surface to heat up by 1°C would have been expected to only have reduced by a small amount. However, the latter time increment increased. Thus, the insulation of 40B-2 must have been higher than the one of intact rebar. This suggests that the corrosion products under the coating served as an insulator for the underlying rebar.   Specimen 84B-2 (Figure 4.125 right) showed a heating time of 133 s/°C and 95 s/°C for the intact and corroded sides respectively. Thus, the heating time of the corroded side was reduced and that of the intact side slightly increased in comparison to the intact ECR. Spalling of the epoxy coating shown in Figure 4.125 (right) reduced the insulation on the corroded side, which increased the thermal conductivity and led to a faster transfer of the heat from the rebar to the concrete in this location.  Specimen 84B-1 resulted in similar heating times for both sides of 119 s/°C and 116 s/°C even though it had only experienced corrosion on one end. However, even though specimens IRT84B-1 and IRT84B-2 resulted in a similar mass loss (Table 4.7 on page 120 and Table 4.8 on page 121), 84B-1 showed only short cracks of the epoxy coated (Figure 4.125 left). Consequently, the corrosion products of 84B-1 had not expanded as much as those of 84B-2. The increased densities of the corrosion products reduced their insulating power and the reduced amount of coating cracks resulted in an increased amount of insulating power of 84B-1 compared to 84B-2. Overall the high density of the corrosion products and short cracks of the epoxy coating of 84B-1 resulted in minimal effects on the thermal response of the concrete specimen. Thus, the heating was very close to the one of intact ECR.   4 Results   210 These findings suggest that the thermal conductivity between the rebar and the concrete was the lowest for the intact ECR, leading to the largest temperature rise of the rebar and the lowest ΔT of the concrete, because the heat could not be transferred easily from the rebar to the concrete. The latter would mean that the epoxy coating thermally insulated the underlying rebar. This was confirmed by the large concrete surface temperature rise of the specimen containing the stripped bar with a similar ΔT of the rebar. The lack of the thermally insulating coating led to a larger thermal conductivity between the rebar and the concrete. Thus, heat could be transferred more easily between the rebar and the concrete, which led to higher concrete surface temperatures. Corroded bars had a similar effect on the concrete surface temperature than the stripped bar because the expansion of the corrosion products led to cracking of the epoxy coating.  The corrosion of ECR resulted in the formation and build up of corrosion products. These corrosion products accumulated under the coating, leading the epoxy to crack. Once the coating had cracked the corrosion products expanded with less restraints. The constraints of the epoxy coating on the expansion of the corrosion products led to increased densities of the corrosion products. An increase in density goes hand in hand with an increase in thermal conductivity and reduction in specific heat capacity. Thus, the epoxy coating likely caused increased thermal conductivities and reduced specific heat of the corrosion products.  Cracking of the epoxy coating signified a loss in thermal insulation of the rebar. Furthermore, the increased density of the corrosion products led to a reduced insulation effect of the rust on the underlying steel.  Thus, coating cracking and rust build up had competing effects on the thermal properties of the system. The loss of insulation resulting from coating cracking dominated in this study. The latter resulted in higher concrete surface temperatures of the corroded and stripped specimens compared to the samples with intact ECR. The heat could more easily be transferred from the steel to the concrete due to the overall reduced insulation effect of the epoxy and rust layer. Thus, the concrete surface temperature increased and the temperature at the rebar ends decreased. Consequently, the applied power required to achieve a temperature of 93.4 °C at the rebar ends increased for the corroded and stripped specimens.   It should be noted, that larger amounts of corrosion could lead to the effect of the corrosion products to dominate and thus leading to thermal measurements contradicting the findings of this study. Furthermore, accelerated corrosion of the ECR took place in a solution before the bars were embedded in concrete. Thus, the constraints on the expansion of the corrosion products were reduced in comparison to the corrosion of ECR in concrete. The latter likely led to lower densities of  4 Results   211 the corrosion products in this study. Additionally, corrosion of ECR in concrete would result in cracks in the concrete and thus a reduction in concrete density. Hence, the corrosion products in this study likely led to reduced thermal conductivities and the concrete to increased thermal conductivities in comparison to field conditions.    5 Conclusions   212 5 Conclusions 5.1 Corrosion Behaviour The effects of sodium chloride, -carbonate and –bicarbonate on the corrosion behaviour of epoxy coated rebar have been studied and compared to the behaviour of uncoated rebar under identical conditions. Assuming the entire curved surface of epoxy coated rebar as the polarized area greatly underestimated the current densities and thus the corrosion rate of epoxy coated rebar. However, assuming only the exposed steel area was accurate only when epoxy coated rebar had not been submerged in a solution for an extended period of time prior to testing or when the solution did not contain any sodium ions. The sodium concentration of the test solutions was related to the size of the polarized area of the long-term specimens. However, other factors affected the polarized area of epoxy coated rebar. Consequently, the increase in the area was not proportionate to the sodium concentration.  Significantly, epoxy coated rebar was found to be more susceptible to corrosion in the presence of NaCl as compared to uncoated rebar under identical conditions. Once the passive layer, formed in the very alkaline environment of 0.9M NaOH, was lost due to the addition of NaCl, uncoated rebar was still protected by its oxide layer, that formed during the hot rolled manufacturing process. However, epoxy coated rebar did not have such a layer and was unprotected at coating holidays.  The 0.01M Na2CO3 solution had a pH of 11 and consequently had a passivating effect on the rebar. The 0.01M NaHCO3 solution, on the other hand, had an almost neutral pH of 8. Consequently, it initially had an accelerating effect on the corrosion behaviour, which decreased over time and eventually even passivated the steel. The initial accelerated corrosion effect was more pronounced for epoxy coated rebar compared to uncoated rebar.  A combination of sodium carbonate and bicarbonate led to a passive layer, that formed slowly but grew continuously over time. This passive layer was very effective in reducing the current densities and corrosion rates and making the open circuit potential more positive. However, this passive layer was very fragile and even small amounts of sodium chloride increased the currents and made the open circuit potential more negative. The passive layer created by the sodium hydroxide on   5 Conclusions   213 the other hand, formed instantaneously but did not grow any further over time. It was very stable and large amounts of NaCl were needed to increase the current densities and shift the open circuit potential to more negative values, indicating an increased propensity to corrosion. 5.2 Accelerated Corrosion The experimental results showed, that an alkaline and neutral environment promoted undercoating and coating holiday corrosion, respectively. This was caused by the cathodic delamination of the epoxy coating, that increased the pH at the steel-epoxy interface. Consequently, a large pH gradient was present between the delaminated area and the coating holiday in neutral solution. The delaminated areas were protected by the pH and the steel at the coating damage was exposed to chloride ions as well as a neutral pH. Thus, corrosion occurred at the coating holiday. Epoxy coated rebar in alkaline solution did not experience an increase in pH from the coating holiday to the delaminated interface. Thus, corrosion initiated and expanded under the coating. However, accelerating cathodic delamination by cathodic polarization prior to the acceleration of corrosion slowed down the lateral expansion of corrosion. However, once corrosion had initiated on the entire surface, the subsequent corrosion rates were not affected by the initial cathodic polarization.  5.3 Corrosion Detection 5.3.1 Conventional Non-Destructive Testing Detection of localized corrosion or corrosion at coating damage was not successful with the half-cell method or the resistivity probe. To a limited extent localized corrosion detection was possible with the linear polarization resistance as well as the ground penetrating radar methods. However, chloride contamination, moisture content and concrete maturity had a significant effect on both measurements.  A combination of the magnitude of the corrosion current determined by the linear polarization resistance method and its standard deviation led to successful identification of corroding epoxy coated rebar. Up to 37.5% and 54.5% of coating holidays were correctly identified by higher corrosion currents in salt free and chloride contaminated concrete respectively. Furthermore, localized high corrosion currents as well as coating damage coincided with a reduction in the amplitude and a delay of the reflected of the ground penetrating radar signal. However, chloride contamination and corrosion have a similar effect on both the linear polarization resistance and ground penetrating radar   5 Conclusions   214 measurements. Thus, in the case of uniform chloride contamination, localized corrosion detection with a combination of the linear polarization resistance and ground penetrating radar is possible. However, corrosion monitoring over time might not be accurate due to changes in the chloride contamination as well as other environmental factors such as moisture content and temperature.  5.3.2 Hall-Effect Sensing Corrosion could successfully be detected by a reduction in the Hall voltage. However, this effect was weak with the current setup. The minimum sensor-magnet spacing and the suitable frequencies were affected by the material of the space holder and the investigated bar type.  A space holder made of PVC led to more consistent minimum spacings and suitable frequencies compared to concrete. This was likely due to the inhomogeneity of concrete as well as the salt and moisture content, which had a significant effect on the Hall voltage at small sensor-magnet spacings and low frequencies. Furthermore, dry conditions led to more stable results between the frequencies and a larger impact of the reinforcing bars on the Hall voltage. Thus, the minimum amount of corrosion that affected the Hall voltage was reduced with a reduction in the moisture content.  Nonetheless, it was shown that corrosion reduced the impact of the rebar on the Hall effect sensing. Thus, corroded bars would likely lead to lower peak bar temperatures in comparison to intact rebar. 5.3.3 Thermal Detection  The thermal sensing experiments led to opposing effects of corrosion of uncoated rebar and epoxy coated rebar. The corrosion of uncoated rebar led to higher bar temperatures and lower concrete surface temperatures. However, corrosion of epoxy coated rebar decreased the bar temperatures and increased the concrete surface temperatures. These observations were the result of the insulating power of the corrosion products and the epoxy coating. While corrosion products and the formation of concrete cracks increased the insulation around the uncoated rebar, the development of cracks in the epoxy coating reduced the insulation of the epoxy coated rebar with increasing amounts of corrosion.  Corrosion of rebar led to the reduction in the ferromagnetic properties of the latter. This suggests, that a corroded bar would lead to a lower peak temperature and consequently concrete surface temperature. Furthermore, corrosion of uncoated rebar was shown to have an insulating effect on the latter, reducing the concrete surface temperature. Thus, the effects of corrosion on the magnetic and thermal aspects of induction heating of uncoated rebar in concrete support each other.   5 Conclusions   215 However, cracking of the epoxy coating led to reduced insulation of the rebar and thus opposed the effect of corrosion on the thermal properties of the rebar. Consequently, coating cracks or spalling increased the concrete surface temperature. Hence, induction thermography is not a suitable non-destructive testing method for corrosion detection of epoxy coated rebar.  5.4 General The degradation of epoxy coated rebar does not only involve steel corrosion but also degradation of the coating itself as well as disbondment of the epoxy coating from the steel substrate. Thus, the impact of environmental factors such as the moisture and salt content as well as the degree of carbonation on the degradation of epoxy coated rebar differs from that of uncoated rebar. Most notable is the unfavourable effect of sodium on the epoxy coated rebar, particularly but not only in combination with chloride such as in marine environments and bridge decks with de-icing salts.  The area of the epoxy coated rebar experiencing water uptake of the coating, water induced adhesion loss and cathodic disbondment had a significant effect on electrochemical experiments. Thus, electrochemical corrosion detection techniques should only be used supplementary. The most promising non-destructive corrosion detection methods for epoxy coated rebar fell in the category of electromagnetism. However, while ground penetrating radar and Hall effect sensing showed encouraging results, supplementary non-destructive testing methods of a different category, such as electrochemistry or thermography, are recommended.     6 Future Research   216 6 Future Research More research of the degradation behaviour of epoxy coated rebar is needed. Electrochemical testing methods, that are independent of the polarized area should be investigated for assessment of the corrosion behaviour of epoxy coated rebar. The extent of the area experiencing adhesion loss and disbondment, particularly of epoxy coated rebar embedded in concrete, needs more research. Furthermore, future research should include the effects of disbondment on non-destructive testing methods in the field of electrochemistry and beyond. A standard test method to accelerate the corrosion of epoxy coated rebar embedded in concrete should be developed. This might include the application of a direct current, in which case a slowly increasing current over time should be considered to account for the development of coating disbondment.  Significantly more research of non-destructive detection of corrosion of epoxy coated rebar is needed. The origin of the more active potential with a decreasing degree of coating damage of corroding epoxy coated rebar is still unclear. While electromagnetic techniques proofed promising, more research in the field of data analysis and software development of the ground penetrating radar method is needed. Furthermore, more sophisticated Hall effect testing setups with an AC powered electromagnet and an array of more advanced Hall sensors should be developed. Hall effect and thermography experiments with epoxy coated rebar, that had corroded inside of concrete, should be performed. Additionally, the effect of different stages of corrosion of epoxy coated rebar on the thermal behaviour should be researched.     References   217 References 3M. 2012. “3M TM Scotchkote TM Fusion Bonded Epoxy Coating 134.” Austin, TX: 3M. ———. 2013. “3M TM Scotchkote TM Fusion Bonded Epoxy Coating 6233P.” Austin, TX: 3M. Abd El Haleem, S. M., E. E. Abd El Aal, S. Abd El Wanees, and A. Diab. 2010. “Environmental Factors Affecting the Corrosion Behaviour of Reinforcing Steel: I. The Early Stage of Passive Film Formation in Ca(OH)2 Solutions.” Corrosion Science 52 (12). Elsevier Ltd: 3875–82. doi:10.1016/j.corsci.2010.07.035. Ahmad, Shamsad. 2009. “Techniques for Inducing Accelerated Corrosion of Steel in Concrete.” The Arabian Journal for Science and Engineering 34 (2): 95–104. Alberto A, and R G Powers. 1996. “Coating Disbondment in Epoxy-Coated Reinforcing Steel in Concrete Field Observations.” In Corrosion 96. Houston, TX: NACE International. Alonso, C., C. Andrade, J. Rodriguez, and J. M. Diez. 1998. “Factors Controlling Cracking of Concrete Affected by Reinforcement Corrosion.” Materials and Structures 31 (7): 435–41. doi:10.1007/BF02480466. Andrade, Carmen, Victor Castelo, Cruz Alonso, and Jose A. Gonzalez. 1986. “The Determination of the Corrosion Rate of Steel Embedded in Concrete by the Polarization Resistance and AC Impedance Methods.” In ASTM Special Technical Publication 906. Philadelphia: American Society of Testing and Materials. Angst, Ueli, Anders Rønnquist, Bernhard Elsener, Claus K. Larsen, and Øystein Vennesland. 2011. “Probabilistic Considerations on the Effect of Specimen Size on the Critical Chloride Content in Reinforced Concrete.” Corrosion Science 53 (1). Elsevier Ltd: 177–87. doi:10.1016/j.corsci.2010.09.017. ASTM International. 2013. “A934/A934M − 13 Standard Standard Specification for Epoxy-Coated Prefabricated Steel Reinforcing Bars 1.” ASTM International. doi:10.1520/A0934. ———. 2014a. “A775 − 07b Standard Specification for Epoxy-Coated Steel Reinforcing Bars.” ASTM International. doi:10.1520/A0934. ———. 2014b. “Astm G59-97 Standard Test Method for Conducting Potentiodynamic Polarization Resistance Measurements.” doi:10.1520/G0059-97R14.2. ———. 2014c. “G5-13 Standard Reference Test Method for Making Potentiodynamic Anodic   References   218 Polarization Measurements.” ASTM International. doi:10.1520/G0005-13E02.2. ———. 2015. “D3936-15 Standard Specification for Fabrication and Jobsite Handling of Epoxy-Coated Steel Reinforcing Bars.” ASTM International. doi:10.1520/D3963. ———. 2016. “A615 Standard Specification for Deformed and Plain Carbon-Steel Bars for Concrete Reinforcement.” ASTM International. doi:10.1520/A0615. Austin, Simon A., Richard Lyons, and Matthew Ing. 2004. “Electrochemical Behaviour of Steel Reinforced Concrete during Accelerated Corrosion Testing.” Corrosion 60 (2): 203–12. Axalta. 2014. “Nap-Gard ® 7-2500 Fusion Bonded Epoxy.” Brown, Michael Carey. 2002. “Corrosion Protection Service Life of Epoxy Coated Reinforcing Steel in Virginia Bridge Decks.” Virginia Polytechnic Institute and State University. Caré, Sabine, and André Raharinaivo. 2007. “Influence of Impressed Current on the Initiation of Damage in Reinforced Mortar due to Corrosion of Embedded Steel.” Cement and Concrete Research 37 (12): 1598–1612. doi:10.1016/j.cemconres.2007.08.022. CMC Commercial Metals. 2015. “Environmental Product Declaration of Concrete Reinforcing Steel.” ASTM International. Concrete Reinforcing Steel Institute - CRSI. 2017a. “Corrosion And Steel Reinforcement - Corrosion Mechanism.” Epoxy Interest Group - CRSI. Accessed September 25. http://www.epoxyinterestgroup.org/use-applications/corrosion-mechanisms/. ———. 2017b. “History of Epoxy-Coated Rebar.” Epoxy Interest Group - EIG. Accessed December 28. http://epoxyinterestgroup.org/about/history-of-epoxy-coated-rebar/. Covino, Bernard S, Stephen D Cramer, Gordon R Holcomb, James H Russell, Sophie J Bullard, and Cheryl L Dahlin. 2001. “Epoxy-Coated Rebar Performance of the Perley Bridge,” no. 1653. Cullity, B. D., and C. D. Graham. 2008. Introduction to Magnetic Materials. 2nded. Wiley-IEEE Press. Dinh, Kien. 2014. “Condition Assessment of Concrete Bridge Decks Using Ground Penetrating Radar.” Dixon, John C. 2007. “Appendix B Properties of Air.” In The Shock Absorber Handbook. John Wiley & Sons, Ltd. doi:10.1002/9780470516430. East, Middle, and N D T Conference. 2007. “A New Application of the Gpr Technique To Reinforced Concrete.” Middle East, no. in mV. Eisenmann, David, Frank Margetan, Chien-Ping T. Chiou, Ron Roberts, and Scott Wendt. 2013. “Ground Penetrating Radar Applied to Rebar Corrosion Inspection,” 1341–48. doi:10.1063/1.4789198. Erdogdu, Sakir. 1992. “Determination of the State of Corrosion of Epoxy-Coated Rebar in Concrete.”   References   219 The University of New Brunswick. doi:10.1017/CBO9781107415324.004. Ghods, Pouria, Rouhollah Alizadeh, and Mustafa Salehi. 2014. United States Patent Application Publication, issued 2014. Griffin, Donald F., and Robert L. Henry. 1964. “The Effect of Salt in Concrete on Compressive Strength, Water Vapor Transmission, and Corrosion of Reinforcing Steel.” Port Hueneme. Han, Jiabin, David Young, Hendrik Colijn, Akhilesh Tripathi, and Srdjan Nešić. 2009. “Chemistry and Structure of the Passive Film on Mild Steel in CO2 Corrosion Environments.” Industrial and Engineering Chemistry Research 48 (13): 6296–6302. doi:10.1021/ie801819y. Harnisch, Jörg. 2012. “Zum Zeitabhängigen Verhalten Elektrochemischer Und Morphologischer Kenngrößen Bei Der Chloridinduzierten Korrosion von Stahl in Beton.” Homeland Security: Science and Technology. 2010. “Aging Infrastructure: Issues, Research, and Technology.” Buildings and Infrastructure Protection Series 1 (December). Hong, S X, W L Lai, and R Helmerich. 2012. “Monitoring Accelerated Corrosion in Chloride Contaminated Concrete with Ground Penetrating Radar.” Proceedings of the 14th International Conference on Ground Penetrating Radar (GPR), 561–66. doi:10.1109/ICGPR.2012.6254927. Incropera, Frank P., David P. DeWitt, Theodore L. Bergman, and Adrienne S. Lavine. 2007. Fundamentals of Heat and Mass Transfer. 6thed. Hoboken, NJ: John Wiley & Cons. doi:10.1073/pnas.0703993104. Iordanova, I, M Surtchev, K S Forcey, and V Krastev. 2000. “High-Temperature Surface Oxidation of Low-Carbon Rimming Steel” 160 (July 1999): 158–60. Jones, Denny A. 1996. Principles and Prevention of Corrosion. Edited by Bill Stenquist, Rose Kernan, and Patricia Daly. Prentice-Hall. Second. New Jersey: Prentice-Hall. Kabir, Shahid, and Ahmad Zaki. 2011. “Detection and Quantification of Corrosion Damage Using Ground Penetrating Radar ( GPR ),” 790–93. Kobayashi, K., and N. Banthia. 2011. “Corrosion Detection in Reinforced Concrete Using Induction Heating and Infrared Thermography.” Journal of Civil Structural Health Monitoring 1 (1–2): 25–35. doi:10.1007/s13349-010-0002-4. Lai, WL, T Kind, and H Wiggenhauser. 2010. “Detection of Accelerated Reinforcement Corrosion in Concrete by Ground Penetrating Radar.” IEEE. doi:10.1109/ICGPR.2010.5550254. Lau, K., A. A. Sagüés, and R. G. Powers. 2010. “Corrosion of Epoxy-Coated Rebar in Marine Bridges - Part 2: Corrosion in Cracked Concrete.” Corrosion 66 (6): 0650021–216. doi:10.5006/1.3452397. Legghe, Elise, Emmanuel Aragon, Lénaïk Béleca, André Margaillan, and Denis Melot. 2009.   References   220 “Correlation between Water Diffusion and Adhesion Loss: Study of an Epoxy Primer on Steel.” Progress in Organic Coatings 66 (3): 276–80. doi:10.1016/j.porgcoat.2009.08.001. Lenntech BV. 2017. “Lenntech.” The Chemical Elements and Water. Accessed September 25. http://www.lenntech.com/periodic/water/overview.htm. Li, L., and A.A. Sagues. 2001. “Chloride Corrosion Threshold of Reinforcing Steel in Alkaline Solutions - Effect of Specimen Size.” Corrosion 57 (1): 19–28. Li, Wei, Bruce Brown, David Young, and Srdjan Nešić. 2014. “Investigation of Pseudo-Passivation of Mild Steel in CO 2 Corrosion.” Corrosion 70 (3): 294–302. doi:10.5006/0950. Maaddawy, Tamer a. El, and Khaled a. Soudki. 2003. “Effectiveness of Impressed Current Technique to Simulate Corrosion of Steel Reinforcement in Concrete.” Journal of Materials in Civil Engineering 15 (1): 41–47. doi:10.1061/(ASCE)0899-1561(2003)15:1(41). Malhotra, V Mohan, and Nicholas J Carino. 2003. “Handbook on Nondestructive Testing of Concrete.” Mammoliti, L. T., L. C. Brown, C. M. Hansson, and B. B. Hope. 1996. “The Influence of Surface Finish of Reinforcing Steel and pH of the Test Solution on the Chloride Threshold Concentration for Corrosion Initiation in Synthetic Pore Solutions.” Cement and Concrete Research 26 (4): 545–50. doi:10.1016/0008-8846(96)00018-X. Millard, S. G., D. Law, J. H. Bungey, and J. Cairns. 2001. “Environmental Influences on Linear Polarisation Corrosion Rate Measurement in Reinforced Concrete.” NDT and E International 34 (6): 409–17. doi:10.1016/S0963-8695(01)00008-1. Mohammed, Tu, and Hidenori Hamada. 2006. “Corrosion of Steel Bars in Concrete with Various Steel Surface Conditions.” ACI Materials Journal 103 (103): 233–42. Moreno, M., W. Morris, M. G. Alvarez, and G. S. Duffo. 2004. “Corrosion of Reinforcing Steel in Simulated Concrete Pore Solutions Effect of Carbonation and Chloride Content.” Corrosion Science 46 (11): 2681–99. doi:10.1016/j.corsci.2004.03.013. Nakayama, Norio, and Akira Obuchi. 2003. “Inhibitory Effects of 5-Aminouracil on Cathodic Reactions of Steels in Saturated Ca(OH)2 Solutions.” Corrosion Science 45 (9): 2075–92. doi:10.1016/S0010-938X(03)00032-5. National Research Council Canada. 2013. “Critical Concrete Infrastructure: Extending the Life of Canada’s Bridge Network.” Construction Innovation 18 (1). Nave, Carl Rod. 2017. “Hyperphysics.” Magnetic Susceptibilities of Paramagnetic and Diamagnetic Materials at 20°C - Magnetic Properties of Ferromagnetic Materials. Accessed September 25. http://hyperphysics.phy-astr.gsu.edu/hbase/Tables/magprop.html.   References   221 NGU Geological Survey of Norway. 1998. “Magnetic Susceptibility Measurements on Concrete Samples.” Trondheim. Nguyen, T., J.B. Hubbard, and J.M. Pommersheim. 1996. “Unified Model for the Degradation of Organic Coatings on Steel in a Neutral Electrolyte.” Journal of Coatings Technology. Nguyen, T., and J. W. Martin. 1996. “Modes and Mechanisms of Degradation of Epoxy-Coated Reinforcing Steel in a Marine Environment.” Durability of Building Materials and Components. ———. 2004. “Modes and Mechanisms for the Degradation of Fusion-Bonded Epoxy-Coated Steel in a Marine Concrete Environment.” Journal of Coatings Technology and Research 1 (2): 81–92. doi:10.1007/s11998-004-0002-6. Omega Engineering Inc. 2017. “Omegabond OB-700 High Temperature Chemical Set Cement Instruction Sheet.” Oshita, Hideki, Hiroaki Horie, Shingo Nagasaka, Osamu Taniguchi, and Shinjiro Yoshikawa. 2009. “Nondestructive Evaluation of Corrosion in Reinforced Concrete by Thermal Behavior on Concrete Surface due to Electro-Magnetic Heating.” JSCE 65 (1): 76–92. Oy, Michiaki. 2003. “Characteristics of Fatigue and Evaluation of RC Beam Damaged by Accelerated Corrosion” 44 (2): 0–5. Poursaee, Amir. 2010. “Corrosion of Steel Bars in Saturated Ca (OH) 2 and Concrete Pore Solution.” Concrete Research Letters 1 (3): 90–97. Powers, Dana. 2009. “Interaction of Water with Epoxy,” no. July: 58. Princeton Applied Research. 2017. “Basics of Corrosion Measurements.” Application Note CORR-1. Accessed September 25. http://www.ameteksi.com/-/media/ameteksi/download_links/documentations/library/princetonappliedresearch/application_note_corr-1.pdf?la=en. Rama Corporation. 2017. “Engineering Guide.” San Jacinto, CA: Rama Corporation. Accessed September 25. http://www.ramacorporation.com/catalog/engineering.pdf. Razeghi, Manijeh. 2009. Fundamentals of Solid State Engineering. Springer. doi:10.1007/978-0-387-92168-6. Rengaswamy, N.S., R. Vedhalakshmi, and K. Balakrishnan. 1995. “Evaluation of Coated Rebar Validity of Short‐term Accelerated Corrosion Tests in Relation to Long‐term Field Evaluation.” Anti-Corrosion Methods and Materials 42 (3): 7–9. doi:10.1108/eb007358. Sagüés, A. A., H M Perez-Duran, and R G Powers. 1991. “Corrosion Performance of Epoxy-Coated Reinforcing Steel in Marine Substructure Service.” Corrosion (Houston, Tex.) 47 (11). NACE   References   222 International: 884–93. doi:10.5006/1.3585202. Sagüés, A. A., R. G. Powers, and R. Kessler. 2001. “Corrosion Performance of Epoxy-Coated Rebar in Florida Keys Bridges,” 1–13. Sagüés, Aa, Kingsley Lau, Rg Powers, and Rj Kessler. 2008. “Corrosion of Epoxy-Coated Rebar in Marine Bridges - A 30 Year Perspective.” Corrosion 66 (6). Sagüés, Alberto A, and Rodney G Powers. 1996. “Coating Disbondment in Epoxy-Coated Reinforcing Steel in Concrete - Field Observations.” In Corrosion 96. Saricimen, H, S K Somuah, N R Jarrah, and O A Ashiru. 1997. “Evaluation of Corrosion Resistance of Fusion-Bonded Epoxy Coated Rebars in Concrete by DC and AC Techniques.” In Corrosion 97. Houston, TX: NACE International. Schneider Bautabellen Fuer Ingenieure. 2006. Neuwied: Werner Verlag. Shi, Xianming, Tuan Anh Nguyen, Prathish Kumar, and Yajun Liu. 2011. “A Phenomenological Model for the Chloride Threshold of Pitting Corrosion of Steel in Simulated Concrete Pore Solutions.” Anti-Corrosion Methods and Materials 58 (4): 179–89. doi:10.1108/00035591111148894. Strategic Highway Research Program - SHRP 2. 2013. “Nondestructive Testing to Identify Concrete Bridge Deck Deterioration.” SHRP 2 Renewal Research. Washington, D.C. Sturgeon, W Joseph, Matthew O Reilly, David Darwin, and Joann Browning. 2010. “RAPID MACROCELL TESTS OF ASTM A775, A615, and A1035 REINFORCING BARS.” Lawrence, Kansas. Sun, Weihua, A. K. Tieu, Zhengyi Jiang, and Cheng Lu. 2004. “High Temperature Oxide Scale Characteristics of Low Carbon Steel in Hot Rolling.” Journal of Materials Processing Technology 155–156 (1–3): 1307–12. doi:10.1016/j.jmatprotec.2004.04.167. Tan, Yong Teck, Sudesh L. Wijesinghe, and Daniel J. Blackwood. 2014. “The Inhibitive Effect of Bicarbonate and Carbonate Ions on Carbon Steel in Simulated Concrete Pore Solution.” Corrosion Science 88: 152–60. doi:10.1016/j.corsci.2014.07.026. U.S. Department of Transportation: Federal Highway Administration. 2017a. “Bridges and Structures.” Deficient Bridges by Functional Classification. ———. 2017b. “Bridges and Structures.” Deficient Bridges by Highway System. http://www.fhwa.dot.gov/bridge/deficient.cfm. ———. 2017c. “Bridges and Structures.” Bridge Replacement Unit Costs 2016. http://www.fhwa.dot.gov/bridge/nbi/sd.cfm. ———. 2017d. “Bridges and Structures.” Deck Structure Type. http://www.fhwa.dot.gov/bridge/nbi/deck.cfm.   References   223 Vollmer, Michael, and Klaus-Peter Moellmann. 2010. “Fundamentals of Infrared Thermal Imaging.” In Infrared Thermal Imaging, 1–72. Weinheim: Wiley-VCH. Watanabe, Takeshi, Huynh Thi Huyen Trang, Kazuki Harada, and Chikanori Hashimoto. 2014. “Evaluation of Corrosion-Induced Crack and Rebar Corrosion by Ultrasonic Testing.” Construction and Building Materials 67: 197–201. doi:10.1016/j.conbuildmat.2014.05.013. Xia, Xinghua, Colin M A Ashruf, Patrick J. French, Joerg Rappich, and John J. Kelly. 2001. “Etching and Passivation of Silicon in Alkaline Solution: A Coupled Chemical/electrochemical System.” Journal of Physical Chemistry B 105 (24): 5722–29. doi:10.1021/jp003208f. Zou, Xiaotian, Alice Chao, Ye Tian, Nan Wu, Hongtao Zhang, Tzu Yang Yu, and Xingwei Wang. 2012. “An Experimental Study on the Concrete Hydration Process Using Fabry-Perot Fiber Optic Temperature Sensors.” Measurement: Journal of the International Measurement Confederation 45 (5). Elsevier Ltd: 1077–82. doi:10.1016/j.measurement.2012.01.034. Zubel, Irena, Irena Barycka, Kamilla Kotowska, and Małgorzata Kramkowska. 2001. “Silicon Anisotropic Etching in Alkaline Solutions IV.” Sensors and Actuators A: Physical 87 (3): 163–71. doi:10.1016/S0924-4247(00)00481-7.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0362873/manifest

Comment

Related Items