Quality Control of FRP Composite Manufacturing and Construction using Infrared Thermography

Transcription

1 Quality Control of FRP Composite Manufacturing and Construction using Infrared Thermography Nagavardhana Kumar Perisetty Problem Report submitted to the College of Engineering and Mineral Resources at West Virginia University in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering Udaya B. Halabe, Ph.D., P.E., Chair Hota V. S. GangaRao, Ph.D., P.E. Hema J. Siriwardane, Ph.D., P.E. Department of Civil and Environmental Engineering Morgantown, West Virginia 2012 Keywords: Infrared Thermography; GFRP; Glass Fabrics; Debonds; Delaminations

2 ABSTRACT Quality Control of FRP Composite Manufacturing and Construction using Infrared Thermography Nagavardhana Kumar Perisetty High performance materials such as Fiber Reinforced Polymer (FRP) composites, which have excellent material properties, are being used for the construction of new structures and repair/rehabilitation of existing structures. Enhanced properties like high strength, high stiffness, light weight and low maintenance costs make FRP composites very desirable compared to conventional materials such as concrete. For existing structures, wrapping of the reinforced concrete or timber members with FRP fabrics can greatly improve the strength and ductility of the members. Though composites have many advantages, subsurface defects such as delaminations, debonds, and voids may be introduced during the initial construction or in-service stage, which can affect the structural integrity and durability of the rehabilitated structure. Hence proper quality control must be ensured during the initial construction process. In addition, timely detection and repair of the subsurface defects is needed during service in order to get best results. Several nondestructive techniques are currently being used for quality control of FRP composite materials. Infrared thermography (IRT) is one of the effective non-contact and nondestructive techniques for both qualitative and quantitative detection of subsurface defects in composites. The technique employs a portable infrared camera which can be used conveniently in both laboratory and field setting. Quality control is also necessary in any composite manufacturing and application process, in order to ensure the durability of the manufactured product (e.g., composite panels, tubes and flanged sections). This study focuses on the quality control aspects of the FRP composite members during manufacturing as well as the condition assessment of the FRP composite wrapped members during field construction and in service. The infrared data from a manufacturing facility and several field bridges presented in this report demonstrate the usefulness of infrared thermography technique in both manufacturing and field settings.

3 ACKNOWLEDGEMENTS I am extremely grateful to my academic and research advisor and Chair of my Advisory and Examining Committee (AEC), Dr. Udaya B. Halabe, for his continuous support and encouragement during my course work and my research. I deeply thank him for his guidance and invaluable suggestions provided during the preparation of my problem report. I would also like to express my sincere thanks to Dr. Hota V.S. GangaRao and Dr. Hema J. Siriwardane, for serving as members of my AEC. I would like to convey my sincere thanks to the external sponsors (USDOT-FRA, WVDOH, and Bedford Reinforced Plastics) for supporting my graduate research assistantship through the Constructed Facility Center, Department of Civil and Environmental Engineering at West Virginia University. I would like to take this opportunity to express my gratitude to my parents (Mr. Prasada Rao Perisetty and Mrs. Vijaya Lakshmi Perisetty) and my brother (Mr. Nagraj Perisetty) who have been very supportive throughout my journey in completing this study. iii

4 TABLE OF CONTENTS ABSTRACT... ii ACKNOWLEDGEMENTS... iii 1 INTRODUCTION BACKGROUND RESEARCH OBJECTIVES AND SCOPE ORGANIZATION LITERATURE REVIEW ON INFRARED THERMOGRAPHY BASICS OF INFRARED THERMOGRAPHY DETECTION OF AIR BLISTERS AND CRACK PROPAGATION IN FRP STRENGTHENED CONCRETE ELEMENTS USING INFRARED THERMOGRAPHY (Hu et al. 2002) Introduction Case Study I: Detection of Air Blisters Case Study II: Prediction of Crack Initiation Results and Discussions Conclusions IRT SURVEY FOR THE QUALITY CONTROL OF FRP REINFORCED RC STRUCTURES (Corvaglia et al. 2008) Introduction Discussions Case Study I: ex Law Courts of Sant Angelo dei Lombardi Case Study II: Highway Reinforced Concrete Single Span Bridge Conclusions iv

5 2.4 EVALUATION OF BONDED FRP STRENGTHENING SYSTEMS FOR CONCRETE STRUCTURES USING INFRARED THERMOGRAPHY AND SHEAROGRAPHY (Taillade et al. 2009) Introduction Nondestructive Evaluation of FRP Bonding by Shearography Infrared Thermography Technique Conclusions HEATING METHODS AND DETECTION LIMITS FOR INFRARED THERMOGRAPHY INSPECTION OF FIBER-REINFORCED POLYMER COMPOSITES (Brown et al. 2007) Experimental Setup and Design Results and Analysis Defect Analysis Conclusions IR THERMOGRAPHY FOR INTERFACE ANALYSIS OF FRP LAMINATES EXTERNALLY BONDED TO RC BEAMS (Valluzzi et al. 2009) Preliminary Thermographic Tests on Concrete Samples IRT Tests on FRP-Strengthened RC Beams Conclusions CHARACTERIZATION OF THE DETERIORATION OF EXTERNALLY BONDED CFRP-CONCRETE COMPOSITES USING QUANTITATIVE INFRARED THERMOGRAPHY (Lai et al. 2010) Introduction Experimental Setup Quantitative Infrared Thermography Results and Discussions Conclusions v

6 2.8 QUANTITATIVE ANALYSIS OF DELAMINATIONS IN GRP PIPES USING THERMAL NDTE TECHNIQUE (Vijayaraghavan et al. 2010) Introduction Experiments Numerical Modeling Results and Discussions Conclusions DETECTION OF SUBSURFACE DEFECT IN FIBER REINFORCED POLYMER COMPOSITE BRIDGE DECKS USING DIGITAL INFRARED THERMOGRAPHY (Halabe et al. 2007) Introduction Experimental Setup Description of laboratory specimen Detection of debonds using infrared thermography Summary and Conclusions EFFECTS OF SOLAR LOADING ON INFRARED IMAGING OF SUBSURFACE FEATURES IN CONCRETE (Washer et al. 2010a) Introduction Experimental Results Thermographic imaging of subsurface deterioration in concrete bridges (Washer et al. 2010b) Conclusions INFRARED THERMOGRAPHY TESTING FOR MANUFACTURING QUALITY CONTROL INTRODUCTION INFRARED TESTING EQUIPMENT vi

7 3.2.1 Description of the Infrared Camera Infrared Testing Results CONCLUSIONS INFRARED FIELD TESTING OF A RC BRIDGE DESCRIPTION OF THE BRIDGE INFRARED TESTING EQUIPMENT INFRARED TESTING AND RESULTS PRIOR TO GFRP WRAPPING (SEPTEMBER 21, 2010) Pier Cap Pier Cap 3 (Bottom Corner Portion) Pier Cap 4 (Part 1) Pier Cap 4 (Part 2) Pier CONCLUSIONS INFRARED FIELD TESTING OF FRP WRAPPED WOODEN RAILROAD BRIDGE COMPONENTS INFRARED THERMOGRAPHY EQUIPMENT INFRARED FIELD TESTING (OCTOBER 29, 2010) Bridge # Bridge # Bridge # INFRARED FIELD TESTING (APRIL 14, 2011) Bridge # Bridge # Bridge # CONCLUSIONS vii

8 6 CONCLUSIONS AND RECOMMENDATIONS CONCLUSIONS RECOMMENDATIONS FOR FUTURE RESEARCH REFERENCES viii

9 LIST OF FIGURES Figure 2.1 Illustration of embedded blister locations (Hu et al. 2002)...6 Figure 2.2 CFRP plate concrete specimens where the blisters are embedded between the composite and substrate (Hu et al. 2002)...6 Figure 2.3 Thermal images taken at various distances for concrete samples wrapped with CFRP (Hu et al. 2002)...7 Figure 2.4 Configuration of test beams (a) Elevation (b) Cross section (Hu et al. 2002)..8 Figure 2.5 Actual photograph of the concrete samples after the CFRP laminate was removed (Hu et al. 2002)...9 Figure 2.6 Thermal image taken during 45-55kN cyclic load range (Hu et al. 2002) Figure 2.7 Thermal image taken after the failure occurred at load range 67.5kN-82.5kN (Hu et al. 2002) Figure 2.8 Actual photograph taken for the test specimen after the failure has occurred (Hu et al. 2002) Figure 2.9 Imposed known defects (Corvaglia et al. 2008) Figure 2.10 Sample preparation (Corvaglia et al. 2008) Figure 2.11 Thermogram of the CFRP sample with imposed defects Figure 2.12 Percentage error in the assessment of defect number 3 for different experimental set-up (Corvaglia et al. 2008) Figure 2.13 Indoor IRT survey of an R.C structure reinforced with GFRP sheets (Corvaglia et al. 2008) Figure 2.14 Outdoor IRT survey of an RC structure reinforced with GFRP sheets (Corvaglia et al. 2008) Figure 2.15 Infrared thermographic survey of a bridge pier reinforced with CFRP and SRP materials (Corvaglia et al. 2008) Figure 2.16 CFRP strip transversal overlapping (Corvaglia et al. 2008) Figure 2.17 Application of SRP reinforcement (Corvaglia et al. 2008) Figure 2.18 Concrete slabs with bonded FRP reinforcements containing different types of defects (Taillade et al. 2009) ix

10 Figure 2.19 Strain measurements on sample 1 for various sizes of defects, and under different levels of partial vacuum (Taillade et. al. 2009) Figure 2.20 Strain measurements on sample 2 with a defect diameter of 40 mm, and for different qualities of the adhesive bond expressed as the ratio between surface of the bonded areas and the surface of a plain PTFE disc (Taillade et al. 2009) Figure 2.21 Experimental setup (Taillade et al. 2009) Figure 2.22 Concrete specimen reinforced with bonded CFRP plates, containing calibrated defects (Taillade et al. 2009) Figure 2.23 Maxima of the thermal contrast taken at different t max (Taillade et al. 2009) Figure 2.24 Thermal contrast v / s. time computed to three zones above the larger defects (blue line for a depth 2mm, green line for a depth 4mm and red line for a depth 6mm (Taillade et al. 2009) Figure 2.25 Defect configuration for Series A specimens (Brown et al. 2007) Figure 2.26 Heat source and camera configuration (Brown et al. 2007) Figure 2.27 Thermal image collected at end of heating (Brown et al. 2007) Figure 2.28 Area identification for defect analysis (Hamilton et al. 2007) Figure 2.29 Constructing T def versus time plots from area parameters (Brown et al. 2007) Figure 2.30 Thermal images for Defect IB (Specimen A-3) (Brown et al. 2007) Figure 2.31 Non uniform heating and weak signals for defects (Brown et al. 2007) Figure 2.32 Comparison of Tmax for different heating methods (Brown et al. 2007).. 36 Figure 2.33 Comparison of normalized T max for different heating methods (Brown et al. 2007) Figure 2.34 Reduced accuracy in area computations (Brown et al. 2007) due to: (a) weak signal (b) non-uniform heating and (c) insufficient pixel resolution Figure 2.35 Area computations for defect A75 (Brown et al. 2007) Figure 2.36 Maximum signal versus radii COV for all detected defects Figure 2.37 Preparation of samples with artificial defects (Valluzzi et al. 2009) x

11 Figure 2.38 Thermographic processing (central strip of rough surface) by various algorithms, indicating main discontinuities (artificial defects are incorporated in label 1) (Valluzzi et al. 2009) Figure 2.39 Setup for bending test on R.C beams (Valluzzi et al. 2009) Figure 2.40 Cross section of beams (Valluzzi et al. 2009) Figure 2.41 Application phases of pretensioned FRP system on underside of beams (Valluzzi et al. 2009) Figure 2.42 Cracking distribution at failure for RC-PrEA (average spacing 215 mm) (Valluzzi et al. 2009) Figure 2.43 Cracking distribution at failure for PRC-PrEA (average spacing 165 mm) (Valluzzi et al. 2009) Figure 2.44 Experimental thermographic tests on beams (Valluzzi et al. 2009) (a) mechanical setup for bending tests; (b) sliding device for thermographic data acquisition; (c) scanning phase; (d) related images in tilted mirror Figure 2.45 Active thermography results of RC-EA and RC-PrEA beams before bending tests, showing main discontinuities at location A (row 1) and location B (row 2); left thermogram at best time during cooling; right: data processed by PCA (Valluzzi et al. 2009) Figure 2.46 Active thermography results of RC-PrEA beam processed by PCA at various loading cycles Figure 2.47 Thermography results of PRC-PrEA beam processed by PCA at various loading cycles (Valluzzi et al. 2009) Figure 2.48 Debonded surface of RC-PrEA beam of figure 2.47, after mechanical test; various kinds of detachments are visible (Valluzzi et al. 2009) Figure 2.49 Visible image of resin layer after detachment of CFRP, with fingerprint of natural defects such as air pockets (see Figure 2.48), which probably occurred during application of resin and laminate (Valluzzi et al. 2009) Figure 2.50 Embedded delaminations and flaws contained within a CFRP-strengthened concrete prism (Lai et al. 2010) Figure 2.51 Thermal history curves of the flaw and sound areas (Lai et al. 2010) Figure 2.52 Thermogram of a flaw and the respective pixel profiles (Lai et al. 2010) xi

12 Figure 2.53 Pixel profile from a flaw to a sound zone (Lai et al. 2010) Figure 2.54 Interfacial delamination (left) and flaw (right) overlaid by 1 mm 2 grid for visual image method (Lai et al. 2010) Figure 2.55 Comparison between actual and apparent flaw sizes (Lai et al. 2010) Figure 2.56 Conditions of the sharp-edge flaw and development of delamination due to exposure to elevated water temperature (Lai et al. 2010) Figure 2.57 Comparison of flaw areas before and after exposures (Lai et al. 2010) Figure 2.58 Comparison of delamination areas after exposures (Lai et al. 2010) Figure 2.59 Details of GRP pipes (Vijayaraghavan et al. 2010) Figure 2.60 Schematic of the experimental setup (Vijayaraghavan et al. 2010) Figure 2.61 Meshed 3-D model from ANSYS (Vijayaraghavan et al. 2010) Figure 2.62 Numerical versus Experimental surface temperature distribution for pipe 1 (Vijayaraghavan et al. 2010) Figure 2.63 Numerical vs. Experimental surface temperature distribution for pipe 2 (Vijayaraghavan et al. 2010) Figure 2.64 Numerical vs. Experimental surface temperature distribution for pipe 3 (Vijayaraghavan et al. 2010) Figure 2.65 Numerical vs. Experimental surface temperature distribution for pipe 4 (Vijayaraghavan et al. 2010) Figure 2.66 Surface temperature profiles across the center line of the delaminations (Vijayaraghavan et al. 2010) Figure 2.67 Comparisons of thermal contrasts of delaminations in pipe 2 and 3 (Vijayaraghavan et al. 2010) Figure 2.68 Variation of the thermal contrasts of delaminations in pipe 3 (Vijayaraghavan et al. 2010) Figure 2.69 Comparison of thermal contrasts of delaminations in pipe 2 and pipe 4 (Vijayaraghavan et al. 2010) Figure 2.70 Experimental setup using digital infrared camera and close up view of the camera (Halabe et al. 2007) Figure 2.71 Various active heating and cooling sources (Halabe et al. 2007) xii

13 Figure 2.72 Front and cross-sectional views of the bridge deck specimen BD1 with wearing surface (Halabe et al. 2007) Figure 2.73 Front and cross-sectional views of the GFRP bridge deck specimens (Halabe et al. 2007) Figure 2.74 Photographs showing the locations of delaminations (Halabe et al. 2007).. 76 Figure 2.75 Surface temperature-time curves for air-filled debonds in specimen BD Figure 2.76 Infrared images of the specimen BD1 with air-filled debonds (Halabe et al. 2007) Figure 2.77 Schematic view and infrared image of bridge deck specimen BD2 showing the air-filled and water-filled debonds (Halabe et al. 2007) Figure 2.78 Surface temperature-time curves for 1 x 1 x 1/16 (25 x 25 x 1.6 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) air-filled delaminations located at the flangeflange junction of bridge deck JD1b (without wearing surface) (Halabe et al. 2007) Figure 2.79 Infrared image of specimens JD1a and JD1b (without wearing surface) with air-filled delaminations at the flange-flange junction (Halabe et al. 2007) Figure 2.80 Infrared image of the bridge deck specimen JD2 with embedded air-filled delamination of size 3 x 3 x 1/8 (76 x 76 x 3.2 mm) at the flange-flange junction Figure 2.81 Surface temperature-time curves for 3 x 3 x 1/8 (76 x 76 x 3.2 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) air-filled delaminations located at the flangeflange junction of bridge deck JD2 (with wearing surface) (Halabe et al. 2007) Figure 2.82 Schematic diagram of concrete bridge deck with corrosion damage Figure 2.83 Diagram of concrete test block with embedded targets at depths of 25 mm (A), 51 mm (B), 76 mm (C), and 127 mm (D) (Washer et al. 2010a) Figure 2.84 Photograph of test setup showing test house, weather stations and test block (Washer et al. 2010a) Figure 2.85 Thermal image of test block showing embedded targets at 25 mm, 51 mm and 76 mm and example locations selected for calculating thermal contrast (Washer et al. 2010a) Figure 2.86 Thermal contrast at embedded targets for sunny day (12/25/2007) indicating contrasts for 25 mm, 51 mm, 76 mm and 127 mm targets relative to the solar load (Washer et al. 2010a) xiii

14 Figure 2.87 (a) Relationship between maximum solar loading and maximum thermal contrast for target B; (b) Relationship between total solar loading area and thermal contrast for target B (Washer et al. 2010a) Figure 2.88 Thermal images and photographs of delaminated concrete in bridge soffit (Washer et al. 2010b) Figure 3.1 Picture of the InfraCAM SD infrared camera Figure 3.2 Setup at the manufacturing plant Figure 3.3 Digital picture and infrared images of the channel section coming out of the manufacturing die Figure 3.4 Digital picture and infrared images of the stacked channel sections Figure 3.5 Tube sections coming out of the manufacturing die Figure 3.6 Infrared image of the tube sections coming out of the manufacturing die Figure 3.7 Digital picture and infrared images of the stacked tube sections Figure 3.8 Infrared image of the defect-free panel section Figure 3.9 Infrared image of the section showing debonding of the GFRP layer Figure 4.1 (a) Deck view and (b) Elevation view of the bridge Figure 4.2 View of the supports Figure 4.3 Infrared testing equipment Figure 4.4 Digital picture and infrared image of the bottom middle portion of pier cap Figure 4.5 Digital picture and infrared image of the bottom corner portion of pier cap Figure 4.6 Digital picture and infrared image of the side portion (part 1) of pier cap Figure 4.7 Digital picture and infrared image of the side portion (part 2) of pier cap Figure 4.8 Digital picture and infrared image of the pier 7 at the joint location (pier cap 4) Figure 5.1 Infrared testing equipment Figure 5.2 Side view of the bridge # Figure 5.3 Digital picture and infrared image of the bottom portion of pile 4 of bent Figure 5.4 Digital picture and infrared image of the middle portion of pile 4 of bent Figure 5.5 Digital picture and infrared image of the top portion of pile 4 of bent xiv

15 Figure 5.6 Digital picture and infrared image of pile 2 of bent Figure 5.7 Side view of the bridge # Figure 5.8 Digital picture and infrared image of pile 3 of bent Figure 5.9 Digital picture and infrared image of pile 4 of bent Figure 5.10 Digital picture and infrared image of pile 2 of bent Figure 5.11 Digital picture and infrared image of the pile 3 of bent Figure 5.12 Side view of bridge # Figure 5.13 Digital picture and infrared image of the pile 3 of bent Figure 5.14 Digital picture and infrared image of the pile 4 of bent Figure 5.15 Digital picture and infrared image of the pile 3 of bent Figure 5.16 Digital picture and infrared image of the pile 1 of bent Figure 5.17 Digital picture and infrared image of the bottom portion of pile 2 of bent Figure 5.18 Digital picture and infrared image of the top portion of pile 2 of bent Figure 5.19 Digital picture and infrared image of the pile 1 of bent Figure 5.20 Digital picture and infrared image of the bent 4 with an old GFRP wrap Figure 5.21 Digital picture and infrared image of the top surface of bent 6 with an old GFRP wrap Figure 5.22 Digital picture and infrared image of the side middle portion of pile 4 of bent Figure 5.23 Digital picture and infrared image of the bottom side portion of pile 4 of bent Figure 5.24 Digital picture and infrared image of the pile 2 of bent Figure 5.25 Digital picture and infrared image of the pile 3 of bent Figure 5.26 Digital picture and infrared image of the pile 4 of bent Figure 5.27 Digital picture and infrared image of the pile 2 of bent Figure 5.28 Digital picture and infrared image of the pile 4 of bent Figure 5.29 Digital picture and infrared image of pile 2 of bent Figure 5.30 Digital picture and infrared image of bent 4 with an old GFRP wrap Figure 5.31 Digital picture and infrared image of the top surface of bent 6 with an old GFRP wrap xv

16 LIST OF TABLES Table 2.1 Displacements of 1.2 m RC beam subjected to static and cyclic loading tests (Hu et al.2002). 8 Table 2.2 Thermophysical properties of selected materials (Shepard 2007) Table 2.3 Material properties (Brown et al. 2007) Table 2.4 Series A Properties (Brown et al. 2007) 28 Table 2.5 Surface temperature increase results for different heating methods (Brown et al. 2007) Table 2.6 Actual and measured areas for select defects (60 second pulse heating) (Brown et al. 2007) Table 2.7 General detectability results (Brown et al. 2007)..41 Table 2.8 Characteristics of samples (Valluzzi et al. 2009)..45 Table 2.9 Main results on beams (Valluzzi et al. 2009) 46 Table 2.10 Tensile properties of CFRP and epoxy resin (Lai et al. 2010) 53 Table 2.11 Definitions, illustrations and methods to quantify flaws and delaminations (Lai et al. 2010)...54 Table 2.12 Location of delaminations (Vijayaraghavan et al. 2010) 60 Table 2.13 Sizes and depth of the delaminations (Vijayaraghavan et al. 2010) 61 Table 2.14 Thermal properties of E-glass fiber and epoxy resin (Vijayaraghavan et al. 2010)..62 Table 2.15 Variation of experimental and numerical results with actual delamination sizes for pipe 2 (Vijayaraghavan et al. 2010).68 Table 2.16 Variation of experimental and numerical results with actual delamination sizes for pipe 3 (Vijayaraghavan et al. 2010).69 Table 2.17 Variation of experimental and numerical results with actual delamination sizes for pipe (Vijayaraghavan et al. 2010) 69 Table 2.18 Time of day each target reaches its maximum and time lag relative to sunrise (Washer et al. 2010a)..88 xvi

17 1 INTRODUCTION 1.1 BACKGROUND Fiber Reinforced Polymer (FRP) composite materials have many advantages over conventional materials and are widely being used for a range of applications in the aerospace and civil infrastructure fields. Composite materials are made from two or more constituent materials with significantly different physical or chemical properties which remain separate and distinct on a macroscopic level within the finished structure. Though composites have been a late addition to the materials in civil engineering, they find themselves in many applications due to their superior properties (light weight, high stiffness, low maintenance cost, non-corrosive nature, etc.), when compared to conventional materials. FRP fabric is one of the widely used materials for repair and rehabilitation of existing structures such as old concrete and timber bridges. The need for the strengthening, repairing or upgrading structures arises due to factors like expired design life, changes in functionality, damage caused by service loads, or environmental effects. The bonded FRP composite fabric acts as an external reinforcement and enhances the strength and ductility of the structure. The bond quality between the FRP fabric and the underlying material is vital for the optimum performance of the composite system and for effective stress transfer between the underlying member and the FRP fabric. In addition to FRP composite fabrics, composite panels such as composite bridge decks are being used to replace deteriorated concrete bridge decks. Other composite members, such as tubes and channels, are also being used for support structures. Quality control of the final product is required for the reliable and safe performance of any new material being used in a structural system. Nondestructive testing (NDT) techniques can be used to assess the quality of the manufactured product in the factory setting and for monitoring the product in the field setting during initial construction and service life. The effectiveness of any nondestructive technique lies not only in detecting the presence of a defect but also providing information about the location, dimensions and the severity of the defect. Infrared thermography (IRT) has been 1

18 found to be an effective nondestructive testing technique to characterize the defects (i.e. voids, debonds, or delamination), by studying the thermal behavior of the material being tested. The term infrared thermography can be defined as a testing method, based on registering infrared part of the radiation spectrum emitted by an object. The radiation spectrum is then converted by a special camera into a color map of the object's surface temperature. Using infrared thermography, it is possible to detect defects that cause alterations of the normal distribution of the surface temperatures by affecting the rate of heat flow through the thickness of the material. So an area just above a defect shows a different surface temperature compared to the surrounding defect-free region. The next step in defect detection is the defect characterization which forms the research area of quantitative infrared thermography. Galietti et al discussed about the characterization of defects at the FRP-concrete substrate interface. In general, the more advanced standards set the limit for acceptable defects, both in terms of number and size (Grinzato et al. 2006). For example, the ACI 440-2R (2002) standard allows defects in the bonding if they are less than 2 in 2 (1300 mm 2 ), or a maximum of 10 delaminations per 10 ft 2 (1 m 2 ). The present work focuses on the quality control aspects of Glass Fiber Reinforced Polymer (GFRP) composites using infrared thermography during the manufacturing stage, initial construction, and service life of the structure. IRT can also help to effectively monitor and control temperature changes in temperature dependent manufacturing process, and ensure the integrity of the manufactured components (e.g., deck panels, tubes and flanged sections). Results, presented and discussed here, prove the effectiveness of using IRT as a tool for quality assurance of composite manufacturing systems. 1.2 RESEARCH OBJECTIVES AND SCOPE The primary objectives of this research are: (a) To conduct a literature review focusing on recent advances in infrared thermography applications for FRP composite structural components and FRP bonded concrete components. 2

19 (b) To evaluate the effectiveness of using infrared thermography to ensure quality control at a manufacturing facility producing GFRP composite components. (c) To demonstrate the usefulness of infrared thermography in the field setting for detection and mapping of subsurface defects in: (i) reinforced concrete bridge members, which are candidates for rehabilitation using GFRP fabrics, and (ii) GFRP bonded timber bridge components (e.g., piles and bents). 1.3 ORGANIZATION This problem report is organized into six chapters. Chapter 1 presents the introduction and background information about infrared thermography along with the objectives and scope of this study. Chapter 2 focuses on the literature review of some of the recent field and laboratory applications of infrared thermography technique in detecting subsurface defects in concrete and composite specimens. Chapter 3 describes the application of infrared thermography technique on composite members during their manufacturing stage in order to assess the quality control. Chapter 4 discusses the feasibility of infrared thermography technique in identifying the presence and extent of defects in the piers and pier caps of a reinforced concrete bridge. Chapter 5 presents the application of infrared thermography for subsurface defect detection in FRP wrapped members of timber bridges. Chapter 6 presents the conclusions and recommendations for future research based on the results of the present study. A list of references cited in this problem report, are presented at the end. 3

20 2 LITERATURE REVIEW ON INFRARED THERMOGRAPHY 2.1 BASICS OF INFRARED THERMOGRAPHY The ability of an object to emit thermal radiation is defined as emissivity. Thermography is the study of temperatures at different locations of an object. Infrared thermography is the technique that uses an infrared imaging and measurement camera to "see" and "measure" invisible infrared energy being emitted from an object which is then converted to a color map of temperature. This color map of temperature is called thermogram. The study of the temperature profile on the surface of the object reveals the presence of any defects it might have and also the characterization of defects. Hence, infrared thermography is a very quick and effective nondestructive technique used to detect subsurface defects such as delaminations, debonds and any subsurface anomalies. Thermal conductivity, thermal diffusivity and thermal effusivity are the two primary properties that influence the thermal behavior of a material. Thermal conductivity is the ease with which heat energy can flow from one point to another within a medium. Thermal diffusivity is defined as the measure of the rate at which a temperature disturbance at one point in a body travels to another point. The propagation of heat through any object is mainly dependent on its diffusivity (Starnes et al. 2003). Thermal effusivity is the resistance of a particular material to increase in its temperature on application of heat (Shepard 2007). There are two basic types of thermography: (a) Passive thermography and (b) Active thermography In passive thermography, the camera is simply pointed at the test specimen and from the thermal image a temperature map is constructed. Active thermography involves heating the surface of the object rapidly using an external heat source and observing how the temperature decays with time. Active approach can be subdivided to many methods like Pulsed Thermography (PT), Lock in Thermography (LT), Step heating (SH) and 4

21 Vibrothermography (VT). Flaws in the material are shown by variations in the temperature decay rate. Maldague (2002) discussed all the fundamental and practical concepts of infrared thermography and all the above mentioned techniques. The basic principle involved here is that the location of the maximum gradient of the temperature field represents the edge of the defect beneath the surface. The procedures here describe the approximation of the defect size by computing the maximum temperature gradient at the defect boundary in a thermal image by Starnes (2003). The following sections discuss the applicability of Infrared thermography in some of the cases published in previous articles and literature. 2.2 DETECTION OF AIR BLISTERS AND CRACK PROPAGATION IN FRP STRENGTHENED CONCRETE ELEMENTS USING INFRARED THERMOGRAPHY (Hu et al. 2002) Introduction Hu et al. (2002) studied about the detection of air blisters at the concrete-frp interface and the prediction of crack initiation in reinforced concrete beams partially reinforced with GFRP laminates and subjected to 3-point bending test until failure Case Study I: Detection of Air Blisters Experimental The test specimen used in this experiment was a concrete beam of size 500 mm x 100 mm wrapped with Carbon Fiber Reinforced Polymer (CFRP) laminates. The infrared imaging equipment, AGEMEA Thermovision 900 System, was used for capturing the images. The minimum precision for this instrument is 2% at 30 C and a resolution of 0.1 C Insertion of blisters into CFRP and substrate interface Plastic rings of depth 1 mm cut from plastic pipes were used to simulate the blisters. The rings were inserted at the interface of Carbon Fiber Reinforced Polymer (CFRP) and concrete specimen. The rings were modeled from commercially available pipes having diameters of 16, 18, 20 and 30 mm respectively. Figure 2.1 shows the schematic view of the beam after the insertion of the blisters. 5

22 Figure 2.1 Illustration of embedded blister locations (Hu et al. 2002) Figure 2.2 CFRP plate concrete specimens where the blisters are embedded between the composite and substrate (Hu et al. 2002) A roller was used to press the rings to the substrate before bonding the laminate, to remove any air bubbles. The finished specimens were allowed to be cured for a week before testing. Figure 2.2 shows the actual sample with the induced defects Detection of blisters at the interface Air has a low thermal conductivity and hence the underlying defects appear as hotter regions compared to the background defect-free regions. The surface with an underlying blister will remain warmer due to lower conductivity. The images were captured using an active thermographic approach by heating the surface of the specimen. Figure 2.3 clearly shows the infrared images that were taken for varying distances between the camera and the surface of the sample. 6

23 Figure 2.3 Thermal images taken at various distances for concrete samples wrapped with CFRP (Hu et al. 2002) Case Study II: Prediction of Crack Initiation Experimental setup A concrete specimen 100 mm x 200 mm x 1200 mm, partially reinforced with GFRP laminates, was placed on a jack and subjected to a three point loading condition as shown in Figure 2.4. The specimen was gradually loaded to a certain level of the ultimate load and later on subjected to cyclic loading with a specified time period. The load peak to load amplitude was set to 20% of the static load and the frequency of vibration was 3 Hz. The displacements were recorded at the center of the specimen after every stage of static and cyclic loading. Table 2.1 gives a detail of the recordings of both the loading conditions. The beam has a length (l) of 1000 mm, overall depth (h) of 200 mm and a width (b) of 100 mm. The effective depth (d), from the top fiber to the center of the reinforcement is 174 mm. The test specimen is shown in Figure

24 (a) (b) Figure 2.4 Configuration of test beams (a) Elevation (b) Cross section (Hu et al. 2002) A series of thermal images were taken during each phase of the loading for detecting the potential failure plane. The generated heat through fretting and friction action at the point of development of cracks indicated the failure plane as a hot spot on the thermogram with a higher temperature compared to the surrounding uncracked regions. Table 2.1 Displacements of 1.2 meters RC beam subjected to static and cyclic loading tests (Hu et al. 2002) Results and Discussions Comparison of the size of blisters from measured and infrared images From the thermal images shown in Figure 2.3, it was evident that even if the distance between the surface of the sample and the transmission line of the IR camera lens was increased upto 20 meters, the air blisters were still clearly identified by the camera. If the interface between CFRP and the concrete substrate had no voids (excluding blister locations) four circular shapes with higher temperature were expected 8

25 to be observed. However the above images showed the distorted shapes indicating poor workmanship in applying the defects. This was proved after the CFRP laminate was removed from the beam. Figure 2.5 shows the image of the beams after the removal of the CFRP. Figure 2.5 Actual photograph of the concrete samples after the CFRP laminate was removed (Hu et al. 2002) The specimens with voids (highlighted areas in the above figure) where the external heat from the surface is trapped were identified. By comparing the actual photograph of the sample before and after testing, the potential of the thermographic technique is justified Prediction of crack initiation The crack formation was expected to be at a faster rate due to the cyclic loading and hence the heat generation occurred through areas where discontinuities were likely to be present. Figure 2.6 shows the thermal images taken for a cyclic load range of kn, where the area highlighted with the white dotted circle shows a higher temperature zone than the surrounding areas due to dissipation of heat energy and it was said that the possible failure plane occurs at a 45 angle to the highlighted triangular region. Figure 2.7 shows the thermal image of the sample taken immediately before the failure occurred. It is observed that the actual failure plane shown in Figure 2.8 was very much in agreement with the predicted failure plane from Figure

26 Figure 2.6 Thermal image taken during 45-55kN cyclic load range (Hu et al. 2002) Figure 2.7 Thermal image taken after the failure occurred at load range 67.5kN-82.5kN (Hu et al. 2002) 10

27 Figure 2.8 Actual photograph taken for the test specimen after the failure has occurred (Hu et al. 2002) Conclusions Based on the above experiments by Hu et al. (2002), the following important conclusions can be made: The blister locations between the GFRP and the concrete could be identified remotely from a distance of 20 meters from the surface. The simulation of the defects reflecting the poor workmanship was clearly identified using infrared thermography. The size of the blisters could be estimated approximately if the distance between the IR camera and the surface of the sample is available. The energy lost due to rubbing action between potential failure planes was identified by IR thermography immediately prior to failure and it provides a warning signal before failure. 2.3 IRT SURVEY FOR THE QUALITY CONTROL OF FRP REINFORCED RC STRUCTURES (Corvaglia et al. 2008) Introduction Corvaglia discussed about the in-situ validation of two reinforced concrete structures retrofitted with FRP material. The study focused on the analysis of a FRP reinforced concrete sample with known defects applied appropriately at the FRP and concrete substrate interface. 11

28 Acceptance criteria and repairing techniques The areas strengthened with FRP should be inspected for detecting some of the following defects, according to the quality control procedure. voids and air encapsulation between concrete and layers of primer, resin, or adhesive and within the FRP system itself; delaminations between layers of FRP system; broken or damaged edges of the FRP system; wrinkling and buckling of fiber and fiber tows; discontinuities due to fracture of fibers, breakage in the fabric, or cracks in precured shells; cracks, blisters, and peeling of the protective coating; resin-starved areas or areas with non-uniform impregnation or wet-out; under cured resin; incorrect fiber orientation The acceptance of defects mainly depends on the size, location and the number of defects in a specific area. Meanwhile, the repair methods depend on the form of defect, type of material used and the level of degradation. Delaminations can be repaired by injection of epoxy resin, Localized FRP laminate cracking or abrasions affecting the integrity of the laminate are categorized as middle level damage and can be repaired by bonding FRP patches with similar characteristics. Major damages such as peeling and debonding of large areas may require removal of the affected area, reconditioning of the cover concrete, and replacing the FRP laminate Infrared thermography procedure calibration According to the ICC Evaluations Service, INC 2003, AC 178, and ACI 440.2R- 2002, a certain amount of defectiveness at the FRP-concrete interface can be accepted. In particular, air void having diameter of 3.2 mm and delaminations of less than 1300 mm 2, with a maximum of 10 delaminations per 1 x 10 6 mm 2, are allowed. 12

29 Experimental setup On the basis of the above criteria, the CFRP reinforced concrete sample was accurately prepared, imposing two types of defects with different shapes and dimensions as shown in Figure 2.9. A commercial halogen lamp with a nominal power of 500 W was used as heating source, with a heating time of 30 seconds. A commercial microbolometric FLIR System with a 320 x 256 focal plane array was used for capturing the infrared images. Figure 2.10 shows the actual sample. Figure 2.9 Imposed known defects (Corvaglia et al. 2008) Figure 2.10 Sample preparation (Corvaglia et al. 2008) The following defects have been imposed on the specimen to be tested. (a) Delaminations, by means of two bonded Teflon foils: Defects no. 5 and 6 (100 mm x 10 mm) Defects no. 1 and 4 (65 mm x 20 mm) Defects no. 7 and 8 (40 mm x 40 mm) Defects no. 2 and 3 (20 mm x 20 mm) 13

30 (b) Lack-of-bonding of the FRP, by means of a simple Teflon foil: Defects no. 13 and 14 (100 mm x 10 mm) Defects no. 11 and 12 (65 mm x 20 mm) Defects no. 15 and 16 (40 mm x 40 mm) Defects no. 9 and 10 (20 mm x 20 mm) The sample surface was analyzed with pulsed thermography (PT). Pulsed phase thermography (or PPT) is a processing technique in which, a heat pulse is generated to heat the specimen and the spectrum of thermal waves sent into the specimen is unscrambled using the Fourier transform of the temperature decay on a pixel by pixel basis thus enabling computation of phase images. As an example, for each pixel (i, j), the temporal decay f(x) is extracted from the image sequence (where x is the index in the image sequence). Next, from f(x), the discrete Fourier transform F(u) is computed (u being the frequency variable). Finally, from the real R(u) and imaginary I(u) components of F(u), the phase is computed. In PPT, it is possible to explore the various frequencies u. After the definition of acceptance criteria and repair techniques for the defects, a MATLAB routine was developed for post processing of thermograpms. The defective area was measured to verify if a defect is acceptable or not and what is the best way to repair it Calibration results Thermographic data were processed with Almond s half maximum contrast method. Almond s method is based on the evaluation of the thermal contrast of the thermogram, defined as the difference between the temperature of the defective region and temperature of the surrounding defect-free region. Also, the pixel must be considered part of the damaged zone if it is characterized by a contrast equal or greater than the half of the maximum contrast relative to the damaged zone (Almond et al. 1997). Thermal contrast can be defined as: 14

31 where T i (t), T i (t 0 ) are the temperatures of the defective regions at times t 0 and time t and T s (t), T s (t 0 ) are the corresponding temperatures for sound regions. Typical thermogram of the CFRP-reinforced sample is shown in Figure Two important set up parameters were investigated: The impulsive heating time, named Heating Time (HT); The time, after the heating step, at which the thermogram should be recorded, named Observation Time (TOb). Figure 2.11 Thermogram of the CFRP sample with imposed defects (Corvaglia et al. 2008) Figure 2.12 shows the partial results for the percentage error in assessing the defect no. 3 for different heating times and different times of observation. Defect no.3: CFRP Figure 2.12 Percentage error in the assessment of defect number 3 for different experimental set-up (Corvaglia et al. 2008) 15

32 2.3.2 Discussions Some interesting remarks can be pointed out: The error, defined as the difference between the estimated and the real number of pixels relative to a defective area, divided by the real number of pixels relative to the same defective area, increases with increasing observation times. Statistical Analysis Of Variance (ANOVA) indicates that the investigated parameters (Time of Observations and Heating Time), have no mutual influence. Thermograms recorded at a Time of Observations of about 20 seconds are not useful. The lower percentage error could be obtained with a heating time period of seconds Case Study I: ex Law Courts of Sant Angelo dei Lombardi Based on the above experimental results a case study of the GFRP reinforced structure was taken up. The plan was a Y shape and the building had 3 levels with an overall span of 10 meters. The structure was mainly investigated with the infrared thermography survey procedure developed using the experimental setup shown before. Surveys were performed with an active thermographic approach in the interior of the structure, using a commercial halogen lamp with a nominal power of 1000W. Figure 2.13 shows the indoor IRT survey of an R.C structure reinforced with GFRP sheets. As seen from the image there are a lot of defects identified. These defects were probably caused due to improper application of the GFRP fabric and also due to the rollers used for removing any air pockets after the application of the reinforcement. The defects were categorized as type 3. Due to high percentage of defective areas (as indicated by the blue squares in the infrared image), it was proposed to remove and reapply the entire FRP system. 16

33 Figure 2.13 Indoor IRT survey of an R.C structure reinforced with GFRP sheets (Corvaglia et al. 2008) In Figure 2.14, a thermogram recorded during an outdoor IRT survey is shown. The outdoor surveys were performed using solar radiation as the main heating source, both during the heating and cooling period. For most part of the GFRP reinforced beamcolumn joint, perfect bonding was identified. However, the IRT procedure was able to identify some small bonding defects on the upper part of the joint. These defects have lower thermal conductivity with respect to the GFRP concrete interface and hence appear hotter than the surrounding defect-free region, during the heating phase (Figure 2.13), and colder during the cooling phase (Figure 2.14). Although the defects were classified at the limit between a type 2 and a type 3, in case of contact critical FRP application (e.g. columns or joints wrapped for confinement), low pressure epoxy injection was suggested as a valuable method for repairing. The inspected area showed a lot of defects of different shapes and sizes probably caused due to improper use and placement of the FRP sheets and the use of the rubber roller (which should help the adhesive to completely penetrate through the fibers of the fabric) and of the aluminium roller with a worm screw (which should help to completely eliminate air bubbles formed during application). The distribution of defects in this case was very significant and it was advised to completely remove and reapply the entire FRP sheeting system. 17

34 Figure 2.14 Outdoor IRT survey of an RC structure reinforced with GFRP sheets (Corvaglia et al. 2008) Case Study II: Highway Reinforced Concrete Single Span Bridge The inspected bridge Torrente Casale, was redesigned in order to satisfy the new traffic demands. Updates to the new seismic codes, led to the recalculation of the bridge capacity for both gravity and seismic loads, in order to account for new design provisions. After the analysis was performed, it was seen that the piers were not adequate to bear the seismic actions. Thus, an innovative rehabilitation technique based on the combined use of Carbon Fiber Reinforced Polymer (CFRP) and Steel Reinforced Polymer (SRP) was adopted. An active infrared thermography technique was used for the evaluation purpose and a commercial halogen lamp of 1000W was used as heating source. Figure 2.15 Infrared thermographic survey of a bridge pier reinforced with CFRP and SRP materials (Corvaglia et al. 2008) 18

35 In Figure 2.15, two types of defects were described. The first type (white square in Figure 2.15, left) was caused by the longitudinal joining of two CFRP sheets. When applying multiple plies of FRP sheets for columns wrapping, or when two strips need to be joined longitudinally, it is necessary to create an overlapping of at least 200 mm. Overlapping is not necessary for the width of the fabric because the different strips of fabric can simply be applied one next to the other. For this study, the strategy of 30 mm overlapping between the two CFRP strips, along the width, was adopted (also evidenced in Figure 2.16, right). According to the IRT survey results, it was concluded that this kind of approach can cause a step effect resulting in a lot of air inclusion all along the column diameter. These defects were suggested to be treated using low-pressure epoxy injection. Figure 2.16 CFRP strip transversal overlapping (Corvaglia et al. 2008) Figure 2.17 Application of SRP reinforcement (Corvaglia et al. 2008) The second type of defect (red circle in Figure 2.15 left and in the white rectangle in Figure 2.17, left) was probably due to the presence of the SRP reinforcement (applied 19

36 between the concrete and the CFRP sheet, as described in Figure 2.17, right) and was classified as a Type 4 defect. However, considering that the debonded zone was located in a reinforced area designed as witness panel (i.e. not resulting from the design of the reinforcement but strengthened only to perform other tests, like direct pull-off test), it was suggested to be repaired with low-pressure epoxy injection Conclusions A reliable infrared thermography based technique was developed as a tool for the evaluation of hidden defects in FRP reinforced concrete structures. Acceptability of bonding defects depends on the size, location, and number of defects in a specific area, while the repair methods depend on the type of material, the form of degradation, and the level of damage. Minor delaminations could be repaired by epoxy resin injection, middle level damages, including localized FRP laminate cracking or abrasions can be repaired by bonding FRP patches (with the same characteristics, such as thickness and ply orientation, as the original laminate) over the damaged area; major damages (including peeling and debonding of large areas) can require removal of the affected area, reconditioning of the cover concrete, and replacing the FRP laminate. The developed IRT procedure was applied for quality control of two R.C structures reinforced with FRP composite. In particular, the developed procedure helped quantify each defect in terms of debonded area and a specific repair technique was suggested for each defect. In cases where defectiveness was very high, it was suggested to remove the existing FRP and reapply the entire system. 2.4 EVALUATION OF BONDED FRP STRENGTHENING SYSTEMS FOR CONCRETE STRUCTURES USING INFRARED THERMOGRAPHY AND SHEAROGRAPHY (Taillade et al. 2009) Introduction Taillade et al. (2009) discussed a method for the characterization of the defects occurring at the FRP-concrete interface. The method was also used to assess the bond quality of the FRP to the concrete substrate. The method is based on two techniques namely shearography and infrared thermography. The theoretical and analytical aspects 20

37 for both the methods were discussed and then the method was demonstrated from the obtained data Nondestructive Evaluation of FRP Bonding by Shearography The shearography technique is basically a speckle interferometric technique that can be applied to the detection of debonded areas in a structure composed of a concrete substrate, one layer of adhesive, and one layer of carbon-epoxy composite. Taillade et al. (2006) made an attempt to assess the bond quality of FRP using shearography. The following sections describe briefly about the principle of the shearography technique, the testing method involved and also the experimental results Principle of the shearography technique The principle of shearography consisting of an interferometer with a video split, is to create the interference of two waves that are submitted to nearly the same random fluctuations in optical path during their trajectories. The difference of these optical phases depends on the deformation of the object. In the case of plane waves, for directions of illumination and observation that are normal to the plate, the phase difference is expressed at the first order by: where is the amplitude of the displacements normal to the object surface, is the illumination laser wavelength and is the shear distance. The phase difference equivalent to noise is roughly 2 /50 (noise including calibration procedure). Without any particular precaution, displacement difference can thus be mapped with a 5 nm uncertainty. According to Taillade, the excitation method involves the application of a partial vacuum (or depressure) ΔP to the surface of the sample by means of a suction cup. The difference in pressure between the blade of air inside the defect and the surface subjected to the stress creates a bump shaped deformation into the defect. The depressure which should be applied to obtain a measurable deformation by shearography, can be very weak 21

38 (only a few Pascal). It depends primarily on the mechanical characteristics of the surface material (elasticity coefficients) and of the width-to-depth ratio of the defect. (a) (b) (c) Figure 2.18 Concrete slabs with bonded FRP reinforcements containing different types of defects (Taillade et al. 2009) (a) sample 1 variable diameters (mm) or (b) sample 2 variable adhesive properties and (c) experimental setup Experimentations and results Figure 2.18 a, 2.18 b show the concrete sample that were prepared with simulated defects. A Teflon disc of 0.5 mm thickness was inserted at the interface of the concrete- FRP, in place of the adhesive materials. Four discs of different diameters (40, 30, 20 and 10 mm) were used in sample 1. Sample 2 had four discs of identical diameters (40 mm) drilled with holes (in number and size represented by the percentage of remaining disc mass) in order to simulate different qualities of adhesive bond. The deformation was visualized through a suction cup with dimensions 180 mm x 180 mm x 70 mm and made of Plexiglas chamber, with a 20 mm wall thickness as shown in Figure 2.18(c). The phase difference for sample 1 was measured through the suction cup. The magnitude of pressure was about 100 hpa ± ΔP/2. Figure 2.19 shows the shearograms for different sizes of the defects introduced into the sample 1. It was observed that the applied depressure increases when the diameter of defects decreases. In the case of sample 2, as shown in Figure 2.20, it was observed that when the percentage of the hole on the disc increases, higher depressure had to be applied in order to measure an optical phase difference of the order of 2π. Furthermore, these results were confirmed by a finite elements analysis. 22

39 (a) (b) (c) (d) ΔP = 8.0 hpa ΔP = hpa ΔP = 26.6 hpa ΔP = hpa Figure 2.19 Strain measurements on sample 1 for various sizes of defects, and under different levels of partial vacuum (Taillade et. al. 2009) (a) Adhesive 0% (b) Adhesive 21% (c) Adhesive 30% (d) Adhesive 45% ΔP = 8.0 hpa ΔP = 53.2 hpa ΔP = 79.8 hpa ΔP = hpa Figure 2.20 Strain measurements on sample 2 with a defect diameter of 40 mm, and for different qualities of the adhesive bond expressed as the ratio between surface of the bonded areas and the surface of a plain PTFE disc (Taillade et al. 2009) Infrared Thermography Technique The basic principle involves the heating of the surface over time period τ and then by following the heat pulse, measuring the temperature distribution on the sample surface by using an infrared camera. A lot of research was done by Galietti et al. (2007), and Lai et al. (2009) in order to assess the bond quality between concrete-frp interface. The experimental setup for the present study is shown in Figure The detection and characterization of the subsurface defects is possible with the emergence of thermal contrast after the application of heat pulse. The depth of the defect can be obtained from the time associated with the maximum contrast. The depth of the defect, D defect can be computed using the equation: 23

40 Experimental setup The test sample was a concrete slab 400 mm x 300 mm in dimensions, manufactured and reinforced by layers of CFRP plates 2 mm thick, as shown in Figure The debonds were introduced by locally replacing the polymer adhesive at the interface by Teflon discs 0.5 mm thick, placed either between the concrete surface and the lower CFRP plate, or between 2 adjacent CFRP layers. The final specimen had embedded discs of 3 different diameters 10, 20 and 30 mm, located at three different depths 2, 4 and 6 mm as shown in Figure Figure 2.21 Experimental setup (Taillade et al. 2009) (a) Top view (b) Profile view Figure 2.22 Concrete specimen reinforced with bonded CFRP plates, containing calibrated defects (Taillade et al. 2009) The principle of the square pulse heating approach involved heating the surface of the composite for some time period and measuring the temperature distribution on the sample surface with an infrared camera. Detection and localization of the subsurface defects were then performed using adequate image analysis approach. The 24

41 characterization of the subsurface defects was achieved by monitoring the thermal contrast between defect-free and defective areas as shown in Figure 2.21 right, after the pulse illumination (thermal relaxation phase). The thermal is expressed by: ) where T sound and T are respectively the temperature above sound and faulty regions. The specimen was subject to a heat pulse for a period of 50 seconds. Figure 2.24 shows thermal images of the sample taken at three different times corresponding to maxima of the thermal contrast. The thermal contrast was computed above the larger defect (diameter = 30 mm). The thermal diffusivity of the carbon/epoxy plate is an unknown parameter. When the case of t max = 1.1 seconds was considered for the defect depth of 2mm, the thermal diffusivity of the composite can be derived from equation 2.3. Thermal diffusivity is defined as the ratio of the thermal conductivity to the heat capacity. It gives a measure of the rate at which heat travels through material. Polymer composite materials usually have intermediate thermal diffusivity that is well suited for non-destructive techniques. Table 2.2 shows the thermal diffusivity of some of the commonly used materials (Shepard 2007). Table 2.2 Thermophysical properties of selected materials (Shepard 2007) Material Diffusivity (m 2 /s) Carbon fiber reinforced polymer * 10-7 Glass Fiber reinforced polymer 1.667* 10-7 Aluminum * 10-7 Steel * 10-7 Glass 6.99* 10-7 Epoxy 1.372* 10-7 The thermal diffusivity in this case for a carbon reinforced polymer was * 10-7 m 2 /s, as shown in the above table. The thermal diffusivity was computed for defects at 4mm and 6mm depths in order to find the time. The experimental values obtained were 4.3 seconds and 11 seconds which could be compared to the theoretical 25

42 values of 4.4 seconds and 10 seconds. Figure 2.23 shows the maxima of the thermal contrast at different t max. Figure 2.24 shows a plot of the thermal contrast versus the time computed for the three zones above the larger defects. The results were very much in agreement with theoretical values but there was an increased deviation observed for the defect at the depth of 6 mm, since measured values were very much close to the ambient noise in that case. Figure 2.23 Maxima of the thermal contrast taken at different t max (Taillade et al. 2009) Figure 2.24 Thermal contrast v / s. time computed to three zones above the larger defects (blue line for a depth 2mm, green line for a depth 4mm and red line for a depth 6mm (Taillade et al. 2009) Conclusions In this study, the principles of shearography and infrared thermography techniques and their application for the evaluation of the bond quality between concrete and external FRP reinforcements were discussed. Results demonstrated that shearography has the advantage of determining not only locations and areas of defects but also allows 26

43 the evaluation of adhesion quality in the case of a partial debonding. Infrared thermography has proved to be a simple method to inspect repaired structures in a qualitative way (detection of the bonding defects) and a further analysis of the thermograms enables to quantify the depth of the defects. 2.5 HEATING METHODS AND DETECTION LIMITS FOR INFRARED THERMOGRAPHY INSPECTION OF FIBER-REINFORCED POLYMER COMPOSITES (Brown et al. 2007) Brown et al. (2007) discussed about three heating methods in his study of the detection of defects in FRP-concrete systems. The study mainly focused on the detection of air voids and epoxy filled holes in FRP system bonded to concrete. The three techniques namely, flash, scan and long pulse were used for detecting the anomalies and a quantitative analysis of the resulting images and data was performed in order to establish the detection limits for each technique Experimental Setup and Design Specimen construction Table 2.3 shows the mix proportion for the construction of the specimen. Four concrete blocks were cast each having dimensions of 12 inch x 6 inch x 2 inch (304.8 mm x mm x 50.8 mm), with a target slump of 3 to 4 inches. This was a non air-entrained mixture. Prior to consolidation, concrete was allowed to cure in the formwork for a period of 2 days. Each specimen received a light sandblasting and the FRP was applied to the surface which was in direct contact with the steel formwork. 27

44 Table 2.3 Material properties (Brown et al. 2007) Simulated defects were introduced at the interface of concrete-frp for all the four specimens. The defects were modeled by drilling several holes on the surface, three at 0.25 inches (6.35 mm), and three at 0.5 inches (12.7 mm) and two at 0.75 inches (19.05 mm) to a depth of 0.25 inches (6.35 mm) into the concrete. These defects depicted bug holes and other surface imperfections. The holes were filled with epoxy tack coat and polyurethane insulating foam. Three holes were left empty to represent air filled defects. A small nylon machine screw (no. 8) was also inserted into the surface of concrete, in each specimen representing an interface bubble. The screw was inserted in such a way that, it protruded 0.12 inches (3.05 mm) above the surface of the concrete before the application of FRP. Table 2.4 shows the dimensions of the interface bubble, measured both in parallel and perpendicular direction to the principle fiber direction. Table 2.4 Series A Properties (Brown et al. 2007) 28

45 The specimens A-1 through A-4 were constructed using 1, 2, 3 and 4 layers of FRP composite respectively. After a light sandblasting to each specimen, a thin layer of epoxy saturant was applied to concrete surface using a 4 inch wide velour paint roller. The epoxy tack coat was applied after one hour from the application of the epoxy saturant. Figure 2.25 shows the finished samples. Figure 2.25 Defect configuration for Series A specimens (Brown et al. 2007) Thermal imaging system The thermal imaging system consisted of an infrared camera operating in the wavelength range of 8 to 12 m. Every pixel of the image was stored as a temperature value and MATLAB or any other proprietary software could be used for post-processing. The infrared camera could save images at a rate of 5 frames/second Heating methods The signal generated by the defects was primarily dependent on the heating phenomenon. Non uniform heat flux may result in thermal gradients in the x-y plane. The uniform heating criterion is met only if one can record the thermal images at the time of heating and also when total area under inspection is heated simultaneously. Two photography flash systems each with an energy rating of 3.3 kj, were used for the flash technique. The specimen was subjected to flash heat for a period of 50 milliseconds. Figure 2.26(a) shows the test setup for this experiment. 29

46 Figure 2.26 Heat source and camera configuration (Brown et al. 2007) (a) Flash heating (plan view) (b) scan heating (c) long pulse heating (plan) and (d) long pulse heating (profile view) Scan heating involves the movement of the heating equipment constantly along the surface of the component and an IR camera is positioned to record the thermal images as the component cools down. Two 500W halogen lamps were used as heating equipment for this technique and the test setup is shown in Figure 2.26(b). The heat source was moved from left to right at a speed of 0.87 in/s (22.1 mm/s). Any point on the specimen was exposed to a heat for a total of 12 seconds. The long pulse heating method configuration is shown in Figure 2.26(c) and Figure 2.26 (d). Four halogen lamps were used as heat source. A square heat pulse was imparted to the specimen for 30, 45 or 60 seconds duration Results and Analysis Three techniques namely flash heating, scan and long pulse methods were used to detect the presence of fabricated defects. Images were also saved for an additional 240 seconds during the cooling phase. The images taken during heating were not considered for analysis as they reflected only the IR energy emitted from the heat source. 30

47 Comparison of background surface temperature increase The variation in the temperature of the defect-free regions is called the background temperature increase. The surface temperature profile for the specimen A-1 (reference sample) was measured in both vertical (L1) and horizontal (L2) directions, passing through the center of the specimen (Figure 2.27). The T max and T mean were measured along each temperature profile and norm was obtained by taking the standard deviation of the normalized temperature increase (Table 2.5). Figure 2.27 Thermal image collected at end of heating (Brown et al. 2007) for (a) flash heating (b) scan heating and (c) long pulse heating (60 second duration) and temperature profiles for (d) vertical line (L1) and (e) horizontal line (L2) Table 2.5 Surface temperature increase results for different heating methods (Brown et al. 2007) 31

48 The scan heating method produced the highest mean temperature increase along both the vertical and horizontal profiles but the major drawback of this method was that the surface under consideration was not heated at the same time. The rate of temperature increase for the specimen and the duration of heat pulse are dependent on the rate of the movement of the heating source along the surface. Up to six specimens were able to be heated during each experiment in case of long pulse heating making it a very effective technique for heating large areas simultaneously. A large temperature gradient was generated in the x-y plane and the largest temperature increase was observed at an area closest to the source. The flash technique provided a temperature increase of 5.2 C and 5.8 C along the horizontal and vertical profiles respectively. But the major shortcoming with this method was that only a small area could be inspected every time, making its scope of application very limited. Figure 2.27(a) shows the thermal images and the temperature profiles generated for the specimen A Defect Analysis Defect signal strength A contrast based approach was used in order to perform quantitative analysis for each embedded defect. The term T def was evaluated by taking the difference between the defective area and the surrounding defect-free area. In this study, an alternate approach was proposed in order to study the variation of T def with respect to time. An area around the defect was identified and highlighted in the form of a rectangle, on the image. The width of this rectangle is one pixel. The location of this rectangle was such that its sides were sufficiently distant from the defect, as shown in Figure Now T def was computed for each rectangle using the equation: where T max is the maximum temperature of the rectangle and T per_avg temperature along the perimeter of the rectangle. is the average 32

49 Figure 2.28 Area identification for defect analysis (Hamilton et al. 2007) (Defect A-75 Specimen A-1).Thermal image and corresponding surface plot for properly defined defect area The variation of T max, T per_avg and with the time (for flash heating) were represented as shown in Figure 2.29(a). The second plot was used to characterize the defect and also to establish the level of detection. There were a number of thermal images and a number of versus time plots mainly because of the number of defects analyzed during the study. Figure 2.30, Figure 2.31(a) show the thermal images for interface bubble (IB) defect and the versus the time plot for the interface bubble defect on specimen A-3. The plot reported a value of 3.25 C at t = 0 seconds and slowly it decayed to 2.5 C at t = 12 seconds. The signal then showed an upward trend until it finally reached an absolute maximum at t = 40 seconds. The false signal between t = 0 seconds and t = 12 seconds was due to minor imperfections and non uniform 33

50 heating. Figure 2.31(b) shows the plot of the same variables as in the above case, starting with a false signal of 1.6 C at t = 0 s, it decreases with a linear slope until t = 15s.The thermal image indicated the presence of a defect but the plot does not give a well defined value for and t max. Figure 2.31(c) shows a plot of minus. When series B was subtracted from series A, the resulting curve C gave an indication about the time when the defect begins to dominate the signal. The minimum for which a defect is classified as detected was 0.2 C. Figure 2.31(d) shows the same plot for an undetected defect E50 in specimen A-3. Figure 2.29 Constructing T def versus time plots from area parameters (Brown et al. 2007) Defects A75, Specimen A-1 (flash heating); (a) T max and T per _ avg versus time and (b) T def versus time Figure 2.30 Thermal images for Defect IB (Specimen A-3) (Brown et al. 2007) (a) t = 0 s; (b) t = t b = 12 s and (c) t = t max = 40 s. 34

51 Figure 2.31 Non uniform heating and weak signals for defects (Brown et al. 2007) (a) versus time for defect IB in specimen A-3(long pulse heating- 45 seconds); (b) versus time for defect E75 in specimen A-3(long pulse heating-45 seconds); (c) versus time for defect E75 with perimeter difference removed; and (d) versus time for undetected defect(e50 in specimen A-3, long pulse heating 45 seconds) Figure 2.32 shows a series of bar diagrams for the selected defects (shown at the right top corner of each figure) where the x-axis denotes the heating method and the y- axis shows for the defect. is dependent on factors like size, depth, material composition of defect, intensity of heat source and duration of heating. was found 35

52 highest for scan heating and lowest for pulse heating. A similar trend was observed for the two, three and four layered specimens. Normalized ( ) is more useful for investigating effects of pulse duration on defect detectability. The results of the normalized for different defects, are shown in Figure Once a suitable was established, the normalized plots indicate how much must be generated to develop the signal for a specific type of defect. Figure 2.32 Comparison of Tmax for different heating methods (Brown et al. 2007) (a) Specimen A-1; (b)specimen A-2; (c) Specimen A-3 and (d)specimen A-4 36

53 Figure 2.33 Comparison of normalized T max for different heating methods (Brown et al. 2007) (a) Specimen A-1; (b)specimen A-2; (c) Specimen A-3 and (d)specimen A Defect area computations One method for determining size of the depth is to draw a line around the boundary of the defect and count the number of pixels. By applying a length factor and measuring the area, the approximate size could be estimated and the method is called boundary trace method (Figure 2.35(a)). After applying the length factor of 1.1 mm/pixel to the area which was bounded by 377 pixels, the area of the defect was estimated to be 440 mm 2. But the true area was 280 mm 2. Additional study on this method focused on results from 60 seconds pulse experiments as shown in Table 2.6. The errors associated with measurements indicated that boundary trace method can over-predict the true area by 60 to 300%. Next the gradient area method of defect characterization was discussed. The gradient image for defect A75 (specimen A1) is shown in Figure 2.35(b). The defect boundary was first identified by considering the pixel with the smallest gradient near the 37

54 center of the defect. A line was constructed between the center and the upper left corner of the area with the defect, and the location of the maximum gradient along this line was saved as a point on the boundary of the defect. A new line was then constructed from the center point to the pixel on the border just below the upper left pixel. This process was repeated to get several points and once the boundary was established the number of pixels contained in the area were added up and converted to physical units using an appropriate factor. Figure 2.34 shows images of the defects that were clearly visible but did not yield good results when gradient area method was applied. For Figure 2.34(a) (defect A75, specimen A-3), was very low (0.6 C) and hence a well defect boundary could not be generated. Defect signals in non-uniform heating regions also hamper the gradient area method. Figure 2.34(b) shows defect E75 of specimen A-3 for a 30 second long pulse heating. The maximum temperature gradient in the defect region subject by nonuniform heating could be influenced by noise. Coefficient of variance (COV) of radius values generated by gradient area method could assess the quality of the defect boundary. COV is defined as the standard deviation divided by the mean. Figure 2.34(a) produced a COV of 0.54 while for the well defined defect in Figure 2.35(b), the value was Figure 2.34 Reduced accuracy in area computations (Brown et al. 2007) due to: (a) weak signal (b) non-uniform heating and (c) insufficient pixel resolution 38

55 Table 2.6 Actual and measured areas for select defects (60 second pulse heating) (Brown et al. 2007) Figure 2.35 Area computations for defect A75 (Brown et al. 2007) (a) boundary trace method (b) gradient area method (c) surface temperature profile and (d) temperature gradient profile 39

56 Proposed method for characterizing detectability A total of 140 versus time plots were generated and to obtain the result set for characterizing defect detectability, a method was proposed. The shape of the plot was the first distinction and four levels of classification were done. A positive slope of t > 0 was assumed for level I and obtained a maximum value at t max (Figure 2.29(b)). Level II starts with negative slope and reaches local minimum at t b and after that curve follows a positive slope until it reaches a local maximum at t max (Figure 2.31(a)). Level III begins with a negative slope and continues the same trend (Figure 2.31(b)). Level IV describes about the undetected defects. Level I defects appeared in images immediately after removing heat source. Level II defects were also well defined on images but the time at which the maximum value occurs must be considered. Level III defects required manipulation of image color scale in order to extract the shape (Figure 2.34(b)). Level IV defects were the undetected defects and additional techniques and procedures may allow their detection possible. COV of the computed radii was the basis for second distinction and a new quantity COV which is the difference between the computed COV of the defects and the inherent COV for the shape of the defect was used. There were three categories A, B and C for this classification, wherein A describes the well defined defects with a COV less than Category B was for moderately defined defects with COV between 0.21 and 0.40 and the last category was for the poorly defined defects with COV between 0.41 and 1.0. The general detectability results for flash heating were shown in Table 2.7 Figure 2.36 shows a plot of versus COV and it had about 92 points showing the defects which were detected. Higher values of COV were likely to occur if max < 2 C. From the plot it was easier to classify defects as: Well defined defects if max > 2 C Poorly defined defects if max < 2 C 40

57 Table 2.7 General detectability results (Brown et al. 2007) Figure 2.36 Maximum signal versus radii COV for all detected defects (Brown et al. 2007) 41

58 2.5.4 Conclusions The study involved three basic methods of heating and helped in the classification of defect detectability. The quality of the defect boundary was also an aspect being discussed. The defect boundary was computed from gradient image and a COV of the defect s radius was used in the classification. Flash heating was able to detect air filled defects in a single layer CFRP system. Scan heating and long pulse heating techniques could detect air filled and epoxy filled defects even in 4 mm thick CFRP systems. But one disadvantage of long pulse heating was that the surface could not be heated uniformly. The ultimate choice of the heating method still depends on the desired objectives. 2.6 IR THERMOGRAPHY FOR INTERFACE ANALYSIS OF FRP LAMINATES EXTERNALLY BONDED TO RC BEAMS (Valluzzi et al. 2009) The bond quality at the interface of the reinforced concrete beams wrapped with CFRP laminates was evaluated by thermographic analysis before loading and under loading conditions, during lab experimentation. Small scale specimens with artificial defects embedded at the concrete-frp interface, were first tested. This was done to select the algorithm for data processing acquired from the thermographic system. From the results obtained, it was shown that infrared thermography is a very powerful tool for qualitative and quantitative analysis of the concrete structures bonded with FRP laminates. Cross-evaluation of crack patterns during bending tests and thermographic results were also discussed Preliminary Thermographic Tests on Concrete Samples Two concrete slabs 400 mm x 400 mm x 50 mm, were strengthened with three CFRP strips. The failure of the first sample was expected and simulated at the FRPadhesive interface. A thick layer of resin was used at the interface for the second sample, to level off the surface for proper placement of laminates. For this sample the failure was expected within the thickness of the resin. Figure 2.37 shows the phases of application of the defects at the interface location. 42

59 Figure 2.37 Preparation of samples with artificial defects (Valluzzi et al. 2009) Positioning of defects on (a) smooth and (b) rough surfaces and (c) application of laminate strips Some of the variables are: the type of defect best representing real debonding (Teflon of various sizes, silicon grease, and nylon for packaging); correct in-depth location of the defect (at interface between CFRP and resin or resin-concrete). Defects were created from Teflon strips 10 mm wide, 10, 25 or 50 mm long, and about 30 µm thick, positioned on a side strip as single layers, about 60 µm thick, placed on the central strip in double layers; silicon grease, about 80 µm thick, covering areas of 20 mm x 9 mm x 50 mm, 20 mm x 9 mm x 30 mm or 20 mm x 9 mm x 20 mm, and nylon squares, 25 mm x 9 mm x 25 mm or 10 mm x 9 mm x 10 mm wide and about 100 µm thick, on the last strip. The equipment used for the preliminary tests consisted of a pair of flash lamps generating 2400 J each in about 10 seconds, and a FLIR ThermaCAM SC3000 (thermal resolution 30 mk). The active approach was used here and the temperature versus time plot was established by maintaining 800 mm distance from the surface. The images were captured at a frequency of 20 milliseconds and hence each surface element could be followed in its cooling phase and compared with the others by means of algorithms. Figure 2.38 label 1 shows the pictorial representation of the differences in results obtained through PCA algorithm (Marinetti et al. 2004), Pulse phase Thermography (PPT) and Thermal Tomography (TT). The images showing imperfections at the interface are shown in Figure Data processing by PCA was suggested for in-situ analysis and is very much recommended in order for better defect detection and quantification. 43

60 Figure 2.38 Thermographic processing (central strip of rough surface) by various algorithms, indicating main discontinuities (artificial defects are incorporated in label 1) (Valluzzi et al. 2009) (a) second principal projection, by PCA (b) second phase map, by PPT (c) map of time of maximum normalized contrast, by TT IRT Tests on FRP-Strengthened RC Beams Thermographic tests were performed on two full-scale beams (10 m long and 300 mm x 90 mm x 500 mm in section) having ordinary steel reinforcement (Figure 2.39 and Figure 2.40). Both beams were strengthened at the underside with pre-impregnated CFRP laminates (1.2 mm thick, 80 mm wide), with a tensile strength of about 2,800 MPa and an ultimate deformation of 1.8%; the elastic modulus was about 166 GPa. Figure 2.39 Setup for bending test on R.C beams (Valluzzi et al. 2009) 44

61 Figure 2.40 Cross section of beams (Valluzzi et al. 2009) (a) ordinary and (b) prestressed reinforced concrete (mm) The mix design of the beams along with the reinforcement and prestressing conditions is shown in Table 2.8 FeB 44K reinforced steel bars with diameters 10 mm, 14 mm and 16 mm were used, with a yield tensile limit of 536 MPa and an ultimate strength of 630 MPa. The half-inch strands used for prestressed beams had a yield limit of 1,693 MPa and reached tensile failure at 1,895 MPa. Table 2.8 Characteristics of samples (Valluzzi et al. 2009) a Final value of deformation including losses (0.05%) The laminates were pretensioned before application and to optimize the adhesive properties of the fibers, the concrete surface was prepared and cleaned by brushing. After a preliminary layer of primer, a thin bed of adhesive resin (1 3 mm) was used to provide a smooth finish to the surface and glue the fibers. The laminate was then positioned on the underside, with its ends were inserted in the slots of the anchoring plates. Table 2.9 shows the loading values of first crack, failure and the relative values of FRP strains. Figure 2.41 shows the application phases of the FRP system on the underside of the beams. 45

62 Figure 2.41 Application phases of pretensioned FRP system on underside of beams (Valluzzi et al. 2009) (a) preparation of surface; (b) positioning of laminate(anchorage zone); (c) end of beam with fixed steel plate anchorage; (d) hydraulic jack acting between sliding anchor and plate Table 2.9 Main results on beams (Valluzzi et al. 2009) The cracking modes at failure for the portions of beams between the two concentrated loads are shown in Figures 2.42 and Note the smaller distance between cracks in the PRC-PrEA beam (for better comparison, the average spacing is included in the captions of the figures). Failure involved large-scale deformations and consequent sudden delamination of the fibers from the concrete substrate, including sliding of the FRP at the anchorage plates. Infrared thermography was applied before loading tests, for preliminary possible detection of defects and, later, during bending, up to 67% of the failure load, in both ordinary reinforced and prestressed beams. 46

63 Figure 2.42 Cracking distribution at failure for RC-PrEA (average spacing 215 mm) (Valluzzi et al. 2009) Figure 2.43 Cracking distribution at failure for PRC-PrEA (average spacing 165 mm) (Valluzzi et al. 2009) The mirror reflected the image of the underside of the beam, which was then captured by the camera (Figure 2.44). A continuous scanning was performed allowing the detection of significant zones and then PCA was applied to the thermographic sequences obtained in these zones. Figure 2.44 Experimental thermographic tests on beams (Valluzzi et al. 2009) (a) mechanical setup for bending tests; (b) sliding device for thermographic data acquisition; (c) scanning phase; (d) related images in tilted mirror 47

64 Two areas on the underside of the beams, one close to the anchorage zone and the other close to the midspan were examined. Optical reflective markers were fixed on relevant points of the beams to highlight the bond conditions at specific areas indicated by Figure 2.45, Figure 2.46 and Figure Figure 2.45 compares the images taken at best visibility time with PCA results. Figure 2.46 shows results at a distance 1 m from the midspan in the RC-PrEA, which was the largest defect area examined. Figure 2.46(a) shows the second component of PCA in unloaded conditions. Figure 2.46(b), 2.46(c) show the PCA processing results at various load steps. RC-PrEA beam, location A: good-quality thermogram RC-PrEA beam, location A: PCA 2 nd component RC-PrEA beam, location B: good-quality thermogram RC-PrEA beam, location B: PCA 2 nd component Figure 2.45 Active thermography results of RC-EA and RC-PrEA beams before bending tests, showing main discontinuities at location A (row 1) and location B (row 2); left thermogram at best time during cooling; right: data processed by PCA (Valluzzi et al. 2009) 48

65 Figure 2.46 Active thermography results of RC-PrEA beam processed by PCA at various loading cycles (a)before loading: PCA 2 nd component (b) 2 nd loading cycle(41.8 kn) : PCA 3 rd component, 27% ultimate strength (c) 3 rd loading cycle(103.7 kn): PCA 3 rd component, 67% ultimate strength (Valluzzi et al. 2009) Similar results are shown in Figure 2.47 for PRC-PrEA beam before loading and during bending. The shape of the defect very similar to the subsequent phases and the size of the defect was approximately 19 cm 2. 49

66 Figure 2.47 Thermography results of PRC-PrEA beam processed by PCA at various loading cycles (Valluzzi et al. 2009) (a) before loading: PCA 2 nd component (b) 2 nd loading cycle(41.8 kn) : PCA 3 rd component, 48%% ultimate strength (c) 3 rd loading cycle(138.2 kn): PCA 3 rd component, 70% ultimate strength Defect evaluation in plane extension was computed with the help of Full-Width Half-Maximum (FHWM) method (Grinzato et al. 2006, and Almond et al. 1994) which computes the extension of the isotherm traced at half the maximum temperature difference between defect and reference areas. The beams were visually inspected after failure and Figures 2.48 and 2.49 report the condition after the testing. 50

67 Figure 2.48 Debonded surface of RC-PrEA beam of figure 2.47, after mechanical test; various kinds of detachments are visible (Valluzzi et al. 2009) (a) peeling involving concrete surface (b) detachment of resin and (c) at interface with concrete (in this case, debonding did not involve concrete) Figure 2.49 Visible image of resin layer after detachment of CFRP, with fingerprint of natural defects such as air pockets (see Figure 2.48), which probably occurred during application of resin and laminate (Valluzzi et al. 2009) Conclusions Subsurface defects at the concrete-frp interface of RC beams were inspected using thermographic techniques. PCA presents a potential method for defect detection especially when hot air is used as heat source. PCA processing is very quick and supports a robust procedure for in-situ applications. Thermography is a very effective technique for identifying debonded areas. The widening of the delaminated areas can be clearly followed during loading. Comparison of the test results with the visual inspections confirmed that thermography is a very reliable method. 51

68 2.7 CHARACTERIZATION OF THE DETERIORATION OF EXTERNALLY BONDED CFRP-CONCRETE COMPOSITES USING QUANTITATIVE INFRARED THERMOGRAPHY (Lai et al. 2010) Introduction Lai et al. (2010) discussed a method to characterize the hidden defects (flaws and delaminations) at the interface of concrete-cfrp components. Flaws are formed during the first application of the CFRP strips, due to poor workmanship and delaminations are formed due to stress concentrations related to chemical/physical degradation of the binding layer. The service life and the durability are mainly dependent on the bond strength between the concrete and CFRP. The field testing usually involves large scale inspection and the conventional methods like hammer tapping and pull-off tests are not ideally suited for large scale inspections. Infrared thermography (IRT) is a widely accepted method of detecting and characterizing hidden defects in composite materials. Giorleo and Moela (2002) made a comparative study between pulsed and modulated thermography applied to glass-epoxy laminates. The defects were characterized by (i) quantitative infrared thermogrpahy and (ii) processing of visual images after removing the CFRP laminates. A relatively simple inverse method was proposed in this study. It is called the two-inflection point algorithm and is used to define the defect boundaries and to estimate the defect areas Experimental Setup Materials and specimens CFRP laminates (60 mm x 200 mm x mm) and concrete prisms (150 mm x 150 mm x 350 mm) were bonded using Sika 300 TM (epoxy resin). Figure 2.50 shows the experimental setup. Table 2.10 shows the results of tensile tests conducted on CFRP strips and Sika 300 TM. OPC ASTM Type 1 cement, river sand, 10 mm and 20 mm size crushed granite with a water/cement ratio of 0.50 was the mix design for concrete prisms. Sika 300 was mixed at the ratio of 3 parts epoxy to 1 parts hardener. The tensile properties of the CFRP and epoxy resin are given in Table

69 Figure 2.50 Embedded delaminations and flaws contained within a CFRP-strengthened concrete prism (Lai et al. 2010) Table 2.10 Tensile properties of CFRP and epoxy resin 53

70 Interfacial bond between CFRP strips and concrete The concrete surface was made rough and cleaned using a high power vacuum cleaner. After laying the CFRP strips on the concrete surface using the binding epoxy, the specimens were allowed to harden for 14 days before testing. Two types of defects were embedded at the interface, (i) sharply edged flaws and (ii) curved-edge delaminations (Table 2.11). The flaws were the air voids occurring due to improper workmanship in applying the epoxy. The delaminations were caused by stress concentrations as a result of exposure to aggressive environment. Three batches of specimens were maintained in three water chambers maintained at 25 C, 40 C and 60 C respectively and the conditions were maintained for about 50 weeks. Table 2.11 Definitions, illustrations and methods to quantify flaws and delaminations (Lai et al. 2010) Quantitative Infrared Thermography A FLIR Prism DS infrared camera with a spectral range of was used for the acquiring the thermal images of the specimen. The grey scale analog images were output to PC since digital interfaces were not available, and the images were then encoded to a sequence of 8-bit digitized and 2-D thermograms or pixel arrays. The process was controlled by in-house software programs developed using IMAQ and LabVIEW. The software was designed to capture images at a frequency of 25 frames/second for a period of 30 seconds. The digitized thermograms had pixel values form This range was assumed to represent the lower (22 C) and upper (40 C) bounds. The temperature accuracy was (40-22)/ C. 54

71 Maximum thermal contrast The heat transfer is always dependent on time. A typical thermal history of the defective and sound areas is shown in Figure Thus the thermal image taken at a particular instant was very essential to distinguish the boundaries of the defect. The thermal contrast can be appropriately defined by the equation 2.6, which is a combination of the temporal and spatial differences. where C(x,y,t) is the spatial thermal contrast at any time t; temperature at which defects were found at a time t; is the spatial is the spatial temperature at which the defects were found initially; and are the same parameters as above for the sound areas. Figure 2.51 Thermal history curves of the flaw and sound areas (Lai et al. 2010) Two-inflection point algorithm for defining flaw boundaries Once the thermogram for the maximum value of C(x,y,t) was selected, the shape of the defect boundary was deduced from the pixel profile as shown in Figure The temperature gradient was close to the steepest gradient computed on the sample surface and this method is called the gradient computation method. There are many algorithms to process the method and in this study the two-inflection point algorithm was used for data 55

72 processing. As shown in the Figure 2.52, each of the column and row pixel profile was segmented at the midpoint to construct two sub-pixel profiles. The subpixel profiles were fitted into a 4 th order polynomial P(x) as shown in Figure The polynomial P(x) was differentiated twice to get a second order equation for which one of the root was the inflection point w boundary of P(x) as shown in equation 2.7. where x = pixel cell; w boundary is the pixel cell defined as the flaw boundary and P(x) is the pixel intensity as a function of pixel cell. The number of pixels from the inflection point to the midpoint of the defect was then counted and it represents the half width of the defect at a particular column or row. Figure 2.52 Thermogram of a flaw and the respective pixel profiles (Lai et al. 2010) Figure 2.53 Pixel profile from a flaw to a sound zone (Lai et al. 2010) 56

73 Digital image analysis of flaw and delamination boundaries The sizes of the defects detected from thermographic studies were compared to the actual sizes by removing the CFRP layer using direct shear test. The visual image analysis was performed to find the actual size of the defects and the size of the delaminations was found by drawing the boundary lines on the CFRP strips initially after open up (Figure 2.54 left). Then the CFRP strip was overlaid by a transparency printed with a 1 mm 2 grid. The sizes were then estimated with the help of commonly available visual image software. The flaw size estimation also followed the same technique except for the first method as the flaw boundary was clearly visible. Figure 2.54 Interfacial delamination (left) and flaw (right) overlaid by 1 mm 2 grid for visual image method (Lai et al. 2010) Results and Discussions Correlation of the apparent and the actual flaw sizes The accuracy in estimating the flaw sizes was 88% and the correlation coefficient was 0.89 as shown in Figure Figure 2.55 Comparison between actual and apparent flaw sizes (Lai et al. 2010) 57

74 Effects of elevated temperatures on flaws and delaminations Figure 2.56 shows the thermograms and open-up images of the specimens before and after exposure to elevated water temperatures. The flaw defect showed very little size variation before and after the test but the delamination which was not present before testing was very distinctly seen in the image after the test. The delaminations mainly occurred because of the degradation of epoxy resin, due to high temperatures, as shown in Figure 2.56(example 2). Figure 2.56 Conditions of the sharp-edge flaw and development of delamination due to exposure to elevated water temperature (Lai et al. 2010) The results shown in Figure 2.57 relate the flaw detection and it can be said that the flaws were not affected by the exposure to elevated water temperatures. It is probably because there was no swelling stress to extend the flaw boundary during the test. Figure 2.58 explains the effect of the temperature variation on the delamination. The attempt to define the boundary of the delamination through the quantitative infrared thermography was not successful due to discrepancies with the results obtained from image processing. The relatively flat pixel profile of the delaminations made it difficult for the clear demarcation of the boundaries. 58

75 Figure 2.57 Comparison of flaw areas before and after exposures (Lai et al. 2010) Figure 2.58 Comparison of delamination areas after exposures (Lai et al. 2010) Conclusions The paper discussed the detection of two types of defects namely flaws and delaminations at the concrete-cfrp interface, using the two inflection point algorithm. The algorithm was used to analyze the thermal signals and hence account for the defect characterization. The quantitative infrared thermography results were compared with the visual image processing results and the accuracy in the flaw size estimation was 88%. It was also concluded that the flaws showed little change but the delaminations were very sensitive when the specimen was exposed to elevated water temperatures. 59

76 2.8 QUANTITATIVE ANALYSIS OF DELAMINATIONS IN GRP PIPES USING THERMAL NDTE TECHNIQUE (Vijayaraghavan et al. 2010) Introduction Vijayaraghavan et al. (2010) made an effort to develop a thermal non destructive technique which could be applicable to assess the delaminations in glass reinforced polymer (GRP) pipes. The study discussed about the analysis and results obtained by active infrared thermographic testing of GRP pipes in reflection mode. In this method, two dimensional finger prints in the form of images were obtained to quantitatively identify the invisible defects. The temperature profiles were compared with the results obtained from numerical modeling by the finite element method (FEM) using ANSYS and then the results from both the methods were compared with implanted defect characteristics Experiments Details of GRP pipes The test specimens used in this study consisted of four GRP pipes (length 500 mm, diameter 300 mm and thickness 5mm) with fabricated defects (delaminations), created using Teflon sheets at the concrete-frp interface. The pipes were manufactured with unidirectional roving of E-glass fiber using filament winding process and the fiber filament was laid at an angle of 55. The pipes have a fiber volume ratio of 78%. The types of delamination implanted in these specimens are shown in Table Other pipes have delaminations as detailed in Table 2.13 and Figure 2.59(a). The thickness of delaminations in the pipes 2 & 3 is 0.2 mm where as the pipe 4 has 0.4 mm thick delaminations. Table 2.12 Location of delaminations (Vijayaraghavan et al. 2010) 60

77 Table 2.13 Sizes and depth of the delaminations (Vijayaraghavan et al. 2010) (a) Schematic diagram (b) Pictorial view Figure 2.59 Details of GRP pipes (Vijayaraghavan et al. 2010) Experimental setup The equipment consisted of a ThermaCAM EX320 of FLIR Systems Inc. having a wavelength in the range of 7.5μm-13μm, thermal sensitivity of 0.08ºC at 25ºC, accuracy of ±2 C, pixels focal plane array (FPA), and a dynamic temperature range of -20 C to +250 C. Two lamps (500W each) were used for heating the specimens prior to acquiring the infrared images. The experimental setup is shown in Figure The duration of heat pulse was set to 5 seconds. The trigger circuit was used to adjust the heating time by switching on/off the heating lamps at a preset time. 61

78 Figure 2.60 Schematic of the experimental setup (Vijayaraghavan et al. 2010) Numerical Modeling Finite element method is a widespread technique that provides solution to several nonlinear, unsymmetrical mathematical relations governed by partial differential equations such as heat transfer equations. The model geometry of the tested sample was defined to obtain a solution to the differential equation. For this study, the GRP specimens with varying sizes were simulated using finite element modeling provided by ANSYS Thermal properties of GRP and Teflon materials Stanley et al. (2003) talked about the thermal properties of a variety of composite materials. Table 2.14 provides the thermal properties of E-glass fiber and epoxy resin. Table 2.14 Thermal properties of E-glass fiber and epoxy resin (Vijayaraghavan et al. 2010) The thermal properties of the materials shown in the table can be combined with some micro-mechanics equations proposed by Stanley et al. (2003) as: 62

79 ) ρ = kg/m 3 C = kj/kg C k 11 = W/m C k 22 = k 33 = W/m C Various properties of GRP pipes are obtained from the above equations. Avdelidis et al. (2004) lists the properties of Teflon material as: ρ t = 2150 kg/m 3 C t = 1043 kj/kg C k t = W/m C GRP pipe model parameters The transient conduction heat analysis of a full GRP pipe specimen (outer diameter 310 mm, inner diameter 300 mm and 500 mm long) simulated with delaminations was performed using ANSYS. The delaminations were located along the same length and at the center of the sample. The pipe was modeled using an anisotropic material for the pipes and the Teflon with an isotropic material Finite element description The meshing was done using 3-D thermal solid elements and 20-node quadrilateral elements were used (Figure 2.61), which is applicable to steady state or transient thermal analysis. The elements have temperature shapes and are suited for modeling curved boundaries (ANSYS Manual 2005). Mapped meshing was used in order to create a fine mesh of the delaminations. The global element length was set to 1 mm. 63

80 Figure 2.61 Meshed 3-D model from ANSYS (Vijayaraghavan et al. 2010) Loading and boundary conditions The input heat flux and convection heat losses were considered as the boundary conditions. Since the input flux was uniform, a uniform input condition was used in the model. The outer top half of the pipe was subjected to a flux of 10,000 W/m 2 for 5 seconds. The inner and the outer bottom half were subjected to convective heat transfer with a convective heat transfer coefficient of h = 10 W/m 2 K. The ambient temperature was set to 28 C Results and Discussions The thermal images of the surface were obtained and the data was used for analysis. Thermal contrast was defined and calculated as the temperature at the center of the delamination minus the temperature in a sound (defect-free) region. The finite element modeling simulation results were obtained for varying delamination sizes and locations, to compare with the experimental thermographic results. (a) Experimental (b) Numerical Figure 2.62 Numerical versus Experimental surface temperature distribution for pipe 1 (Vijayaraghavan et al. 2010) 64

81 (a) Experimental (b) Numerical Figure 2.63 Numerical vs. Experimental surface temperature distribution for pipe 2 (Vijayaraghavan et al. 2010) (b) Experimental (b) Numerical Figure 2.64 Numerical vs. Experimental surface temperature distribution for pipe 3 (Vijayaraghavan et al. 2010) (a) Experimental (b) Numerical Figure 2.65 Numerical vs. Experimental surface temperature distribution for pipe 4 (Vijayaraghavan et al. 2010) 65

82 The surface temperature distributions obtained for the four pipes, from both experimental and numerical model are shown in Figure 2.62 to Figure The surface temperature profiles across the length of the pipe, from both experimental and numerical analysis are shown in Figure It can be seen that the results obtained from both analysis are very much in agreement in terms of their temperature variation pattern Location and sizes of delaminations For the pipe 1(Figure 2.59(a) and 2.59(b)), the temperature profile was very much uniform. The average thermal contrast across the center line, for both experimental and numerical methods, was 0.24 C and 0.28 C respectively. Some non-uniform temperature distributions at the location of simulated delaminations appear in case of pipe 2 (Figure 2.63 and Figure 2.66(b)). The results of the maximum thermal contrast obtained experimentally at delaminations 1, 2, 3 and 4 were calculated to be 1.38 C, 1.39 C, 1.59 C and 1.78 C respectively. The numerical analysis yielded values of 1.31 C, 1.34 C, 1.52 C and 1.65 C. From the images it was observed that the sizes of the delaminations obtained from numerical analysis could be compared to the actual sizes. The delamination sizes were estimated by considering on the temperature distribution across the horizontal and vertical axis at the center of the delamination. A variation in the temperature from the average temperature of the surrounding defect-free region indicated the boundary of the delaminations. Similar approach was followed to locate and estimate the sizes of defects in pipes 3 and 4. 66

83 (a) For pipe 1 (b) For pipe 2 (c) For pipe 3 67

84 (d) For pipe 4 Figure 2.66 Surface temperature profiles across the center line of the delaminations (Vijayaraghavan et al. 2010) The variation of the results obtained from both the methods were then computed with reference to the actual sizes as shown in Tables 2.15, 2.16 and The percentage of error decreases with an increase in the size of the defect for pipes 2 and 4. The percentage error value for pipe 2 was greater than that for pipe 4 and this may be due to the thinner defect in pipe 2, which results in a reduced thermal contrast. For pipe 3, the error percentage increased as the size of the defect increased and this may be due to variation in depth. Also it is interesting to note that the thermal contrast was decreasing from defect 1 to 4, in case of the pipe 3. Table 2.15 Variation of experimental and numerical results with actual delamination sizes for pipe 2 (Vijayaraghavan et al. 2010) 68

85 Table 2.16 Variation of experimental and numerical results with actual delamination sizes for pipe 3 (Vijayaraghavan et al. 2010) Table 2.17 Variation of experimental and numerical results with actual delamination sizes for pipe (Vijayaraghavan et al. 2010) Figure 2.67 Comparisons of thermal contrasts of delaminations in pipe 2 and 3 (Vijayaraghavan et al. 2010) Depth of delamination The depth of the delaminations was varied in pipe 3 and compared with those in pipe 2 keeping the other parameters constant. The thermal contrast values for the defects 69

86 in both the pipes were plotted as shown in Figure The thermal contrast variation was very much constant for pipe 2 and the maximum variation was C indicating that the defects were almost at the same depth. But for pipe 3, when the delamination depth is 1 mm from the outer surface, the thermal contrast was 4.67 C whereas for 2 mm deep delamination, the value was 1.89 C. The 3 mm and 4 mm deep delaminations produced thermal contrast values of 0.81 C and 0.36 C respectively. From the Figure 2.68, the thermal contrast decreases in a polynomial fashion with increasing depth as expressed in equation The defects near to the heating source were identified quickly with an increased thermal contrast while the deeper defects were detected slowly with a reduced thermal contrast. Figure 2.68 Variation of the thermal contrasts of delaminations in pipe 3 (Vijayaraghavan et al. 2010) Thickness of delaminations Figure 2.69 shows the plot of the experimental thermal contrast results for delamination in pipe 2 and pipe 4. It shows that thermal contrast values for all the defects in pipe 4 are greater than those for pipe 2 irrespective of their sizes. The differences in thermal contrast for both the pipes at the defect locations were 0.74 C, 0.81 C, 0.9 C and 70

87 0.77 C which is due to the thermal resistance of the defect. The thermal resistance of the delamination increases as the thickness of the delamination increases. Figure 2.69 Comparison of thermal contrasts of delaminations in pipe 2 and pipe 4 (Vijayaraghavan et al. 2010) Conclusions The study mainly focused on the characterization of the delaminations in GRP pipes using the thermographic technique, both experimentally and numerically. The sizes of the delaminations obtained from the thermal images were very much closer to the actual sizes of the defects. The defects closer to the outer surface were identified quicker with an increased thermal contrast. The thermal resistance of the delaminations increases as their thickness increases which leads to an increase in the thermal contrast. The study justifies that infrared thermography can be a very effective technique in characterizing the subsurface defects in GRP pipes. 71

88 2.9 DETECTION OF SUBSURFACE DEFECT IN FIBER REINFORCED POLYMER COMPOSITE BRIDGE DECKS USING DIGITAL INFRARED THERMOGRAPHY (Halabe et al. 2007) Introduction About 25% of the bridges in the US are considered to be structurally deficient of functionally obsolete and the rehabilitation of these bridges is estimated to be very expensive process (USDOT-FHWA). Hence, innovative materials like fiber reinforced polymer are being used as the replacement for old concrete decks due to their many advantages. Subsurface delaminations or debonds are some of the defects which might be formed during construction or during the service life of the FRP bridge decks. Hence, evaluation of such materials is necessary to ensure construction quality and structural integrity. The condition assessment of the bridges is very important to ensure the durability of the FRP bridge decks. Non-destructive evaluation is being considered as a very effective method of condition assessment of old structures. Infrared thermography is found to be one of the portable non destructive techniques which uses infrared images for the data processing and data presentation is also very simple. The present research focuses on the study of laboratory specimens with known defects. Field tests were also performed to demonstrate the capability of Infrared thermography for testing FRP bridge decks Experimental Setup Infrared thermography works on the principle that the rate of heat conduction through a defect free region and a defective region is different. Thus when the specimen containing possible defects is subject to a heat pulse or when it is cooled down, the variations in temperature show up at different regions. A ThermaCAM S60 (FLIR Systems) infrared camera was used to capture the infrared images after the heating or cooling source has been removed. The camera allows capturing of radiometric images continuously at a frame rate of 60 Hz. The infrared images can be analyzed by a laptop computer using associated processing software. The infrared camera has a sensitivity of 0.06 o C. A 1500W quartz heater was used as the heating source. Figure 2.70 shows the shows the laboratory setup of the infrared camera. Figure 2.71 shows the other equipment 72

89 used for heating purposes. Figure 2.71(a) shows the quartz heater as described above. Figures 2.71(b) and 2.71(c) show the 3 x 3 (910 mm x 910 mm) 1500W heating blanket and a carbon dioxide (CO 2 ) tank fitted with a spray nozzle. This was used as a cooling source in some of the experiments. The solar radiation was used as a heat source for testing large decks in the field. Figure 2.70 Experimental setup using digital infrared camera and close up view of the camera (Halabe et al. 2007) (a) 1500W (b) 1500W heating blanket (c) liquid CO 2 quartz heater tank Figure 2.71 Various active heating and cooling sources (Halabe et al. 2007) 73

90 2.9.3 Description of laboratory specimen GFRP bridge deck modules composed of E-glass fibers (continuous strand roving and triaxial fabrics) and either polyester or vinyl ester resin combination, were used in this study. The fiber volume fractions for the polyester and vinyl ester decks are about 45 and 50%, respectively. Some of the decks were coated with 3/8 (9.5 mm) thick wearing surface. Two polypropylene sheets with an air pocket were used to make the debonds. The water-filled debonds were made by enclosing water within plastic pouches. The simulated debonds were placed on the top surface of the GFRP deck, before the application of wearing surface. The specimen BD1 shown in Figure 2.72 was a bridge deck module of plan size 24 x 12 (610 x 305 mm), flange thickness of 0.5 (12.7mm) and web thickness of 0.35 (8.9 mm). This deck specimen had an overall depth of 8 (203 mm). One side of the deck specimen consisted of two air-filled defects of sizes 2 x 2 (51 x 51 mm) and 3 x 3 (76 x 76 mm) and the other side consisted of two defects of sizes 1 x 1 (25 x 25 mm) and 1/2 x 1/2 (13 x 13 mm) between the wearing surface layer and the deck. All these debonds had a thickness of 1/16 (1.6 mm). Figure 2.72 Front and cross-sectional views of the bridge deck specimen BD1 with wearing surface (Halabe et al. 2007) 74

91 Bridge deck specimen BD2 has a plan size 24 x 12 (610 x 305 mm), flange thickness of 0.45 (11.4 mm) and web thickness of 0.4 (10.2 mm). The overall depth of this specimen was 4 (102 mm). Two water-filled debonds of sizes 2 x 2 x 1/16 (51 x 51 x 1.6 mm) and 3 x 3 x 1/8 (76 x 76 x 3.2 mm), and an air-filled debond of size 2 x 2 x 1/16 (51 x 51 x 1.6 mm) were placed on the surface of the GFRP deck specimen before it was overlaid with the wearing surface layer. Specimens JD1 and JD2 had plan sizes of 15 x 8 (381 x 203 mm), flange thickness of 0.45 (11.4 mm) and web thickness of 0.4 (10.2mm) with an overall width of 4 (102 mm). The middle flange joint area in one of the modules had fabricated delaminations embedded in it (Figure 2.73). The modules were joined together with pliogrip resin so that the delaminations were embedded at an average depth of 0.32 (8.1 mm) from the top surface. The thickness at the flange-to-flange junction of the specimens was measured to be about 0.6 (15.2 mm). (a) GFRP deck specimen without (b) GFRP deck specimen without overlay overlay Figure 2.73 Front and cross-sectional views of the GFRP bridge deck specimens (Halabe et al. 2007) Figure 2.74 shows the delaminations on the top and bottom flange joint areas JD1a and JD1b. The delamination in the top flange joint area had dimensions of 3 x 3 x 1/20 (76 x 76 x 1.27 mm). In the bottom flange, two delaminations of sizes 2 x 2 (51 x 75

92 51 mm) and 1 x 1 (25 x 25 mm) were embedded, both with thicknesses of 1/16 (1.6 mm). The specimen JD1a and JD1b did not have any wearing surface overlay. Specimen JD2 had a delamination of size 3 x 3 (76 x 76 mm) in plan and 1/8 (3.2 mm) thickness in the middle of the flange joint area (top flange). The top flange was covered with a wearing surface with a thickness of about 3/8 (9.5 mm). The delamination was located at an average depth of 0.69 (17.5 mm) from the top surface. (a) Delamination of size 3 x 3 x 1/20 (b) Delaminations of sizes 1 x 1 x 1/16 (76 x 76 x 1.27 mm) in specimen JD1a (25 x 25 x 1.6 mm) and 2 x 2 x 1/16 (51 x 51 x 1.6 mm) in specimen JD1b Figure 2.74 Photographs showing the locations of delaminations (Halabe et al. 2007) Detection of debonds using infrared thermography Air-filled debonds The debonds were clearly shown as hot spots with higher temperatures compared to the surrounding defect-free regions. The heating time was about 150 seconds and the heater was placed at a distance of 0.2 m from the surface of the specimen. The infrared images were recorded for a particular time interval in order to develop the surface temperature-time curves. The infrared images were captured for every 10 seconds after the heat source was removed. The temperature patterns of the defective and defect-free regions were tabulated and curves were developed with temperature on Y-axis and time on X-axis as shown in Figure Figure 2.76 shows the infrared image of BD1. 76

93 Figure 2.75 Surface temperature-time curves for air-filled debonds in specimen BD1 of sizes1 x 1 x 1/16 (25 x 25 x 1.6 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) (Halabe et al. 2005) (b) Infrared image of specimen BD1 with air-filled debonds of sizes 3 x 3 x 1/16 (76 x 76 x 1.6 mm) and 2 x 2 x 1/16 (51 x 51 x 1.6 mm) (a) Infrared image of specimen BD1 with air-filled debonds of sizes 1 x 1 x 1/16 (25 x 25 x 1.6 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) Figure 2.76 Infrared images of the specimen BD1 with air-filled debonds (Halabe et al. 2007) 77

94 The curve in Figure 2.75 was obtained for the 1 x 1 (25 x 25 mm) and 1/2" x 1/2" (13 x 13 mm) debonds in specimen BD1. Images with reasonably good contrast between the defective and defect-free areas were available for about 300 seconds duration for the 3 x 3 (76 x 76 mm) and 2 x 2 (51 x 51 mm) debonds and about 200 seconds duration for the 100 x 100 (25 x 25 mm) and 1/2 x 1/2 (13 x 13 mm) debonds, after the heating source had been removed. After a particular period from the time of removal of heating source, the deck approaches thermal equilibrium due to heat conduction to surrounding defect-free regions Water-filled debonds Figure 2.77 shows the schematic view of the specimen BD2 and the corresponding infrared image. The surface temperature above the air-filled debond (white area in infrared image) was higher than the temperature above defect-free area (temperature difference of 5.7 C). As observed from the infrared image, the water-filled debonds were shown as lower temperature zones compared to the air-filled debonds. This is because the thermal diffusivity of water (0.14 x 10-6 m 2 s -1 ) is very near to the thermal diffusivity of the GFRP (0.13 x 10-6 m 2 s -1 ) along the flange depth. The air-filled debonds have a higher thermal diffusivity (33 x 10-6 m 2 s -1 ) which causes the higher temperature difference. The temperature difference between the water-filled debonds (cold spots) and the defect-free regions for the 3 x 3 (76 x 76 mm) debond was found to be 23.9 C and 22.5 C for the 2 x 2 (51 x 51 mm) debond. There was a bright spot above the 3 x 3 (76 x 76 mm) debond and this is because of the presence of air bubbles inside the plastic pouch used for making the debond. The infrared images revealed that the boundaries of the air-filled debond were more clearly defined when compared to the water-filled debonds. 78

95 Figure 2.77 Schematic view and infrared image of bridge deck specimen BD2 showing the air-filled and water-filled debonds (Halabe et al. 2007) Detection of delaminations using infrared thermography The specimen JD1 was made with defects representing delaminations. The specimen was heat for about 180 seconds before acquiring the infrared images. The temperature-time curves were developed for the 2 x 2 and 1 x 1 delaminations as shown in Figure Figure 2.79 shows the infrared image of the specimen without wearing surface and having delaminations of sizes 1 x 1 x 1/16 (25 x 25 x 1.6 mm), 2 x 2 x 1/16 (51 x 51 x 1.6 mm) and 3 x 3 (76 x 76 x 1.3 mm). The thermal conductivity of the GFRP (0.3 and 0.38 W m -1 C -1 in the perpendicular and parallel directions respectively) is higher than the thermal conductivity of air (0.024 W m -1 C -1 ). Hence the defect-free regions transfer heat at a faster rate and have a lower temperature. The maximum thermal contrast for the 3 x 3 x 1/20 delamination was obtained after 140 seconds from the time of removal of the heat source. The maximum thermal contrast for the 2 x 2 and 1 x 1 delamination was obtained after 120 seconds and 110 seconds respectively. 79

96 Figure 2.78 Surface temperature-time curves for 1 x 1 x 1/16 (25 x 25 x 1.6 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) air-filled delaminations located at the flangeflange junction of bridge deck JD1b (without wearing surface) (Halabe et al. 2007) (a) Infrared image of the specimen JD1b with delaminations of sizes 1 x 1 x 1/16 (25 x 25 x 1.6 mm) and 2 x 2 x 1/16 (51 x 51 x 1.6 mm) (b) Infrared image of the specimen JD1a with delaminations of sizes 3 x 3 x 1/20 (76 x 76 x 1.3 mm) Figure 2.79 Infrared image of specimens JD1a and JD1b (without wearing surface) with air-filled delaminations at the flange-flange junction (Halabe et al. 2007) 80

97 The specimen JD2 had a 3/8" (9.5 mm) thick wearing surface and hence the time of heating for this specimen was about 300 seconds. Figure 2.80 shows the infrared images of the specimen before and after placing the wearing coating. The temperature difference between the defective and defect-free areas for the two tests were found to be 13 C and 6.3 C, respectively. (a) (b) Figure 2.80 Infrared image of the bridge deck specimen JD2 with embedded air-filled delamination of size 3 x 3 x 1/8 (76 x 76 x 3.2 mm) at the flange-flange junction before placing the wearing surface overlay and (b) after placing the wearing overlay surface (Halabe et al. 2007) The surface temperature-time curves were obtained for the air-filled delamination of size 3 x 3 x 1/8 (76 x 76 x 3.2 mm) and defect-free area for specimen JD2 with wearing surface are shown in Figure As the depth of the location of the delamination is higher (13 mm), the heat conduction through the thickness of the specimen towards the surface takes a longer time. This meant that an infrared image with a good thermal contrast could be obtained even after 600 seconds from the time of removal of the heat source. Air-filled delaminations located at higher depths were replaced with water-filled delaminations and the tests were repeated as discussed above. This did not produce favorable results as the thermal diffusivity of the GFRP and water were very close. 81

98 Figure 2.81 Surface temperature-time curves for 3 x 3 x 1/8 (76 x 76 x 3.2 mm) and 1/2 x 1/2 x 1/16 (13 x 13 x 1.6 mm) air-filled delaminations located at the flangeflange junction of bridge deck JD2 (with wearing surface) (Halabe et al. 2007) Summary and Conclusions Infrared thermography proved to be an excellent technique for the detection of air-filled debonds of size 1/2 x 1/2" x 1/16 under 3/8 (9.5 mm) thick wearing surface in GFRP bridge deck specimens. Also, air-filled delaminations of size 1 x 1 x 1/16 (25 x 25 x 1.6 mm) at flange-flange junction could be detected in specimens without a wearing surface. The infrared image for a 3" x 3" x 1/8" (76 x 76 x 3.2 mm) air-filled delamination at the flange flange junction of a GFRP deck specimen with 3/8" thick (9.5 mm) wearing surface did not show a very prominent boundary of the delamination, but could confirm the presence of a subsurface delamination. All the specimens used in the study were representative of field applications. From the surface temperature-time curves it was concluded that the image with the maximum thermal contrast was obtained immediately after removing the heat source. Also it was concluded that the best visual contrast in the infrared images could be obtained when the temperature difference between defective and defect-free areas is about 1/10 th of the background defect-free temperature. 82

99 2.10 EFFECTS OF SOLAR LOADING ON INFRARED IMAGING OF SUBSURFACE FEATURES IN CONCRETE (Washer et al. 2010a) Introduction Concrete is the most widely used material for the construction of bridges and many other structural and sub-structural members. The structural concrete deteriorates with time and environment. Corrosion of embedded steel is one of the serious problems in reinforced concrete. This is mainly due to the intrusion of chloride into the concrete. Chlorides cause the steel to convert into iron oxide and expand. Figure 2.82 shows the cracking pattern in concrete at the level of embedded steel which leads to delamination. Figure 2.82 Schematic diagram of concrete bridge deck with corrosion damage (Washer et al. 2010a) Hammer sounding is the most commonly used method to detect the presence of subsurface defects. The hollow sound produced when the hammer impacts on the surface indicates the presence of subsurface delaminations. Chain drag method is another method to detect the presence of subsurface defects. In this method a steel chain is dragged along the surface of the area under inspection in order to create acoustic tones. But both these methods have many shortcomings. Due to the latest advancements in imaging technologies, thermographic imaging has emerged as a very powerful technique in order to rapidly identify the presence of subsurface defects in concrete structures. The technique works on the principle of heat conduction through a defective and a defect-free region. This shows up a temperature difference shows up on the surface of the region when it is subjected to heating or cooling. The infrared image of the surface can clearly show the thermal contrast indicating the presence of subsurface anomalies. The defect regions show up as hot spots or cold spots depending on the surrounding environmental conditions. 83

100 This study focused on the effect of the solar loading on the ability of infrared thermography in the detection of subsurface features in concrete. A concrete test block was employed for this purpose with embedded targets. The quantitative analysis was performed based on the temperature gradients that appear on the surface. The effect of the depth of the anomaly and the time of detection for obtaining maximum contrast was discussed Experimental A large test block of 2.4 m x 2.4 m with a thickness of 0.9 m was used as the test specimen for the present study. Styrofoam sheets depicting subsurface defects were embedded within the concrete block during construction. The sheets were 300 mm x 300 mm in area and 13 mm thick. The targets were attached at depths of 25 mm (A), 51 mm (B), 76 mm (C) and 127 mm (D), on the north and south faces (2.4 m x 2.4 m), and 25 mm (A) and 76 mm (C) on the east and west faces (0.9 m x 2.4 m). Figure 2.83 shows the schematic view of the test block with the embedded targets. Figure 2.83 Diagram of concrete test block with embedded targets at depths of 25 mm (A), 51 mm (B), 76 mm (C), and 127 mm (D) (Washer et al. 2010a) The concrete block was supported on a 0.25 m thick concrete footing. An infrared camera (FLIR S65) was mounted in a test house located 9 m opposite to the concrete block. It has a sensitivity of 0.08 C and a range of -40 C to 1500 C. The thermal images of the surface of the block were captured for every 10 minutes, 24 hours a day. An on-site 84

101 weather station was also setup in the area to record the temperature of the environment, wind speeds, relative humidity and solar loading in W/m 2. Figure 2.84 shows the photograph of the test setup. A desktop computer was also setup in the test house to store the images for further processing and analysis. Figure 2.84 Photograph of test setup showing test house, weather stations and test block (Washer et al. 2010a) Fig 2.85 shows the infrared image of the test block surface facing the south side. The three bright spots visible in the image were the 3 targets at the depths of 25 mm, 51 mm, and 76 mm. This color contrast was used to distinguish qualitatively between areas of damaged concrete and sound concrete. The thermal contrast, which is defined as the temperature difference between sound areas and the defective areas, was determined considering the two points E and F as shown in Figure T contrast = TE TF 2.13 The thermal contrast was determined for all the targets and this was parameter used for studying the behavior of time varying thermal contrast. It was noted that the surface temperature of the block was uniform over majority area of the block as can be seen in Figure

102 Figure 2.85 Thermal image of test block showing embedded targets at 25 mm, 51 mm and 76 mm and example locations selected for calculating thermal contrast (Washer et al. 2010a) Results The testing data was collected between November 2007 and January The solar loading was the main factor affecting the thermal contrast on the south side. Corrosion induced delaminations occur commonly at a depth of about 50 mm and hence target B was selected for most part of the analysis. Curves were established to study the effect of solar loading on thermal contrast over a 24 hour period (Figure 2.86). The testing day was very sunny with constant sun throughout the daylight hours. For maximum solar load the thermal contrast for target B was greater than 5 C. The targets appeared as cold spots in the thermal image during night as they cool down much faster than the sound areas. From the curve it was noted that the maximum thermal contrast occurs at about 2:00 P.M while the maximum solar loading was recorded just after noon. As the target depth increased, the thermal contrast increased later in the day which means that there was time lag for the heat transfer to take place through the depth of the block and develop contrast. 86

103 Figure 2.86 Thermal contrast at embedded targets for sunny day (12/25/2007) indicating contrasts for 25 mm, 51 mm, 76 mm and 127 mm targets relative to the solar load (Washer et al. 2010a) In order to determine the time of occurrence of maximum contrast for each target, the test results for 89 days were taken into consideration. Both the time of occurrence of maximum contrast and the time lag relative to sunshine were tabulated as shown in Table 2.18 (local sunrise times taken from It was found that the time required for maximum thermal contrast to occur was the same regardless of the minimum contrast threshold used to sort the data. The delay for target B to develop maximum thermal contrast was approximately 6 hours 15 minutes, while the deeper target like target D required about 9 hours. There exists a linear relationship between the time of maximum thermal contrast and depth of targets. For identifying the days of significant thermal contrast at the targets, a threshold of 1 C was applied to the data. This threshold helped in identifying subsurface defects in the field inspection where the thermal contrast is not expected to be as great. For this threshold it was observed that approximately 70% of the days met the criteria. If the threshold is increased to 2 C, about 63% of the 89 days met the criteria. 87

104 Table 2.18 Time of day each target reaches its maximum contrast and time lag relative to sunrise (Washer et al 2010a) The relationship between the solar load and the thermal contrast was developed by comparing the maximum thermal contrast with the maximum thermal load occurring on that day. Figure 2.87(a) shows the relationship between the maximum solar loading, measured in W m2, and the maximum thermal contrast determined for that day. For target B, the correlation coefficient between maximum solar load and thermal contrast was This was due to the effect of periodic cloud cover which block the heat radiation, and also partly due to wind speeds and temperature variations during day and night. To compensate the evaluate the effect of cloud cover, the total solar energy was taken into consideration which can be determined by Es = S dt 2.14 where Es = total solar loading (kw h m 2 ) and S = measured solar loading (W m 2 ) Figure 2.87(b) shows the improved correlation when the total solar energy was taken into consideration with a correlation coefficient of The minimum threshold of 1 C was applied to know the amount of solar energy required to develop contrast. It was found that a minimum contrast of 1 C occured for 90% of the days with atleast 0.7kW h m 2 of solar loading, This value was obtained by making an intercept of a horizontal line drawn at 1 C, with the trend line in the figure, and calculating the percentage of days remaining above 0.7 kw h m 2. 88

105 Figure 2.87 (a) Relationship between maximum solar loading and maximum thermal contrast for target B; (b) Relationship between total solar loading area and thermal contrast for target B (Washer et al. 2010a) Thermographic imaging of subsurface deterioration in concrete bridges (Washer et al. 2010b) Field Testing Thermal imaging is one of the commonly used techniques for the inspection of bridge decks which directly receive the solar loading. The more problematic areas are the soffit regions where the application of the technique is limited. The present field study 89

106 provides the details of the technique employed for the detection of subsurface delaminations in the soffit region of a concrete bridge. Thermography is proved to be an effective technique for the early detection of such defects in order to avoid problems like spalling of concrete. Figure 2.88 Thermal images and photographs of delaminated concrete in bridge soffit (Washer et al. 2010b) (a) Thermal image during cooling trend (night time), (b) Thermal image during warming time (daytime), (c) soffit as it appeared at the time of thermal imaging, and (d) soffit after delaminated concrete was removed by striking it with a hammer Figure 2.88 shows the infrared images of the soffit on the south-side of a 4 lanes interstate highway bridge. Figure 2.88(a) shows the infrared image of the soffit portion captured at about 6:40 A.M in the morning. The image clearly shows the delaminated area as a colder spot with a temperature of ~22.5 C compared to the surrounding concrete with a temperature of 25.2 C. This was because of the lack of sufficient heating of the 90

107 deck and underlying regions of the deck at the beginning of the day. Figure 2.88(b) shows the infrared image of the same region as shown in Figure 2.88(a) but at a different orientation. This image was obtained at 10:20 A.M which indicated a positive thermal contrast as the bridge started getting the solar load. The temperature difference between the defective concrete and the sound concrete was determined to be ~1.7 C. Figure 2.88(c) shows the digital picture of the same portion of the bridge at the time of obtaining the infrared images. Figure 2.88(d) shows the digital image of the same portion of the bridge after the loose concrete is removed with the help of a 2 lb hammer or a wedge tool. Some of the important points that can be made from this study are: (a) Traffic control was not necessary to scan for damaged areas of the soffit. Traffic control was necessary while knocking down the loose concrete as the mass is located directly above the traffic lane. (b) The delamination was identified as a region with positive thermal contrast while the day progresses after overnight cooling. (c) The delamination depth increased upto 2 inches (51 mm) at inner zones from a 0 inch depth at the edges. (d) The visual examination of the bridge proved that the deterioration of the bridge was very extensive but the scanned portion of the soffit showed a much higher positive thermal contrast. The concrete in this portion was easily separated when it was struck with a hammer Conclusions The effect of solar loading on the embedded targets at different depths in a concrete block was studied here. The thermal contrast behavior due to environmental conditions was evaluated in order to optimize the applicable conditions for infrared thermography. The study discussed the effect of the solar load in terms of the maximum solar load, and the total solar loading energy, on the occurrence of thermal contrast for the embedded targets, over the course of a day. It was concluded that the ideal time for detection of defects at a depth of 25 mm is about 5:40 hours after sunrise. For the targets at higher depths of 127 mm, the time of detection was about 9:00 hours from the time of sunrise on a sunny day. 91

108 For the 51 mm deep target, the correlation of total solar loading with the maximum thermal contrast was much better when compared to the maximum solar loading versus the maximum thermal contrast. This was due to the effect of periodic cloud cover and effects due wind speeds and temperature variations during day and night. These findings were very useful for the application of infrared thermography in the field settings. Results from the field testing proved the effectiveness of the thermographic imaging at locations where the solar loading was not available as required. The condition assessment of the soffit areas of a bridge proved the presence of delaminated concrete which could result in spalling of concrete and the danger of the mass of concrete falling on the traffic below. The traffic control was not required for scanning the underlying portions of the bridge. 92

109 3 INFRARED THERMOGRAPHY TESTING FOR MANUFACTURING QUALITY CONTROL 3.1 INTRODUCTION This chapter discusses the application of Infrared thermography testing during the manufacturing (pultrusion) stage of Glass Fiber Reinforced Polymer (GFRP) structural components. The testing was conducted at the Bedford Reinforced Plastics (BRP) manufacturing plant in Bedford, PA on July 8, Three types of GFRP components were tested, namely, 4 inch (101.6 mm) wide C sections, 3.5 inch (88.9 mm) square tube sections, and panel sections. The following sections provide a brief description of the infrared equipment used, followed by the details of the infrared tests conducted on the specimens. 3.2 INFRARED TESTING EQUIPMENT This study used the InfraCAM SD infrared camera for capturing the infrared images (Figure 3.1). The description of the camera is given in the next section. The control temperature inside the manufacturing chamber (die) for the components was very high (~ 250 o C) and the components coming out of the die were very hot and the heat distribution was very uniform. Hence there was no need for using an external source for heating the components. Figure 3.1 Picture of the InfraCAM SD infrared camera 93

110 3.2.1 Description of the Infrared Camera Figure 3.1 shows the picture of the InfraCAM SD infrared camera manufactured by FLIR systems. The camera measures the infrared radiation emitted from an object and converts it to an equivalent temperature value in accordance with the Stefan-Boltzmann law. The thermal images that the camera produces are directly saved on a SD memory card which stores thousands of images in standard radiometric JPEG format. With the new SD card technology, you no longer need to be tethered to a laptop computer. Also, this camera comes with a much lower price tag (~$3500 in year 2009) which makes it lot more affordable than the $50,000 price tag for high-end infrared cameras (e.g., ThermaCAM S60). The InfraCAM SD infrared camera is one of the lightest infrared thermal imaging camera that is currently commercially available and weighs only 1.21 pounds. The camera is dust and splash proof. The camera has a built in 24 o lens and meets IP 54 standards and withstands harsh industrial environments. It can detect infrared radiation in the spectral range of 7.5 to 13 microns. The camera is capable of making temperature measurements in the range of -10 C to +350 C (+14 F to +662 F). The thermal images of InfraCAM SD are clearly displayed on the large 3.5 color LCD with 240 by 240 pixels (the actual detector is 120 by 120 pixels). The minimum focus distance of the infrared camera is 0.3m. Thermal sensitivity of this infrared camera is 0.1 o C. It is possible to capture and store images on a removable SD flash card. The images that the camera produces can be analyzed either in the field by using the real-time spot temperature measurement marker built into the camera software, or in a computer using FLIR Systems Quickreport software. The spot temperature measurement option offered by the software enables temperature measurement corresponding to any point in the field or later in the laboratory. The area feature provides average temperature over a small area and has the advantage of minimizing the random noise associated with the various pixels Infrared Testing Results Channel sections The pultruded channel sections were 4" (101.6 mm) wide and cut to 20 feet (6.1 m) in length. These channel sections are used for ladder rails, horizontal supports, pump housing units, stair treads and pedestrian bridges. They also find applicability in large 94

111 span constructions. The channel sections coming out of the manufacturing die were very hot and the heat distribution was very uniform. Hence no external heat source was required for acquiring the infrared images. Figure 3.2 shows a picture of the setup at the manufacturing plant. Figure 3.2 Setup at the manufacturing plant Figure 3.3 shows the digital picture and infrared images of the channel section coming out of the manufacturing die. The infrared image shows the temperature at the location of the spot marker as 168 o C. The more or less uniform color of the image shows that the section is defect-free. (a) Picture of the channel section (b) Infrared image showing the channel section free of defects Figure 3.3 Digital picture and infrared images of the channel section coming out of the manufacturing die 95

112 Figure 3.4 shows the digital piture and infrared images of the channel sections which were stacked up after cutting them into required size. Figure 3.4(b) shows one of the channels as a brighter spot. This is due to the fact that the channel section which appears as a brighter spot has a higher temperature than the other sections. This channel was the last one coming out of the manufacturing die and showed up as a hotter region with an average temperature of 62.6 C whereas the channel sections which were pultruded earlier had cooled down to a lower temperature (~ 48 C). (a) Picture of the stacked channel sections (b) Infrared image of the stacked channel sections Figure 3.4 Digital picture and infrared images of the stacked channel sections Tube sections The square tube sections are typically pultruded in sizes of 1 x 1 x 1/8 (25.4 mm x 25.4 mm x 3.2 mm) up to 4 x 4 x 3/8 (102 mm x 102 mm x 9.6 mm). The components tested in this study were 3.5" x 3.5" (89 mm x 89 mm) square tubes which could be used in handrails, caged and straight ladders, structure framings and as column sections. Figure 3.5 shows the sections coming out of the manufacturing die. As mentioned in the previous section, no external heat source was required for heating these components they already had uniform heating in the pultruding die. 96

113 Figure 3.5 Tube sections coming out of the manufacturing die Figure 3.6 Infrared image of the tube sections coming out of the manufacturing die No defective areas were found in the case of tube sections as seen in the infrared image in Figure 3.6. This indicated that excellent quality control was ensured throughout the manufacturing process. The infrared image in Figure 3.6 shows higher temperature for the part of the component near the die and lower temperature away from the die due to atmospheric cooling with time. Figure 3.7 shows the picture and infrared images of the stacked up tubes. Figure 3.7(b) indicates the average temperature values at three locations. The outermost tube on the right side shows a higher temperature (46.8 o C) 97

114 since this was the most recently pultruded component in the stack. On the other hand, the tube sections at the bottom of the stack show a lower temperature (39.4 o C). (a) Picture of the tube (b) Infrared image of the sections tube sections Figure 3.7 Digital picture and infrared images of the stacked tube sections GFRP panel sections This section demonstrates a comparison between infrared images from defect-free and defective GFRP panels. Figure 3.8 shows the infrared image of a defect-free panel. The brighter lines indicate the web portions of the panel (i.e., the flange-web junction). The surface temperature above the web location was about 66.4 C whereas the surface temperature above the flange appeared to be relatively lower with an average temperature of 56.9 C. 98

115 Figure 3.8 Infrared image of the defect-free panel section Figure 3.9 shows the infrared image of a trial run of a GFRP panel with a very significant defect which was visible while the component was coming out of the manufacturing die. At the defect location, the top layer of the panel had peeled off due to initial set-up problems during the pultrusion process. This was a new manufacturing process and the infrared images were acquired during the test run. The die set up was quickly adjusted to pull a defect free panel. Figure 3.9 Infrared image of the section showing debonding of the GFRP layer 99

116 3.3 CONCLUSIONS Infrared thermography testing can be an effective tool in ensuring the quality control during the GFRP component manufacturing process. The channel and tube sections tested in this study were pultruded using an established process, hence all the components tested using infrared thermography were found to be defect-free. It is important to note that the infrared thermography technique used here can detect defects as small as 10 mm x 10 mm in plan size. However, defects with the size of a "needle hole" in the channel and tube sections cannot be detected using the current technology. The larger defect such as peeling off of the GFRP layers in the panel sections occurring due to initial set-up problems during the pultrusion process could be detected using infrared thermography. This was a new manufacturing process and the die set up was quickly adjusted to pull a defect free panel. 100

117 4 INFRARED FIELD TESTING OF A RC BRIDGE This chapter describes the infrared field testing conducted on a reinforced concrete (RC) bridge in Huntington, West Virginia. This bridge is a candidate for rehabilitation with GFRP fabrics in the near future. The details of the bridge and results of the infrared thermography testing for the detection of underlying defects in the RC members prior to their rehabilitation are presented in the following sections. 4.1 DESCRIPTION OF THE BRIDGE The Madison Avenue bridge in West Virginia is a four-lane bridge which is part of the US 60 highway between Charleston and West Huntington. Figure 4.1 shows the deck view, and elevation view of the bridge. The average temperature between the testing hours of 11:00 A.M. and 1:00 P.M. on the testing date (September 21, 2010) was 25 C and it was a sunny day. (a) South Side North Side Figure 4.1 (a) Deck view, and (b) Elevation view of the bridge (b) 101

118 (a) (b) (c) Figure 4.2 View of the supports (a) Support at higher elevation (north side), (b) Support at lower elevation (south side), and (c) Close up view of the severely damaged lower support 102

119 The bridge is supported by 8 piers (4 on each side - north and south) and 4 pier caps (2 on each side) as shown in Figures 4.2(a) and 4.2(b). The geometry of the bridge was such that the north end was at a higher elevation and the south end was at a lower elevation. The support on the south side, shown in Figures 4.2(b) and 4.2(c), was at a lower height with rain water seeping from this end. Hence the bridge was severely damaged at this end, and the infrared thermography testing was also focused at this end. 4.2 INFRARED TESTING EQUIPMENT Figure 4.3 (a) shows the portable infrared camera which was used for the field testing. The description of the camera and its features has already been provided in the previous chapter (Section 3.2.1). Figure 4.3 (b) shows the shop heater that was used for heating the potential critical areas of the piers and the pier caps. This heater was a 1500W quartz heater consisting of two heating rods. The heater could provide high intensity heat with uniform heating by moving it back and forth over the surface under inspection. (a) InfraCAM SD TM infrared camera (b) Shop heater Figure 4.3 Infrared testing equipment 4.3 INFRARED TESTING AND RESULTS PRIOR TO GFRP WRAPPING (SEPTEMBER 21, 2010) It was necessary to heat the RC members using the shop heater prior to acquiring the infrared images. The heater was moved back and forth from a distance of 1 ft (0.3m) 103

120 over the test surface in order to ensure uniform heating. The heating time for each location was about 15 minutes as the test area consisted of a very large mass of concrete. After the heat source was removed and the infrared image was acquired, the defective regions showed up as hot spots (areas with higher temperatures) compared to the surrounding defect-free regions. This is because the regions with subsurface defects do not conduct heat quickly as compared to the defect-free regions. So the surface temperature above the defects is higher Pier Cap 3 Figure 4.4 shows the digital and infrared pictures of the bottom middle portion of the pier cap 3 (highlighted in Figure 4.2(b)). The digital picture (photograph) in Figure 4.4 (a) shows the spalling of concrete. The infrared image in Figure 4.4 (b) was acquired at this location and the image shows the hot spots along the crack lines. The highlighted regions indicate the surface temperatures at the critical areas (as high as 44.9 C) and the defect-free area (30.3 C). (a) Picture of the bottom portion of pier cap 3 (b) Infrared image of the bottom portion of pier cap 3 Figure 4.4 Digital picture and infrared image of the bottom middle portion of pier cap Pier Cap 3 (Bottom Corner Portion) This portion of the pier cap 3 is near the left side of pier 6 (shown in Figure 4.2(b)). Figure 4.5(a) shows a digital picture of the region with loose concrete mass and several cracks. This portion was captured with the infrared camera and Figure 4.5(b) 104

121 shows that most of the region has defects, as suggested by the hot areas with surface temperatures over 34 C. The area was also subjected to tap testing in order to verify the results obtained from infrared thermography. The tap hammer also suggested the presence of hollow mass of concrete. (a) Picture of the bottom corner portion of pier cap 3 (b) Infrared image of the bottom corner portion of pier cap 3 Figure 4.5 Digital picture and infrared image of the bottom corner portion of pier cap Pier Cap 4 (Part 1) The side portion of the pier cap 4 (Figure 4.2(b)) was also inspected for detecting debonded mass of concrete. Figure 4.6(a) shows the digital picture of the defective region clearly demarcated with the help of infrared thermography. This also proves the ability of infrared thermography to clearly identify the extent of underlying defective regions. In this case, the defective zone had an average temperature of 43.7 C and the major part of the debonded concrete region had an average temperature of 40.2 C as compared to a value of 33.3 C for the defect-free regions. It should be noted that the debonding is not visible on the surface (Figure 4.6(a)) but was captured using infrared thermography (Figure 4.6(b)). 105

122 (a) Picture of the side portion (part 1) of pier cap 4 (b) Infrared image of the side portion (part 1) of pier cap 4 Figure 4.6 Digital picture and infrared image of the side portion (part 1) of pier cap 4 (a) Picture of the side portion (part 2) of pier cap 4 (b) Infrared image of the side portion (part 2) of pier cap 4 Figure 4.7 Digital picture and infrared image of the side portion (part 2) of pier cap 4 106

123 4.3.4 Pier Cap 4 (Part 2) This region was identified adjacent to part 1 of the same pier cap and had similar problem of concrete with subsurface debond. The tap hammer tests conducted in this region indicated hollow sound and coincided very well with the defective region detected using infrared thermography as shown by the area marked with white boundary in Figure 4.7(a). The average temperature for the defect (hot spot) was found to be 47.5 C (Figure 4.7(b)). The second hot spot had an average temperature of 42.9 C. On the other hand, the defect-free zone had an average temperature of 32.7 C Pier 7 This pier was the only one which had crack propagating from the pier cap. The portion of the pier near the joint with the pier cap was tested using both infrared thermography and tap hammer tests to identify the underlying defective region. Figure 4.8(a) shows the cracked region. But infrared thermography (Figure 4.8(b)) indicated the presence of a large inner void due to peeling of the external concrete layer. The crack line to the right in the infrared image showed up as a relatively hotter zone (with an average temperature of 36.9 C) compared to the other zones. The tap hammer test also produced hollow sound in the defective region, thus supporting the infrared thermography results. (a) Picture of the pier 7 at the joint location (b) Infrared image of the pier 7 at the joint location Figure 4.8 Digital picture and infrared image of the pier 7 at the joint location (pier cap 4) 107

124 4.4 CONCLUSIONS The infrared thermography testing of the Madison avenue bridge was conducted in order to identify the regions which needed to be rehabilitated by wrapping them with Fiber Reinforced Polymer fabrics. Some of the interesting points that can be concluded from this field infrared thermography testing are as follows: In addition to the identification of defects which are visible to the naked eye, infrared thermography serves as a very effective technique to identify the presence of underlying defects (debonds) that may not be visible on the surface. It is possible to mark the extent of the defect propagation area using infrared thermography even though the entire damaged area may not be visible on the surface. This information about the extent of damage area is crucial in optimizing the GFRP wrapping process. The heating time needed for each location in the reinforced concrete components was about 15 minutes as compared to a 1 to 2 minute heating time for debond detection in most GFRP bonded composite specimens. This is because concrete has a large thermal mass and the components were quite thick. It is important to note that not all defects can be detected using infrared thermography. For example, defects with very small size or deeper defects may not be detected by the infrared thermography technique. Therefore, absence of anomalies (hot spots) in the infrared image does not necessarily mean that the test object is completely defect-free. 108

125 5 INFRARED FIELD TESTING OF FRP WRAPPED WOODEN RAILROAD BRIDGE COMPONENTS This chapter focuses on the application of infrared thermography for monitoring wooden railroad bridge components wrapped with Glass Fiber Reinforced Polymer (GFRP) fabrics. The infrared thermography testing was conducted on October 29, 2010 and April 14, 2011 for FRP wrapped wooden piles and bents (pile caps) of three wooden bridges in Moorefield, WV. The three bridges that were tested are numbered as Bridge #32.8, #35.7 and #36.7 by the Federal Railroad Administration (FRA). The following sections describe the infrared thermography equipment and the test results. 5.1 INFRARED THERMOGRAPHY EQUIPMENT Figure 5.1 (a) shows the portable infrared camera which was used for the field testing. The description of the camera and its features has already been provided in chapter 3 (Section 3.2.1). Figure 5.1 (b) shows the shop heater that was used for heating the possible critical areas of the piles and pile caps (bents). This heater was a 1500W quartz heater consisting of two heating rods which could provide high intensity and uniform heating of the surface under inspection by moving the heater continuously back and forth over the test surface from a distance of 1 ft (0.3 m). (a) Picture of the InfraCAM SD TM (b) Shop heater infrared camera Figure 5.1 Infrared testing equipment 109

126 5.2 INFRARED FIELD TESTING (OCTOBER 29, 2010) Bridge #32.8 In the bridge #32.8, two GFRP composite wrapped piles were tested. The numbering of the bents was based on the directions. The bents were numbered 0 to 6 from north to south. Figure 5.2 shows a side view of the bridge. Each bent is supported on 4 timber piles. The pile 4 of bent 5 and pile 2 of bent 6 were initially wrapped with GFRP. The wrapped portions of these piles were subjected to infrared thermography testing. The shop heater was used to heat the surface of the critical areas of the piles. The heating time was approximately 1 to 2 minutes in order to get the desired thermal contrast. After the heat source was removed, the infrared images were acquired with the help of the infrared camera for further analysis. Figure 5.2 Side view of the bridge # Pile 4 of Bent 5 (a) Bottom portion Figure 5.3(a) shows the digital picture of the bottom portion of pile 4 of bent 5. Approximately 50% of the pile was wrapped with GFRP fabric and hence the infrared images were acquired by dividing this wrapped region into 3 portions namely, bottom, middle and top portion. Figure 5.3(b) shows the infrared image of the bottom portion of the pile with the hot spots (highlighted areas). The average temperatures of the defective 110

127 and defect-free regions were obtained. The hot spots (defective regions) had higher average temperatures of 44.7 C and 43.9 C compared to the surrounding defect free regions which had an average temperature of 39.4 C as shown in Figure 5.3(b). (a) Digital picture (b) Infrared image Figure 5.3 Digital picture and infrared image of the bottom portion of pile 4 of bent 5 (a) Digital picture (b) Infrared image Figure 5.4 Digital picture and infrared image of the middle portion of pile 4 of bent 5 111

128 (b) Middle portion Figure 5.4 shows the digital picture and infrared image of the middle portion of pile 4 wrapped in GFRP fabric. There was a visible bump as shown by the marked region in Figure 5.4(a). The infrared testing of this region confirmed the presence of minor defects as shown by the two high temperature spots in Figure 5.4(b). These hot spots had higher temperatures of 54.8ºC and 54.4ºC whereas the surrounding defect-free regions had an average temperature of 48.2ºC, indicating the presence of defects. (a) Digital picture (b) Infrared image Figure 5.5 Digital picture and infrared image of the top portion of pile 4 of bent 5 (c) Top portion Figure 5.5 shows the digital picture and infrared image of the top portion of the pile 4 wrapped in GFRP composite fabric. When the digital picture and the infrared image were compared at the top point where the GFRP reinforcement ends, there was no proper bonding of the GFRP fabric with the underlying timber pile. Hence, the infrared image shows a hot spot (39.9 C) in the top region. Also, the air void at the lower portion of Figure 5.5(b) had an average temperature of 44.7 C when compared to the surrounding defect-free region's temperature of 24.3 C. 112

129 (a) Digital picture (b) Infrared image Figure 5.6 Digital picture and infrared image of pile 2 of bent Pile 2 of Bent 6 This was an interior pile of the extreme end support, on the south side of the bridge. The infrared thermography testing for this portion was difficult because the accessible area for heating the pile uniformly was very limited. Figure 5.6 shows the digital picture and the infrared image of the pile wrapped in GFRP fabric. The infrared image in Figure 5.6(b) shows the presence of a hot spot with a high temperature of 45.2 C and the surrounding defect-free temperature has an average temperature of 25.8 C Bridge #35.7 The bridge #35.7 was supported over 7 bents and each bent was supported on four piles. The bents were numbered from 0 to 6 from north to the south direction as shown in Figure 5.7. A total of four piles in this bridge were wrapped with GFRP composite. These piles were numbered as pile 3 and pile 4 of bent 0 and pile 2 and pile 3 of bent 4, and were subjected to infrared thermography testing in order to identify the presence of any defects. The following sections provide the details of the infrared thermography testing on these four piles. 113

130 Figure 5.7 Side view of the bridge # Pile 3 of Bent 0 Figure 5.8 shows the digital picture and infrared image of the pile 3 of bent 0. As observed from the digital picture, the access area was very small and hence the heating of the pile was difficult. However, the infrared image was still obtained (Figure 5.8(b)) and shows that there are a number of bright spots which indicate debonded areas under the wrap. These debonds were also confirmed using the tap hammer tests. The bright spot had an average temperature of 45.8 C while the surrounding defect-free region had an average temperature of 35.9 C. (a) Digital picture (b) Infrared image Figure 5.8 Digital picture and infrared image of pile 3 of bent 0 114

131 Pile 4 of Bent 0 Figure 5.9 shows the digital and infrared images of the pile. The infrared image shows some debonded areas (hot spots) with higher temperatures compared to the surrounding regions. The average temperatures for the debonded regions were 38.1 C and 37.5 C as compared to the defect-free region which has an average temperature of 30.3 C. (a) Digital picture (b) Infrared image Figure 5.9 Digital picture and infrared image of pile 4 of bent Pile 2 of Bent 4 Figure 5.10(a) shows the digital picture of the pile 2 of bent 4. The highlighted area in this picture was found to be a defective region using infrared thermography. The infrared image shown in Figure 5.10(b) shows a debond (hot spot) with a temperature of 44.3 C when compared to 33.6 C for the defect-free region. The highlighted region was also tested with the help of a tap hammer, which produced hollow sound in this region indicating the presence of air-filled debond. 115

132 (a) Digital picture (b) Infrared image Figure 5.10 Digital picture and infrared image of pile 2 of bent Pile 3 of Bent 4 Figure 5.11 shows the digital picture and infrared image of the pile 3 of bent 4. The digital picture in Figure 5.11(a) clearly shows that the pile wrapped in GFRP had many imperfections in the underlying timber surface. As a result, after wrapping the pile with GFRP fabric, many small air-filled voids were formed. These small voids were clearly identified using infrared thermography. Figure 5.11(b) shows the infrared image of the pile with many surface imperfections. The irregular distribution pattern for higher and lower temperature regions in the infrared image clearly indicates the presence of small air gaps at many locations. The average temperature of the highlighted defect free region was found to be 38.6 C whereas the hot spots (air-filled voids) had temperatures as high as 52.4 C. 116

133 (a) Digital picture (b) Infrared image Figure 5.11 Digital picture and infrared image of the pile 3 of bent Bridge #36.7 The bridge #36.7 was supported on 8 bents and each bent was supported by 4 piles. Figure 5.12 shows a side view of the bridge. As observed from the picture there was a small body of water underneath the bridge and many of the piles were partiallly submerged in water. The infrared thermography testing of this bridge proved to be a very quick and effective technique in order to identify the subsurface defects in the GFRP wrapped piles. The pile 3 and pile 4 of bent 0, the pile 3 of bent 2, pile 1 and pile 2 of bent 3 and the pile 1 of bent 4 were wrapped with GFRP fabrics in Summer Also, infrared thermography tests were performed on bent 4 and bent 6 which were wrapped with GFRP fabric much earlier, in Summer Pile 3 of Bent 0 Figure 5.13(a) shows the digital picture of the pile 3 of bent 0. The highlighted areas were identified as defective regions with the help of infrared thermography. The defective region at the top was a bump and shows up as a hot spot in the infrared image (Figure 5.13(b)). Both the highlighted defective regions had an average temperature of 25.0 C as compared to the relatively colder defect-free regions which had an average temperature of 19.7 C. 117

134 Figure 5.12 Side view of bridge #36.7 (a) Digital picture (b) Infrared Image Figure 5.13 Digital picture and infrared image of the pile 3 of bent Pile 4 of Bent 0 Figure 5.14(a) shows the digital picture of the pile 4 of bent 0 wrapped with GFRP. As observed from the picture there were some air voids (debonds) present after wrapping the pile with GFRP and this was clearly identified with the help of infrared thermography (Figure 5.14(b)). The defective regions had an average temperature of 39.4 C compared to defect-free regions with an average temperature of 24.6 C. These 118

135 regions were tap tested which also indicated the presence of debonds similar to the results from infrared thermography. (a) Digital picture (b) Infrared image Figure 5.14 Digital picture and infrared image of the pile 4 of bent Pile 3 of Bent 2 Figure 5.15 shows the digital picture and infrared images of the pile 3 of bent 2. Figure 5.15(a) shows the presence of an air void in the upper part and a metal bolt in the lower part. Infrared thermography clearly detected the air void and it was shown by the hot spot with a temperature of 62.3 C in Figure 5.15(b). The metal bolt being a good conductor of heat dissipates the heat energy through its length very quickly and shows a relatively lower surface temperature of 21.6 C Pile 1 of Bent 3 Figure 5.16(a) shows the digital picture the pile 1 of bent 3. The bottom portion of the pile (middle part of the picture) was severely damaged. Hence this portion of the pile was wrapped with more layers of GFRP fabric for additional strength. After the heat source was removed and the infrared image was acquired (Figure 5.16(b)), it was observed that there was a bright spot in this portion with an average temperature of 45.4 C indicating the presence of air void. 119

136 Metal Bolt (a) Digital picture (b) Infrared image Figure 5.15 Digital picture and infrared image of the pile 3 of bent 2 (a) Digital picture (b) Infrared image Figure 5.16 Digital picture and infrared image of the pile 1 of bent Pile 2 of Bent 3 (a) Bottom portion The bottom portion of this GFRP wrapped pile, shown in Figure 5.17(a), had a lot of surface imperfections before the application of GFRP. Also, the improper application of resin caused formation of small air voids. This was clearly identified with the help of infrared thermography. The infrared image in Figure 5.17(b) shows the presence of 120

137 small air voids distributed across the wrapped section of the pile. The surface temperatures for some of the regions with air voids are clearly demarcated in Figure 5.17(b). (a) Digital picture (b) Infrared image Figure 5.17 Digital picture and infrared image of the bottom portion of pile 2 of bent 3 Metal Bolt (a) Digital picture (b) Infrared image Figure 5.18 Digital picture and infrared image of the top portion of pile 2 of bent 3 (b) Top portion Figure 5.18(a) shows the digital picture of the top portion of the pile mentioned in the previous section. There was an existing metal bolt driven into the timber pile before 121

138 the application of GFRP and the bolt caused the bump on the outer surface. As metals are good conductors of heat, the infrared image of the wrapped pile after heating (Figure 5.18(b)) revealed the metal bolt as a cold spot with a temperature of 17.5 C. The infrared image also revealed that the top portion of the pile had a few air-filled debonds. The hot spot identified in the top portion had a temperature of 32.3 C. (a) Digital picture (b) Infrared image Figure 5.19 Digital picture and infrared image of the pile 1 of bent Pile 1 of Bent 4 Figure 5.19 shows the digital picture and infrared image of the pile 1 of bent 4. The infrared thermography of this pile was difficult to conduct as the pile was submerged in water and the area was not easily accessible. Hence, this technique could not be applied effectively. The infrared image shows hot spots at the center of the image with higher temperatures but the thermal contrast was not really good. The hot spots at the center of the image shown in Figure 5.19(b) were identified as debonds, with an average temperature of 45.0 C, but the image could not clearly show the boundaries for the debonds. The defect-free regions had an average temperature of 36.7 C Bent 4 (wrapped with GFRP in summer 2000) Figure 5.20 shows the digital picture and infrared image of the bent 4 which had previously been wrapped with GFRP fabric in summer The digital picture in Figure 5.20(a) clearly shows the deteriorating condition of the wrap. After the heating 122

139 source was removed and the infrared image of the bent was acquired, it was observed that there was a higher temperature zone (hot spot) in the bent, located near the bent-pile junction, as indicated in Figure 5.20(b). This was due to the debonding of the GFRP fabric in this region. The hot spot had an average temperature of 44.5 C as compared to 30.8 C for the defect free regions located away from the debonded region. (a) Digital picture (b) Infrared image Figure 5.20 Digital picture and infrared image of the bent 4 with an old GFRP wrap Bent 6 Top Surface (wrapped with GFRP in Summer 2000) Figure 5.21 shows the digital picture and infrared image of the top surface of the bent 6. The digital picture in Figure 5.21(a) shows that the GFRP surface has severely degraded due to environmental effects including the direct ultraviolet (UV) exposure from solar rays. Figure 5.21(b) shows the infrared image that was obtained after heating the top surface of the bent for about 2 minutes. The infrared image shows very few hot spots on the surface of the bent. One of the highlighted regions in the image shows a debond with a higher temperature of 40.3 C while the surrounding defect-free region has a temperature of 31.7 C. In spite of the fact that the digital picture in Figure 5.21(a) shows considerable UV degradation, the infrared image in Figure 5.21(b) indicates that most of the wrapped region is debond-free. This indicates that the GFRP wrap is functioning very well in terms of providing the confinement action to the underlying timber bent. 123

140 (a) Digital picture (b) Infrared image Figure 5.21 Digital picture and infrared image of the top surface of bent 6 with an old GFRP wrap 5.3 INFRARED FIELD TESTING (APRIL 14, 2011) Infrared thermography tests were conducted on the same bridge locations as mentioned in the previous sections, on a different day (April 14, 2011), in order to compare the condition of the GFRP wrapped regions and determine if additional debonds had formed between the previous and the current test dates Bridge # Pile 4 of Bent 5 (side middle portion) The wrapped portion of this pile was divided into two zones in order to obtain the infrared images. Figure 5.22 shows the digital picture and infrared image of the middle portion of the pile. It should be noted that this represents the side of the pile and not the front part shown in Figures 5.3, 5.4, and 5.5 in the previous section. The digital picture in Figure 5.22(a) shows the imperfection caused during the wrapping process. As observed from the infrared image in Figure 5.22(b), this imperfection resulted in a bright spot at the center portion of the image indicating the presence of a defect. The average temperature for this region was found to be 43.2 C which is much higher compared to 35.3 C for the defect-free surroundings. 124

141 (a) Digital picture (b) Infrared image Figure 5.22 Digital picture and infrared image of the side middle portion of pile 4 of bent Pile 4 of Bent 5 (side bottom portion) Figure 5.23 shows the digital picture and infrared image of the bottom portion of the pile 4 of bent 5. As seen from the infrared image in Figure 5.23(b), the temperature in the entire image is quite uniform with an average temperature 31.6 C. The tap meter was later used to test this region and the readings were very similar over the entire region indicating a defect-free region. (a) Digital picture (b) Infrared image Figure 5.23 Digital picture and infrared image of the bottom side portion of pile 4 of bent 5 125

142 Pile 2 of Bent 6 Figure 5.24 shows the digital picture and infrared image of the pile 2 of bent 6. As seen from the digital picture, there is a metal bolt in the bent portion which is seen as a cold spot in the infrared image with an average temperature of 28.2 C. The portions at the junction of the pile and bent appear to have small air voids and hence show up as brighter spots with average temperatures of 61.4 C and 60.4 C. These temperatures are significantly higher compared to the 50.3 C temperature of the defect-free region. When the infrared image in Figure 5.24(b) is compared with the infrared image in Figure 5.6(b) from the previous test date, it can be seen that both the images show the debond in the pile at the same location thus proving the consistency of the results obtained by infrared thermography. The comparison of Figure 5.24(b) with Figure 5.6(b) also indicates that an additional debond has begun to develop at the bent-pile junction. (a) Digital picture (b) Infrared image Figure 5.24 Digital picture and infrared image of the pile 2 of bent Bridge # Pile 3 of Bent 0 Figure 5.25 shows the digital picture and infrared image of the pile 3 of bent 0. From the digital picture (Figure 5.25(a)), it is evident that the timber pile did not have a smooth surface before the wrapping was done. Hence, from the infrared image it was evident that air voids were formed when the wrapping was done. Figure 5.25(b) shows 126

143 the infrared image of the pile with a bright spot showing the presence of a debond. The debonded region has an average temperature of 44.4 C as compared to a non-defective region at the bottom, which has an average temperature of the 32.5 C. The highlighted critical area demarcated using infrared thermography was then tap tested and the digital tap meter produced hollow sounds indicating the presence of voids. This was in agreement with the infrared thermography results. Figure 5.8 shows the same pile which was tested earlier. The comparison of Figure 5.25(b) with Figure 5.8(b) shows that the debond size has increased between the previous test on October 29, 2010 and the current test on April 14, (a) Digital picture (b) Infrared image Figure 5.25 Digital picture and infrared image of the pile 3 of bent Pile 4 of Bent 0 Figure 5.26 shows the digital picture and infrared image of the pile 4 of bent 0. As seen from the digital picture in Figure 5.26(a), this pile had a very low clearance as it was at the extreme support location. Hence, the uniform heating of the pile was difficult to achieve. The infrared image in Figure 5.26(b) shows a much brighter area at the top corresponding to the bent portion and a cold spot in the bottom part. The uniform heating could not be achieved at the bottom portion of the pile, so debonds could not be detected. 127

144 (a) Digital picture (b) Infrared image Figure 5.26 Digital picture and infrared image of the pile 4 of bent 0 (a) Digital picture (b) Infrared image Figure 5.27 Digital picture and infrared image of the pile 2 of bent Pile 2 of Bent 4 Figure 5.27 shows the digital picture and infrared image of the pile 2 of bent 4. After the heating source was removed and the infrared image was obtained, it was observed that there were some bright spots (Figure 5.27(b)) indicating the presence of airfilled debonds. As observed from the infrared image, the average temperature for the debonded regions were found to be 41.3 C, 41.5 C and 42.0 C which are much higher 128

145 when compared to surrounding defect-free regions which had an average temperature of 33.0 C. These debonded areas were demarcated using a chalk (Figure 5.27(a)) and tested with digital tap testing meter which produced hollows sounds when these debonds were tapped. Figure 5.27(b) is taken at a closer distance compared to Figure 5.10(b) and the comparison of these two figures indicated the formation of additional debonds between the two test dates Bridge # Pile 4 of Bent 0 Figure 5.28 shows the digital picture and infrared image of the pile 4 of bent 0. The digital picture in Figure 5.28 shows a surface imperfection. The area was examined using infrared thermography for the possibility of any subsurface defects. The infrared image in Figure 5.28(b) shows a very large hot spot with a very clear boundary confirming the presence of a large debonded region. The average temperatures within the debonded regions were found to be 39.8 C and 39.1 C (in different portions) as compared to the surrounding defect-free region which has an average temperature of 30.9 C. The comparison of Figure 5.28(b) with Figure 5.14(b) indicates that the debonded area is increasing in size. (a) Digital picture (b) Infrared image Figure 5.28 Digital picture and infrared image of the pile 4 of bent 0 129

146 Pile 2 of Bent 3 Figure 5.29 shows the digital picture and infrared image of the pile 2 of bent 3. The pile had a metal bolt driven in it before the GFRP wrapping was done. Hence after the wrapping, there was an air gap created around the metal bolt. The infrared thermography testing of the pile clearly identified the metal bolt as a cold spot and the surrounding air gap was shown as a brighter spot. The metal bolt conducts heat quickly through its length and appears as a cold spot in the infrared image. The metal bolt had an average temperature of 19.6 C and the surrounding air gap had a higher temperature of 29.4 C thus indicating a significant thermal contrast. The surrounding defect-free region had an average temperature of 25.8 C which is lesser than that of the air gap but higher than the metal bolt temperature in the infrared image. As the image in Figure 5.29(b) was acquired from a closer distance, it can be compared to Figure 5.18(b) and both the images shows the presence of metal bolt and some debonds just above the metal bolt. Metal Bolt (a) Digital picture (b) Infrared image Figure 5.29 Digital picture and infrared image of pile 2 of bent Bent 4 Figure 5.30 shows the bent which was wrapped with GFRP in summer The digital picture in Figure 5.30(a) shows the deteriorating condition of the wrap. Figure 5.30(b) shows the infrared image of the bent that was acquired after heating the surface with the quartz heater for about 2 minutes. The infrared testing revealed bright spots near 130

147 the junction of the bent and the pile and also at the center portion of the bent (this debond was not visible to the naked eye). The defects had average temperatures of 59.3 C and 58.1 C when compared to the defect-free region which had an average temperature of 46.3 C. The crack line at the junction was also clearly visible in the infrared image. The comparison of Figure 5.30(b) with Figure 5.20(b) showed that additional debonded areas were formed in the upper portion of the bent (with surface temperature of 59.3 C). (a) Digital picture (b) Infrared image Figure 5.30 Digital picture and infrared image of bent 4 with an old GFRP wrap (a) Digital picture (b) Infrared image Figure 5.31 Digital picture and infrared image of the top surface of bent 6 with an old GFRP wrap 131