DEVELOPMENT OF INFRA-RED THERMOGRAPHY NDT DETECTION OF DEFECTS IN CONCRETE AND STEEL STRUCTURES EXTERNALLY BONDED WITH CFRP SYSTEMS By Jawdat Mustafa Kamal Tashan B.Sc. Eng. (Hon) M.Sc. Eng. A thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy Faculty of Engineering and Industrial Sciences Swinburne University of Technology 2012
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
DEVELOPMENT OF INFRA-RED
THERMOGRAPHY NDT DETECTION OF
DEFECTS IN CONCRETE AND STEEL
STRUCTURES EXTERNALLY BONDED WITH
CFRP SYSTEMS
By
Jawdat Mustafa Kamal Tashan
B.Sc. Eng. (Hon)
M.Sc. Eng.
A thesis submitted in fulfillment of the requirements for the degree of
Doctor of Philosophy
Faculty of Engineering and Industrial Sciences
Swinburne University of Technology
2012
III
To all people who made life at this stage of civilization, in
the hope that this work will contribute
Summary
V
SUMMARY
Carbon fibre reinforced polymer (CFRP) composites are currently used externally to
retrofit and strengthen concrete and steel structures. One of the most important
requirements of CFRP- strengthened structures is the bond at the interface surface.
Bond defects can have a significant influence on the behaviour of the CFRP composite
structure. Different non-destructive tests were used previously to detect these defects.
This research investigates the ability of infra-red thermography (IRT) non-destructive
techniques (NDT) to detect different defects involving unbond areas, debond areas,
delamination, wet areas and cracks that may occur at the CFRP-substrate bond surface.
The literature review covers the background of the IRT approaches and techniques
employed in different applications. A review of the different CFRP applications and
their related installation methods used currently to retrofit different civil engineering
applications is presented, and summaries and evaluations of current studies that utilize
IRT to detect CFRP-concrete bond defects are outlined.
A total of 32 CFRP strengthened concrete and steel samples were constructed and tested
in this study. Artificial bond defects with different shapes and sizes were implanted
under CFRP composites. The defects involve unbond, delamination and debond areas
created at the bond line. Groove defects were embedded on the concrete surface of
selected specimens to verify the capability of IRT NDT to detect humidity. Cracks of
different sizes were generated at the concrete surfaces of several specimens to
investigate the technique in crack detectability. CFRP fabrics of different types were
used in the strengthening process of concrete and steel specimens. CFRP laminates were
also used in different combinations. Single and multiple-layers in the CFRP system
were adopted in the retrofitting designs.
The experimental work was divided in two major studies: qualitative and quantitative
infra-red thermography assessments. The qualitative tests were conducted with IR
detector type FLIR B200. Passive and active IRT were developed. Lamps of 2000 watts
were used as excitation sources in the active thermography approach. The qualitative
Summary
VI
results showed that the IRT is suitable for the detection of bond defects. The results also
showed that humid areas at the bond interface can be recognized by means of IRT
NDT. Generally, the qualitative thermography test results make this technique a
candidate for rapid detection and especially for bond and debonding defects in the bond
zone in single CFRP systems (fabric or laminate) and the substructure (concrete or
steel). The results indicate that for the purposes of in-depth defect characterization,
qualitative thermography is not recommended.
The second phase of the experimental work focused on the IRT quantitative approach.
A total of 32 specimens were tested during this phase, and different excitation systems
were employed. The quantitative studies were subcategorized into eight parts, and each
part addressed a different task. These tasks involved: emissivity evaluations, the
investigation of different bond defects and crack detection. Moreover, water presence
detectability was examined, and different heating inputs were studied. Precise
measurements of defect sizes and IRT error elimination studies were performed in the
quantitative studies. The overall results show high defect detectability and reasonable
accuracy in defect size identification. The experimental results provide guidelines that
can help thermographers to conduct efficient IRT NDT involving thermal input that can
be used to generate the designed thermal response with minimum thermal detection
during the IRT NDT.
Numerical analyses were then conducted to simulate and gain a better understanding of
the key parameters that have the most influence on the thermal response of a defect
within retrofitted surfaces. First, verification studies of the experimental and numerical
results were performed. There was a very good correlation between the empirical results
and the simulated FE analyses. Two 3D models were built using ANSYS 13 finite
element software analysis. One was for a concrete specimen strengthened externally
with a single fabric sheet which had a bond defect and the other was attached with
double CFRP sheets. Parametric studies involving material thermal properties, material
thickness and thermal input loads were carried out for both models. The results of these
numerical studies can serve as guidelines for thermographers to enable them to design
the thermal load input to achieve desired thermal responses.
Acknowledgments
VII
ACKNOWLEDGMENTS
This work would not have been possible without the help and contributions of others.
First, I would like to express my great appreciation to my main supervisor, Prof. Riadh
Al-Mahaidi for his enthusiasm, patience, encouragement and support throughout my
research. The support and guidance of my co-supervisor, Prof. John Wilson, is also
greatly appreciated. Their continuous inspiration, guidance and advice on my research
have been invaluable.
I would like to express my sincere gratitude to Monash University staff members Mr.
Long Goh, Mr. Jeffrey Doddrell, Mr. Alan Taylor and Ms. Jenny Manson for their help
and willing assistance with the laboratory phase of this study. Dr. Alex McKnight
assisted by proofreading the final version of the thesis.
I would also like to thank my colleague, Mr. Asghar Habibnejad for his tremendous
support in the experimental program.
I am indebted to my wife Ava Sidiq Mamkak for her patience, sacrifice, support and
understanding.
I would like to thank my mother, Mrs. Najla Albaiaty, Mr. Ali Tashan, Mr. Tariq
Tashan, Mr. Muard Tashan, and Ms. Gihan Tashan for their constant encouragement
and love throughout the course of my life.
Declaration
IX
DECLARATION
The candidate herein declares that the research work presented in this thesis contains no
material which has been accepted for the award of any other degree or diploma in any
university or other institutions. I affirm that to the best of my knowledge, the thesis
contains no material previously published or written by another person, except where
due reference is made in the text in the thesis.
Jawdat Tashan
Table of contents
XI
TABLE OF CONTENTS
SUMMARY ............................................................................................................................................... V
ACKNOWLEDGMENTS ..................................................................................................................... VII
DECLARATION ...................................................................................................................................... IX
TABLE OF CONTENTS ......................................................................................................................... XI
LIST OF FIGURES ............................................................................................................................ XVII
LIST OF TABLES ............................................................................................................................. XXIX
LIST OF NOTATIONS ..................................................................................................................... XXXI
2.2.4.1 Planck’s law .................................................................................................................................. 11
4.3.2.2 Air blower ................................................................................................................................... 104
4.5.1 Part 1: Emissivity value validation of the FRP using IRT ................................................ 123
4.5.1.1 Test set-up ................................................................................................................................... 124
4.5.4.2 Air blower excitation system ...................................................................................................... 194
4.5.4.3 Summary of Part 4 experimental program .................................................................................. 204
4.5.5 Part 5: Infra-red errors and noise .................................................................................... 205
4.5.5.1 Errors in IRT ............................................................................................................................... 205
4.5.5.2 Noise in the IRT .......................................................................................................................... 216
4.5.6 Part 6: IR detection of the presence of water.................................................................... 221
4.5.6.1 Summary of Part 5 ...................................................................................................................... 228
4.5.7 Part 7: Long-Pulsed IRT and Lockin thermography approaches ..................................... 229
APPENDIX A .......................................................................................................................................... 333
APPENDIX B .......................................................................................................................................... 337
LIST OF PUBLICATIONS ................................................................................................................... 343
Table 5.16 Epoxy specific heat simulations 283 to 289 ............................................... 304
Table 5.17 Epoxy conductivity simulations 290 to 295................................................ 306
Table 5.18 Concrete specific heat simulations 296 to 302 ........................................... 306
Table 5.19 Concrete conductivity simulations 303 to 308............................................ 307
Table 5.20 Double CFRP thickness simulations 309 to 315......................................... 308
Table 5.21 Epoxy thickness simulations 316 to 322 ..................................................... 310
Table 5.22 Concrete thickness simulations 323 to 326 ................................................. 311
Table 5.23 Thermal load simulations 327 to 341.......................................................... 312
List of notations
XXXI
LIST OF NOTATIONS
C = Thermal contrast
C (t) = Thermal contrast at specific time
Cmax = maximum thermal contrast
Ctmax = time that meet the peak of the thermal contrast
co = speed of light in vacuum
E = total emissive power
Eλ = spectral emissive power
Eλb = spectral emissive power for a blackbody
h = Planck’s constant
i , j = the x and y positions in an image of N ×M pixels
k = Boltzmann’s constant
n = constant refractive index
q = the input heat flux in watts per metre square
T = absolute temperature
T (t)defect = surface temperature above the subsurface defect at specific time
T (t)background = surface temperature in the surroundings defects-free area at specific time
t = time in seconds
Tambient = the ambient temperature
Tg = epoxy glass transition temperature
tmax = time for the maximum thermal signal
tmin = time for the minimum thermal signal
ΔT = thermal signal
ΔT (t) = thermal signal at specific time
ΔTmax = maximum thermal signal
ΔTmin = minimum thermal signal
ε = total emissivity
ε (T,λ) = spectral emissivity
λ = wavelength
µ = the mean of the noise distribution.
List of notations
XXXII
σ = Stefan-Boltzmann constant
Introduction
1
1 CHAPTER ONE: INTRODUCTION
1.1 Background
The use of carbon fibre reinforced polymer (CFRP) composites is expanding widely in
the strengthening of concrete and steel structures in civil engineering applications.
CFRP retrofit systems are two-phase materials that consist of micro-scale carbon fibres
saturated in a polymer matrix. The retrofitting can be applied with different types of
CFRP. Most CFRP products are applied to external surfaces of the structure to offer
additional strengthening. CFRP bars are also widely employed in structural concrete
members. This CFRP product can be used by grouting the bars with epoxy to provide
the required bonding forces within the existing structure.
The advanced properties of CFRP materials, involving their high strength, high
durability, high resistance to deterioration and light weight, have encouraged engineers
and manufacturers to employ these products in different industries, including aerospace
engineering and marine applications. CFRP systems have begun recently in civil
engineering structures to take the place of traditional methods of strengthening
structures like attaching external steel sections to existing concrete structures. Most of
the traditional methods of strengthening require the use of heavy steel sections that are
not easy to install at the site and may corrode easily when exposed to the weather.
According to the American Concrete Institute Committee 440 report (ACI Committee
440 2008), the advanced properties of CFRP composite materials make these products
ideal for use in different retrofitting processes in concrete structures, to enhance the
flexural and/or shear capacity of the structural member. However, the structural
mechanism and performance of these composite materials are still not fully understood.
The success of the strengthening or rehabilitation process with CFRP is crucially
dependent on the bonding conditions between the CFRP system and the substrate
structure. Bond defects due to improper CFRP application, delamination and cracking
can reduce the integrity and compatibility of the composite structure strengthened with
CFRP applications. The bond between adhesive and substrate structure is one of the
load path steps in the strengthening system, and it needs to be strong enough to transfer
Chapter One
2
the stresses to the carbon fibre materials adequately. If the retrofitted structures contain
these kinds of defects, the system will not provide the desired additional strength, and
the designed CFRP- system performance, durability and expected lifetime of the
strengthened structure will be under question. For these reasons, a process to detect and
study bond defects and to evaluate the installation quality of externally bonded CFRP
applications to civil engineering structures is urgently needed.
Different non-destructive methods have been used in bonding CFRP systems in
aerospace and mechanical applications. However, civil engineering structures differ
from other applications. Therefore, there is a need for a reliable and efficient method to
identify and detect bond defects and delamination of CFRP composites applied in civil
engineering structures.
There are several common non-destructive testing (NDT) methods to evaluate material
integrity and the overall composite structural consistency in civil engineering
applications. Nevertheless, because CFRP systems lack magnetism and electrical
resistance, some traditional non-destructive methods face major complexities in the
evaluation and detection of bond defects and delamination between CFRP and concrete
structures. According to the ACI 440 committee, several methods can be applied to
detect CFRP composite bonding defects, including acoustic emission, ultrasound, laser
shearography and infra-red thermography nondestructive tests methods. Acoustic
emission captures stress elastic waves produced by the development of cracks in
structures. Damage severity can by estimated through the study and analysis of these
waves. However, this method has limited capability to be applied in the field due to
reading errors that come from the noisy atmosphere of most civil engineering sites.
Ultrasound is a method which depends on injecting the structure with echo pulses and
receiving the reflected waves. These waves convey substrate defect data and provide
quantifiable information about the overall state of the structure. In spite of the
widespread use of this method in aerospace and mechanical applications, the use of this
method in civil engineering field conditions is limited for similar reasons to acoustic
methods. Moreover, because of the CFRP material's high attenuation [around 0.6
dB/mm (W. Hillger, R. Meier and Henrich 2004)] these materials have to be inspected
Introduction
3
with narrow band pulses and low frequencies. All these difficulties in meeting field
conditions requirements narrow the acoustic and ultrasound nondestructive methods
which can be applied widely to civil engineering applications. The laser shearography
method functions by projecting a laser beam onto the investigated surface and recording
images via a shearography camera. The method has promising abilities in terms of its
defect and flaw detection abilities, but, the high cost of the equipment is the major
reason that limits its use in civil engineering projects.
Infra-red thermography (IRT) nondestructive testing (NDT) has been suggested for the
detection of substrate defects and anomalies in CFRP-concrete and CFRP-steel
structures. The method is based on capturing the emission of infra-red radiation from
the investigated surfaces. Anomalies and defects under these surfaces can be localized
and observed in the thermal images (thermograms) with different temperature patterns
to the sound surrounding areas. IRT NDT can overcome the drawbacks and functional
difficulties of other nondestructive methods, including irrelevant sound information
coming from noisy field conditions. Moreover, the IRT equipment costs are reasonable.
IRT is easy to perform in different field conditions and can be used to evaluate and
inspect large areas. These advantages make IRT NDT a promising method for civil
engineering observation processes that can be executed effectively in most CFRP
strengthening applications.
1.2 Research objectives
Infra-red thermography has been promoted as an efficient method for the evaluation of
structural system integrity. Previous researchers have studied the use of IRT to detect
defects and anomalies at the FRP/concrete interface. However, most previous studies
have focused on qualitative IRT rather than quantitative assessment. A fully
comprehensive assessment of quantitative thermography in civil engineering
applications has not yet been provided. The application of the IRT in concrete and steel
structures strengthened with CFRP systems needs further investigation. Moreover, there
remains a lack of detailed scientific studies of the best test configuration and inspection
techniques for the thermographic evaluation of structures. If this NDT method is to
Chapter One
4
become widely used for the detection of bond defects and delamination in external FRP
composite bonded to concrete structures, a standard method with acceptable reliability
is required. The development of such a method requires a full understanding and deep
analysis of the parameters and factors controlling temperature re-distribution, heat flow
and radiation behaviours on the CFRP-substrate bond zone.
This thesis concentrates on experimental and numerical studies to develop a standard
methodology for the application of non-contact IRT NDT to assess concrete and steel
structures strengthened externally with different CFRP composites.
1.3 Research phases
Multiple approaches were presented in this research study. The research started with a
literature survey of IRT NDT and its application in CFRP strengthening in civil
engineering projects.
The next part of the study involved qualitative IRT studies applied to controlled-defect
specimens.
The third phase of this research investigation drew on the data gathered from an
extensive laboratory experimental program that using quantitative IRT techniques.
The final phase involved generating a finite- element numerical model to study the
different parameters influencing thermal responses in the IRT testing.
1.4 Thesis outline
This dissertation consists of six chapters, including this introductory chapter. Chapter 2
presents literature review including all the existing knowledge on IRT technology,
CFRP materials and systems and the use of IRT to evaluate bond defects in CFRP
retrofitted structures. Chapter 3 reports laboratory experimental work using the
qualitative IRT approach, and the deficiencies and drawbacks of this approach. Chapter
4 reports the results of a quantitative IRT laboratory experimental program on CFRP-
Introduction
5
strengthened concrete and steel specimens. The results of different quantitative studies
are reported in this chapter to help establish a standard for the use of IR NDT to detect
bond defects in structures strengthened externally with different CFRP products.
Chapter 5 presents a numerical approach to the study and assessment of the behaviour
of existing thermal models of retrofitted specimens. In addition, finite element modeling
is adopted to predict the thermal responses for other circumstances. A parametric study
is reported to examine the major factors influencing defect detection. Finally, in Chapter
6, major conclusions from this research are presented, with recommendations for future
studies.
Literature review
7
2 CHAPTER TWO: LITERATURE REVIEW
2.1 Introduction
Numerous studies on the detection of defects in the bonding area between FRP and
structures have been conducted using IRT NDT, and different approaches, parameters
and characteristics have been adopted. The literature review in this chapter is devoted to
IRT, FRP strengthening systems, and the inspection of defects in FRP composite
structures.
This chapter is divided into three main parts. The first reviews the principles of IRT
NDT and its applications; the next addresses the use of FRP composites for
strengthening structures; finally, the third part reviews the detection of FRP bond-
interface defects by IRT.
2.2 Infra-red thermography
2.2.1 Background
From the beginning of civilization, light was always an important issue in human life.
Man was curious about light and even gave it religious significance. Methods and
instruments for observing light have been recorded from early written history. One of
the oldest instruments in the world is the Nineveh Lens. It was discovered in northern of
Iraq, in deposits dated to 722 B.C. It was used as a lens to concentrate the sunlight (Kett
1958). At the beginning of the 1st century Ptolemy studied different properties of light.
He investigated the refraction of light for a series of materials with high transparency in
his book “Optics” (Ptolemy and Smith 1996). In 1021 Ibn Al-Haytham Alhazen
provided for the first time an explanation of twilight (Sabra 1989), and observed light
through a pinhole camera. As early as 1310 Dietrich von Freiberg gave the positions of
the primary and secondary colors of the rainbow. Most of light’s properties had been
highlighted and formulated after Willebrord van Roijen Snell stated his law of light
refraction, and Issac Newton delivered his Hypothesis of Light theory during the 1600s.
Chapter Two
8
The experiments of William Herschel in 1800 led to the discovery of infra-red radiation
(Herschel 1800a). On 13 March 1793 he accidentally discovered the planet Uranus
(Maldague and Moore 2001). During his work as an astronomer he tried to use a prism
to keep his eyes undamaged when examining the sun. This led him to discover infra-red
rays. He employed a glass prism to scatter the sunlight onto a number of mercury
thermometers. During his examination of the separated light he found that just beyond
the red colour, where there was no visible light, the thermometer recorded the highest
temperature. He concluded that there are invisible rays beyond the red colour of visible
light. He named these rays “the solar and the terrestrial rays that occasion heat”
(Herschel 1800b). Herschel demonstrated that the spread of these rays depends on the
medium or object properties. By using the newly invented thermocouple, Ampere stated
that both infra-red and visible light were the same phenomenon and had the same
optical characteristics (Hindle 2008).
There were several achievements during the nineteenth century after Herschel’s
experiments. The first infra-red image was created in 1840 by Herschel's son Sir John
Frederick William Herschel. He uses an evaporograph (Maldague and Moore 2001). In
1900 Max Planck formulated his law of radiation. Major improvements in the infra-red
industry sector were made during and after World War II. Most of these patents were
for military purposes, such as the detection of soldiers, ships and torpedoes (Maldague
and Moore 2001). Later, many innovations were applied in the medical, scientific and
environmental industries. Infra-red thermal imagers begun to supplied commercially in
the 1960s, and a giant leap in infra-red detection capability took place in the 1980s
when array detectors (a combination of several single detectors) were adopted and
integrated with microprocessors. This improvement significantly enhanced the
efficiency of infra-red capture and helped the upgrading of infra-red detection devices
with the ability to capture images more swiftly. The digital technology revolution has
significantly facilitated IRT. Advances in the control and calibration of infra-red devices
using computers and the in the capacity to manage and acquire infra-red data and
analyze infra-red images have promoted the use of IRT NDT.
Literature review
9
2.2.2 Fundamentals of infra-red radiation
Heat can transfer in a medium or between bodies by conduction, convection, radiation
or a combination of these. Conduction is the spread of heat energy whenever a
temperature difference exists between two solid materials in contact or among parts of a
material. Convection involves the mass movement of a fluid or gas molecules over a
distance. Radiation occurs when a material emits energy over a distance through a
material, fluid, gas or vacuum. The transfer of energy in electromagnetic wave form is
called radiative heat transfer (Bejan and Allan 2003).
All surfaces above absolute zero temperature emit electromagnetic radiation through the
movement of atoms. This radiation occurs when an electric charge accelerates. The
object's temperature and the surface conditions will influence the radiation spectrum and
intensity. The energy of the atomic particles will increase when the object’s surface is
heated. The atomic particles agitate thermally in a chaotic manner, which generates a
form of radiant electromagnetic energy known as infra-red radiation. The frequencies of
waves produced from this infra-red energy are located between the microwave and
visible light on the electromagnetic spectrum, as shown in Figure 2.1. The wavelengths
will be beyond the red visible light, from around 700 nm to 1 mm where the microwave
range begins. This infra-red range is subdivided into further regions. The International
Commission on Illumination (CIE) places the infra-red in three bands (Byrnes 2009):
IR-A (from 700nm to 1.4µm), IR-B (from 1.4µm to 3µm) and IR-C (from 3µm to
1mm). An international standard for the boundaries of the infra-red sub-regions is not
yet available. Different infra-red band classifications are available in astronomy that set
the IR regions in three bands (near, mid and far infra-red) with wavelengths from 700
nm to 350 µm (NASA ipac 2007). Another subdivision considers the infra-red
detector’s sensor response. However, the most accepted and common subdivision is
illustrated in Figure 2.1 and the infra-red bandwidth is distributed as follows:
Near-infrared (NIR): its wavelength varies from 750 nm to 1.4 μm
Short-wavelength infrared (SWIR): wavelength ranges from 1.4 μm to 3 μm
Mid-wavelength infrared (MWIR): 3–8 μm is the spectrum range of this IR
band. Intermediate infrared (IIR) is another name for this wavelength.
Chapter Two
10
Long-wavelength infrared (LWIR) wavelength is between 8 and 15 μm. Most
passive thermography works in this region of the infra-red rays.
Far infrared (FIR) band has wavelengths beyond 15 μm up to 1,000 μm.
Figure 2.1 Infra-red wavelength ranges
When the object is subjected to infra-red radiation, the part absorbed by the object will
convert to heat. Radiation intensity as a type of heat transfer is measured in watts per
square metre (W.m-2), and as mentioned above, depends on the temperature and the
object’s surface conditions and nature. Usually infra-red radiation has a constant
wavelength at a specific temperature range. At higher temperatures, the wavelengths of
the radiation intensity are shorter, while the band wavelengths become longer when
temperatures are low. All materials change their internal energy continuously at a
molecular level by emitting and absorbing photons and electromagnetic waves. Most the
visible light behaviours are applicable to infra-red radiation. Thermal radiation is
emitted in all directions: it reflects, moves in straight lines, bends, interferes, is
absorbed, and travels in an ideal vacuum at the same speed as visible light (≈
1,079,253,000 km/hour) (Maldague and Moore 2001). The absorbed part of thermal
radiation will transfer to heat and increase the surface temperature of the material.
Literature review
11
Energy radiation is exchanged continuously between surfaces and objects, even when
the surfaces and bodies are in temperature equilibrium.
2.2.3 Fundamentals of IRT NDT
IRT is a method which reads the emitted electromagnetic radiation from the object’s
surface or medium of interest. There are two modes of measuring the temperature:
contact and non-contact. The contact mode is commonly by means of sensors attached
to the object of interest. These sensors measure the temperature as electrical signals.
Thermocouples, thermistors, integrated circuit sensors, and resistance temperature
detectors are the most common transducers (Maldague and Moore 2001). Temperature
can be measured in a non-contact mode by using different kinds of sensors and detectors
including photonic detectors, quantum detectors, pyroelectric detectors, and infra-red
imaging devices. Most of these measurements are based on fundamental principles of
thermodynamic relationships. Infra-red imaging “thermometers” are the most widely-
used form of non-contact temperature measurement. The wide temperature ranges that
these imagers cover make them appropriate for use in many different applications.
However, the variety in these devices is based on the test environments and targets for
which the infra-red imager is designed. For that reason, it is essential for persons who
conduct IRT testing to have a very good understanding of the thermal test environment.
A testing program must take into consideration many parameters and factors before
infra-red testing can be conducted. However, the major task for the thermographer will
be interpretations after the collection of the desired results. Infra-red thermal detector
measurements are exposed to different kinds of faults including surface emissivity,
reflections and fluorescence. Special precautions need to be taken to reduce errors in
thermograms (thermal images) to minimize misreading of results.
2.2.4 Theoretical principles
2.2.4.1 Planck’s law
The thermal radiation that leaves a material’s surface is called the emissive power, and
it is measured per area of that surface. The total sum of emitted energy over the entire
spectrum is called total emissive power (E). The energy power at a given frequency is
Chapter Two
12
called spectral emissive power (Eλ). Many factors affect the total emissive power
including material surface properties, surface original temperature, and material type.
The ideal material that does not reflect any radiation is called blackbody (ASTM E 1965
2003). The surface of this blackbody is a perfect absorber which can absorb all radiation
in any wavelength and direction. Apart from being a perfect absorber, the blackbody is
also a perfect emitter. At a particular temperature and wavelength, no surface can emit
energy greater than a blackbody. In his law, Max Planck quantified the blackbody’s
emissive power, as shown in Equation 2.1 (Planck and Masius 1914; Bejan and Allan
2003):
( )
[ (
) ]
Equation 2.1
where,
Eλb = spectral emissive power for a blackbody (W/m3),
h = Planck’s constant (6.626×10-34 J.s)
co = The speed of light in vacuum
λ = wavelength (m),
T = absolute temperature (K),
n = constant refractive index (equal 1 in vacuum),
k = Boltzmann’s constant (1.3806 × 10-23 J/K).
By simplifying Equation 2.1:
[ (
) ] [ ]
Equation 2.2
where,
C1 = the first radiation constant (2πhco2) = 3.7419 × 10-16 (W/m2),
C2 = the second radiation constant (hco/k) = 0.01438769 (m.K).
Literature review
13
Figure 2.2 shows Equation 2.2 for a range of different wavelengths and temperatures. It
reveals that radiation energy is a function of the wavelength for a specific temperature.
In the figure the wavelength is in µm and the blackbody emissive power has been
plotted in W/m2.µm.
Figure 2.2 Spectral blackbody emissive power (ASM 1992)
As shown in this figure, the wavelength that corresponds to the maximum emissive
power is related to the absolute temperature. The maximum of Equation 2.2 is known as
Wien’s displacement law:
C3 = λmax. T = 0.028978 m.K Equation 2.3
where C3 is known as the third radiation constant.
Because the value of the [exp (C2/λT)] in Equation (2.2) is significantly greater than 1 in
infra-red thermography applications, Equation 2.2 can be re-introduced as:
Chapter Two
14
( )
[ (
)] [ ]
Equation 2.4
Equation 2.4 is known as Wien’s law. It provides approximate values for the original
equation.
By integrating Equation 2.2 over the entire spectrum length the total emissive power for
a blackbody can be shown as:
∫
[ (
)] [ ]
Equation 2.5
Resulting in
Equation 2.6
where σ is the Stefan-Boltzmann constant and has the value of ( 5.67051× 10-8
W/(m2.K4)). Equation 2.6 is known as Stefan-Boltzmann’s law and it calculates the
radiation emitted from an ideal blackbody surface.
2.2.4.2 Emissivity
Emissivity (ε) is a variable defined as the ratio of the electromagnetic radiation emitted
from a surface to the radiation that would be emitted from an ideal blackbody at the
same temperature. Emissivity of all materials is measured on a scale between 0 and 1.
Blackbody has an emissivity of 1. All other materials have absorptive values of less
than 1. The spectral distribution and the emissive power value are the factors that make
the difference in the spectral emissive power between a real material and a blackbody.
Figure 2.4 shows the effect of emissivity on radiation intensity. The figure shows that at
all temperatures and wavelengths, grey bodies have similar emissivity distributions but
less emissivity than blackbody. All other materials that have different distributions (not
similar to the grey body pattern) over the wavelength are defined as spectral radiators.
However, many materials exhibit approximately grey body behaviour.
Literature review
15
Figure 2.3 Emissivity effect on radiation from surface of emissivity ε with hypothetical
intensity (Maldague and Moore 2001)
The incoming radiation on a surface might depart in a specular or diffuse manner.
Figure 2.5 illustrates the reflection of both manners. These two manners apply for both
emittance and reflectance radiations. The radiation may also be reflected in a manner
between them, as shown in Figure 2.5b. Diffuse emittance has no favored directions,
and the angle of the incoming ray (α) in Figure 2.5c is assumed to not affect the
outgoing direction (Lienhard 1981). The radiation departs blackbodies diffusely.
Figure 2.4 Specular and diffuse radiation reflection [Reproduced from Lienhard (1981)]
Radiation emitted in all directions from material surface is known as hemispherical
spectral emissivity. The hemispherical spectral emissivity of a grey body is defined as:
Chapter Two
16
( ) ( )
( ) Equation 2.7
where,
ε (T,λ) = spectral emissivity,
Eλ = spectral emissive power for grey body (W/m3).
The total hemispherical emissivity of a real material is defined as the ratio of the total
emissivity on the material surface to that of an ideal blackbody at identical temperature,
( ) ( )
( ) Equation 2.8
where,
ε (T) = total emissivity at specific temperature,
E (T) = total emissive power for grey body (W/m3) at specific temperature.
From Equations 2.7 and 2.8 it can be noted that emissivity is a function of the
wavelength and temperature. However, emissivity at the same time is a function of the
material surface properties. Rough surfaces have higher emissivities than smooth
surfaces. These smooth materials are more difficult to test thermally than materials with
higher emissivities (Maldague and Moore 2001). Coated surfaces have different
emissivities depending on the coating properties. From Equation 2.6 and 2.8 the total
emittance of grey body at a particular temperature can be measured as shown in
Equation 2.9.
( ) Equation 2.9
The measurement of infra-red thermal radiation is influenced by many different
parameters. The material absorptivity, emissivity and reflection properties influence the
thermal reading continuously, even if the material is in a condition of thermal
Literature review
17
equilibrium. In addition, several features affect thermal detector performance and cause
errors in the thermal reading and results, including noise and atmosphere conditions.
For any thermal test, all these factors lead to thermal reading errors and need to be taken
into consideration by the thermographer during the IR test and in the analysis of the
results.
2.2.5 Infra-red thermography techniques
Many techniques are applied in IRT NDT; however, the most generally recognized
approaches that are used in different applications are passive and active techniques. The
test used depends in both techniques on the difference in temperature between the target
object Ttarget and its ambient. However, in the active approach the test is conducted with
an external heat source applied to the investigated surface. In contrast, a thermal steady-
state procedure is usually required in the passive technique.
IRT testing involves temperature and heat flow measurement to detect and calculate
defects or failures within materials. To interpret the temperature level and temperature
changes on a test specimen, a fundamental knowledge of the heat transfer pattern and
thermal properties of the test material is essential.
IRT imaging is the non-contact, non-destructive mapping of thermal behaviour on the
target test surface. Thermal imaging equipment is available in numerous conformations
and with varying degrees of complexity (Maldague and Moore 2001). The maps
recorded by thermal imaging equipment are usually termed thermograms. The
thermographer should have expertise in heat flow and infra-red radiation and must be
familiar with the thermal imaging equipment’s capability and functioning in order to
acquire the best thermal image and to enhance the analysis of the thermograms.
2.2.6 Passive techniques
In passive thermography materials are tested naturally, without applying any external
heat flow or using external excitation systems. No heating or cooling is applied to the
material. The approach depends on the natural difference in the temperature pattern
Chapter Two
18
between the material and the surrounding ambient. The evaluation of a material
according to its temperature distribution depends on its ideal temperature value, the rate
of temperature change, and the actual difference between the material and the ambient
or a reference.
Astronomy is the field of science where IRT started, and infra-red technology has
enhanced astronomical observations and encouraged qualitative assessment of telescope
performance. Figure 2.5 shows a comparison of the universal galaxy M51 imaged with
the Spitzer Space Telescope and an image of the same galaxy taken by the Herschel
Space Observatory which was launched in May 2009 with a state-of-the-art infra-red
imager. The Herschel Space Observatory has the ability to provide three colour far-
infra-red images of different wavelengths. The Herschel infra-red images reveal
structures that cannot be discerned in the Spitzer image (European Space Agency
2011a).
Figure 2.5 M51 imaged with the Spitzer Space Telescope and an image of the same galaxy taken by the Herschel Space Observatory (European Space Agency 2011b)
Passive thermographic testing is generally used to monitor the production and different
stages of manufacturing where non-standard temperatures may indicate potential errors
or problems. Different materials and applications have been tested using this approach
such as metal fabrication and steel quality, glass production and bottle forming (Wilson
1991), and welding quality control (Nagarajan, Banerjee, Chen and Chin 1992;
Nagarajan, Wikle and Chin 1992) (i.e. tracking of seams and checking their quality).
Literature review
19
Figure 2.6 shows the IR thermogram used as a tool to evaluate the quality of the
welding process in a railway.
Figure 2.6 Thermogram of railway weld (Khauv 2011)
IRT recently been applied to micro-scale industries. IR detectors are used in electronics
manufacturing product lines to monitor if there are any abnormalities within the
product, as shown in Figure 2.7. The IR images in this figure were captured with a
SC7600-M FLIR infra-red imager with G3 lenses that have zoom capability up to 5µm
to detect and evaluate microchip electronic connections.
Figure 2.7 Microchip connection checking using IRT (Khauv 2011)
Passive IRT is also used in the evaluation and rehabilitation of historical buildings. This
NDT is commonly used to investigate and evaluate the whole building structure and the
Chapter Two
20
structure beneath the plaster surfaces. Figure 2.8 shows a thermal image captured by a
FLIR team to diagnose problems with the Basilica of the Sacred Heart in Paris.
Figure 2.8 IR image of the Sacred Heart building in Paris
Passive techniques can also be used to evaluate insulation systems in buildings (Lyberg
and Ljungberg 1991) and monitor the maintenance of these buildings. For example,
water leaks as a serious problem that IRT can detect. The infrared detectors can
recognize the presence of water easily due to the differences in thermal properties
between water and building materials. Problems including water leaking into the
building through windows, sliding doors at balconies or even roofs can be monitored
using IR testing. Figure 2.9 shown as the IR image of water leaks in the ceiling of a
building in Chicago using a FLIR T300 infrared imager. The early detection of these
faults can minimize the repair process and cut costs.
Literature review
21
Figure 2.9 IR diagnosis of water leaks in ceiling (Chicago Infrared Thermal Imaging
Inc. 2011)
Passive IRT has also been used to investigate furnaces and heating structures to
diagnose the causes of heat losses (Ljungberg 1997). General thermal building
performance can also be investigated by this technique. Heat losses can be formalized
and estimated by adopting passive IRT techniques (Vavilov, Anoshkin, Kourtenkov,
Trofimov and Kauppinen 1997). Gas emission tracking and detection are usually
carried out in a passive testing scenario (Ljungberg and Jonsson 2002b). Figure 2.10
shows the tracking of emissions by thermal images in field. Although there are some
gases that cannot be distinguished by IRT imaging, the passive approach can be
supported with heated or cooled backgrounds to solve the problem of gas invisibility in
the thermal images. This application provides a valuable solution for the monitoring of
gas leaks in gas pipe- lines.
Figure 2.10 Gas leak thermography test from a pipe buried at 80 cm depth (Ljungberg
and Jonsson 2002b)
Chapter Two
22
More recently, passive thermographic techniques have been used to investigate and
calibrate greenhouse heating systems and to indicate any abnormality during plant
growth, as shown in Figure 2.11 (Ljungberg and Jonsson 2002a).
Figure 2.11 Infra-red sensor for control of the leaf temperature, Thermograms indicate
deficiencies in the gas-IR heating system (Ljungberg and Jonsson 2002a)
Applications of the passive approach are numerous. It has been employed in medicine
in the last two decades, and it has become a very efficient tool for medical and
veterinary applications. Thermal imaging is an effective means to detect anomalies and
abnormalities that cannot be identified with the naked eye, or even X-rays and
ultrasound in some circumstances. Thermographic devices allow the early diagnosis of
illnesses related to blood circulation problems, and the identification of problems
connected with rheumatology, neurology, orthopedics, and sinusitis. It has been shown
to be very efficient in sports medicine for the diagnosis of neuromusculoskeletal
damage (Meditherm Inc. 2009). Figure 2.12 shows how thermal imaging can assist with
the location of health problems. Because each part of the body has a particular
thermographic pattern, the observation of differential heat patterns helps oncologists to
monitor breast health and to diagnose breast cancer in the early stages (Head, Lipari,
Wang and Elliott 1997; Lipari and Head 1997). Figure 2.12-c shows a thermographic
cancer inspection of a woman’s breast. This approach is considered risk-free compared
with other tumor detection methods such as mammography and X-rays.
Literature review
23
(a) (b)
(c)
Figure 2.12 Health problems diagnosed by IR thermal imaging, (a) Diagnosis of jaw problem (Meditherm Inc. 2011a) ; (b) Football player with stress fracture (Meditherm
Inc. 2011b) ; and (c) Breast thermography diagnosis (Meditherm Inc. 2011c)
IRT detection helps art historians to check pentimento and painting alterations in
masterpieces beneath the surface of the painting. This process can help to distinguish
originals from copies and to study the previous trials of the drawing or the artist’s
guidelines. Figure 2.13 reveals the under-drawing infra-red image of the DaVinci
masterpiece “The Virgin of the Rocks”.
Chapter Two
24
Figure 2.13 The Virgin of the Rocks under-drawing infrared image
In meteorology, weather satellites equipped with infra-red technology scanning in the
range of 10.3 to 12.5 µm facilitate the calculation of water and land temperature, and
cloud monitoring. The Australian region infrared satellite image issued by the
Australian Bureau of Meteorology at 11:37 am EST Sunday on 28 August 2011 is
shown in Figure 2.14. Infra-red satellite images are used in weather warnings and
predictions. For example, people can receive advance warnings about possibly severe
hurricanes. Figure 2.15 shows the IR satellite image of hurricane Irene at 12 pm on
Sunday, 28 of August 2011 before hitting New York City. Such information helped the
New York City government to give the order for the evacuation off residents well
before the hurricane’s arrival.
Literature review
25
Figure 2.14 Australian region infrared satellite image (Australian Bureau of
Meteorology 2011)
Figure 2.15 Hurricane Irene arrives in NYC (The City of New York 2011)
Passive IRT techniques are also used in biology, for the detection of forest fires, the
monitoring of road traffic and for military purposes (Maldague 1993), as shown in
Figures 2.15 to 2.19.
Chapter Two
26
Figure 2.16 Infra-red biological application: Brazilian free-tailed bat (Center for
Ecology and Conservation Biology-Boston University 2011)
Figure 2.17 Aerial fire IR mapping (Khauv 2011)
Figure 2.18 Load traffic IR monitoring (Khauv 2011)
Literature review
27
Figure 2.19 US Navy IR imagery taken from a U.S. NavyP-3C Orion maritime patrol
aircraft, assisting in search and rescue operations for survivors of the Egyptian ferry Al Salam Boccaccio 98 in the Red Sea (U.S. Navy 2006)
Figure 2.20 High speed IR detector image for machine gun testing (Khauv 2011)
From all the above applications and uses, the passive approach is recommended in the
industry sector because it provides enhanced quality during the production process. The
use of this infra-red technology in civil engineering applications will reduce expenditure
on rehabilitation and repair operations and minimize the amount of energy consumed.
Chapter Two
28
In addition it has the potential to be used for other applications because of its accuracy
and speed.
2.2.7 Active technique
The active IRT technique generally depends on the fundamental principle that heat
transfer in material is changed by the presence of material discontinues or the
occurrence of debond and cracks. Alterations in heat transfer appear as different
temperature patterns on the surface of material subjected to external heat flux. Because
of the differences in surface temperatures, areas with underlying defects will appear
with different temperatures (hot or cold spots) with respect to the surroundings area.
Figure 2.21 illustrates the mechanism used to localize hot spots. If a constant heat flux
is applied to a homogenous surface that has no defects, the increase in the surface
temperature should be uniform in distribution. Therefore, if the surface has any kind of
anomaly or defect, such as delamination, cracks, and voids, it will affect heat flow
through that material (Malhotra and Carino 2004).
Figure 2.21 Hot spot localization
To investigate materials using this technique, an external heat source is required to be
integrated as an excitation system during thermal imaging. This approach is one of the
most popular thermal stimulation methods in infra-red thermal techniques. The term
“active thermography” is used as an encompassing term for all non-destructive
Subsurface defect
Subsurface defect
Hot spot
External Applied Heat
Literature review
29
evaluations carried out with thermal cameras and external excitation heat sources (Shull
2002). However, the three major active thermography techniques are:
Pulsed thermography,
Step heating thermography,
Lockin thermography.
2.2.7.1 Pulsed thermography technique (PTT)
The pulse thermography active procedure is based on exposing the material surface to a
short temperature simulation and recording the temperature pattern on the surface of the
heated material as thermal images. After short thermal injection the temperature on the
material surface alters quickly because of the material’s diffusivity properties and
radiation. The alteration in the rate of diffusion due to the presence of discontinuities
and defects will make these areas appear with different temperatures with regard to the
defect-free neighbouring areas observed with an IR thermographic imager. The areas of
discontinuities will appear with different temperatures relative to the non-defected areas
at the surface in the thermal image. Due to the test’s high speed and accuracy, infra-red
PTT is a very common method in the active approach (Vavilov and Maldague 1994).
There are several different active IR PTT test configurations and setups. Figure 2.22
shows the active test set-up by line, point and surface. Each type of configuration has its
advantages and disadvantages. The advantages of line pulse infra-red thermography for
instance include the homogeneity of the thermal simulation on the investigated area, and
continuous control over the heat transit. Nevertheless, this kind of test cannot be
employed on the entire surface. The line heating sources involve flashing lamps, laser
beams, or even air jets. This set-up is recommended for the inspection of cracks parallel
to the heating line (Lesniak 1995). Line pulse configurations are illustrated in Figure
2.22a. The point infra-red test involves heating the inspected point by a spot heat light
beam. This type of set-up is suggested for the IRT investigation of limited localized
areas. Like the line setup this configuration is not suitable for the inspection of entire
surfaces. Figure 2.22b shows the point test set-up. Figure 2.22c shows pulse IRT by
surface inspection. Although various heating sources can be used for this configuration,
Chapter Two
30
lamps and scanning lasers are the most common. The capability to test the entire surface
is the most important feature of this set-up. However, the homogeneity of the external
heating distribution is still a challenge during the thermogram analysis of this
configuration.
Figure 2.22 IR pulsed thermography test configurations, (a) line method, (b) point
method and (c) surface method
Cold thermal sources can be used if the material that needs thermal investigation is
already in a hot ambient. Sources like water line jets, ice or cold air jets follow the same
fundamental principles. The thermographic test is based on the variation between the
test material and the ambient, whatever that difference is.
There are two basic methods of observation for any infra-red active technique:
reflection (one-sided) or transmission (two-sided). Figure 2.23 shows both methods in
reflection and transmission configurations in the defect detection phase. In the reflection
method the excitation sources and the thermal detector are positioned on the same side
of the inspected target. The defects will appear as a hot spot, as shown in Figure 2.23c.
The thermal image captured by this test method offers higher resolution than the
transmission method. However, the reflection method’s ability to detect deep defects is
very low. In contrast, the transmission method reveals defects as cold spots in the
thermograms, as shown in Figure 2.23d. Thermograms obtained by this method provide
good information regarding the detection of deep defects while their resolution is
usually low. However, the signal observed in both methods will have the same
Field of view and observation area
Infrared Detector
Heating Source
Specimen under investigation
`
Processing
Line Heating Source
Infrared Detector
Specimen under investigation
Direction of scanning
`
Processing
`Processing
Infrared Detector
Spot Heating Source
Specimen under investigation
(a) (b) (c)
Literature review
31
behaviour. Figure 2.24 illustrates heat sources and the infra-red recorded wave shapes in
the PTT approach.
(a) (b)
(c) (d)
Figure 2.23 Schematic of (a) Reflection observation method (One-sided); (b) Transmission observation method (Two-sided); (c) Reflection observation and hot spot
image; (d) Transmission observation and cold spot image
Figure 2.24 Pulsed heat and IR recorded waves in pulsed thermography approach
Chapter Two
32
The active pulsed thermography technique is very widespread because inspection
requires short capture times (Vavilov and Maldague 1994), although the resolution
limitation for deep reading is its main drawback.
2.2.7.2 Step heating thermography
The step heating thermography technique involves monitoring the target surface for the
period of application of pulsed heating. This approach usually does not require high
heat. Temperature calibration as a function of time is one of the major features of this
approach (Aamodt, Spicer and Murphy 1990). The blur in the thermal image can be
reduced by using step heating thermography, which makes the detection of deep
material defects and discontinuities easier (Osiander and Spicer 1998). This technique is
also used to determine material thermal properties such as conductivity. The possibility
of early thermal calibration is the main feature of this method in respect to the pulsed
thermography technique. However, the decision of whether to test material using a
pulsed or step heating thermography approach usually depends on the accessibility of
heating sources and the capability to control and generate heat waves in steeply manner.
2.2.7.3 Lockin thermography technique (LTT)
The basic idea of the lockin thermography active technique is to generate thermal waves
within the tested material and monitor the surface closely (Busse 1994; Gerhard and
Busse 2006). This approach was derived from photothermal radiometry (Kanstad and
Nordal 1979). Thermal waves can be generated externally by optical periodical
illumination, for instance, by laser beams and halogen lamps, or internally by subjecting
the tested material to modulated acoustic waves. The lockin active technique allows
better energy control over the inspected surface. However, this approach normally takes
more time than pulsed thermography because the experiment must be conducted for
each depth of the specimen (Clemente Ibarra-Castanedo, Stéphane Guibert, Jean-Marc
Piau, Xavier P. V. Maldague and Abdelhakim Bendada 2007). This can be performed
by examining the material over a wide range of different frequencies. This active
technique has applications in coating thickness measurement, and sub-surface defect,
anomaly and discontinuity detection (Rantala 1996). The general test configuration of
Literature review
33
the lockin thermography active technique is illustrated in Figure 2.25. The introduction
of different frequencies in this approach leads to better analysis with respect to depth
and noise (Gerhard and Busse 2006). A laser beam is used to introduce modulated
thermal waves into the inspected material. Modulated halogen lamps can take the place
of the laser beam to provide low frequency thermal waves simultaneously to the entire
investigated area. At the same time as the thermal wave injection, an infra-red detector
monitors and captures the thermal wave’s response and decomposes it by a lockin
amplifier to extract the amplitude and the modulation phase. Figure 2.26 shows the
lockin setup with both laser beam and lamp. Sinusoidal thermal injected wave and the
infra-red recorded wave shapes produced by the lamp are shown in Figure 2.25.
Figure 2.25 Sinusoidal input wave and IR recorded wave in LTT approach
Figure 2.26 Basic locking thermography set-up, laser beam and lamp (Gerhard and
Busse 2006)
Chapter Two
34
Generation of thermal waves can be introduced internally by the simulation of elastic
modulated waves. The mechanical energy will change to heat due to the collision of the
internal free surfaces with defects, small discontinuities or even micro-cracks (Clemente
Ibarra-Castanedo, Stéphane Guibert, Jean-Marc Piau, Xavier P. V. Maldague and
Abdelhakim Bendada 2007). Ultrasonic waves are used because of their efficient ability
to transform into heat, and these waves will not increase the stress on the mechanical
discontinuities (Maldague and Moore 2001). The temperature surface map can be
provided by using an infra-red thermal camera or by coating the inspected structure with
temperature-sensitive liquid crystals (Broutman, Kobayashi and Carrillo 1969).
However, infra-red cameras are more flexible because there is no need for surface
preparation as in the liquid crystal system. Figure 2.27 illustrates the lockin
thermography technique with ultrasonically-modulated internal simulation. This
technique is applicable for revealing cracks in metals, detecting damaged areas in
laminates, and identifying corrosion in metals (Salerno, Wu and Busse 1997). A
comparison of optical and ultrasonic lockin thermography waves is shown in Figure
2.28. The ultrasonic scenario shows a potential capability to detect deeper defects with
respect to optical lockin. This is because the thermal waves generated in this scenario
have to transmit only half the distance (between the discontinuity and the surface) than
with optical means.
Figure 2.27 LTT set-up with ultrasonically modulated internal simulation
Processing
Ultrasonic transducer
Controlling
Amplifier
Infrared Camera
Literature review
35
Figure 2.28 Two means of generation of thermal waves in LTT
It is important to point out that more than one active technique can be used in the same
thermography test. For instance, one technique can be employed for general scanning
and once the discontinuity regions are detected, another scenario can be adopted for
deep inspection. Moreover, these techniques can be linked. Pulsed phase thermography
(PPT), for example, is a technique which links the pulsed and lockin thermography
active approaches. In the PPT technique a special thermal wave with specific frequency
is generated to target a specific material’s depth which will make a frequency-to-
frequency analysis similar to the lockin analysis based on pulsed thermography data.
This approach was introduced (Maldague and Marinetti 1996) to merge the advantages
of both the pulsed and lockin thermography techniques.
In summary, a collection of active infra-red thermography techniques to detect
subsurface anomalies and discontinuities is available for a wide variety of applications.
The nomination of the most adequate procedure depends on the particular application
and the availability of experienced staff and experimental resources.
2.2.8 Noise in IRT
Noise can be defined as unwanted signals that arise in infra-red thermography reading
(Hudson 1969). Noise can be categorized into two main kinds: fixed pattern noise and
random noise. Fixed pattern noise refers to noise that has individual patterns, whereas
Subsurface defect
Subsurface defect
Thermal wave
Ultrasonic wave
Mat
eria
l sur
face
Optical mean to generate
thermal waves for Lockin
thermography
Ultrasonic source to generate thermal waves for Lockin
thermography
Chapter Two
36
random noise has independent signal values to the following or preceding values in
terms of position or time which do not follow any determined pattern. Noise can be
defined according to its probability density function, which describes how often a
particular value of the random variant is detected (Maldague and Moore 2001). A
histogram noise population is usually calculated to predict the probability density
function. As the histograms usually show Gaussian distribution, Gaussian distribution is
often assumed in noise processes in infra-red thermography analysis. However, there is
still a chance of non-Gaussian noise occurring. Different filters are used to reduce noise
effects. The most common filters employed in noise processing are Gaussian,
neighbourhood averaging, Butterworth, median and harmonic filters.
To identify the noise content shown in infra-red images it is necessary to analyze two
images at pixel level (Haddon 1988). If the two thermal images have the same scene
under the same condition then noise will appear as the differences between the two
images. The ratio of signal power to noise power is defined as the signal-to-noise ratio
(average power image / average power noise), which can be evaluated from the
following equation (Maldague and Moore 2001):
∑ ∑
Equation 2.10
where,
√∑ ∑ ( )
Equation 2.11
| | Equation 2.12
i , j = the x and y positions in an image of N ×M pixels,
µ = the mean of the noise distribution.
Literature review
37
2.2.9 Errors in IRT
Radiation heat flow is a complex process. Any radiation measurement is subject to a
number of possible sources of error that can mislead image interpretation. These
potential errors are the result of radiation transmission across a medium that splits the
infra-red detector and the tested material surface. In that medium a part of the radiant
energy may be absorbed or change its direction. For that reason it is essential to have
knowledge of the properties of the medium as well as the surface properties of the
material. The errors that can affect infra-red measurement can be categorized into three
main groups (Childs 2001):
Process characterization errors involving: surface emissivity, reflections, and
fluorescence.
Transmission path errors involving: absorption, scattering, size of object effects and
vignetting.
Signal processing errors.
Emissivity is already identified for most materials; however, attention should be given
to the surface preparation and finishing of the material. Surface conditions such as
oxidization or polishing alter the emissivity value of the material. Several techniques
exist to increase material emissivity in terms of coating and surface modification.
Different techniques are employed to overcome low emissivity (Maldague and Moore
2001).
Recognizing and avoiding reflections from the background atmosphere is essential in
infra-red thermography recording to minimize errors. Background reflections are
defined as all undesired reflections from external sources that reflect on the surface of
the investigated material. Figure 2.29 shows the background reflections error. During
infra-red thermal capture, the infra-red detector is usually not able to distinguish
between the thermal radiations emitted from the heated material’s surface and the
background radiations that reflect on the same surface. The probability of occurrence of
background radiation reflection is increased for low emissivity materials and if the test
surface is not a plane (Maldague and Moore 2001).
Chapter Two
38
Although background reflections are commonly due to external sources hotter than the
target, reflective error from colder sources should also be taken into consideration. On
the other hand, background radiation from external sources will be hardly noticeable in
the thermal images if the medium of the test is heated well above these external sources
(Childs 2001). The elimination of these background reflections depends on their nature;
if it is point source reflection, the theromgrapher can relocate the infra-red detector until
its best position is identified. The thermographer can also block the line of sight
between the source and the surface. For significant extended source background
reflections, one possible solution to minimize undesired reflection is by shielding the
infra-red detector from these external radiation sources. Figure 2.30 illustrates the use of
a shield as a solution to minimize the background radiation reflections.
Figure 2.29 Background reflection [Reproduced from Childs (2001)]
Figure 2.30 Shielding the test to minimize the significant background reflection
[Reproduce from Childs (2001)]
Material surface
Reflected radiation
Infrared
Detector
Radiation emitted from material surface
Target
In this position the thermal detector is shielded from the additional source of
radiation by the target
Additional source of background radiation
In this position the thermal detector measures both the emitted and the
reflected radiation on the target
Literature review
39
Transmission path errors take place while the radiation is passing the medium between
the infra-red detector and the target investigated surface. Atmospheric effects on infra-
red measurements are complex due to the presence of various gases in the air (which is
the general medium between detector lenses and tested objects), and the differences in
concentration of these gases. Infra-red transmitted energy that crosses the air medium
may be subject to absorption or scattering at various levels which leads to errors in the
infra-red thermal reading. The nature of the medium will determine the number and
severity of these errors. The transparency of air is not 100 percent. All rays and
radiation crossing air will have some part of the transmitted radiation that will be
absorbed. The majority of the absorption in air is due to the presence of water vapour
(H2O), carbon dioxide (CO2) and ozone (O3). However, transmittance is heavily
dependent on radiation wavelength, reading distance, and meteorological conditions
(Maldague and Moore 2001). Figure 2.31 shows the transmittance percentage of these
gases with respect to wavelength.
Figure 2.31 The main gases responsible for infra-red radiation absorption. Atmospheric
transmittance (Maldague and Moore 2001)
From Figure 2.31 it is clear that the transmission patterns flow in a special manner
dependent on the application conditions. For that reason and to maximize the
transmittance percentage, each infra-red detector has specific band infra-red
wavelengths with which it can work, as shown in Figure 2.32. The wavelength range of
these devices is usually related to the application. For most infra-red investigations in
Chapter Two
40
civil engineering, the efficient infra-red spectrum ranges are in the windows of LWIR
and MWIR.
Figure 2.32 IR windows in the spectrum
As the solid particles suspended in the medium, such as dust and smoke have grey body
performance, it is essential for the thermographer to avoid dusty environments (Childs
2001). In addition, these solid particles accumulate on the infra-red detector lenses and
block the radiation or even cause damage to the device. Every infra-red device has its
usage and operational requirements in terms of the humidity, temperature and
environmental conditions in which it can work.
Vignetting is defined as obstruction of the field of view (Childs 2001). The field of view
is the image size with respect to the detector lens scanning angle. It is important to
remove any body that can cause a reduction in the amount of radiation recorded by the
infra-red device.
The last source of error is the probability of error during the recording of the thermal
data. Good quality control throughout IRT testing plays a key role in reducing this kind
of error. To minimize errors of this type, it is recommended to perform important IRT
NDTs twice. Further concerns can be reduced by conducting each test individually by
different thermographers.
Literature review
41
2.2.10 Qualitative and quantitative thermography
Infra-red detector performance is the heart of any infra-red NDT. Its capability in terms
of qualitative or quantitative measurement is the most essential feature of any infra-red
detector. Qualitative thermography is a process by which thermal images exhibit an
infra-red radiation map of the target surface, uncorrected for target, instrument and
media characteristics (Maldague and Moore 2001). Therefore, qualitative infra-red
detectors cannot provide thermograms with accurate temperatures. However, qualitative
detectors can be used for many applications when temperature accuracy is not crucial,
and the development of qualitative detectors means that they are of modest cost
compared with quantitative detectors. In contrast, quantitative infra-red images show the
distribution of the infra-red radiance on the surfaces, correct for target, instrument and
media characteristics and present a true temperature map of the tested surface.
Other parameters affect infra-red detector performance. These parameters control the
process of instrument selection. The infra-red thermography camera will be selected on
the basis of its features according to the application so that it will perform adequately.
Temperature range, temperature sensitivity, speed of response, spectral range,
repeatability, working distance and total field of view are the main performance
characteristics of radiation thermometers.
In this study, both qualitative and quantitative non-destructive infra-red thermography
tests were applied. Thermal sensors were used with advanced uncooled infra-red
detectors to detect differences in temperature (if any) on the surface of interest.
2.3 FRP system and materials
2.3.1 Background
The use of composite materials to enhance structural performance is not a new concept.
The Babylonians used straw to reinforce mud structures, as in the Dur-Kurigalzu
ziggurat in old Mesopotamia near Baghdad. Heavy modern industries in different
sectors like naval, aerospace, and the military always demand new composite materials
Chapter Two
42
with lighter weight and better strength. Carbon fibre composite materials started being
used in Japan and Europe in the mid 1980s (Nanni 1999). In the last decades there has
been an increasing tendency in civil engineering applications to develop new materials
that have better qualities and superior performance. Fibre composite materials are one
of these advanced new materials that are starting to be applied to concrete, steel and
masonry structures. FRP materials have advanced performance in the construction of
civil engineering structures in terms of the following:
High strength
High ductility
High resistance to deterioration
High durability
Low cost
Light specific weight
Small thickness that does not change the volume
Design freedom.
FRP composites are produced by embedding continuous fibres in a resin matrix which
combines the fibres. The fibres as a main load-bearer give the FRP composite its
strength and stiffness to resist different loads. Polymer matrix or resin ensures loads
have homogenous distribution between the fibres. Standard carbon fibre-reinforced
polymer (CFRP) composite is a combination of materials formed of unidirectional
continuous micro-fibres and adhesive matrix. A diagram of a CFRP uni-directional
fibres structure with its component materials is shown in Figure 2.33. The micro-scale
carbon fibres are arranged in one direction in this CFRP type. An electronic scanning
magnification of this CFRP type is shown in Figure 2.34. The scanning electron
microscope enlarged the image in this figure 150 times to reveal the arrangement of
fibre.
Literature review
43
Figure 2.33 Representation of CFRP materials [ Reproduced from Nanni (2004)]
Figure 2.34 Scanning Electron Microscope (SEM) image of CFRP fabric
an ideal blackbody surface at the same temperature. The surface emissivity value plays
a major part in the accuracy of the IR surface temperature reading. The more precise the
determination of emissivity, the more accurate is the surface temperature acquired by
IRT NDT.
4.5.1.1 Test set-up
Portions of concrete-FRP surfaces in Specimens 2, 4, 5, 8, 13, and 18 were painted
black to simulate a blackbody which has a known emissivity value. According to the
ASTM E 1933 standard, concrete-FRP specimens are required to have a minimum of 10 oC temperature difference, hotter or cooler, than the ambient temperature (ASTM E
1933-99a 2005). An oven was used to heat specimens and to generate the 10 oC
difference between specimens and the room temperature. Figure 4.15 shows Specimen
13 inside the oven. The oven raised specimen temperatures in a homogenous pattern
varying from 25 oC to 10 oC but remaining well below the epoxy glass transition
temperature (Tg). IR thermograms were recorded immediately after the specimen was
removed from the oven. Natural cooling was monitored to exclude the measurement of
emissivity values when the difference in temperature between the specimen’s surface
and room temperature was less than 10 oC.
Figure 4.15 Concrete-CFRP specimen inside oven
Parametric adjustments of the data processing unit were performed according to the
thermal properties of the known painted part of the specimen. IR images were recorded
Quantitative IRT experimental laboratory program
125
and monitored on both modified and original portions of the specimens’ surfaces. The
known emissivity of the painted part was input in the IR software for the modified
painted portion. Then emissivity of the original surface was then obtained by adjusting
the input value of the emissivity until the IR camera detected the same temperature as
the modified painted surface. Figure 4.16 shows the original and modified painted parts
of Specimen 2. This process was repeated five times for each specimen and the average
emissivity reading was recorded.
Figure 4.16 Thermogram of Specimen 2 shows the modified surface for emissivity test
4.5.1.2 Emissivity values
Test results were recorded for Specimens 2, 4, 5, 8, 13, and 18. The IR software Image
Processor ProII was used to adjust the emissivity values on different areas of the
surface of interest. The measured emissivity values of the tested specimens at 10 ºC
above the calculated room temperature for the unpainted parts of the specimens varied
from 0.96 to 0.98 for the carbon FRP fabric and for the laminate FRP composite the
emissivity value was around 0.92, as shown in Table 4.5. This process was repeated five
times for each of the six tested specimens. The average emissivity readings for the
CFRP were 0.97 and 0.92 for fabric and laminate system respectively.
Painted area
Chapter Four
126
Table 4.5 Emissivity values of IRT tests
Specimen IRT run #1 IRT run #2 IRT run #3 IRT run #4 IRT run #5
2 0.98 0.97 0.98 0.97 0.98
4 0.96 0.97 0.96 0.95 0.97
5 0.89 0.93 0.91 0.92 0.92
8 0.97 0.96 0.96 0.97 0.96
13 0.96 0.98 0.97 0.97 0.98
18 0.97 0.96 0.96 0.97 0.96
Areas in most of the specimens with CFRP laminates were painted with a thin matt
black coating with an emissivity value of about 0.97 before performing the
thermographic investigations in order to calibrate and record each specimen’s surface
emissivity value.
Observation angles can affect emissivity values noticeably. All the emissivity
experiments in this part followed as far as possible the same angle that was used in most
of the IR experiments conducted in this study.
4.5.1.3 Summary
Knowledge of the precise surface emissivity is required to calculate the actual surface
temperature. In applying IRT NDT for the detection of subsurface defects, knowing the
accurate value of the emissivity is not essential to detect and/or characterize the defect,
or even determine the defect size. This because detection depends on the defect’s
thermal signal and/or thermal contrast, and both of these parameters are emissivity-
independent (i.e. both temperature above the defect and background temperature in the
defect-free area have the same emissivity). Nevertheless, it was necessary to measure
the emissivity values of both CFRP fabric and laminate to compare the surface
temperatures on different defects and to compare the surface temperature according to
experimental and finite element simulation results.
Quantitative IRT experimental laboratory program
127
4.5.2 Part 2: Using PTT to detect different bond defects
During the qualitative thermography tests presented in Chapter 3, it was noticed that all
the unbonded areas and debonding defects implanted under a single CFRP fabric were
detectable. However, defects beneath multiple layers were not easily identified. In this
part of the experimental program, all specimens were investigated thoroughly. A total
of 381 IRT tests was conducted on the 32 specimens. Each test involved analyzing 600
thermogram images. For each individual defect, the surface temperature above the
defect and the defect-free areas was recorded. This stage addressed the following:
detection of unbonding areas, debond detection, far detection and transmission IRT
observation.
4.5.2.1 Unbond defect detection
The detectability of unbond defect is influenced by several factors, including the size of
the defect, the depth of the defect, the number of composite material layers and the
properties of the CFRP composites and substructure. From the thermal images of
specimens, it is possible to detect and locate the unbond areas in different CFRP
systems. However, the aim of this part of the experimental quantitative IR program was
to develop a deeper understanding of the detection procedure.
Figure 4.17a demonstrates the IR images of Specimen 1. The thermogram results show
that the bond defects were very detectable under a CF 130 CFRP fabric composite. Six
regions of interest (ROIs) were localized as measurement functions at defects UB011,
UB012 and UB013 to analyze the IR reading of Specimen 1 thermograms. Figure 4.17a
illustrates the locations of the specimens’ subsurface defects. The ability to detect
defects is represented by the value of the thermal signal (ΔT) calculated from Equation
4.1. Figure 4.17b shows defect UB011 thermal signals versus time with the excitation
source positioned at different distances. From the results in Figure 4.17b, it can noted
that the unbonded thermal signal in this specimen followed Pattern A. The maximum
thermal signal showed immediately after the excitation source was turned off and the
shutter closed. The recommended IRT site design that shows the maximum thermal
signal was when the heat source was positioned at 0.5 m from the specimen’s surface
and the input thermal interval pulse wave was 5 s. IR tests performed at less than the 0.5
Chapter Four
128
m distance or more than the 5 s pulse duration showed an increase in the maximum
temperature on the CFRP surface to over 60 oC. During the IR test, the CFRP’s surface
temperature was monitored to ensure that it did exceed the glass transition temperature
of the epoxy. The mechanical properties of the resin matrix degrade and suddenly
change when its temperature increases beyond its glass transition temperature (Tg). The
Tg of the applications used in CFRP strengthening systems are in the range of 55 to 70 oC (CEB-FIP Bulletin 14 2001).
(a) Thermal image
(b) Defect UB011 thermal responses at different distances
Figure 4.17 Defects in Specimen 1
-1.00.01.02.03.04.05.06.07.08.09.0
10.011.012.0
0 10 20 30 40
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB011-1s at 50 cm
ΔT-UB011-1s at 70 cm
ΔT-UB011-1s at 100 cm
ΔT-UB011-1s at 120 cm
Quantitative IRT experimental laboratory program
129
As mentioned in Section 4.4 above, the size of the ROI to study the surface temperature
on the thermogram can vary. The most important factor in ROI size is that it should
represent enough pixels to characterize the temperature suitably on the ROI. Figure 4.18
shows the difference in the signals with two different ROI designs adopted for defect
UB011. As illustrated in Figures 4.18a and 4.18b, the sizes of the ROI rectangles
differed considerably. However, the differences in the signals collected from these two
ROIs were negligible at less than 1 oC, due to the selection of Design 1 of the ROI that
was set exactly on the unbonding area. The average temperature was collected for most
of the ROIs in this study; however, some defects were designed not to have equal
degrees of deterioration, such as the debonding in Specimens 3, 26 and S3. These
defects were designed with ROIs that collected the maximum temperature within the
ROI rectangle. It was also found that by reducing the size of the ROI, the difference
between choosing an average or maximum ROI rectangle will be eliminated. All of the
ROIs applied to the specimens in this research were chosen very carefully to represent
the artificial defect type. The sizes of these ROIs differed from flaw to flaw. Small
defects were designed with ROIs that covered most of the defect to supply enough
pixels in the ROI area. Larger defects were set with ROIs not covering the entire defect,
and only a reasonable ROI within the defect area was selected. However, the most
critical issue was to select the area of ROI that showed the defect clearly.
(a) ROI1 design (b) ROI2 design
Chapter Four
130
(c) Signals of ROI1 and ROI2 designs of UB011
Figure 4.18 Defect UB011 thermal responses at different ROI sizes
Figure 4.19 reveals that even with a very short pulse duration of 1 second, the IRT
detection system is still able to read differences of more than 12 oC between the defect
area and the surrounding defect-free area for the single CF130 fabric layer. Further
analysis shows that by increasing the input heat flux, the maximum thermal signal rises
lineally, as shown in Figure 4.20. The rate of (ΔTmax / input heat flux) increases with the
increase in heating pulse interval (Tashan and Al-Mahaidi 2012). The results in Figure
4.20 show the input heat flux required to attain the desired thermal signal in the IR tests.
The maximum thermal signal of 5 s pulse interval is described by Equation 4.3 where q
is the input heat flux in watts per square metre.
-2.0
2.0
6.0
10.0
14.0
18.0
22.0
0 10 20 30 40 50 60
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB011-5s at 50 cmΔT-UB011-1s at 50 cmΔT-UB011-1s at 50 cm-ROI2ΔT-UB011-5s at 50 cm-ROI2
Quantitative IRT experimental laboratory program
131
ΔT(q)max = 0.032 q – 5.746 Equation 4.3
Figure 4.19 Defect UB011 thermal responses at different pulse intervals
Figure 4.20 Heat flux versus maximum thermal signal in Specimen 1 for different pulse
intervals
-2.0
2.0
6.0
10.0
14.0
18.0
22.0
0 10 20 30 40 50 60
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB011-1s at 50 cmΔT-UB011-3s at 50 cmΔT-UB011-5s at 50 cm
0
100
200
300
400
500
600
700
800
900
1000
0 5 10 15 20 25
Hea
t Flu
x (w
/m2 )
∆Tmax (oC)
1 s
3 s
5 s
Chapter Four
132
Similar equations connecting the output thermal responses with the input applied heat
flux intensity can assist the thermographer to design appropriate IRT test configurations
in terms of detectability level.
The detection of unbond defect under different kinds of carbon fabrics was investigated
with Specimens 24 and 27. Both specimens were strengthened with unidirectional
CF130 and CF140 CFRP MBrace fabrics as shown in Figure 3.11. Active IRT PTT was
performed on these specimens to examine the effect of changing CFRP physical
properties (i.e. fabric thickness, fabric directions) on the thermal detection of the same
defects.
Four ROIs where analyzed thermally in Specimen 24’s defects. The first two regions
were to study defect UB241 which was embedded under CF130 CFRP type, as shown
in Figure 4.21, while the other ROIs were assigned to record the thermal response of
defect UB242 implanted in the CF140 CFRP fabric-concrete bond zone.
Figure 4.21 Defects in Specimen 24 thermogram
The results in Figure 4.22 show that for the same pulse duration time, the thermal signal
detection is enhanced by increasing the input heat flux. The thermal signals of Specimen
Quantitative IRT experimental laboratory program
133
24 defects follow Pattern A with very high values for both defects. The UB241 defect
under the CF130 fabric shows a considerably higher ΔT (about 25% more) compared to
the UB242 signal when the heat source was applied at 50 cm with 3 s heating interval.
The difference between the CF130 and CF140 defects was reduced to less than 10%
when different heating intervals were applied, as shown in Figure 4.22b for heating at
50 cm. Both defects had almost the same behaviour after the heat source was turned off.
The thermal signal faded 20 s from the beginning of the IR test when the heating was
applied for 3 s. However, this fading duration is related to different parameters involved
pulse duration and substructure material. Figure 4.22b shows that signals for both
UB241 and UB242 faded after 10 s when the pulse was at 1 s. When the pulse was
longer, at 5 s, the signals recorded zero.
The results of Specimens 24 and 27 for the detection of the same unbonded area under
different CFRP fabric types confirm that the detection of defects is enhanced by the
reduced CFRP composite thickness. The detection of both UB241 and UB271 detection
was better than UB242 and UB272, because the CFRP fabric above the first two faults
was CF130 which is 33 % less thick than the CF140 on UB242 and UB272.
(a) 3 seconds pulse duration, heat source at 50 and 70 cm
-2
2
6
10
14
18
0 5 10 15 20
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
UB242- 3s at 50cmUB241- 3s at 50cmUB242- 3s at 70cmUB241- 3s at 70cm
Chapter Four
134
(b) 1 and 5 seconds pulse durations
Figure 4.22 Infra-red signals of Specimen 24 defects
The results from the IR analysis of Specimen 24 confirm that, by increasing the input
heat flux, the maximum thermal signal rises lineally, as shown in Figure 4.23. For the
CFRP CF130 used in UB241, similarly to Specimen 1, the rate of (ΔTmax / Input heat
flux) increases with the increase in heating pulse interval. However, for pulses of 1 s the
rate was not perfectly linear, due to the short time available to capture the IR image and
the few IR frames recorded during the 1 s pulse length. Figure 4.23a shows these
increasing. The results in Figure 4.23b present the input heat flux versus maximum
thermal signal for defect UB242. The slopes of the linear relationships between the heat
and the maximum signals do not change for this defect and follow the same increase
rate. That could be due to the CFRP type of CF140 which have thicker section compare
to the CF130, and have different fabrics waving pattern. However, maximum signal
during the 1 s pulse duration shows also a non perfect linear behaviour that was pointed
up in Figure 4.23a. Figure 4.23 shows that the maximum signals in CF140 defects are
lower than defects under CF130 CFRP fabric. CFRP CF140 is thicker than CF130,
which allow the layer to transfer the heat slightly faster and then register lower signals.
This lower signals result might be also due to the different waving CFRP patterns and
-2
2
6
10
14
18
22
0 5 10 15 20
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
UB242- 5 s at 50 cmUB241- 5 s at 50 cmUB242- 1s at 50cmUB241- 1s at 50cm
Quantitative IRT experimental laboratory program
135
the choosing of the ROI that can effect on the IR image analysis. This can highlight the
task’s hardness of making comparing between different CFRP materials.
From Figure 4.23 it can be noted that for CF140 type pulses with heat flux less than 450
W/m2 are produce ΔTmax less than 2.5 oC, which is very small temperature to well
recognition of a defect. While for the CF130 the minimum input heat that can provide
more than 2.5 oC as thermal signal is 300 W/m2. The relation between the input heat
flux and the pulse interval are affecting by different parameters involve the angle of the
lamp, and the ambient temperature. For that reason, in the concrete -CFRP fabric
system, to provide a well observed detection, heat wave injection with less than 500
W/m2 is not recommended. Usually this 500 W/m2 wave is generated when the
excitation lamp located at 1.2 m from the test object.
(a) Defect UB241
0
300
600
900
1200
1500
1800
0 5 10 15 20
Hea
t flu
x (w
/m2 )
∆Tmax (oC)
UB241-1sUB241-3sUB241-5s
Chapter Four
136
(b) Defect UB242
Figure 4.23 Heat flux versus maximum thermal signal in Specimen 24 for different pulse intervals
Unbonded area defects under multiple CFRP fabric layers were examined by PTT IRT
on Specimen 6. Defects UB063 and UB064 were identified clearly. Defect UB064
(under double CF140 sheets) had a smaller thermal signal compared with UB063.
Figure 4.24 indicates that, by increasing the distance between the heat source and the
investigated surface, the ΔTmax ratio of a defect under a single CFRP layer to a defect
under a double layer increases. The maximum thermal signal detection under a single
CFRP layer is just above double that of the of ΔTmax UB064 beneath double CFRP
layers when the heat source is positioned at 50 cm. By increasing the heat excitation
source distance to 1.2 m, the ratio of ΔTmax between single and multi layer rises to 400
%, as shown in Figure 4.24.
0
300
600
900
1200
1500
1800
0 5 10 15 20
Hea
t flu
x (w
/m2 )
∆Tmax (oC)
UB242-1s
UB242-3s
UB242-5s
Quantitative IRT experimental laboratory program
137
(a)
(b)
Figure 4.24 Thermal signals of defects in Specimen 6: (a) UB063, (b) UB064
Equation 4.2 was used to calculate the thermal contrast of Specimen 6 defects. Figure
4.25 shows the IR contrast results with the heat source located at 50 cm and pulses of 5
s were injected. As shown in the figure, the noise level in the contrast values is low until
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
0 20 40 60 80
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB063-5s at 50cm
ΔT-UB063-5s at 70cm
ΔT-UB063-5s at 120cm
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 20 40 60 80
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB064-5s at 50cmΔT-UB064-5s at 70cmΔT-UB064-5s at 120cm
Chapter Four
138
it reaches the maximum contrast Cmax level when the excitation heat lamps are turned
on. Immediately after the lamps are turned off, the level of noise increases gradually
until the test ends. Figures 4.25a and 4.25b demonstrate the difference between C values
at different excitation distances with the same pulse interval. The figure show that, the
noise level is decreased by increasing the distance between the lamps and the
investigated surface. To determine the maximum contrast and its corresponding time,
the contrast smooth curves were calculated as shown in Figure 4.25.
(a)
(b)
Figure 4.25 Thermal contrast of Specimen 6 with 5 s pulse: (a) excitation at 50 cm, (b) excitation at 120 cm
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
6.00
0 50 100 150
Con
trast
Time (s)
UB063-5s at 50cm
UB064-5s at 50cm
-1.00
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
0 50 100 150
Con
trast
Time (s)
UB063-5s at 120cm
UB064-5s at 120cm
Quantitative IRT experimental laboratory program
139
Defect UB063 contrast signals are shown in Figure 4.26 for 5 s PTT applied from
different distances. The maximum contrast values are very high for these heating waves.
Maximum thermal contrast reaches a value of 5.71 when the excitation source is
mounted 50 m from the tested specimen. The behaviour of the contrast responses
follows the same pattern for the same pulse period with different lamp distances. When
the lamps’ location is fixed, the pattern of contrast responses at different pulse durations
shows high noise when the pulse duration is short, as demonstrated in Figure 4.27. The
contrast wave time decay is increased by the increase pulse length. Figure 4.26
demonstrates the value of C reaches 1.5 for defect UB063 after 29 s, 31 s, 38 s, and 59 s
from the IR test commencement when the lamps are positioned at 50 cm, 70 cm, 100
cm, and 120 cm respectively.
Figure 4.26 Contrast of UB063 with 5 s pulses at different distances
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
6.00
0 50 100 150
Con
trast
Time (s)
UB063-5s at 50cm
UB063-5s at 70cm
UB063-5s at 100cm
UB063-5s at 120cm
Chapter Four
140
Figure 4.27 Contrast of UB063 with 1 m distance at different pulses
Unbonding artificial defects under CFRP laminates composite systems were
investigated in Specimen 5. Figure 4.28 illustrates these defects. As shown in the figure,
unbonding defect UB051 covered by a single layer of the laminate is easily detected.
UB052 with two CFRP laminates is a little harder to detect compared with UB051.
Figure 4.42 presents the three-dimensional profile of the debonding fault in Specimen 3.
The hot spot appears with temperatures increasing gradually towards the middle of the
debonding area where the trapped heat reaches its peak (Tashan and Al-Mahaidi 2009).
This is a clear indication of the absence of bonding at this implanted deficiency. The 3-
D profile of the temperature variation gives an indication of the severity of debonds
within defect zones. The reflections on the CFRP fabric surface can mislead the reading
of the thermograms, but software filters can be used to reduce these reading errors. A
Gaussian filter (5×5) shows good results in eliminating the spiky errors when applied to
the 3D IR shown in Figure 4.42a. As shown in Figure 4.42b, the Gaussian filter alters
slightly the maximum temperature of the IR image. As shown in Figure 4.42b, the peak
temperature in the debond area was shifted by 0.8 oC.
(a)
Chapter Four
156
(b)
Figure 4.42 Three dimensional profile of DB031: (a) before applying Gaussian filter, (b) after applying 5 ×5 Gaussian filter
Specimen 26 was fitted with a fabricated debonding area similar to Specimen 3’s
artificial fault. However, the CFRP fabric used in Specimen 26 was Type CF140, while
Specimen 3 was strengthened with CF130. The differences in the CFRP fabric
properties of these two specimens and in the debonding area sizes that were generated in
a random way lead to different IR results for these two specimens. The maximum
thermal signal for DB031 is three times that for DB241, as shown in Figure 4.43.
Moreover, they follow different curve patterns, as DB261 shows Pattern B, whilst
DB031’s defect signal shows Pattern A.
For different pulse durations with different excitation source distances, Specimen 26
shows the same Pattern B signals. Figure 4.44 illustrates these signals. The gap between
the maximum ΔT is bridged by decreasing the input heating and shortening the duration
of the heating pulses. The pulse of 5 s from 50 cm in Figure 4.44 was noticed to have a
signal of 2.5 oC even after the end of the thermal test at 100 s.
Figure 4.45 confirms that the contrasts are noisier than the signals. For that reason, the
contrast responses required more smoothing in the construction of Figure 4.45. From
Quantitative IRT experimental laboratory program
157
the results, it is observed that the noise level is high when the surface receives more heat
from the near lamps, as shown in the difference between the contrasts after the end of
the pulse in Figure 4.45a. At 50 cm excitation distance, the smoothed maximum contrast
Cmax decreases from 8.7 when the pulse is applied for 5 s to 7.2 for 1 s pulse interval.
The C values shown in Figure 4.45b for 50 cm and 1 s pulse durations display more
noise compared to the 5 s pulse length shown in Figure 4.45a.
Figure 4.43 Specimens 3 and 26 debonding responses
Figure 4.44 Debond DB261 signals
-2.0
3.0
8.0
13.0
18.0
23.0
0 50 100 150
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-DB261-5s at 50cm
ΔT-DB031-5s at 50cm
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0 50 100 150
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-DB261-5s at 50cmΔT-DB261-5s at 70cmΔT-DB261-1s at 50cmΔT-DB261-1s at 70cm
Chapter Four
158
(a)
(b)
Figure 4.45 Contrast of DB261: (a) at 5 s pulse, (b) at1 s pulse
Debonding in steel was investigated by testing Specimen S2. Figure 4.46 describes the
DBS21 thermal signals captured at 5 s pulse phase. All thermal responses in steel show
a Type A thermal signal pattern. The ΔTmax is affected considerably by heat flux
-1.00
1.00
3.00
5.00
7.00
9.00
11.00
13.00
15.00
0 50 100 150
Con
trast
Time (s)
C -DB261-5s at 50cm
C -DB261-5s at 70cm
-1.00
1.00
3.00
5.00
7.00
9.00
11.00
0 50 100 150
Con
trast
Time (s)
C -DB261-1s at 50cm
C -DB261-1s at 70cm
Quantitative IRT experimental laboratory program
159
intensity level. By changing the location of the heat source from 50 cm to 70 cm, the
maximum thermal signals drop by half approximately, as shown in Figure 4.46. The
debond defect inserted in the steel-CFRP system fabric has higher ΔTmax compared with
the corresponding defect in the concrete-CFRP system. However, after reaching the
peak point at ΔTmax the signal of the debond defect attached to steel reduces sharply
compared to the defect in concrete-based structure.
Figure 4.46 Steel Specimen 2 thermal signals
Figure 4.47 offers the comparison between DB031 and DBS21 defect signals. From this
figure, it can be seen that the difference between thermal signals fading in concrete and
steel is dependent on the pulse duration.
-2.0
3.0
8.0
13.0
18.0
23.0
28.0
0 50 100 150
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-DBS21-5s at 50cmΔT-DBS21-5s at 70cmΔT-DBS21-5s at 100cmΔT-DBS21-5s at 120cm
Chapter Four
160
Figure 4.47 Comparison of Specimens’ 3 and S2 debonding signals
The polynomial smoothing contrasts of DBS21 are shown in Figure 4.48. The Cmax is
higher compared to DB031 and DB261, due to the larger size of the air pocket within
the Specimen S2 defect zone. The contrasts for different pulses show similar behaviour
with different intensities. The time when Ctmax reaches the peak of the contrast was
found to be immediately after the end of the pulse when the lamps were turned off. The
noise level increased gradually towards the end of the IR test.
-2
3
8
13
18
23
28
0 50 100 150
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-DB031-5s at 50cmΔT -DB031-5s-100cmΔT-DBS21-5s at 50cmΔT-DBS21-5s at 100cm
Quantitative IRT experimental laboratory program
161
Figure 4.48 Thermal contrast for Specimen S2
Defects inserted in bi-directional CFRP fabric show similar thermal signals to defects in
uni-directional fabrics. Figure 4.49a shows the thermal signals for the debonding defect
in Specimen 13 at pulse intervals of 1 s, 3 s, and 5 s recorded when the lamps were at
distances of 0.5 m, 0.7 m, 1 m, and 1.2 m. From the figure it can be concluded that even
with the greater thickness of the TYFO BCC (± 45o) fabric at 0.55 mm, the technique
still provides good thermal signals. As shown in Figure 4.49b, the linear relationship
between input heat flux and maximum signal is confirmed for this type of bi-directional
fabric.
-6.00
-4.00
-2.00
0.00
2.00
4.00
6.00
8.00
10.00
12.00
0 50 100 150
Con
trast
Time (s)
Poly. (C -DBS21-5s at 70cm)
Poly. (C -DBS21-3s at 70cm)
Poly. (C -DBS21-1s at 70cm)
Chapter Four
162
(a)
(b)
Figure 4.49 Defect DB131 (a) thermal signals at different pulse and distances, (b) heat flux versus maximum thermal signal for DB131 at different pulse intervals
Debonding with different defect thicknesses was investigated in Specimens 19, 20 and
21. Table 4.6 summarizes the maximum signal detection for all debonding artificial
0
2
4
6
8
10
12
14
16
18
20
0 30 60 90 120 150
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
DB131-1s at 50
DB131-1s at 70
DB131-1s at 100
DB131-1s at 120
DB131-3s at 50
DB131-3s at 70
DB131-3s at 100
DB131-3s at 120
DB131-5s at 50
DB131-5s at 70
DB131-5s at 100
DB131-5s at 120
0
200
400
600
800
1000
1200
1400
0 5 10 15 20 25
Inpu
t Hea
t Flu
x (W
/m2 )
∆Tmax (oC)
DB131- 1 sDB131- 3 sDB131- 5 s
Quantitative IRT experimental laboratory program
163
defects that were inserted with different thicknesses in these specimens. Due to the
small thickness of air pockets in defects DB191 and DB192, which were less than 0.25
mm, it was impractical to remove the epoxy material totally from the debond zone and
no air pocket were generated within the debonding surface. The epoxy works as a
bridge in these two defects which transfers the heat from the CFRP fabric to the
concrete subsurface. For this reason, the signals in these defects have higher values
compared to other corresponding debond defect signal values in Table 4.6. From the
analysis of Specimen 19, it can be seen that debonding with less than 0.25 mm thickness
works in an exceptional way. Because of the narrow debonding, there is no lack of
epoxy within the debonding. This means that the epoxy layer thickness was increased in
these defects which produced higher ΔT within the debonding regions. This kind of
debonding defect which arises with no air pocket within the areas does not act in a
similar way to fully debonded or fully unbonded defects.
Debond defect DB201 has a thickness close to that of DB211, and both defects show
similar maximum signal values, as shown in Table 4.5. However, DB211 with a
thickness 0.1 mm larger than DB201, shows as expected, a slightly larger signals of the
ΔTmax. The thickness of DB212 defect is double that of DB211, and a difference in
maximum limit signals between these two defects was noticeable. The average
enhancement in detection between Specimen 21 debond areas was about 285% at 1 s
pulses, 180% at 3 s pulses and 159% at 5 s pulses. Although the detection improvement
at 1 s was high, the values of maximum thermal signals were very low.
Chapter Four
164
Table 4.6 Debonding defects summary
Debonding ID
defects and
thickness
(mm)
ΔTmax (oC)
at 50 cm 70 cm 100
cm
120
cm
DB191(0.1)
1 s 9.4 7.1 4 2.7
3 s 16.4 12.5 6.9 4.8
5 s 19.4 16 8 5.6
DB192(0.25)
1 s 12.7 6.7 3.1 1.9
3 s 20.9 11.3 4.9 3.3
5 s 24.5 12.2 5.6 3.7
DB201(0.4)
1 s 3.2 1.1 0.8 0.5
3 s 7 4 2 1.1
5 s 10.5 7 3.9 1.9
DB211(0.5)
1 s 3.2 1.9 1 0.7
3 s 7.9 4.6 2.6 1.8
5 s 11.9 7.2 4.3 2.7
DB212(1)
1 s 9.2 5.7 2.9 1.9
3 s 16.1 8.8 4.6 2.9
5 s 21.4 12.2 6.3 3.8
The ability of the IRT to identify delamination defects was studied by testing Specimens
16, 6, 7, and 13. Specimen 16 was constructed with an artificial delamination defect, as
shown in Figure 3.11-16. In spite of the three CFRP composite layers on the surface of
this concrete specimen, the delamination defect between the double FRP laminates was
very detectable for applied heating intensities imposed for different pulse durations. The
thermal image in Figure 4.50a exhibits defect DL162’s shape and location in Specimen
16. Figure 4.50b show that the signal was more than 2.5 oC, even for short pulses at 1 s
from half a metre. The ΔTmax with exposure of the CFRP surface for 5 s was just below
5 oC, which is a good signal for the location of potential flaws in the bonding zone. It
was noticed that, by reducing the input heat wave when the lamps are positioned around
Quantitative IRT experimental laboratory program
165
1 m, the signals are weak to unacceptable. The signals for pulse intervals from 1 s to 5 s
at 1 m show very low values at less than 0.5 oC, due to the effect of installing multi-
CFRP layers above the delamination DL162 which hinder heat wave transmission and
produce shallow thermal responses.
The thermal contrasts calculated during the IR analyses of delamination under multi-
layers of CFRP composites follow Pattern B as shown in Figure 4.14. Figures 4.50c and
4.50d highlight contrast registers its maximum values almost at the end of the IR test.
This makes the value of the contrast unreliable, especially with the amount of noise that
increases towards the end of the IR test. As shown in Figure 4.50c, the maximum
contrast captured for this defect was 3.6. However, due to the unacceptable noise level
this C value is inappropriate.
(a)
Chapter Four
166
(b)
(c)
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
0 50 100 150
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-DL162-5s at 50cmΔT-DL162-5s at 100cmΔT-DL162-1s at 50cmΔT-DL162-1s at 100cm
-1.00
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
0 50 100 150
Con
trast
Time (s)
C -DL162-5s at 50cm
C -DL162-5s at 100cm
Quantitative IRT experimental laboratory program
167
(d)
Figure 4.50 Defect DL162: (a) location of DL162, (b) thermal signals, (c) contrast at 5 s, (d) contrast at 1 s
The maximum signals of the artificial delamination defects in Specimens 6, 7 and 13 are
shown Table 4.7. By studying the two delamination areas of Specimen 6, it can be noted
how the size of the delamination area can influence the surface temperature distribution,
when the larger DL061 defect area records higher signals than the DL062 delamination
for all lamp distances and pulse intervals. The average improvement for the detected
ΔTmax of DL061 and DL062 was between 222%, and 207 % for intervals from 1 s to 5
s.
Delaminations in bi-directional CFRP fabrics were investigated with defects DL072 and
DL132. Specimen 7’s defect DL072 thermal results are shown in Table 4.7. The data
show a higher thermal maximum signal than the delamination underlying a uni-
directional fabric in Specimen 16, possibly due to the increase in the delamination
thickness of DL072. The delamination in Specimen 13 shows very similar values of
ΔTmax to Specimen 7. The only small alteration of the values between these specimens’
defects was due to the fabric design, as DL132 is between two TYFO BCC (± 45o)
sheets, while DL072 is between uni-directional CF140 and bi-directional CFRP fabric
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
0 50 100 150
Con
trast
Time (s)
C -DL162-1s at 50cm
C -DL162-1s at 100cm
Chapter Four
168
layers. However, this difference between the values of DL132 and DL072 was expected
to be the opposite, with DL072 being expected to have the high pattern of ΔTmax. This
small increase in DL132 thermal signals might be related to different parameters
including the rough surface preparation of Specimen 13, which created a pointy
concrete surface that contacted the CFRP with less epoxy and helped heat to transfer
faster to the substrate structure.
Table 4.7 Summary of maximum thermal signals for delamination defects
Defect ID
ΔTmax (oC)
at 50 cm 70 cm 100
cm
120
cm
DL061
1 s 7 6.6 2.6 1.8
3 s 13 9.2 4.9 3.3
5 s 15.7 11.4 6.5 4.5
DL062
1 s 6.1 4.3 1 0.5
3 s 11.7 5.5 1.6 1
5 s 14 6.9 2.4 1.6
DL072
1 s 7.1 3.5 1.3 0.9
3 s 13.8 7.1 3.6 2.5
5 s 18.3 9.6 5.3 3.5
DL132
1 s 8.4 3.6 1.6 0.8
3 s 14.7 7.7 3.5 3.2
5 s 22.8 11.7 5.3 4
4.5.2.3 Far distance IR detection
Tests were conducted at distances to explore the opportunity of carrying out these IR
tests from far distances. The same active IR tests were applied to Specimen 1 at
different pulse intervals. The IR camera was mounted at 5 m and 10 m from the
specimen while the heating lamp was positioned at 70 cm. According to the IR camera
features, the view of field can detect an area of 6 mm2 from 10 metres, as shown in
Quantitative IRT experimental laboratory program
169
Figure 4.1b. For that reason, the maximum distance at which the IR camera can detect
the defect and read size correctly is 10 m.
The results reveal that even though that the distance between the specimen surface and
the IR camera was increased up to 10 m, the location, shape and size of the fabricated
defects under the CFRP fabric were still observed and identified with proportional
defects’ sizes. The IRT NDTs thermograms achieved from far distances are shown in
Figure 4.51.
(a) Image captured from 5 m distance
(b) Image captured from 10 m distance
Figure 4.51 Thermal image of Specimen 1
Chapter Four
170
Far distance detection was investigated in Specimen 1 in 6 IRT tests. Both tests were
performed with active PTT. Heat load pulses with intervals of 1 s, 3 s, and 5 s were
applied. Figure 4.52 shows the thermal responses of these six IRT tests. Results of both
camera distance locations follow the same pattern for each pulse interval. All IR images
show encouraging results in terms of accuracy of defect size measurement and detection
with a minimum of 5 oC difference between the defect and its surrounding area at a
minimum pulse interval of 1 s. However, the thermograms captured 10 m from the
object show higher ΔTmax compared to IR images recorded at 5 m. This may due to the
increase of the transmission line between the IR and the investigated surface which
leads to increased errors in the emissions readings.
Figure 4.53 reveals the three thermal responses of defect UB011 captured from 0.7 m, 5
m and 10 m from the specimen’s surface. The 4 oC difference between the readings at
0.7 m was because of a different IR analysis at the pixel level and different camera
angle.
(a) Captured from 5 m distance
0.0
5.0
10.0
15.0
0 10 20 30 40
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
UB011 at 5 sUB011 at 3 sUB011 at 1 s
Quantitative IRT experimental laboratory program
171
(b) Captured from 10 m distance
Figure 4.52 Thermal responses of Defect UB011
Figure 4.53 UB011 signals captured from different distances
Verification of the ability to conduct the IRT NDT from far distances can help in
applying IRT tests in the field. It is obvious that IR detector cannot be located close to
all structures on sites. The distances that allow reliable results in this section are
0.0
5.0
10.0
15.0
0 10 20 30 40
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
UB011 at 5 sUB011 at 3 sUB011 at 1 s
-2.0
2.0
6.0
10.0
14.0
18.0
22.0
0 10 20 30 40
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB011-5 s at 10 mΔT-UB011-5 s at 5 mΔT-UB011-5s at 70 cm
Chapter Four
172
reasonable distances that offer the possibility of testing most structures that needs to be
tested thermographically in the field. However, unwanted emittance which affects
thermograms that captured from far distances may be a problem that will need to be
solved. Usually a filter that attached to the IR cameras can help overcome unwanted
emittance.
The passive approach is most appropriate technique for IRT from far distances in site
conditions. It is recommended to carry out far passive IRT just after the sun-rise or after
the sun-set, when temperature has the maximum chance.
4.5.2.4 Transmission observation IRT
Cold spots can form, as indicated in Figure 2.23 when transmission observation method
is applied in PTT IRT. Concrete specimens are too thick to capture any signals by
means of transmission IRT, and the results of IR tests using this technique show no
thermal responses when applied to selected concrete specimens. For that reason, it is not
feasible to inspect defects inserted in concrete specimens with the transmission
observation method Specimens with steel substrate are more appropriate for the
employment of transmission PTT. Three steel specimens were tested using this
technique to explore PTT with transmission observation technique. However, unbonded
defects with a single sheet of CFRP fabric in Specimen S1 were not identified using this
method. Unbonding defects on the CFRP laminate-steel zone in Specimen S4 were
localized and detected with very small thermal responses. Figure 4.54 illustrates defect
UBS41’s thermal responses. The IR results show that the unbonded area beneath FRP
laminate is noticeable; however, the values of the maximum thermal signal and contrast
are small compared to the signals and contrasts obtained by applying the reflection
observation method. The negative IR values in the figures below reveal the principal
cold spots generated by applying this transmission detection method.
Figure 4.54a presents the thermal signal response with a pulse period of 10 s and the
lamp mounted at 0.7 m from the surface of Specimen S4. The ratio of the pulse interval
to thermal signals is very high when the specimen is observed by the transmission
scheme, being only 4 oC when recorded as ΔTmax with 10 s pulses. The steel specimen
Quantitative IRT experimental laboratory program
173
thickness is 3 mm. For steel sections strengthened with CFRP laminate and more than 3
mm thick more time is required for the injection of the heat pulse. The contrast value is
small, being less than 0.75 at Cmax, as shown in Figure 4.54b. The results show that the
noise level in the transmission observation method is at minimum. The contrast appears
as a smoothed curve in the figure below, even after the pulse of 10 s end, due to the
stability of the temperature distribution in the specimen using this transmission method.
The noise level was slightly high at the beginning of the test when the pulse was
The experiments that conducted in this part involved the study of defect detection using
PTT. Artificial defects of unbonded areas, debonding, and delamination were examined
using PTT using the reflection observation technique. The transmission IR observation
method was also chosen for selected specimens. Thermal responses of defects
underlying single and/or multiple-CFRP fabric and laminate composites were evaluated
at different pulse durations and different lamp distances. The first set of IR experiments
focused on unbonded defects. Unbonded defects covered with different types and layers
CFRP fabric were investigated and the effect of increasing the fabric thickness was
examined. Thermal response curves of unbonding defects under single and double
CFRP laminate were also constructed. The experimental runs also included unbond
flaws under an arrangement of CFRP fabrics and laminates. Finally assessments of
defects inserted in CFRP- concrete and CFRP-steel systems were carried out.
The second experimental set performed emphasised debonding and delamination
detection by using PTT IRT. Irregular artificial debonding defects under different CFRP
fabrics and different substrate structures were evaluated. Three dimensional profiles of
the debond areas were constructed to study the severity of debonding within the flaw.
Delamination between CFRP systems was inspected in fabrics, laminates and
combination of both.
Applying the IRT from far distances was also studied. IR runs were conducted to study
the ability to capture consistent thermograms from different distances and up to 10 m
from the tested objects. Far distance detection reliability was analyzed for unbond
defects.
Finally, IRT NDTs adopting the transmission observation method were applied in this
part of the quantitative experimental program. Steel specimens only were chosen to be
tested using this observation method.
The quantitative experiments reported in this part present several interesting
conclusions, of which the following is a summary:
Quantitative IRT experimental laboratory program
175
For unbond, debond and delamination, the thermal signals decrease with
increasing thickness of CFRP composites.
The noise level in the thermal contrast is higher than the thermal signals. A
smoothing process is required to find the C versus time relationship using a
moving average and polynomial algorithm. The noise level is low until Cmax.
The level of noise then increases gradually towards the end of the test. The noise
level is decreased by increasing the distance between the lamps and the
investigated surface.
Most bond defect thermal signals follow Pattern A when the defect is located
under CFRP fabric. Defects underlying laminates Perform with A or B.
The maximum thermal signal is captured immediately after the excitation source
is turned off and the shutter closed for all defects in the CFRP fabric systems.
Flaws in the laminate-CFRP composite show their ΔTmax not immediately at the
end of the pulse, but after a short time, was due to the different thermal
properties of the CFRP fabric and laminates. The time range of this period was
different from defect to defect, according to the design, the specimen and/or IR
test setting.
The IRT PTT test proves that detection of different bond defects can be
achieved even with pulse intervals of 1 s. However, other fast PTTs with higher
pulse lengths at 3 s and 5 s show higher signals and contrasts in the thermal
analyses.
For unbond defects under different CFRP fabric, the maximum thermal signal
increases lineally with increasing input heat flux.
For different CFRP fabrics the maximum thermal signals decrease with the
increase of the fibre thickness.
To generate well-recognized detection for bond defect the input heat flux is
recommended to be greater than 500 W/m2 and the pulse length more than 1 s.
The maximum thermal signal is proportional to the number of CFRP layers. It
decreases to about half with the increase of CFRP fabric sheets to 2 layers.
Chapter Four
176
The rate of thermal signal fading is greater in defects under a single CFRP layer
than multi-layers and the fading rate for fabrics is higher than for CFRP
laminates.
Bond defect detection does not depend only on the CFRP composite design and
system, but also on the substrate material. For identical pulse lengths, defects
with concrete substrate show greater thermal responses than those with steel.
However, due to extensive heat capture when IR is conducted with more than 5 s
pulse intervals and with very high injected heat waves (when the lamp is close,
up to 0.5 m), defects in steel systems reveal higher signals than in concrete.
The 3-D profile constructed for debonding defects is a very efficient tool to
determine the severity of the unbonding within the debonding zone.
The size of debonding air pockets effect the thermal response.
The maximum thermal signal increase nonlinearly with increasing debonding
region thickness.
By increasing the number of CFRP layers, the contrast of a delamination will
produce unacceptable noise levels and provide irrelevant C values.
The maximum thermal signal increases by increasing the delamination area.
Rough surface specimen preparation alters the IR reading, and may present
irrelevant spots due to spiky point formation in the bonding zone.
The technique shows an excellent ability to detect defects from 10 m accurately.
However, IR thermograms from far distances contain a high level of unwanted
emittance due to the long transmission line between the IR detector and the
object.
In the transmission observation method, specimens need more time to generate
well-identified thermal signals. Cold spots appear with negative signals. The
noise in the thermal contrast appears at the commencement of the test and
decreases towards the end.
4.5.3 Part 3: Defect size measurement
Knowledge of the precise size of unbond, debond areas and delaminations helps the
assessment and evaluation of the integrity of the entire structure that has been retrofitted
with CFRP systems. This assessment and monitoring can lead to reduced stress from
Quantitative IRT experimental laboratory program
177
over-loading and keep the structure well beyond the serviceability limit. At the same
time, reading the size of defects accurately can help radically with repair and
maintenance. This section investigates the ability of IRT to determine subsurface defect
sizes with high accuracy. Defects in Specimens 1, 2, 4 - 9, 16, 17, 24, 27, S1, and S3 -
S5 were measured using active PTT IRT. These defects were located under different
CFRP materials. Halogen lamps were used as the excitation system during the tests. The
IR detector was positioned 70 cm from the investigated objects.
The defect size is determined by analysis of the thermal image at pixel level.
Measurement area functions provide excellent defect size measurement, by drawing an
ROI around the defect boundaries and calculating the number of pixels inside the ROI.
The size of a defect can then be calculated by translating the pixels to their
corresponding equivalent size and / or area. Figure 4.55 demonstrates the pixels
calculation analysis for the measurement of defect UB011 in Specimen 1. The lengths
of the lines in this IR image were as follows: Line 1 145 pixels, Line 2 68 pixels, and
Line 3 440 pixels. To find the equivalent length ratio for each pixel, the specimen’s
know distance was used. Line 3 of 440 pixels was equal to 300 mm. Then each pixel in
this thermogram is equal to 1.467 mm. By converting the size of defect UB011 of this
specimen in lines 1 and 2, the calculated defect size is 98.86 × 46.36 mm, representing
with great accuracy the actual size of 100 × 50 mm.
Figure 4.55 Defect sizes measurement in Specimen 1
Chapter Four
178
The defect size and area can also be established and verified using the boundary outline
method. In this method, an area measurement function is used to draw round the
boundaries of the defect. The number of pixels is then calculated and converted to the
corresponding area. The boundary outline method is very useful to calculate defects that
have irregular shapes. Figure 4.56 shows the calculation of the size of defect DB031.
The reference size of the specimen’s 300 mm dimension was taken to determine the
image length/pixel ratio. By converting the 505 pixel length of Line 1, it was found that
each pixel equals 0.59406 mm. That means each 1 pixel square will identify 0.353 mm2.
ROI 2 in Figure 4.56a was set to the debonding defect DB031 via the boundary outline
method. The area of this defect was analyzed at pixel level. The measurement of the
pixel number was 30651 pixels, which is calculated to be 108.17 cm2.
(a) (b)
Figure 4.56 Boundary outline method for defect area measurement- Specimen 3
The defect size measures were exactly the actual size, however, these measurements can
vary for many reasons. The most important factor that affects the defect size
measurement is the selection of the defect boundaries, which depend completely on the
thermographer’s judgment. Figure 4.56b shows the selected boundary in Specimen 3’s
debond defect. As shown in the figure, the decision as to the edge where the analyst can
consider the defect boundary to be located is not an easy task. The other major factor is
the time of the thermogram frame that the analyst selects to calculate the size of the
defect. To have an accurate defect size there is a need to analyse the specific IR image
captured at tmax or immediately after it. Factors include the colour scale of the
Quantitative IRT experimental laboratory program
179
thermogram and the angle between the surface and the IR camera view line can also
play a role. To obtain a very accurate defect shape and size using this method of pixel
area calculation, it is recommended to have an IR test design where the IR detector is
perpendicular to the tested surface. In this way, the error of the angle of view will be
eliminated. However, this option was not always available during the entire IR test
programs
The thermograms of Specimens 1 and 2 illustrate that the sizes of the unbonded defects
under a single CF130 and CF140 layer matching exactly the actual embedded defects’
sizes, as shown in Figures 4.54 and 4.58. As shown in the latter figure, it is very clear
that the resin crosses the designed boundaries of unbonded defect UB021. For that
reason, the size of this defect width was measured precisely at the desired position.
Figure 4.57 Measuring defects in Specimen 1 in mm
50
50
50
50 100
Chapter Four
180
Figure 4.58 defect size of UB021 in mm
Opposite fibres setting is usually used when the structure is strengthened with more than
one layer of CFRP composites. This may lead to reduced ability to read accurately the
defects’ sizes due to heat diffusion caused by the fibres’ opposite alignment. However,
the pixels size readings matched the real defect sizes with good accuracy, even when
double sheets of CFRP were used and attached in opposite fabric directions. Figure 4.59
exhibits the dimensions of the UB081 defect that was retrofitted with double CF140
sheet with opposite fibres direction alignments. In this defect size reading, the
boundaries of the defect were not easy to determine clearly, possibly due to the
difference in the CFRP fabric direction of the two layers in this specimen. Furthermore,
the detection of the size measurements of UB071 and DL072 which were constructed
by attaching 0.55 mm CFRP bi-directional fabric to the top of CF140 fabric was very
accurate, as shown in Figure 4.60.
100 70
Quantitative IRT experimental laboratory program
181
Figure 4.59 Specimen 8 defect sizes in mm
Figure 4.60 Specimen 7 defect measurements in mm
Defect size determination in steel specimens was also precise. Nevertheless, selecting
the best IR image needs a punctual frame analysis to decide the frame with maximum
thermal signal and clear surface temperature distribution. Figure 4.61a illustrates the
five defects in steel Specimen S1. The size pixel reading shows the accurate defect size.
However, the angle of the IR with respect to the surface altered the sizes slightly. For
very accurate size reading of defects, thermograms should be captured perpendicularly.
Defect sizes are influenced significantly by the capture time of thermograms. For steel
specimens, the signal fading rate was high compared to concrete, which allowed a short
time for the IR analyst to read the defect size precisely. Figure 4.61b shows an ROI that
was drawn to collect thermal information on defects UBS11 and UBS14 in Specimen
100
180220
70
Chapter Four
182
S1. Several 3-dimensional profiles of this region of interest were constructed at different
times after the end of the 1 s pulse from 50 cm distance. Both defects read 8 oC thermal
signals immediately after the end of the pulse, as shown in Figure 4.61c, however, the
shapes of the defects were not clear, due to the increase in the defect-free area
neighbouring the defect. After 3.25 s from the pulse end, the signal reduced but the
shape detection increased, as shown in Figure 4.61d. Figure 4.61e shows that when the
ΔT reaches to 4 oC at 4.25 s from pulse end, defect shapes become easier to determine.
After 7 s the signals become around 2 oC, but with clear defect size dimensions for
UBS11 and UBS14.
Quantitative IRT experimental laboratory program
183
(a) Specimen S1 defects (b) ROI of defects UBS11 and UBS14
(c) 3-D profile at t = 0 s after pulse end (d) 3D-profile at t = 3.25 s after pulse end
(e) 3D- profile at t = 4.5 s after pulse end (f) 3-D profile at t = 7 s after pulse end
Figure 4.61 Steel Specimen S1surface temperature profiles at different times
20.0
24.0
28.0
32.0
36.0
40.0
Surface Temperature (oC)
UBS11
UBS14
20.022.024.026.028.030.032.034.036.038.040.0
Surface Temperature (oC)
UBS11
UBS14
20.022.024.026.028.030.032.034.036.038.040.0
Surface Temperature (oC)
UBS11
UBS14
20.021.022.023.024.025.026.027.028.029.030.0
Surface Temperature (oC)
UBS11
UBS14
Chapter Four
184
The measurement of defect and anomaly sizes beneath CFRP laminate is a major
challenge. The 1.4 mm thickness of the CFRP laminate is one of the main reasons why
it is hard to observe these defects. The IRT tests performed on Specimens 5, 9, 16, 17
and S4 prove that the technique is able to measure with high accuracy the different
defects in the CFRP laminate concrete and steel bond zones. Figures 4.62 to 4.65
illustrate these measurements. Defects UB051 and UB052 were calculated with
acceptable accuracy. Figure 4.62 illustrates these two defect sizes and highlights that the
actual defect areas does not match exactly the rectangular areas shown in Figure 3.11-5.
Again, that is due to crossing the epoxy during the application of the CFRP laminate.
Unbonded area UB091 size measurement is shown in Figure 4.63. The size of this
defect was very small and it was located close to the CFRP laminate edge where
excessive epoxy used to attach the laminate can mislead the interpretation of the size
defect reading in the thermogram. However, the reading of the defect dimension was
very accurate.
Figure 4.62 Specimen 5 thermogram measurements in mm
7070
80
Quantitative IRT experimental laboratory program
185
Figure 4.63 Specimen 9 defect size in mm
The artificial defect UB161 was not able to be measured precisely, as shown in Figure
4.64, mainly due to the three layers of the CFRP on top of the defects. The resolution of
the defect boundaries was not as good as the defect size reading under a single CFRP
layer. Defect DL162 in the same specimen was determined with good accuracy
compared to the unbond defect of UB161. This may be attributed to the different
number of layers above each of Specimen 16’s defects.
The defect size under CFRP laminate can be easily misanalysed due to the laminate’s
properties. The size of groove defect GR171 was detected with the wrong
corresponding area. Figure 4.65 shows how the IR image size reading is not identical to
the actual size of the embedded GR171 defect. The groove was guaranteed to be empty
from excessive epoxy. The only interpretation for this wrong size reading at the bottom
of the groove under the CFRP laminate is the groove end in the concrete, which was not
a sharp edge. Figure 4.65c shows the smooth edge at the end of the groove cut in the
concrete surface. This might make the heat transfer faster in this area and lead to the
misreading of the groove size shown in Figure 4.65b.
50
25
Chapter Four
186
Figure 4.64 Specimen 16 defects measurement
(a) (b)
(c)
Figure 4.65 Groove size detection in GR171: (a) the actual size of the groove under the CFRP laminate, (b) the measured detected defect, (c) groove end details at the concrete
surface
Quantitative IRT experimental laboratory program
187
4.5.3.1 Summary of Part 3 experimental program
This part of the quantitative experimental program was designed to answer whether IRT
can determine the precise shape and size of the detected defect. The tests studied
various kinds of artificial implanted defects including unbond, debond, grooves in
concrete surface and delamination. After full IR analysis several conclusions were
reached as follows:
The thermographer’s judgment in the selection of defect boundaries can play a
major role in the defect’s calculated area and shape.
The accurate area and shape of the defect depend considerably on the IR image
capture time, i.e. how many seconds after the end of the pulse the IR image was
captured.
Having the IR detector positioned at a perpendicular angle with respect to the
investigated surface is preferable to help to calculate the size and shape of the
nominated defect very accurately. However, border dimensions of defects can
still be read precisely by means of the proportional method when a perpendicular
IR imager position cannot be achieved.
Defects with ΔTmax values less than 2 oC do not generate well-defined
boundaries. Shape and size cannot be determined accurately in this case.
Increasing the number of the CFRP layers over the defect reduces the ability to
calculate the defect area accurately.
Setting the multi-CFRP fabrics in different fibre directions reduces the ability to
calculate the defect area.
The defect size calculation in CFRP-steel system needs higher IR frame rate
analysis due to the high speed of heat wave fading in steel substrate.
Defect shape and size under laminate CFRP system are harder to calculate than
under the fabric systems.
The technique shows that defect sizes of unbond under multi-CFRP fabric layers
cannot be measured precisely.
The exact size of the groove in the concrete-CFRP laminate bond surface is
undetermined.
Chapter Four
188
4.5.4 Part 4: Excitation system design
The main aim of this experimental program was to investigate the efficiency of using
different excitation systems. Heating tungsten halogen lamps in spot and flood modes
were compared. In addition, an air blower excitation system was investigated. Different
kinds of implanted artificial defects were subjected to these excitation schemes and
observed thermographically using quantitative IR testing. The experimental program in
this part was conducted through 44 IR tests for lamps with different heating shape
functions. The second heating scheme was achieved by applying a total of 39 PTT IRT
runs using an air blower excitation system.
4.5.4.1 Lamps heating modes
The heating tungsten halogen light lamps used in the design as an excitation source to
generate heat waves in the active thermography had maximum capacity of 2000 watts
with varibeam capability. The light beam can vary from spot to flood mode. The spot
mode was utilized for most of the IR tests to generate the surface detection shown in
Figure 2.22c with high thermal responses. However, studies were performed using both
spot and flood modes in this part of the quantitative active thermography program.
As shown in Figure 4.66, the injected heat waves struck the surface with non-
homogenous behaviour. The heat wave was designed to hit the centre of the specimen,
and for that reason the UB051 and UB052 artificial defects in the centre of that
specimen have higher thermal signals compared to the off-centre defects GR053 and
GR054. Moreover, within the same UB051 defect, the area close to the centre of the
specimen (the epicentre of the heat wave) has a higher temperature than the unbonded
areas far from this point, as shown in Figure 4.66c.
Quantitative IRT experimental laboratory program
189
(a) (b)
(c)
Figure 4.66 Thermograms of Specimen 5 (a) before the test, (b) during the heat pulse, and (c) 1s after the heat pulse
The spot mode highlights the maximum response causing it to generate a larger heat
wave within the defect zone, while the flood lighting mode helps the thermographer to
draw the boundaries of the defect clearly. Figure 4.67 illustrates the two types of heating
light modes generated by means of tungsten halogen lamps on Specimen 24. As shown
in Figure 4.67a, the maximum temperature was recorded in the centre of the specimen
surface where the heat wave was designed to strike. This was an advantage for
enhancing the detectability of the unbonding defect. At the same time, it could mislead
the thermographer’s analysis, especially during the location of the ROIs. Choosing a
large ROI with the ability to record the maximum temperature during the IR sequence
run can cause misinterpretation of the location of a defect, particularly with the presence
of small hot spots unrelated to the subsurface defects. However, this challenge
Chapter Four
190
commonly faces the thermographer. For that reason, special care needs to be taken in
the selection and design of the ROI in IR analysis. One of the methods adopted in this
research to overcome this problem was to select a small rectangular ROI in the
investigated unbonded area and to record the average temperature within this small
ROI. The locations of these ROIs were usually selected to be not within the area in the
centre of the specimen where the heat wave can register a very high temperature.
However, this was not always possible, especially when the artificial defect was inserted
in the middle of the specimen.
Heat waves applied in flood distribution documented the defect boundaries clearly. No
obvious nonrelated hot spots misled the detection of the unbonding fault. However, the
thermal signals captured in this mode were much lower than in the spot mode. The flood
distribution of the light beam works perfectly if the investigated area is large, but
examining large areas needs more uniform heat waves.
(a) (b)
Figure 4.67 Specimen 24 after 1 s of pulse (a) using the spot light mode, (b) using the flood light mode
For the unbond defect in Specimen 2 the differences in thermal responses between the
two excitation light modes are marked, as shown in Figure 4.68. The flood mode
maximum signals reduce by 40% over the spot phase at different pulse lengths, mainly
due to the difference in heat intensities and heat distribution of each light mode. The tmax
of both modes for the same pulse duration was the same as shown in Figure 4.68.
Quantitative IRT experimental laboratory program
191
Moreover, the fading time for both mode signals where the signals record around zero
values was very similar.
Figure 4.68 Thermal responses of UB021 in spot- and flood-lighting modes
In the debonding area the difference between these two light-distribution modes
increases. Figures 4.69a and 4.69b reveal the homogeneity of the temperature surface
distribution of defect DB031. As shown, the spot mode concentrates all heat in the
middle of the specimen, while the flood mode distributes the heat uniformly over the
entire surface. The detection for DB031 in Figure 4.69a was easier to determine the
boundaries of the debonding zone than boundary of the same defect that shown in
Figure 4.69b. To find the edges of the defect precisely by means of flood-distributed
heat wave, it is necessary to apply the wave for a medium to long duration.
-2.0
3.0
8.0
13.0
18.0
23.0
28.0
0 10 20 30 40
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
ΔT-UB021-5s at 50cm-spot modeΔT-UB021-5s at 50cm-flood modeΔT-UB021-1s at 50cm-spot modeΔT-UB021-1s at 50cm-flood mode
Chapter Four
192
(a) (b)
Figure 4.69 Specimen 3 during pulse time (a) using the spot-light mode, (b) using the flood-light mode
The difference in the thermal signals recorded for Specimen 3 is shown in Figure 4.70a.
From this figure it can be seen that both light modes have the same thermal signal
pattern (Type A). However, there was a great difference in the detectability. The
enhancement of the maximum signals was more than 160% for different setting of the
IR configuration at 1 s and 5 s from 50 cm. The thermal contrast difference again was
smaller than the differences in the thermal signals. In debonding detection it takes more
time to create higher recognition. Figures 4.70b and 4.70c show the smoothed contrast
behaviour in debonding defect DB031 in spot and flood modes at 1 s and 5 s pulse
durations. The improvement in the detectability of the maximum contrast was 50% by
using the spot mode introduced at 1 s and 5 s durations.
From the comparison of heating schemes, the results show that for concrete
strengthened with CFRP composites, long PTT enhances the detection of defects
generally. The improvement in the thermal signal reading and the analysis of defects in
the concrete-laminate bond surface is appropriate in terms of the total temperature on
the surface. This detection enhancement suggests that long PTT should be utilized in
IRT assessment of concrete structures strengthened with CFRP laminate. Artificial bond
defects in CFRP fabric-concrete composites show high increases in the thermal signals
captured when long PTT is adopted. However, this increase raises the surface
temperature to more than the epoxy glass transition limit. The increase in pulse duration
was found to be more efficient and to assist in the detection process when the long
pulses are applied from far distances. An excitation system tested at 0.5 m showed a
high increase in ΔTmax values for both unbond and debond defects covered with a single
CFRP fabric. This increase in the signals is inappropriate because of the unacceptable
rise in the investigated surface’s temperature. For artificial bond defects in the concrete-
multi CFRP fabric layers, the PTT with long pulses enhances detectability substantially
with an adequate increase in the surface temperature which does not reached Tg limit of
the epoxy.
0.0
5.0
10.0
15.0
20.0
0 10 20 30 40
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
DBS21-0.05Hz
0
5
10
15
20
25
30
35
40
0 20 40 60 80
Ther
mal
sig
nal Δ
T (o C
)
Time (s)
DBS21-0.025Hz
Quantitative IRT experimental laboratory program
239
One of the main advantages of using the long pulse duration heating scheme is that the
increase in the thermal signal of the defect means that the size and shape can be
established easily. The higher signals lead to better defect size and shape determination.
Using the lockin thermography technique, the results show that the ΔTmax in concrete
unbonding defects is raised by increasing the sinusoidal wave cycles. Steel unbond
defects show no evidence of this rise in the thermal signal peak points. In general, at the
end of the cycles the value of signals does not normalize and level off totally. This ΔT
value is decreased by reducing the frequency rate and it is higher in concrete than steel
substrate. Low frequency provides better detection for defect at the same depth.
Debonding defects in both concrete and CFRP fabric systems show very high signals
with the LTT heating scheme. However, is not recommended to apply LTT for
debonding surface defects with air pockets due to the high rise in the surface
temperature over the defect area.
4.5.8 Part 8: Detection of cracks
The final investigation in the quantitative experimental program was to detect cracks in
the concrete surface beneath CFRP applications. Deep spalling was also under
examined in several specimens. Active PTT was used in this study. Figure 4.100 shows
the schematic of the IRT set-up applied to the specimens. The crack defect area in the
concrete surface will appear with different temperatures relative to the defect-free areas
at the surface in the thermal image. However, due to the small sizes of the cracks,
detection was expected to be difficult.
Chapter Four
240
Figure 4.100 Schematic of IRT for crack detection
Cracks of three types were manufactured in concrete specimen surfaces using three
methods: wide straight grooves, fine curved grooves and loading cracks. Wide straight
grooves 3.6 mm wide and 13.2 mm deep were designed in Specimens 10 and 15 to
investigate the ability of IRT to detect cracks under thick multi-CFRP fabrics and
laminates. Figure 4.101a shows Specimen 10's artificial grooves constructed to study
the identification of wide cracks through multi-CF 130 fabric sheets. Fine curved
grooves were produced during the construction of the concrete specimen. During the
making of the concrete specimens, fine plastic sheets were inserted in the mould with
controlled thickness and depth. After the initial concrete setting, the plastic sheets were
removed carefully to prevent any changes in the artificial crack widths. However, all
crack sizes were checked before the application of CFRP. Loading cracks were
generated in specimens 11, 12, and 14 by three points loading. Loading cracks were
closer to the crack sizes that can occur in real life situations. Figure 4.101b reveals
CR141 and CR142 loading cracks generated in Specimen 14 before attaching the CFRP
sheet.
Quantitative IRT experimental laboratory program
241
(a) Specimen 10
(b) Specimen 14 before CF130 fabric application
Figure 4.101 Artificial crack generation
Two lines were chosen as ROI to reveal the thermal results of IR analysis of Specimen
10's artificial cracks. Figure 4.102a shows the location of these ROIs. They were chosen
to be away from the specimen’s centre to avoid the irrelevant increase in the
temperature within the ROI line profile caused by the pulse hitting the centre of the
specimen. CR101 and CR102 were covered with a single sheet of CF130 fabric, while
double CF130 sheets were attached to cracks CR103 and CR104. The cracks under a
single fabric sheet were very detectable from 50 cm and 70 cm and for all pulse
durations, as shown in Figures 4.102b to 4.102g. As expected, by increasing the
distance and reducing the pulse duration, crack detection was weakened. IR analysis of
Chapter Four
242
pulses applied from 1 m and 1.2 m are present in Appendix B. Figures 4.102b and
4.102e highlight the extent to which surface temperature can be affected by changing
the lamp position by 20 cm. The temperature detected on cracks dropped more than 10 oC when the lamp location moved from 50 cm to 70 cm. The lamp distance or the input
heat flux were expected to be more crucial parameters when using IRT to investigate
finer cracks. For 5 s pulse intervals, theCR102 crack shows a slightly higher
temperature compared to CR101. It is true that both cracks have exactly the same
dimensions and their width is identical, but CR102 was designed to be 20 mm closer to
the centre of the specimen where the pulse heat was planned to strike, as shown in
Figure 3.11-10. That made the received heat at CR102 greater than at CR101 and
caused the difference in surface temperature shown in Figure 4.102b. For pulses with 3
s and 1 s periods the effect of non-identical alignment for these two cracks was
negligible. This provides an interesting guideline for thermographers, they cannot
compare two defect areas (even if both have the same dimensions) unless many
conditions apply including the location of the target of the pulse wave.
(a) ROIs in IR image
CR101 CR102
CR103 CR104
Quantitative IRT experimental laboratory program
243
(b) At 5 s from 50cm (c) At 3 s from 50cm
(d) At 1 s from 50cm (e) At 5 s from 70cm
(f) At 3 s from 70cm (g) At 1 s from 70cm
Figure 4.102 Cracks CR101 and CR102 profile trends
Cracks CR103 and CR104 were covered with two CF130 fabric layers. Pulses with 5 s
and 3 s from 50 cm and 70 cm were able to generate identifiable temperatures
differences on these cracks, as demonstrated in Figure 4.103. However, these
temperature differences were small and faded faster compared to CR101 and CR102.
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Single CF130-pixels
CR102CR101
Chapter Four
244
IRT analysis of pulses from 1 m and 1.2 m are present in Appendix B. Thermal signals
were not reliable for pulses from 1 m and 1.2 m distances.
(a) At 5 s from 50cm (b) At 3 s from 50cm
(c) At 1 s from 50cm (d) At 5 s from 70cm
(e) At 3 s from 70cm (f) At 1 s from 70cm
Figure 4.103 Cracks CR103 and CR104 profile trends
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels
CR103CR104
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels
CR103CR104
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels 0
510
1520
25
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels
CR103CR104
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels
CR103CR104
05
1015
2025
20.0
30.0
40.0
50.0
1
51
101
151
201Time (s)
Surface Temperature (oC)
ROI - Double CF130-pixels
Quantitative IRT experimental laboratory program
245
Cracks under CFRP laminates, even wide cracks of 3.6 mm, were unable to provide
acceptable thermal signals. Figure 4.104a reveals the thermal signals of artificial cracks
CR153 and CR155 under laminate composite in Specimen 15. The results of this figure
illustrate that the maximum crack thermal signal that can be detected in CR153 is about
1.8 oC for the FRP combination of CF140 and laminate when the lamp is mounted at 0.5
m. CR155 IRT with 5 s pulse and from 50 cm provides a maximum thermal signal just
above 2 oC. Both of these values are considered too small to recognize defects. From the
results, it can be concluded that fine cracks under laminate CFRP are hard to detect.
Due to the good length of cracks in general, the thermographer can sometimes evaluate
potential cracks visually from IR images even with small thermal signals. For example,
CR155 can be seen in the thermogram in Figure 4.104b. However, this identification is
dependent on the colour temperature scale used in the IRT analysis.
(a) Thermal signals
0.0
0.5
1.0
1.5
2.0
2.5
0 10 20 30 40 50 60
Ther
mal
Sig
nal Δ
T (o C
)
Time (s)
CR153 at 5s from 50 cm
CR155 at 5s from 50 cm
Chapter Four
246
(b) Thermal image
Figure 4.104 Cracks in Specimen 15
The IR results of Specimen 25 reveal a number of imperfections in the bonding that can
be read from the temperature distribution on the surface of interest. The irregularity of
the hotspot areas in the thermogram shown in Figure 4.105, may be due to the rough
surface preparation and imperfections in the CFRP installation. Detection was unrelated
to crack location and size. The results of Specimen 25 do not show the real values of
thermal signals. The rough surface preparation of the concrete before the application of
CFRP sheet can cause many small point hotspots in the thermograms and lead to
misinterpretation of the defect's location and size.
Figure 4.105 Specimen 25 IR image
CR155
CR153
Quantitative IRT experimental laboratory program
247
Figure 4.106 shows the surface temperature 3-D profile of the ROI line designed to
investigate cracks in Specimen 12. The ROI in the specimen thermograms is shown in
Figure 4.106a. The measured width of the loading CR121 crack was 0.4 mm in its
narrowest part; however, it did not have the same width over the entire length of the
crack. The IR image in this figure shows that the crack size was wider than 0.4 in the
middle of the specimen, although the ROI was chosen to be in an area were the crack
has the minimum width of 0.4 mm. Figures 4.106b to 4.106g demonstrate the ROI
temperature profile for different pulse length durations and from different lamp
locations. From the IR results in Figure 4.106b to 4.106d, the differences between
detected temperatures over ROI1 for pulses of 5 s , 3 s, and 1 s lengths and from half a
metre distance can be seen. From this lamp distance pulses of 3 s and more can provide
good detectability of this size crack for about 5 s after the end of the pulse. Pulses of 1 s
show poor capability to identify the CR121 defect.
The good detectability when applying the 5 s pulse from 50 cm is reduced when the
lamp is positioned further away. The difference in temperature of CR121 and the
surrounding defect- free area reduces considerably by more than 10 oC when the lamp
location is shifted from 50 cm to 70 cm. This shows that the recognition of fine cracks is
very much dependent on the pulse amount and duration. Pulses with 3 s and less could
not reveal the crack clearly when the lamps were mounted more than 50 cm away,
while pulses of 5 s can cause recognizable differences in over crack temperatures from
1.2 m. The signal is extended differently for each different pulse length. In general,
longer pulse length generates a longer thermal signal. All pulse ranges create short
detection times in IRT investigation, when none of the pulses and/or lamp distance
designs experience signals readable for more than 10 s, as illustrated in Figure 4.106b to
4.106g. The crack size detected in thermograms was 0.8 mm. However, the thermal
signal responses were extended for no longer than 5 s after the pulse end. The short
period of the signal might force the analyst to minimize the time for frame analysis.
Figure 4.106d shows that even for 1 s pulse duration and 50 cm lamp position, the
technique is able to detect this fine crack, but with a very small thermal signal value.
Pulses from that distance with longer time periods show higher signal values, as shown
in Figures 4.106b and 4.106c. As revealed in Figures 4.106f and 4.106g, for this crack
size 5 s pulses provide inappropriate thermal signals when the lamp is placed further
Chapter Four
248
than 1 m. Pulses of less than 3 s and applied from further than 70 cm show no good
thermal responses for this crack.
(a) Location of ROI 1 in the Specimen 12
(b) At 5 s from 50cm (c) At 3 s from 50cm
(d) At 1 s from 50cm (e) At 5 s from 70cm
CR121
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
Quantitative IRT experimental laboratory program
249
(f) At 5 s from 100cm (g) At 5 s from 120cm
Figure 4.106 ROI thermal data in CR121 crack
Crack CR141 in Specimen 14 shows a similar surface temperature response to CR121,
because the ROI were positioned on crack CR141 where it was 0.8 mm wide, and
CFR121 was generated with the same width size. The comparison of the surface
temperature behaviours of these two cracks, as can be seen from Figures 4.106 and
4.107, in terms of maximum temperature and length of the signal lead to the conclusion
that they also have the same depth besides their identical width. The CR142 crack with
width of 0.4 in this specimen was undetectable in all pulse designs, as shown in Figure
4.107.
(a) At 5 s from 50cm (b) At 5 s from 70cm
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
0 5 10 15 20 25
20.0
30.0
40.0
50.0
151
101151
201251
301
351
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR121
05
1015
2025
20.0
30.0
40.0
50.0
60.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142
CR141
05
1015
2025
20.0
30.0
40.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142CR141
Chapter Four
250
(c) At 5 s from 100cm (d) At 5 s from 120cm
(e) At 3 s from 50cm (f) At 1 s from 50cm
Figure 4.107 ROI thermal data of Specimen 14
Generally the heat wave should be designed to strike perpendicularly the centre of the
surface of interest to provide as homogenous a temperature distribution as possible.
However, different angles of heat waves were tested to study if they can improve crack
detection. The best IR recognition in terms of crack patterns and sizes was when the
heat wave hit the surface of interest off-centre and at a 60o angle to the specimen's
surface. Figure 4.108 shows the schematic of the IRT configuration to enhance crack
identification.
05
1015
2025
20.0
30.0
40.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142CR141
05
1015
2025
20.0
30.0
40.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142CR141
05
1015
2025
20.0
30.0
40.0
50.0
60.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142
CR141
05
1015
2025
20.0
30.0
40.0
50.0
60.0
1
51
101
Time (s)
Surface Temperature (oC)
ROI 1-pixels
CR142
CR141
Quantitative IRT experimental laboratory program
251
Figure 4.108 IRT configuration to improve crack detection
An IRT inspection was conducted of cracked reinforced concrete Specimen 11
strengthened with two strips of single CFRP MBrace CF130. Concrete cracks were
observed in the IR images recorded. The thermogram in Figure 4.109 shows the cracks
in concrete that divided the specimens into three slices with different temperatures. A
hot strip was observed at the middle between the two major cracks in the
CFRP/concrete specimen. This may be related to the crack depth which met the
reinforcement mesh and caused spalling in the concrete middle strip. As shown in
Figure 4.109, if the crack is moderately deep, it may act as an obstacle to the heat flow
reaching the areas far from the external heat source. In Specimen 11 the heating sources
were directed towards the specimen’s surface at an angle of 60 degrees to the horizontal
level at the top and the bottom edge of the specimen. Figure 4.109a shows that the
cracks generated from loading were deep enough to form spalls in the concrete and to
put a stop to the heat transfer in this specimen. The thermogram shows that the middle
slice had a 2.7 oC temperature difference from the neighbouring areas. A 3-D surface
temperature profile was produced to enhance the cracked area in this specimen, as
shown in Figure 4.109b. The spike in the temperature profile at one edge of the
specimen is due to the angled position of the heating source. Spall in concrete was easy
to detect due to the hot spot area formed in the entire concrete segment that fractured
from the concrete surface. The IRT was unable to evaluate the severity of the spall in
general. The middle spall between CR111 and CR112 was fixed within the concrete
specimen. Spalls Specimens 22 and 23 were unidentifiable by IRT techniques. PTT and
long PTT were applied to these specimens to investigate the capability of IRT to locate
Chapter Four
252
spalling in deep concrete. However, none of these techniques was suitable to produce a
recognizable thermal response, possible because concrete's thermal properties can easily
dampen the heat wave.
(a) Thermogram (b) 3D profile
Figure 4.109 Specimen 11 thermal results
Measurement of the cracks was also conducted in this part of study. Major cracks like
CR102 were detected and measured with very high accuracy. However, to measure that
crack it was essential to position the IR detector perpendicular to the investigated
crack's surface, as shown in Figure 4.110a. The line ROI above the crack shows a value
of 3.7 after pixels conversion. The error measurement reading was less than 0.1 mm
which is very good. The crack size in Specimen 12 was too fine to be measured with
this thermogram pixel resolution. Fine cracks of 1 mm and less can show inaccurate size
readings. The crack in Figure 4.110b was 0.8 mm wide; however, the IR image size
reading showed that the crack width was 0.9 mm. This error in measurement may be
due to different reasons, but mainly to the pixel resolution which was not sufficient to
represent this small size. Crack CR111was also too small to be measured accurately.
The cracks in Specimen 11 generated a spall in the concrete. In such cases the crack will
usually be very hard to measure. The location of the crack in this instance is very
detectable, but the measurement of its size is not possible.
Quantitative IRT experimental laboratory program
253
(a) CR102
(b) CR121
(c) CR111
Figure 4.110 Crack measurement from thermograms
4.5.8.1 Summary and findings
The results of an experimental study have been presented in this section to investigate
the ability of IRT NDT to detect and measure cracks between CFRP fabrics and
Chapter Four
254
concrete specimens. PTT was adopted. The experiments show that the technique is
capable of detecting the locations and sizes of major cracks quite adequately, and the
sizes and shapes of cracks up to 0.8 mm can be identified with high accuracy. The
detection and measurement of cracks in the CFRP concrete bond zone are significantly
dependent on the pulse interval and the distance between the external heat source and
the surface of interest.
4.6 Guidelines for quantitative IRT NDT
The data collected from the results are not sufficient for the development of a
mathematical relationship for thermal signal maximum values as a function of pulse
interval, CFRP material type (laminate or fabric, or type of fabric weave), and CFRP
layers for the different defects investigated. However, the data provide information
about the input pulse durations that need to allocated for each defect type and for
different CFRP composites. The following points are guidelines to help thermographers
to perform IRT PTT.
It is essential for theromgraphers to avoid performing IRT NDT in dusty
environments, as the solid particles suspended in the medium have grey body
performance.
Thermographers should mover the IR imager device until they obtain the best IR
view and angle that show the minimum reflection on the investigated surface.
It is recommended to conduct PTT IRT with short pulse lengths (1 s) for general
scanning and once the discontinuity regions are detected, a full PTT IRT with
appropriate pulse intensity and duration is recommended for deep inspection.
The flood mode of heating is recommended when a large area is under
evaluation, or it can be used as a first IR test in advance of a second detailed test
with spot mode to indicate the areas that need more investigation.
The pulse duration length and lamp distance should be designed according to the
type of CFRP application. For example, for single-layer CFRP fabric, even 1 s
can detect unbonding or debonding in the concrete or steel bonding zone. Table
4.9 shows proposed guidelines for minimum pulse durations for each lamp
Quantitative IRT experimental laboratory program
255
distance for all CFRP applications and combinations tested in this quantitative
research.
The experimental results show that the minimum heat flux intensity that should
be provided to generate the minimum thermal signal when the excitation lamp is
located at 1.2 m from the test object is 500 W/m2.
The IR detector should be positioned at a fixed distance during the test. This
distance should be designed with respect to the potential defect size. Small sizes
need closer IR images to determine the actual size of the defect with respect to
the field of view of the IR camera.
Isolating shutters should be used during IR testing to eliminate undesirable
radiation from the excitation source after it is turned off.
The probability of background radiation reflection is increased for low
emissivity materials and if the test surface is not a plane. The thermographer
needs to take these factors into account in field tests.
From the IR results, a 2 oC minimum is a reasonable value for a thermal signal
to detect an anomaly or defect. With this value of the signal, the size and the
shape of the defect can be characterized adequately.
It is recommended to apply pulses with an intensity that ensures a rise in the
investigated surface’s temperature compared to the background to alleviate the
effects of undesired reflection from objects surrounding the IRT test scene.
The results of the IR quantitative tests can help to provide pulses designs for
different substrates and different CFRP composites. The pulse design guidelines,
shown in Table 4.9, are proposed thermal pulse inputs that can be considered
when conducting a quantitative PTT IRT NDT.
To minimize the influences of unwanted emission from surrounding objects, it is
recommended to heat the investigated surface to a temperature 10 oC higher than
the objects in the background.
To provide good detection of water it is necessary to supply a high pulse for a
good length of time. Long PTT is recommended.
The guidelines categorize pulses mainly according to defect type, CFRP system under
test and substrate material. The 4th column in the table represents the excitation lamp’s
distance from the surface investigated. The recommended pulse interval range is
Chapter Four
256
provided in the last column. These proposed pulse duration ranges offer an upper and
lower boundary of pulse duration for each distance of the lamp to detect all bond defects
in the CFRP-structures investigated in this study.
Table 4.9 IR recommended thermal inputs for different CFRP composites
Defect type CFRP system Substrate
material
Lamp distance
(cm)
Recommended
range pulse length
(s)
Unbonding Single fabric CF130 Concrete
50 1 – 3
70 1 – 3
100 1 – 3
120 1 – 3
Unbonding Single fabric CF140 Concrete
50 1 – 3
70 1 – 3
100 1 – 3
120 3 – 5
Unbonding Double fabric CF140 Concrete
50 3 – 5
70 3 – 5
100 3 – 5
120 >5
Unbonding Single laminate Concrete
50 1 – 3
70 3 – 5
100 >5
120 >5
Unbonding Double laminate Concrete
50 3 – 5
70 3 – 5
100 >5
120 >5
Unbonding Single fabric and single
laminate combination Concrete
50 1 – 3
70 3 – 5
100 3 – 5
120 >5
Unbonding Single fabric and double
laminate combination Concrete
50 3 – 5
70 >5
100 >5
120 >5
Quantitative IRT experimental laboratory program
257
Unbonding Single fabric CF130 Steel
50 1 – 3
70 1 – 3
100 1 – 3
120 3 – 5
Unbonding Single laminate Steel
50 3 – 5
70 3 – 5
100 >5
120 >5
Debonding Single fabric CF130 Concrete
50 1 – 3
70 1 – 3
100 3 – 5
120 3 – 5
Debonding Single fabric CF140 Concrete
50 1 – 3
70 1 – 3
100 3 – 5
120 3 – 5
Debonding Single fabric CF130 Steel
50 1 – 3
70 1 – 3
100 3 – 5
120 3 – 5
Debonding Single fabric 45 bi-
directional Concrete
50 1 – 3
70 1 – 3
100 1 – 3
120 3 – 5
Delamination laminate Concrete
50 1 – 3
70 3 – 5
100 >5
120 >5
Delamination Fabric CF140 Concrete
50 1 – 3
70 1 – 3
100 3 – 5
120 >5
Delamination Fabric 45 bi-directional Concrete
50 1 – 3
70 1 – 3
100 3 – 5
120 >5
Chapter Four
258
Night-time is the best time to conduct an IR test in the field, because unwanted
reflection radiation that might come from objects surrounding the investigated surface
will be minimized. However, it is sometimes very difficult to eliminate the radiations
from surrounding objects in the field. In this case, the effect of the surrounding objects
should be taken into consideration during the IR analysis of the recorded images. There
is no signal standard that can be applied, and normally it depends on the object's
temperature and emissivity.
With all the above guidelines there still remain limited specifications and studies for the
applications of IRT in the field conditions, and site conditions play a pivot role in IR
readings. It is obvious that the temperature at the time of IR testing affects the
temperatures of surfaces under test. Cloudy skies, high winds and surface moisture also
affect the radiation recorded by the IR decoder.
Numerical analysis
259
5 CHAPTER FIVE: NUMERICAL ANALYSIS
5.1 Introduction
The numerical analysis of IRT NDT for testing concrete specimens strengthened
externally with CFRP fabric and laminates was the second component of the research
program. This chapter presents the outputs of using the finite element method (FEM) as
an analytical tool to simulate, investigate and study different parameters that affect the
thermal detection of different defects. The numerical modeling and parametric studies
were used to predict IR results and evaluate potential IR test procedures. Different
laboratory circumstances and testing scenarios were applied in the FEM analysis.
Numerical analyses were used to study the influence of several different factors. Single
parameter studies were conducted using FEM. Models of bond defects were mimicked
in the simulation FEM analyses for defects covered with single and double CFRP
fabrics. Different parameters, including the thermal properties of different materials,
layer thicknesses and thermal input loads, were investigated.
5.2 FEM studies of bond defects in single CFRP fabric
5.2.1 Modeling
5.2.1.1 Geometry
Extensive parametric studies involving FEM analyses were conducted. The modeling
involved a study of different parameters that affect the detection of bond defect in
concrete-CFRP system. All the analytical simulations presented in this study were
executed using FE software ANSYS 13.
Concrete Specimen 2 with a single CF140 fabric sheet was used. The artificial defect in
this specimen was in the form of an unbonded strip at the middle of the bond zone
between the substrate structure and the CFRP composite 70 mm wide along the
specimen length, as shown in Figure 3.11-2. A full 3-D model was constructed to
simulate this specimen. The concrete dimensions were 300 mm wide, 300 mm length,
Chapter Five
260
and 50 mm depth. The single carbon fibre sheet was CF140 0.25 mm thick. The epoxy
resin layer was MBrace saturant 0.9 mm thick. The thermal properties and materials
densities used in the modeling are shown in Table 5.1. The concrete material properties
assigned to model the FE simulation substrate structure were the same properties used
to construct this specimen in the laboratory. The carbon fabric thermal properties were
as shown in Table 5.1, were estimated from data sheets provide by the CFRP
manufacturer (MBrace). The thermal properties of air were assigned to model the
unbond defect, adopted from the ANSYS materials library. The air void was presented
at the defect location between the concrete and the CFRP fabric.
FE simulations were set up to investigate the effect of changing the CFRP conductivity.
Table 5.15 summarizes the simulation results from runs 272 to 282which analyzed the
conductivity variation from 6 W/m.oC to 16 W/m.oC at 5 s pulse duration. The influence
was very small with less than 1 % for the entire range of variation.
The results indicate that the maximum thermal signals on the CFRP surface are
decreased slightly by the increase in the thermal CFRP conductivity factor in a linear
trend with the 5 s pulse. There is no change in tmax values over the investigated
conductivity range. Similarly to the single CFRP conductivity investigation, that small
influence of changing the CFRP thermal conductivity over the thermal signal was due to
the small thickness of the CFRP layers. A comparison of the changes in the thermal
signals of CFRP conductivity in single and double CFRP systems reveals that the
maximum thermal signals is increased by the increase of the conductivity contrary to
the single CFRP system for the same pulse interval. This is mainly due to the effect of
the additional CFRP layer and its epoxy resin which raiser the heat to travel less easily
than above the defect in the single CFRP.
Table 5.15 Double CFRP conductivity simulations 272 to 282
Run
#
Pulse interval
(s)
Conductivity
(W/m.oC) ΔTmax (oC)
272 5 6 7.589
273 5 7 7.594
274 5 8 7.6
275 5 9.38 7.609
276 5 10 7.614
277 5 11 7.62
278 5 12 7.628
279 5 13 7.634
280 5 14 7.64
281 5 15 7.646
282 5 16 7.652
Chapter Five
304
5.3.3.2 Influence of epoxy resin material thermal properties
Changes in the specific heat of the epoxy layers beneath the two CFRP fabric sheets are
presented in the simulation analyses from 283 to 289. Table 5.16 and Figure 5.21 show
the results of these simulation runs. The epoxy specific heat varied in these runs from
1600 J/kg.oC to 1900 J/kg.oC. From the results, it can be seen that the maximum thermal
signal is decreased linearly by the increase of the epoxy specific heat. Figure 5.21b
compares the changing rates in the thermal signal of single and double CFRP layers. It
can be seen from this figure that the influence of changing epoxy properties is higher in
the double system compared to the single system due to the increase in the number of
epoxy layers. The rate slope is also changed for the same reason, as the epoxy layer
above the defect changes the thermal signal slope rate. As shown in Figure 5.21, the
maximum thermal signal reduce linearly with the increase of epoxy specific heat. The
maximum change was about 6.26 % (with less than 0.7 oC) for epoxy specific heat
greater than 1900 J/kg.oC. The time for maximum thermal signal was fixed at 6.85 s and
not affected by the change of the epoxy specific heat.
Table 5.16 Epoxy specific heat simulations 283 to 289
Run
#
Pulse interval
(s)
Specific heat
(J/kg.oC) ΔTmax (oC)
283 5 1600 7.872
284 5 1650 7.739
285 5 1700 7.609
286 5 1750 7.484
287 5 1800 7.363
288 5 1850 7.245
289 5 1900 7.132
Numerical analysis
305
(a) (b)
Figure 5.21 (a) Maximum thermal signals versus different specific heats of epoxy, (b) Changing rates for both single and double layers of CFRP
Similarly to the CFRP conductivity study of the single CFRP sheet, FE simulations 290
to 295 were conducted to examine the effects of changing the conductivity of epoxy
over the range from 0.17 W/m.oC to 0.22 W/m.oC. The results of these simulation runs
are presented in Table 5.17. The maximum change in ΔTmax was 3.7 %. However, the
change in temperature was slight at less than 1 oC. The change in the epoxy conductivity
leads the surface temperature to rise in the defect-free area, which causes an increase in
the thermal signal. In the CFRP double system, by comparing the changes in ΔTmax due
to changes in CFRP and epoxy conductivities, it can be seen that the effect of modifying
epoxy conductivity is slightly higher than changing the CFRP conductivity. The time
for the maximum thermal signal was not affected by the change of the epoxy
conductivity values.
y = -0.0025x + 11.81R² = 0.9992
7
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
8
1500 1600 1700 1800 1900 2000
ΔT m
ax(o C
)
Epoxy specific heat (J/(kg.oC))
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
1500 1600 1700 1800 1900 2000
Cha
nge
in Δ
T max
(%)
Epoxy specific heat (J/(kg.oC))
Single CFRPDouble CFRP
Chapter Five
306
Table 5.17 Epoxy conductivity simulations 290 to 295
Run
#
Pulse interval
(s)
Conductivity
(W/m.oC) ΔTmax (oC)
290 5 0.17 7.406
291 5 0.18 7.509
292 5 0.19 7.609
293 5 0.2 7.706
294 5 0.21 7.801
295 5 0.22 7.892
5.3.3.3 Influence of concrete substrate material thermal properties
Studies of the effect of changing the substrate concrete specific heat on the thermal
signal were carried out in runs 296 to 302. Similarly to the concrete investigations in
Parametric Study 2, the concrete specific heat varied from concrete stone specific heat
at 76 J/kg.oC to the light concrete at 1000 J/kg.oC. Table 5.18 illustrates these
simulation results. The results show that changing the concrete specific heat has very
slight influence on the detected thermal responses with less than 0.5 oC difference over
the entire range. These small changes in the maximum thermal signal were showed a
linear trend. The time for the maximum thermal signal was not influenced by change of
the concrete specific heat.
Table 5.18 Concrete specific heat simulations 296 to 302
Run
#
Pulse interval
(s)
Specific heat
(J/kg.oC) ΔTmax (oC)
296 5 760 7.602
297 5 800 7.609
298 5 840 7.617
299 5 880 7.624
300 5 920 7.63
301 5 960 7.636
302 5 1000 7.642
Numerical analysis
307
Studies of the conductivity of concrete were conducted over a range from 1.3 W/m.oC to
1.8 W/m.oC. Simulation runs from 303 to 308 were conducted to investigate the effects
of changing the concrete conductivity factor, and the results of these simulation runs are
exhibited in Table 5.19. The effect of the change is very small at less than 0.02 oC for
the entire range of conductivities studied. The tmax shows no change for all different
concrete conductivities for the same heating pulse duration.
Table 5.19 Concrete conductivity simulations 303 to 308
Run
#
Pulse interval
(s)
Conductivity
(W/m.oC) ΔTmax (oC)
303 5 1.3 7.604
304 5 1.4 7.607
305 5 1.5 7.609
306 5 1.6 7.612
307 5 1.7 7.614
308 5 1.8 7.617
5.3.4 Parametric Study 7: Thickness of materials
This study highlighted the effects of the change in layer thicknesses of CFRP fabric,
epoxy and concrete. The study was subdivided in three run-sets to study the influence of
changing thicknesses of CFRP, epoxy and concrete. For all sets, the thermal input heat
flux intensity was fixed at 1055 W/m2 at 5 s pulse length.
5.3.4.1 CFRP layer thickness
These studies focused on the range from 0.25 mm to 0.55 mm. Both CFRP sheets
covering the defect were changed together, meaning that if the first layer was 0.3 mm
then the 2nd layer had the same thickness of 0.3 mm. During the seven simulation runs
the thicknesses of the epoxy layers and concrete substrate were fixed at 0.5 mm and 50
mm respectively. Table 5.20 illustrates the effects of changing CFRP thicknesses on the
thermal signals. The maximum thermal signal decreases by the increase in the CFRP
Chapter Five
308
layers thicknesses. The maximum thermal signal detectability deteriorates down to 36 %
when the CFRP thickness is increased to 0.55 mm at of 4.8 oC.
The thicker fabric layers of CFRP show smaller ΔTmax in a nonlinear trend, as shown in
Figure 5.22a. From the result shown in Figures 5.22b and 5.22c, the time for the
maximum thermal signal is increased linearly by increasing of CFRP thickness. The rate
of tmax change increases by the increase of the CFRP layers, as shown by a comparison
of Figures 5.13b and 5.22c. The rate of Δtmax was increased by 0.171 s per 0.1 mm and
0.2 s per 0.1 mm for the CFRP single and double sheets respectively.
Table 5.20 Double CFRP thickness simulations 309 to 315
Run
#
Pulse interval
(s)
CFRP fabric
thickness (mm) ΔTmax (oC) Change (%)
309 5 0.25 7.60 0
310 5 0.3 7.04 -7.4
311 5 0.35 6.46 -15.0
312 5 0.4 5.97 -21.5
313 5 0.45 5.54 -27.0
314 5 0.5 5.18 -31.9
315 5 0.55 4.86 -36.1
(a) (b)
y = 11.586x2 - 18.468x + 11.514R² = 0.9998
2
3
4
5
6
7
8
9
10
0.2 0.3 0.4 0.5 0.6
ΔT m
ax(o C
)
CFRP thickness (mm)
0
1
2
3
4
5
6
7
8
9
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Ther
mal
Sig
nal ∆
T (o C
)
Time (s)
0.25 mm0.3 mm0.35 mm0.4 mm0.45 mm0.5 mm0.55 mm
Numerical analysis
309
(c)
Figure 5.22 Double CFRP layers simulation (a) Maximum thermal signal versus CFRP thicknesses; (b) Thermal signals versus time; (c) Time of maximum thermal signals
5.3.4.2 Epoxy layer thickness
Analyses of simulations were performed to examine the influence of change in the
epoxy thickness layer on the thermal signal detected under two CFRP layers. The
thickness of epoxy varied from 0.3 mm to 1.5 mm. Pulses of 5 s of 1055.56 W/m2 were
applied to the top of the 2nd CFRP sheet. Table 5.21 illustrates the results of changing
epoxy thickness in the 1st CFRP-concrete bond zone and in the bond surface between
the 1st and the 2nd CFRP fabrics layers. The results show that, by increasing the epoxy
resin layer thickness, the maximum signal is decreased. Similar to the results of the
single CFRP layer system, changing the epoxy thickness has less influence than
changing the CFRP thickness. Simulation 316 shows that the narrower resin layer helps
to present higher ΔTmax in a sharp non-linear trend, as shown in Figure 5.23. By
increasing the epoxy thickness to 1 mm and more, the change in ΔTmax becomes
negligible at less than 1 oC, as shown in runs 319 to 322 in Table 5.21. The signal
reached only 4 oC at the 1.5 mm thickness of epoxy.
y = 2x + 6.35
6.6
6.7
6.8
6.9
7
7.1
7.2
7.3
7.4
7.5
0.1 0.2 0.3 0.4 0.5 0.6t m
ax(s
)
CFRP thickness (mm)
Chapter Five
310
Table 5.21 Epoxy thickness simulations 316 to 322
Run
#
Pulse interval
(s)
Epoxy thickness
(mm) ΔTmax (oC) Change (%)
316 5 0.3 11.19 47.1
317 5 0.5 7.60 0
318 5 0.7 5.80 -23.6
319 5 0.9 4.87 -35.9
320 5 1.1 4.38 -42.3
321 5 1.3 4.23 -44.3
322 5 1.5 4.07 -46.3
Figure 5.23 Maximum thermal signal versus epoxy thickness
5.3.4.3 Concrete layer thickness
Runs from 323 to 326 were designed to analyze the influence of changing the concrete
substrate thickness. The thickness of concrete varied in these runs from 30 mm to 200
mm. ΔTmax showed negligible change when the concrete thickness varied, whilst tmax
showed no change at all. Table 5.22 shows that the percentage change in the maximum
detected thermal signal was approximately 0.1 % when the concrete was reduced to 30
y = 7.8298x2 - 19.366x + 15.862R² = 0.9766
0
2
4
6
8
10
12
0.2 0.5 0.8 1.1 1.4 1.7
ΔT m
ax(o C
)
Epoxy thickness (mm)
Numerical analysis
311
mm. The results of this analysis emphasize the minor effect of concrete thickness on the
thermal signal detected and confirm the reliability of the adiabatic boundary conditions
assumed in all parametric studies presented in this chapter.
Table 5.22 Concrete thickness simulations 323 to 326
Run
#
Pulse interval
(s)
Concrete
thickness (mm) ΔTmax (oC)
323 5 30 7.597
324 5 50 7.609
325 5 100 7.614
326 5 200 7.601
5.3.5 Parametric Study 8: Thermal loads and periods
In this parametric study, simulations with different intensity pulses applied to the top
surface of the 2nd CFRP fabric were analyzed. The same modeling sizes, thermal
properties, thermal boundaries conditions and cooling that applied in the previous
studies were used in this study. The effect of changing the heat flux intensity was
studied in simulation runs 327 to 341, and the results are presented in Table 5.23. Pulses
of 5 s and different heat flux intensities from 444 W/m2 to 2000 W/m2 were applied.
The results shown in Figure 5.24a indicate that the maximum thermal signal increases
linearly with the increasing the heat applied to the specimen. The rate of increase in the
double CFRP system was much smaller than the rate of increase in the single fabric for
the same thermal inputs. A comparison of Figures 5.15 and 5.24a shows this difference.
The time for maximum thermal signal is independent of the injected heat wave as it is
not affected by changing the value of the input heat wave intensity within the same
pulse interval. For all curves of different thermal loads in Figure 5.24ba the tmax
remained at 6.85 s.
Chapter Five
312
Table 5.23 Thermal load simulations 327 to 341
Run #
Pulse
interval
(s)
Input heat flux
(W/m2)
Input heat
flux (W)
ΔTmax
(oC)
327 5 444.44 40 3.2
328 5 555.55 50 4.0
329 5 666.66 60 4.8
330 5 777.77 70 5.6
331 5 888.88 80 6.4
332 5 1000 90 7.2
333 5 1111.11 100 8.0
334 5 1222.22 110 8.8
335 5 1333.33 120 9.6
336 5 1444.44 130 10.4
337 5 1555.55 140 11.2
338 5 1666.66 150 12.0
339 5 1777.77 160 12.8
340 5 1888.88 170 13.6
341 5 2000 180 14.4
Numerical analysis
313
(a) (b)
Figure 5.24 (a) Thermal signal versus input heat flux; (b) Thermal signal versus time of different input heat flux
5.3.6 Summary and findings
The investigations described in Section 5.3 focused on studying the different potential
parameters that may affect the thermal responses of bond defects covered with double
CFRP layers during IRT testing. Detection can be represented in different parameters,
however, the most important thermal response feature that represents the detectability
level is the maximum thermal signal on the investigated surface of the defect area and
the time for that thermal signal. A bonding defect under double CFRP layers was
modeled and investigated. Different parameters were investigated after the results were
verified first by the corresponding thermal responses from the experimental program. It
was noticed that pulses with durations of 1 s and 3 s generate thermal signals with small
values for defects under double CFRP sheets. For that reason, pulses with 5 s only were
applied in these studies.
The 5th parametric study involved the verification of the simulation and experimental
thermal results of unbond defects under a double CFRP CF140 fabric. The results of the
simulated model were very close to the experimental results for all imposed pulse
duration phases. The difference between the experimental and the simulated maximum