NDT XT

NONDESTRUCTIVE
TESTING METHODS AND

NEW APPLICATIONS
Edited by Mohammed Omar
Nondestructive Testing Methods and New Applications

Edited by Mohammed Omar
Published by InTech
Janeza Trdine 9, 51000 Rijeka, Croatia
Copyright 2012 InTech

All chapters are Open Access distributed under the Creative Commons Attribution
3.0
license, which allows users to download, copy and build upon published articles
even for
commercial purposes, as long as the author and publisher are properly credited,
which
ensures maximum dissemination and a wider impact of our publications. After this
work
has been published by InTech, authors have the right to republish it, in whole o
r part, in
any publication of which they are the author, and to make other personal use of
the
work. Any republication, referencing or personal use of the work must explicitly
identify
the original source.
As for readers, this license allows users to download, copy and build upon publi
shed
chapters even for commercial purposes, as long as the author and publisher are p
roperly
credited, which ensures maximum dissemination and a wider impact of our publicat
ions.
Notice
Statements and opinions expressed in the chapters are these of the individual co
ntributors
and not necessarily those of the editors or publisher. No responsibility is acce
pted for the
accuracy of information contained in the published chapters. The publisher assum
es no
responsibility for any damage or injury to persons or property arising out of th
e use of any
materials, instructions, methods or ideas contained in the book.
Publishing Process Manager Ivona Lovric
Technical Editor Teodora Smiljanic
Cover Designer InTech Design Team
First published February, 2012
Printed in Croatia
A free online edition of this book is available at www.intechopen.com
Additional hard copies can be obtained from [email protected]
Nondestructive Testing Methods and New Applications, Edited by Mohammed Omar

p. cm.
ISBN 978-953-51-0108-6
Contents
Preface IX
Part 1 General Nondestructive Testing
Methods and Considerations 1
Chapter 1 Nondestructive Inspection Reliability: State of the Art 3
Romeu R. da Silva and Germano X. de Padua
Part 2 Innovative Nondestructive Testing
Systems and Applications 23
Chapter 2 SQUID Based Nondestructive Evaluation 25
Nagendran Ramasamy and Madhukar Janawadkar
Chapter 3 Applications of Current Technologies
for Nondestructive Testing of Dental Biomaterials 53
Youssef S. Al Jabbari and Spiros Zinelis
Chapter 4 Neutron Radiography 73
Nares Chankow
Chapter 5 Flaw Simulation in Product Radiographs 101
Qian Huang and Yuan Wu
Chapter 6 Study of Metallic Dislocations by Methods of Non
Destructive Evaluation Using Eddy Currents 127
Bettaieb Laroussi, Kokabi Hamid and Poloujadoff Michel
Chapter 7 Magnetic Adaptive Testing 145
Ivan Tom and Gbor Vrtesy
Part 3 Concrete Nondestructive Testing Methods 187
Chapter 8 Elastic Waves on Large Concrete Surfaces for
Assessment of Deterioration and Repair Efficiency 189
D. G. Aggelis, H. K. Chai and T. Shiotani
VI Contents
Chapter 9 Ultrasonic Testing of
HPC with Mineral Admixtures 221
R. Hamid, K. M. Yusof and M. F. M. Zain
Chapter 10 Imaging Methods of Concrete
Structure Based on Impact-Echo Test 235
Pei-Ling Liu and Po-Liang Yeh
Preface
The Nondestructive testing science is a broad field that covers variety of testi
ng methods
and applications, in addition to the associated pre and post processin
g mathematics. In
terms of methods and techniques the Nondestructive testing modalities rely on di
fferent
physical phenomena such as the electromagnetism, the acoustic emission,
the thermal
emission and the penetration of high-energy radiation through materials and stru
ctures.
This diversity in the Nondestructive testing tools is only matched by
its fields of
application, which covers the testing of civil and mechanical structures and com
ponents,
the online monitoring of manufacturing processes and products, and a w
ide array of
medical applications that include dental and veterinary medicine.
This book will seek to introduce several Nondestructive testing embodim
ents to
address different testing techniques that rely on several physical phen
omena while
addressing the wide range of its applications. This is done in an ef
fort to highlight
several types of the Nondestructive evaluations and its ability to acc
ommodate
multitudes of fields and tests. Also the manuscript will explain the
different
mathematical and statistical processing techniques used in pre-processing the ac
quired
data in terms of noise reduction, data compression and signal conditio
ning; in
addition to processing the signals and correlating it with the propert
ies of the
materials or structures that are being tested. Sections of this book
will be solely
dedicated to new applications or to using innovative NDT technologies.
The specific Nondestructive techniques addressed in this book include;
the magnetic
adaptive testing, the ultrasonic testing methods, the Neutron Radio-grap
hy, the
Superconducting Quantum Interference Device SQUID sensor based testing r
outines.
The text will also include chapters to discuss the testing reliability
and validation
studies. This book is structured in three main sections; mainly a section on the
General
Nondestructive Testing Methods and its Specific Considerations, a sectio
n on
Innovative Nondestructive Testing Systems and Applications, and finally a sectio
n on
the Concrete Nondestructive Testing Methods.
Dr. Mohammed Omar
Clemson University,
International Center for Automotive Research CU-ICAR, Greenville, SC
USA
X Preface
Part 1
General Nondestructive Testing
Methods and Considerations
1
Nondestructive Inspection
Reliability: State of the Art
Romeu R. da Silva
1
and Germano X. de Padua
2
1
Federal University of Rio de Janeiro,
2
Petrleo Brasileiro S.A. (PETROBRAS),
Brazil
1. Introduction
In health, there are numerous types of tests for the identification of pathologi
es in patients.
Some questions that can be brought up: How accurate are these tests? What are th
e "losses"
of a medical report error if the patient has a serious health problem and it can
not be detected
by the examination chosen? On the other hand, if the patient has no
problem and the
medical report shows positive? What consequences are there in a medical report e
rror?
If we imagine that the medical risks assumed in inaccurate reports ma
y lead to serious
consequences, which can happen with the result of an inspection of eq
uipment without
reliability? Unlike the medical field instead of a fatal case, there
may be multiple fatalities,
environmental damage, irreparable financial losses, etc.
There is several non-destructive inspection methods used to evaluate th
e integrity of
industrial equipment and thus raise several questions. What are the mo
st reliable? Which
ones provide lower risk of decision? There is an ideal method for a
given type of
equipment? A more reliable inspection method also costs more? Some of these ques
tions are
answered in the study of methods for estimating the reliability of No
ndestructive Testing
(NDT), area of scientific research that has been the focus of many i
nvestments in recent
decades, aiming mainly to provide greater operational reliability of eq
uipments from
different branches of industries.
PoD curves may become a powerful tool for quantifying the performance
of inspection
techniques, as well as inspectors and can be used to:
Establish criteria for project acceptance;
Set up maintenance inspection intervals;
Qualification of NDT procedure;
Performance verification of qualification of persons;
Qualify improvements in NDT procedures.
Considering the thematic importance and the increasing trend of investment proje
cts aimed
at better understanding the reliability of NDT methods, this chapter has the mai
n objective
of making an approach on the state of the art studies of the reliab
ility of non-destructive
inspection to be used as the first bibliographic guidance to future r
esearches. Firstly, it
Nondestructive Testing Methods and New Applications 4
covers topics of major theoretical techniques used in the estimation o
f reliability curves.
Then, some of the most relevant research publications in the area of
reliability of NDT are
commented in their main results. It must be noted that this work doe
s not exhaust all the
literature produced; there are other references that can be studied to
obtain detailed
information on this research topic.
2. Methods for reliability assessment
2.1 PoD - Probability of Detection curves
Its supposed that the first PoD (Probability of Detection) studies aros
e by the end of 60s
and beginning of 70s, when most of studies were from aeronautic industry. At tha
t time, it
was realized that the question what is the smaller detectable discontin
uity with NDT
methods? was less appropriate than what is the larger not detectable discontinuity
?.
Currently, the most used method to determine the reliability and sensi
tivity of a NDT
technique is through the assessment of probability of detection curves.
A PoD curve
estimates the capacity of detection of an inspection technique in regard to disc
ontinuity size.
In the ideal technique, the PoD for discontinuities smaller than estab
lished critical size
would be zero. In the other hand, discontinuities greater than critical size wou
ld have PoD
equal 1, or 100% of probability of detection. In such ideal technique, would not
happen what
we know as False Positive (rejection of acceptable components) or False Negative
(approval
of defective components). However, in real situation, PoD curves do no
t have an ideal
behavior, presenting regions of False Positive and False Negative. Figure 1 illu
strates a real
and ideal PoD curve [2].
Fig. 1. Pattern of real and ideal PoD curves [2].
These curves are commonly constructed empirically. The most known metho
d is Round
Robin Testing (RRT), where a group of inspectors proceed a nondestruct
ive examination of
test pieces with artificial defects, simulating real defects that may be found i
n welded joints,
for example. Artificial defects are fabricated in various dimensions. P
oD curves may be
drawn from results of one inspector or based on a group of inspectors [2-4]. Two
issues need
Nondestructive Inspection Reliability: State of the Art 5
to be highlighted in this RRT methodology, the first is the amount of test piece
s necessary to
guarantee statistic reliability of the estimated curve, and second is
the complexity of
obtaining artificial defects in dimension, location and characteristics
as similar as real
defects. In welding, for instance, only skilled, experienced and well trained we
lders are able
to produce defective welds in such way that simulate real situations
of inspections that
provide representative results of PoD.
At First European-American Workshop of reliability (Berlin, June 1997),
a model of reliability
was proposed, which recognize three functions connected to the reliabil
ity of a
nondestructive testing technique: intrinsic capacity of the system, char
acteristics of specific
applications and human factor. Thus, its suggested that reliability of a NDT tech
nique will
never be higher than that idealized. The reliability of a technique, when applie
d to a specific
type of defect, may be represented by following concept:
Rc = (IC) g (PA) (EF) (1)
Where,
Re is the total reliability of the system.
f (IC) is function of intrinsic capacity of the NDT system;
g (PA) is function of parameters applied (access, surface finishing etc.);
h (HF) is function of human factor (skills, training, experience etc.).
By this concept, the function f is associated to intrinsic capacity o
f the specific inspection
technology in ideal conditions. In case of any noise (deviation of ideal conditi
ons), the ideal
reliability is going to be reduced as function of g nature. When the
re are human factors
associated to manual inspection, reliability is reduced, according to f
unction h. Automatic
inspections are free of these factors, due to this fact, often provid
e higher probability of
detection [2].
The PoD of a discontinuity sizing a is determined as the average of p
robability of
detection for all discontinuities sizing a. A PoD curve is constructed
from the average of
PoD for each dimension of discontinuity. Normally, a confidence level is associa
ted, since it
is estimated in function of a finite sample space. The length is the
dimension commonly
used, although the height (internal defect) or depth (surface defect)
may be used as
well [2-4].
Difficulties in fabricating a number of test pieces high enough, frequ
ently provide a poor
sample space. Due to this, there are various statistic models used to estimate P
oD curves [2-
5]. These models run data obtained from two types of analysis: o versus o and hit
/miss [1-5].
According to Carvalho [2], some NDT techniques connect a signal with
response o to a
real dimension o of the discontinuity. Nevertheless, some inspection techniques
do not size
the defects, the response is only detected or not. The analysis hit/m
iss get useful due to its
simplicity. Both methods may be used to implement PoD curves, however, different
results
are obtained when applied to the same data set.
Figure 2 shows a scheme presented by Carvalho [2] to describe the methodology of
analysis
o versus o. Observe that a defect sizing o in a welded test piece c
ause a signal with
magnitude o on the ultrasonic apparel during examination.
Fig. 2. Scheme of method a versus a to implement PoD curves [2].
An inspection procedure may be prepared with two purposes:
1. Detecting defects with any dimension, or detecting defects within specific d
imension, or
even detecting a specific type of defect;
2. Ratify the inspected part is free of defect, or if the inspected part is fre
e of defects larger
than specific dimension, or even if the part is free of specific type of defect.
A practical procedure to prepare PoD curves, from aerospace industry, may be sum
marized
as follows:
1. Fabrication of test pieces containing high amount and various types of defec
ts;
2. Proceed inspection of test pieces using proper technique;
3. Record the results as function of defect dimensions;
4. Plot PoD curve as function of defect dimension.
Nevertheless, prior fabrication of test pieces, it is necessary to hav
e the answer to the
questions: which defect dimension will be used, length, width or depth
? What is the range
of defect dimension will be investigated, 1 to 9 millimeters for exam
ple? How many
intervals are necessary within the range of dimension? [5].
To stipulate the number of test pieces, two important issues must be
considered. First, the
amount of test pieces shall be great enough to estimate PoD curve an
d the limit of
confidence interval. Second, the sample space shall be great enough to
determinate the
statistic parameters of PoD curve that provide better data adjustment.
2.1.1 Statistic model for hit/miss
For analysis of Hit/Miss cases, various statistic distributions have be
en proposed.
Distribution log-logistics or log-probability was found to be more suitable and
function PoD(a)
may be written as follows [5]:
Po =
c
n
3
_
Inc-
o
]
1+ c
n
3
_
Inc-
o
]
(2)
Where a is a defect dimension, and e are average and standard devi
ation, respectively
[5]. Equation 2 can be written as follows:
Po =
c
(o+Inc)
1+ c
(o+ Inc)
(3)
It is simple to reach the equation 4:
ln[
Po(u)
1-Po(u)
= o +[ lno (4)
Where =
u
[
and o =
n
[3
.
Thus
ln(probobility) ln (o). (5)
2.1.2 Statistic model for data of response signal
Concerning to response signal of the inspection technique, it is consi
dered a linear relation
between ln o and ln a, where a is the dimension established of the
defect [5]. This relation
may be represented by equation 6:
ln (o) = o
1
+ [
1
ln(o) + y (6)
Where y is the error with normal distribution, presenting average equa
l zero and standard
deviation constant and equal o
1
. Equation 6 represent normal distribution of ln (o) centered
at p (o) and deviation o
y
2
, where,
p (o) = o
1
+[
1
ln (o) (7)
PoD (a) function for the NDT response signal (ln (o)) may be presented as follows
:
Po (o) = Probobility(ln (o) > ln (o
th
)) (8)
Where ln (o
th
) is the limit of defect evaluation [3].
Using statistic pattern simbology, the PoD function for the response sig
nal of NDT may be
represented by equation 9:
Po(o) = 1 F _
In(u
th
)- (u
1
+[
1
In(u))
c
y
_ (9)
Where F is a continuous cumulative distribution function.
Using the symmetric property of normal distribution:
Po (o) = F j
In(u)-
c
[ (10)
Which is a cumulative log-normal distribution, where p(o) =
Inu
th
-u
1
[
1
and the standard
deviation o =
c
y
[
1
. The parameters o
1
, [
1
e o
y
are estimated through the maximum
verisimilitude method. Such function is often used on analysis Hit/Miss as well
[5].
2.1.3 Estimation of PoD curve parameters
To estimate PoD curve parameters using hit/miss method, it is recommended that d
imension
of defects being uniformly distributed from the smallest to largest di
mension of interest,
containing at least 60 defects. For signal response analysis, it is r
ecommended, at least, 30
defects [5].
2.1.4 Confidence interval of PoD curve
For a hit/miss analysis, a confidence interval of 95% is usually appl
ied, it is necessary a
minimum of 29 defects on each dimension range of study, taking into
account that the
number of discontinuities detected follows a binomial distribution. It
can be interpreted as
29 test pieces containing one defect each. Thus, as an example, if a
n analysis requires 6
ranges of dimensions, it is going to be necessary, at least, 174 test pieces, in
creasing costs for
fabrication of test pieces to estimate PoD curve and confidence intervals correc
tly [5].
As stated previously, a confidence interval may be calculated, assuming it follo
ws a normal
distribution, through the equations 11 and 12.
P jz [
u
2

x -
c n
z [
u
2
[ = 1 o (11)
j x z [
u
2
c
n
, x + z [
u
2
c
n
[ (12)
Where is the significance level, is is average and is standard deviation.
Figure 3 shows a didactic example of 95% confidence interval ( =5%).
2.1.5 General aspects of experimental PoD curves
Experimental PoD curves are plotted when a high volume of inspection data were o
btained
experimentally. They can be applied in projects that include fabricatio
n of test pieces
containing defects with controlled characteristics, such as type, dimens
ions and location.
Another application is to equipments which inspection history is fully
recorded from the
same reference block containing well known defects.
For fabrication of test pieces, a significative number of artificial d
efects is necessary to
provide a sample space that enable estimation of the curves. To reproduce the fi
eld situation
as feasible as possible, many inspectors and defects characteristics shall be us
ed.
Fig. 3. Example of PoD curve with 95% confidence level [2].
The main advantage, in this case, is obtaining the curves without application of
mathematic
models. Based only on the detection rates obtained, which result is t
he closest to the field
inspection. On the other hand, the disadvantage is the high number of
experimental tests
required, what increases the cost of project and may extend it a lot.
2.1.6 General aspects of PoD curves modeled through experimental data
When only a few numbers of experimental data is available, due to in
sufficient number of
test pieces or inspection data, it is possible to plot a PoD curve through a mat
hematic model.
Thus, we can, for example, extrapolate defects out of the dimension s
cale inspected. The
main advantages of this methodology are low cost, easiness and readine
ss. The
disadvantage is that in case of extrapolation of larger defects inspec
tion data to smaller
defects, the PoD obtained may be too low, what do not represent real situation.
But, this is
the most employed method.
2.1.7 Mathematic simulation of PoD curves
Recently, modeling of PoD curves has increased considerably. The low computation
al cost of
simulation, compared to fabrication of test pieces, acquisition of reso
urces for inspection and
use of equipments, is driving to amplify the use of this methodology
[2, 5]. Furthermore,
modeling of a PoD may provide a study of inspection parameters before
its execution, and
enable an evaluation of False Positive rates. In this chapter, only M
onte Carlo Simulation
Method will be approached, despite other methodologies to simulate PoD curves ar
e available.
2.1.8 Monte Carlo simulation method
The Method Monte Carlo (MMC) is a statistic method applicable to stoc
hastic simulations,
suitable to other areas such as physics, mathematics and biology. MMC
has been used a
long time in order to obtain numeric approaches of complex functions.
This method is
typically used to generate observations of any distribution of probabil
ities and use of
sampling to approximate the interest function.
The application of Monte Carlo simulation to estimate PoD of a defect
may be obtained
through the equation 13.
Po() = E|I(x, y)] = ] ] I(x, y)
x,
dt
0
du
0
(x, y)JxJy (13)
Where D is the diameter of defect, x and y are random variables associated to the
position
of the center of circular defect, f
x,y
(x,y) is the density function of probability for both
variables, E[_] is the expected or average value. In an ultrasonic inspection fo
r example, the
elements dt and da are distance between each probe and distance betwe
en two data
acquisitions, respectively. As the center of defect may be located ran
domly in a rectangle
(inspected area), the function f
x,y
(x,y) is given by two normal distributions, one for each
coordinate of center of defect, as follows:

x,
(x, y) =
1
du
1
dt
(14)
I(x, y) is an indicator of inspection function, it assumes value 1 if the defect
was detected and
0 if not detected. In case of ultrasonic examinations [2], detection
is considered successful
when an overlap between defect and ultrasonic beam occur and the ampl
itude of the echo
produced by defect is larger than reference curve. The simulation of
test pieces is
accomplished by random definition of the center of defect (x, y), acc
ording to equation 14.
Then, results of inspection of simulated defects, detected or not detected, shal
l be considered
for the study. Equation 13 may be rewritten as follows:
Po() =
I(x
i
,
i
)
N
N
=1
(15)
Where N means the number of simulations (simulated test pieces), it must be grea
t enough to
provide statistic reliability of the results. According to literature [
6], the error rate of results
obtained through equation 15, considering a confidence level 95%, is given by eq
uation 16.
crror = 2uu_
1-Po
N.Po
(%) (16)
From equation 16, it is possible to determinate the number of simulations necess
ary to reach
the error level wished. Details of Method Monte Carlo are provided by
Carvalho [2] and
Ang [6].
3. ROC curves (Receiver ou Relative Operating Characteristic)
The ROC curves are well known in theory of signal detection and acce
ssed on technical
referenced of pattern recognition [7-10]. These curves are result of relation be
tween number
of false positives (FP), abscissas axis, and number of true positives (TP), ordi
nates axis. Alike
PoD, reliability is given by area under the curve. Reliability of technique is b
etter as much as
higher values of TP and lower values of FP. Ideal reliability is encompassed in
a 100% of a
square area, according to didactic example of figure 4.
The probability of detection, or in other word, the probability of True Positive
is:
Po = P(IP) =
1P
1P+PN
(17)
where FN is the value of False Negative.
The probability of False Alarm or False Positive (FP):
P(FP) =
PP
1N+PP
(18)
where TN is the value of True Negative [7, 8].
The ROC curves have some advantages compared to PoD curves. One of these advanta
ges is
the evaluation of false positive index, which are not taking into acc
ount when PoD curves
are plotted. Certainly, these indexes are very important for nondestruc
tive testing. Just
supposing a situation when false positive may imply in an unnecessary
emergency shut
down of an equipment or operating unit. In the other hand, a worse
situation would be a
false negative that may start up defective equipment, elevating the ri
sk of a catastrophic
occurrence, causing damages to facilities, environment and human deaths.
Fig. 4. Example of ROC curve.
4. Bibliographic review
4.1 Experimental PoD
4.1.1 Manufacture of specimens
The manufacture of specimens with artificial defects can be considered
an art, as they
should be induced in order to represent, in location, size and shape defects tha
t occur in the
reality of manufacturing processes and equipment operation. At this point, the m
ain focus is
to address some techniques of manufacturing well-done defects in materi
als in order to
produce specimens for estimation of inspection reliability, which can a
lso be used for
training and certification of NDT personnel.
a. Fatigue cracks
For metal alloys, fatigue cracks are initiated and grown under controlled condit
ions with the
purpose of construction of PoD curves. Fatigue cracks have particular
characteristics, they
are economical to produce and constitute a challenge for detection. Th
ese cracks can be
initiated, for instance, through a notch. The controlled growth of the
crack can be
accomplished by loading constant at approximately 70% of the yield str
ength of the
material, or by fatigue test, monitoring its growth with methods such
as ultrasound by
TOFD (Time of Flight Diffraction). The notch should be removed from th
e original specimen
before the inspection process to allow a correct measurement of the crack only [
11].
b. Welding defects
Silk s book [1], in chapter 3, contains an item devoted to descriptio
n of the main causes of
welding defects, which do is not scope of the proposed work.
Another interesting work is that of Bullough et al [12] describes a model for es
timating the
distribution of defects in submerged arc welding in equipment of nuclear industr
y.Potential
defects are mainly: lack of fusion, solidification cracking in weld metal hydrog
en cracking in
the HAZ and weld metal. It also estimated the probability of the presence of def
ects in weld
inspection after manufacturing by calculating the probability of defect formatio
n versus the
probability of not detecting [12].
The table 1 below contains some recommendations to produce controlled
defects in welded
plates resulting from the experience of welders of the SENAI RJ Technology Weldi
ng Center.
Type of Weld Defect Recommendations
Lack of root fusion
Welding with lower amperage/Welding only one side of root
face.
Lack of fusion in the wall
groove
Place a piece of graphite on the wall and make the filling.
Lack of Penetration
Use a rod thicker than the root gap/ Put a piece of material
(carbon steel) at the root gap.
Excessive Penetration Welding with higher amperage.
Surface crack
Add copper, aluminum or cobalt before welding the face
reinforcement.
Internal Crack
During the filling step of the weld beam, create a notch
through a cutting disc or saw blade thinness. Afterwards,
finish the face reinforcement.
Crack in the Root
The same procedure for internal crack, but carried out in the
root step.
Porosity
Lower gas flow (Ar for GTAW). For a 7 mm diameter nozzle,
the flow of gas is recommended 8 l/minute. You can use a
flow 3l/minute to generate porosity. To generate porous in
Shielded Metal Arc Weld, the best practice is to weld in direct
polarity.
Face Undercutting
Apply high amperage / increase the speed of welding. For
GTAW and SMAW, weld with angle 15 or 30 .
Table 1. General Recommendations to produce typical welding defects.
4.1.2 Estimation of experimental PoD
One of the first projects of reliability of NDT method was the Program for Inspe
ction of Steel
Components (PISC) in mid-70s, which has been initiated in order to as
sess the ability of
defect detection by method of ultrasound in the walls of pressure vessels of up
to 250mm in
the nuclear industry [13-15]. Several ultrasound procedures existing at
that time were
strictly applied to the results of inspections, which resulted in PoD with low v
alues [13-15].
However, some inspectors could also use the procedures they wanted, thus achievi
ng much
better results in terms of detection for the same defects analyzed. I
n the PISC II and III
programs, the project drew on more flexible procedures. The results sh
owed that
characteristics of defects such as shape and geometry are more relevan
t to the POD when
compared to other physical parameters. They also concluded that there were some
mistakes
in the ASME code, however, the most relevant contribution was made to
a detailed
evaluation of NDT techniques for detecting and sizing of defects [13-15].
Another important program was titled NORDTEST, which was developed in
Scandinavia
by the Netherlands Institute of Welding (NIL), the ICON (Inter Calibra
tion of Offshore
Nondestructive Examination) and TIP (Topside Inspection Project). The ma
in object of this
project was to compare the manual method of ultrasonic with the method applied t
o X-ray
inspection of welded plates carbon-manganese steel with thickness smaller than 2
5 mm. The
results also were used to establish acceptance curves (curves (1-POD)
versus height of the
defect) [13]. This project also compared the technique of inspection b
y manual ultrasonic
with the automated inspection assisted by processing techniques (such as focusin
g system),
certifying that the computerized inspection result in a PoD significant
ly higher than the
manual inspection [16].
The UCL (University College London) conducted a project in the offshor
e area for the
preparation of PoD from fatigue cracks in tubular joints in the mid
90s. The aim was to
compare the probability of defect detection of these cracks by the method of mag
netic particles
with the method of eddy currents, as well as the method of using ultrasonic Cree
ping waves.
Which has reached PoD between 90 and 95% for cracks larger than 100mm[5, 17]?
In the 90 s, the Netherlands Institute of Welding (NIL) has issued a report with
the results of
a project to study the reliability of the method of mechanized ultras
onic, among other
methods, to detect defects in welded plate of 6 to 15mm thick. The
results proved that the
mechanized method and TOFD (Time of Flight Diffraction) technique have
probability of
detection much higher than the manual method (60-80% of PoD compared to about 50
% of
the manual). The mechanized method is also more effective in sizing of defects [
2, 18].
Carvalho [2] employed the method of ultrasonic pulse-echo both in manu
al and in
automated form, as well as the automatic method of TOFD. More informa
tion about
inspection procedures can be obtained in [2]. The inspection by the p
ulse-echo manual
technique was carried out by five (5) inspectors duly certified by AB
ENDI - Brazilian
Association of Nondestructive Tests and Inspection, recognized by SNQC - Nationa
l System
of Qualification and Certification, in accordance with ISO 9712. Thus, the PoD
curves were
constructed from an average of 75 defects (samples), since each length
was repeated 15
times. Each set was replicated by this bootstrap technique [19] in 1500 a new se
t containing
75 samples. The average probability of detection of each length was e
stimated for each of
these sets. The 1500 PoD values were arranged in ascending order by
choosing a 95%
confidence interval [2].
By figures 5(a) and (b), it is possible to certify that it reaches
close to 100% detection for
defects larger than 20 mm for both LP (Lack of Penetration) and LF (Lack of Fusi
on) classes.
Figure 5 (c) shows that for defects in lengths of about 12 mm, the class LP has
a higher value
of PoD, whilst that from it value, the opposite happens. The integrated curve sh
ows that the
class LP has higher PoD (77%) than LF (63%). Carvalho [2] concluded
that it must be
explained by the fact that the defect LP is usually located in the
root of weld and can
therefore be detected by both sides of the beam. As for high dimensional defects
, LP can be
confused with the background echo, which does not happen with the LF,
this may justify
the higher PoD of LF from a given value of length.
Fauske et al [20] and Verkooijem et al [21] also concluded in their
work that automatic
inspection allows PoD values higher than manual inspection by ultrasonic. The fi
rst reached
the value of 80% PoD for automatic detection of cracks of 10mm with
a depth of 1 mm,
while the manual inspection resulted in only 60%. Verkooijem [21], who
worked with the
classes LP, inclusion of slag and porosity, reached 83.6% probability
of detection with
automated pulse-echo ultrasound, 52.3% for manual technique and 82.4% f
or TOFD
technique.
Carvalho [2] also concluded that automatic inspection provide PoD value
s much higher
than manual inspection by ultrasonic. The graphs in Figure 5, which a
lso include PoD
curves for each inspector used in the tests show that the PoD of automated syste
ms (pulse-
echo and TOFD) is a typical case of ideal PoD, where there is a cr
itical size defect below
which there is no detection. Discussions on the performance of the in
spectors can be
found in [2].
(a) (b)
Fig. 5. PoD curves for defects LF and LP, respectively [2].
Figure 6 below shows the results of Carvalho [2] for what is called
PoS (Probability of
Sizing), which is a graph where the x-axis represents the expected si
ze (projected) of the
defect, and the y-axis represents the size found by the inspector. Thus, a point
located at y =
x, if the scales are the same in Cartesian axes, means accuracy in
sizing of the defect
(discontinuity). By the result obtained, it became evident that there
was overestimation in
most of the results. It is relevant to emphasize that this overestimation was ve
ry significant,
because defects with 3, 5, 7 and 10mm have been scaled up to 20mm. Carvalho [2]
discusses
the cost-benefit of this behavior, from the point of view of the operational saf
ety it is good,
on the other side it can cause an unnecessary shut down of equipment operation,
which will
result in interrupted profit [2].
(a) (b)
Fig. 6. Probability of sizing for the LF and LP defect classes [2].
4.2 Simulation of PoD curves
The sonic intensity of a divergent ultrasonic beam decreases in relati
on to the center.
Carvalho [2] emphasizes that the sonic distribution of beam divergence follows t
he equation
19, which actually describes a bell-shaped function, as illustrated in Figure 7.
s(x, y) = S
0
. cxp _o[x
2
+ y
2
b
] (19)
S
0
is the intensity at the center of the sonic beam and a and b constants determined b
y
data supplied by the manufacturers of transducers [2].
In addition to the decrease in intensity due to the sonic beam diver
gence, there is the
attenuation caused by the absorption and dispersion of the wave, thus,
considering a
distance "d" of the transducer, the equation that models the distribution is:
S
d
(x, y, J) = c
-u.d
. S
0
. k(J). cxp _o[x
2
+y
2
b
] (20)
where,
s
d
(x,y,d) = sound intensity encountering a point (x,y) at a distance d from the t
ransducer;
= material attenuation coefficient;
d = distance between the transducer and the point of interest (see Figure 7);
k (d) = factor to maintain total sound intensity of an ideal materia
l ( = 0) constant at any
depth considering the variation of the ultrasonic beam aperture.
Fig. 7. Sonic intensity distribution as a function of beam divergence [2].
In conventional ultrasonic method, the reference curves are constructed
with standard
blocks containing defects of known dimensions to determine whether there is a de
fect in the
material examined. It is also possible to estimate this reference curv
e numerically by using
equation 21 below, which is the integral of the equation 20 over the area of the
defect.
S
= c
-u.d
. S
0
. k(J
x
2
+
2
<
c
2
). exp_o[x
2
+y
2
b
] (21)
In relation to a situation of inspection, as shown in Figure 8, ther
e is an incidence onto a
defect diameter at a distance d from the transducer, the total inten
sity sonic inspection,
depending on the position of the sonic beam in relation to the defect, is obtain
ed by equation
22, which is compared with equation 21 to determine whether or not detection.
S
= c
-u.d
. S
0
. k(J
A
). exp_o[x
2
+y
2
b
] (22)
Where A is the area of the defect reached by the ultrasound beam (Figure 8).
Fig. 8. Illustration showing an ultrasound beam focusing on a defect of area A.

After the normalization process in relation to the pattern flaw c at d position,
the R valued
is:
R =
cxp_-u[x
2
+
2
b
]dxd
A
cxp_-u(x
2
+b
2
)
b
]dxd
_x
2
+j
2
<
c
2
(23)
Then, if
curve
R R the flaw is detected, where R
curve
is the value of the employed reference
curve, that is, R
curve
= 1 for the 100% reference curve, and so on. After the simplificat
ion R
becomes independent of s
0
and k(d) [2].
As a practical result, Carvalho [2] obtained the curves shown in Figu
re 9. Figure 9 (a)
represents the result of PoD for a flat plate considering various distances betw
een transducers
(a) (b)
(c)
Fig. 9. PoD curves obtained by numerical simulation [2].
of different diameters and defects. The results show that as greater t
he distance between the
transducers, the PoD is lower. In contrast, as larger the diameter of the defect
, PoD is higher.
The reference curve used was 50%. Figure 9 (b) presents the results obtained for
evaluation by
considering two conditions of reference curve: 20% and 50%. The PoD curve obtain
ed for 20%
was higher than 50%, what was expected since it is a condition of m
ore conservative
detection. Since the figure 9 (c) represents the results obtained for
variation of the distance
from the transducer to the defect: 25mm and 50mm. The distance of 50mm resulted
in higher
PoD, according to Carvalho [2], due to greater divergence of the beam
at this distance.
However, Carvalho [2] points out that this simulation was not consider
ed the fact that as
greater the distance, greater the attenuation that will be the sonic intensity.
Another important
aspect is that the practice of ultrasonic inspection by the inspector
is based on amplitude
curves (measured in dB) for the detection of defects.
4.3 ROC curves
There are few works of development of ROC curves on nondestructive te
sting. One of the
most important work is Fcsks [7, 8] for the construction of ROC curves
of radiographic
films. This study aimed to estimate the reliability of human factors
in the evaluation of
radiographs, using the RRT method (Round-Robin Test) with inspectors of
laboratories in
Croatia, Hungary and Poland. A set of films was selected at BAM (Fed
eral Institute of
Materials Berlin) who provided technical and scientific support to the implement
ation of the
project. Thirty eight radiographic films were used, containing 206 weld
defects of different
types and sizes. These films were scanned with the scanner LS85 SDR
Lumisys. The
digitized radiographic images were used as the basis for preparation o
f templates of the
reports. With the digital images, series of identical films were print
ed in a laser printer
AGFA Scopix so that each inspector could review the same films. The
inspectors received
instructions, a procedure of reports and special forms to fill in the
result. The identity of
each inspector was preserved through identification codes. At the end,
a total of 60
inspectors from different laboratories were used. The best result of d
etection was 88.3%
(probability of true positive) and 12.9% probability of false positive,
when all defects have
been considered (even below the lower acceptance criterion). In an ass
essment that
considered only major defects that the level 1 of acceptance of EN 12517:1998,
the PoD was
85.2%, and only 2.1% for false alarms (positive). Fcsk [7] concluded th
at the results in
general were not good, with high values of false alarms, indicating a need for b
etter training
and qualification of inspectors.
4.4 Other references
In addition to the publications already mentioned, it is relevant high
light some other very
interesting, as the publication of the Department of Defense [23] whic
h is a complete
theoretical approach of the reliability of NDT, including chapters on
method and not just
PoD curves, but also the construction of ROC curves [23].
Another interesting publication is the Meeker [24] that presents a step
by step how to
implement a POD curve by Hit / Miss and o x a methods, which is a
practical and fast
tutorial. In addition to the MATLAB tools of programming (Mathworks),
there is the
STATUS program (Statistical Analysis Tools for Ultrasonics Test System)
to prepare PoD

curves by the two methods of analyzing, as well as analysis of PoS
(Probability of Sizing).
Another program is the mh1823 that also allows the construction of curv
es and
performing reliability analysis (freely available).
There are several other publications which are listed below, which alt
hough not directly
addressed in this work, especially not to make the technical discussio
n of the subject too
repetitive, since many are similar to the methodology presented by the work desc
ribed, they
can be analyzed because present other applications of experimental situations [2
5-45].
4.5 General discussion
In general, the human factor is usually the cause of most failures t
o detect discontinuities,
and that even well qualified and experienced inspectors make mistakes.
Thus, automatic
inspection methods allow significantly higher probability of detection a
s the ultrasonic
method. However, it must be noted that sometimes the automatic inspect
ion is still
economically impracticable. Another important issue that a PoD analysis
can support a
choice when there are two possible methods of inspection of a given
equipment, for
example, if a method A cost 50% of a method B and if your PoD is 30%
of B PoD ,
your actual cost is greater than B.
The experimental methods require high costs to be carried out, as wel
l as time, as they
demand the production an enough quantity of specimens to allow the correct estim
ation of
relevant statistical parameters of the reliability curves and their con
fidence intervals.
Another important point here is that the preparation of specimens with
defects properly
controlled, as in the case of welding defects, is an art a bit rare to find skil
led professionals
to do so. The defects made should reproduce the maximum of reality i
n terms of size,
morphology and location in the weld. These methods employ mathematical
tools for
estimation of curves and log-normal function is most often employed fo
r this purpose,
although other functions can also be used. Considering the lack of samples in ma
ny studies,
some researchers also resort to the techniques of simulation data, as the bootst
rap technique
to generate new data sets and produce the needed confidence intervals.
There is now a series of works focusing on numerical simulation of r
eliability curves as a
solution to the disadvantages presented by the method of preparation o
f specimens.
Nevertheless, despite the cost advantages and speed, it is extremely difficult t
o reproduce in
a simulation all variables that are embedded in a real inspection and
which are not easily
quantified, such as those related to the human factor. In Mores study
[46], there are
descriptions of 39 variables that are relevant to an inspection by ul
trasonic. Considering
advances in tools and simulation programs, this methodology should gain
more space
compared to the experimental methods.
The probability of detection of defects has been used to meet the re
quirements for
equipment design, and has worked as a tool for comparing the performan
ce capabilities of
various NDT methods, procedures and guidance of qualified professionals.
5. Conclusions
The reliability of nondestructive testing has been the subject of impo
rtant research projects
over the past 40 years, many of them highlighted in this work, which
can reach the main
conclusions:
Most of the works present experimental methodologies of specimens manuf
acture. In this
case, one should always look for the minimum number of defects generated in orde
r to get
the better estimation of PoD curves and their confidence intervals.
The preparation of samples requires extreme skill, in particular, the welded sp
ecimens
must contain defects that are the most realistic possible with ideal
dimensions to
regulatory acceptance criteria used.
Some functions are usually employed to model mathematically the PoD curve obtai
ned
with some experimental data, and the Log-Normal function is the normal
ly chosen by
the researchers.
Due to the disadvantages of experimental methods related to the time,
cost and
logistics, the simulation of PoD and ROC curves can be a solution, h
aving already
carried out several experiments in this direction. However, it is necessary to e
mphasize
that is very difficult to reproduce in simulation, either physical or
numerical
(simulations of ultrasonic or radiographic tests, for example), several factors
that occur
in a real inspection, especially those related to the human factors, which gener
ally make
a major influence on PoD. But with the advances of the simulators, i
t is possible that
this disadvantage be mitigated in a near future.
The methodologies for estimating the reliability of NDT are extremely
important for
modern industry and will increasingly focus investments.
6. References
[1] Silk, M G. The Reliability of Non-destructive Inspection. Assessing
the Assessment of
Structures under Stresse. Adam Hilger, Bristol. 1987.
[2] Carvalho, A. A. Confiabilidade de Tcnicas de Ensaios No Destrutivos n
a Inspeo de
Dutos Utilizados na Indstria do Petrleo. Tese de Doutorado. COPPE. Rio
de
Janeiro, 2006. (In portuguese).
[3] Carvalho, A. A., Sagrilo, L. V. S., Silva, I. C., et al. The P
OD Curve for the Detection of
Planar Defects Using a Multi-Channel Ultrasonic System. Insight, Vol. 44, N. 10,
p.
689-693, 2002.
[4] Carvalho, A.A. Rebello, J.M.A. Silva, R.R. Sagrilo, L.V.S. Reliabil
ity of the Manual and
Automatic Ultrasonic Technique in the Detection of Pipe Weld Defects. Insight, v
ol.
48, n.11, November 2006.
[5] Georgiou, G.A. PoD Curves, their Derivation, Application and Limitati
ons. Vol. 49, n.7.
Insight, p. 409-413, 2007.
[6] Ang, A.H.S. Tang, W.H. Proabability Concepts in Engineering Planing
and Design
Decision, Risk an Realibility, vol. 3, New York, Editora Jonh Wiley & Sons, 1984
.
[7] Fcsk, F. Muller, C. Reliability of Routine Radiographic Film Evaluation An Ex
tended
ROC Study of the Human Factor. 8
th
ECNDT, Barcelona 2002.
[8] Fcsk, F. Muller, C, Scharmach, M. Measuring of The Reliability of
NDE. 8
th
International Conference of the Slovenian Society for Nondestrutive Test
ing.
Application of Contemporary Nondestrutive Testing in Engineering. September 1-
3, Slovenia, pp. 173-180.
[9] Schoefs, F., Clment, A., Nouy, A. Assessment of ROC curves for ins
pection of random
fields. Structural Safety. Vol. 31, p.409-419, 2009.
[10] Nockemann, C. Heidt, H. H. Thomsen, N. Reliability in NDT: ROC S
tudy of
Radiographic Weld Inspections. C. Nockemann, H. Heidt and N. Thomsen.
NDT&E International, vol. 24, n.5, October 1991.
[11] http://www.asnt.org/publications/materialseval/basics/jan98basics/jan98basi
cs.htm
(acesso em 10/05/2011).
[12] Bullough, R., Dolby, R.E., Beardsmore, D.W., Burdekin, F.M., Schnei
der, C.R.A. The
Probability of Formation and Detection of Large Flaws in Welds. Intern
ational
Journal of Pressure Vessels and Piping. Vol. 84, p.730-738.
[13] Lamaitre, P., Kobl, T.D. Summary of the PISC Round Robin Results
on Wrought and
Cast Austenitic Steel Weldments, Part I: Wrought to Wrought Capability
Study.
International Journal of Pressure Vessels & Piping. Vol. 69, pp. 5-19, 1996.
on Wrought and
Cast Austenitic Steel Weldments, Part II: Wrought to Cast Capability S
tudy.
on Wrought and
Cast Austenitic Steel Weldments, Part III: Cast to Castt Capability St
udy.
[16] PoD/PoS Curves for Nondestrutive Examination Offshore Technology Re
port,
2000/018. Prepared by Visser Consultancy Limited for the Health and Sa
fety
Executive (HSE).
[17] NDT of Thin Plates Evaluation of Results. Nil Report. NDP 93-38
Rev.1. (in Dutch),
1995.
[18] Dover, W.D., Rudlin, J.R. Results of Probability of Detection Tria
ls. Proc. IOCE 91,
Aberdeen, 13-16 October 1992.
[19] Efron, B., Tibshirani, R. J., AN Introduction to the Bootstrap, 1a
Edio,New York,
Editora Chapman & Hall/CRC, 1993.
[20] Fauske, T. H., Dalberg, P., Hansen, A. Ultrasonic Crack Detection Trials. 1
5th WCNDT -
World Conference on Nondestructive Testing. Roma, October, 2000.
[21] Verkooijem, J. The Need for Reliable NDT Measurements in Plant Ma
nagement
Systems. 7th ECNDT - European Conference on Nondestructive Testing.
Copenhagen, May, 1998.
[22] Bartholo, P.U. Modelagem da Probabilidade de Deteco do Ensaio Ultras
snico e
Avaliao da Influncia de Inspetores e Tipos de Defeitos. Projeto Final de
Curso.
Universidade Federal do Rio de Janeiro, RJ, Brasil, 2008. (In portuguese).
[23] MIL-HDBK-1823. Non-Destructive Evaluation System Reliability Assessment, 19
99.
[24] Meeker, W. Q. and Escobar, L. A. Statistical Methods for Reliabil
ity Data, John Wiley
and Sons, New York.
[25] Forsyth, D. S., Fahr A., On the Independence of Multiple Inspections and the
Resulting
Probability of Detection, Quantitative Nondestructive Evaluation, Iowa, July, 200
0.
[26] Wall, M. Wedgwood, F. A. Burch, S. Modeling of NDT Reliability (POD) and A
pplying
Corrections for Human Factors. 7th ECNDT European Conference on
Nondestructive Testing. Copenhagen, May, 1998.
[27] Forsyth, D. S. Fahr A. Leemans, D. V. et al. Development of POD from In-Ser
vice NDI
Data. Quantitative Nondestructive Evaluation. Iowa, July, 2000.
[28] Topp, D. A. Dover, W. D. Reliability of Non-Destructive Inspection
Test. Anlisis de
Riesgo y confiabilidad Estructural de Instalaciones Marinas. Mxico, dicie
mbre,
2001.
[29] Olin, B.D, Meeku, Willian Q. Application of Statistical Methods to
Nondestructive
Evaluation. Technometrics, vol. 38, n.2, 1996, p. 95-112.
[30] Wall, M. Burch, S. Lilley, J. Human Factors in POD Modelling and
Use of Trial Data.
Insight, Vol. 51, N. 10, October 2009.
[31] Rummel, W D. Recommended practice for a demonstration of non-destr
uctive
evaluation (NDE) reliability on aircraft production parts. Materials Evaluation V
ol.
40. August 1982.
[32] Generazio, E.R. Directed Design of Experiments for Validating Probability o
f Detection
Capability of NDE Systems. Review of Quantitative Nondestructive Evaluat
ion.
Vol.27, p. 1693-1700, 2008.
[33] Burch, S.F., Stow, B.A., Wall, M. Computer Modelling for the Pred
iction of the
Probability of Detection of Ultrasonic Corrosion Mapping. Vol. 4, n.12,
Insight,
pp.761-765, December 2005.
[34] Gandossi, L., Simola, K. Derivation and Use of Probability of Det
ection Curves in the
Nuclear Industry. Insight, Vol. 52, n. 12, December, 2010.
[35] Sweeting, Trevor. Statistical Models for Nondestructive Evaluation.
International
Statistical Review. Vol. 63, n.2, p. 199-214, 1995.
[36] Wall, M., Burch, S.F., Lilley, J. Review of Models and Simulators for NDT.
Vol. 51, n.11,
Insight, November 2009.
[37] Hong, H.P. Reliability Analysis with Nondestructive Inspection. Structural
Safety, Vol.
19. N.14, pp. 383-395, 1997.
[38] Bullough, R., Burdekin, F.M., Chapman, O.V.J., Green, V.R., Lidbury
, D.P.G, Pisarski,
H., Warnick, R.G., Wintle, J.B. The Probabibility of Large Defects in Thick-Sectio
n
Butt Welds in Nuclear Components. International Journal of Pressure Vessels and
Piping, Vol. 78, p. 553-565, 2001.
[39] Kazantsev, I.G., Lemakien, I., Salov, G.I., Denys, R. Statistical
Detection of Defects in
Radiographic Images in Nondestructive Testing. Signal Processing. Vol. 82, p. 79
1-
801, 2002.
[40] Wall, M., Burch, S.F., Lilley, J. Simulators for NDT Reliability (PoD). Ins
ight, Vol. 51, p.
612-619, 2009.
[41] Wall, M.,Wedgwood, F.A., Burch, S. Modeling of NDT Reliability (PoD) and Ap
plying
Correction for Human Factors. 7
th
European Conference on Nondestructive
Testing. Copenhagen, 1998.
[42] Forsyth, D.S., A., Leemans, D.V., et al. Development of PoD from In-service
NDI Data.
Quantitative Nondestructive Evaluation, EUA, 2000.
[43] Topp, D.A., Dover, W.D. Reliability of Nondestructive Inspection Te
st. Anlisis de
Riesgo y Confiabilidad Estructural de Instalaciones Marinas, Mxico, 2001.
[44] Simola, K., Pulkkinen, U. Models for Nondestructive Inspection Data
. Reliability
Engineering and System Safety. Vol. 60, p. 1-12, 1998.
[45] Serabian, S. Ultrasonic Probability of Detection of Subsurface Flaw
s. Materials
Evaluation. Vol. 40, p. 294-298, 1982.
[46] Mor, J.D. Aplicao de Lgica Fuzzy para Avaliao da Confiabilidade Humana
nos
Ensaios No Destrutivos Tipo Ultrassom. Tese de Doutorado. COPPE/UFRJ, Ri
o
de Janeiro, 2004.
Part 2
Innovative Nondestructive Testing
Systems and Applications
2
SQUID Based Nondestructive Evaluation
Nagendran Ramasamy and Madhukar Janawadkar
Indira Gandhi Centre for Atomic Research, Kalpakkam
India
1. Introduction
Nondestructive testing (NDT) or Nondestructive evaluation (NDE) refers t
o techniques,
which are used to detect, locate and assess defects or flaws in mate
rials or structures or
fabricated components without affecting in any way their continued usef
ulness or
serviceability. The defects may either be intrinsically present as a r
esult of manufacturing
process or may result from stress, corrosion etc. to which a material or a compo
nent may be
subjected during actual use. It is evident that techniques to detect critical fl
aws before they
have grown unacceptably large are of vital importance in the industry
for in-service
inspection, quality control and failure analysis. There are several NDE techniqu
es of which
one of the widely used techniques is based on eddy currents. However,
the conventional
eddy current technique has the drawback that it can detect flaws upto a certain
depth under
the surface of the conducting specimen under investigation and is not
suitable for locating
deep subsurface defects. Such limitations can often be overcome with t
he use of high
sensitivity SQUID sensor. Nondestructive evaluation of materials and str
uctures using low
temperature SQUID (LTS) as well as high temperature SQUID (HTS) has been propose
d and
the potential of the technique has been demonstrated during the last
two decades (H.
Weinstock, 1991 & G.B. Donaldson et al, 1996). The SQUID based NDE o
ffers many
advantages such as high sensitivity (~ 10 to 100 fT/\Hz), wide bandwi
dth (from dc to 10
kHz), broad dynamic range (>100 dB) and its intrinsically quantitative nature; d
isadvantage
of this technique is that the SQUID sensor operates only at cryogenic
temperatures which
makes it relatively expensive. However, despite the expensive cryogen a
nd associated
inconvenience in handling, SQUID sensors find a niche in areas where
other NDE sensors
fail to achieve the required performance (H.-J. Krause & M.V.Kreutzbruck, 2002).
The SQUID based NDE systems have been developed and utilized in sever
al areas of
application. At university of Strathclyde, system based on SQUID sensor has been
used for
the detection of flaws in steel plates (R.J.P. Bain et al., 1985, 19
87; S. Evanson et al., 1989).
Weinstock and Nisenoff were the first to demonstrate the possible use of SQUID s
ensors for
the study of stress strain behaviour in a ferromagnetic material (H.
Weinstock & M.
Nisenoff, 1985, 1986). The SQUID sensors have been used for the detec
tion of tendon
rupture in pre-stressed steel tendons of concrete bridges through magne
tic flux leakage
method (G. Sawade et al., 1995; J. Krieger et al., 1999). Marco Lang et al evalu
ated the fatigue
damage of austenitic steel by characterizing the formation of martensit
e due to quasi-static
and cyclic loading with the help of the SQUID based measuring instrument (M. Lan
g et al.,
2000). SQUID sensors have been successfully used for the detection of ferrous in
clusions in
26
aircraft turbine discs by Tavrin et al. (Y.Tavrin et al., 1999). The presence of
such inclusions
may initiate cracks in these critical parts and may eventually lead t
o failure. The magnetic
inclusions in the nonmagnetic alloy of the turbine discs have been in
vestigated by pre-
magnetizing the turbine discs and probing their remanent field using t
he SQUID sensor.
Since the sensitivity of the SQUID is very high (under 5 fT/Hz) and
remains constant
down to frequencies as low as 1 Hz, these sensors are widely utilize
d for the detection of
deep sub-surface flaws in conducting materials through low frequency ed
dy current
excitation to take advantage of increased skin depth at low frequencie
s. The SQUID based
eddy current NDE technique plays a major role in detecting deep subsu
rface defects
embedded in thick multilayered aluminum structures used in aircraft lap joints.
In this chapter we describe the working principle of the DC SQUID se
nsor, schemes
generally used to couple an external signal to the SQUID sensor, cons
truction of a SQUID-
based NDE system by using a low temperature DC SQUID sensor and its associated r
eadout
electronics developed in our laboratory. The SQUID system in our laboratory has
been used
for the measurement of deep subsurface flaws by inducing eddy currents
in conducting
materials at relatively low frequencies. Detailed experimental studies have been
carried out
for the determination of optimum eddy current excitation frequencies for
the flaws located
at different depths below the top surface of an aluminum plate. This
system has also been
used for the measurement of extremely low content of magnetic o - fe
rrite in the 316L(N)
stainless steel weldment specimens subjected to high temperature low cycle fatig
ue (LCF).
2. SQUID sensor
The SQUID (Superconducting Quantum Interference Device) is an extremely
sensitive
sensor for magnetic flux and its output voltage is a periodic functio
n of applied magnetic
flux with the periodicity of one flux quantum u
0
(= h/2e = 2.07 x 10
-15
Wb). To utilize the
SQUID sensors for real applications, Flux Locked Loop (FLL) readout el
ectronics has been
developed in our laboratory to linearise the periodic output voltage of the SQUI
D. By using
SQUID sensor and its associated readout electronics, it is possible to
detect a change in
applied magnetic flux whose magnitude is much less than one flux quantum. SQUID
sensor
can measure any physical quantity that can be converted into magnetic
flux, and has been
used, for example, for the measurement of magnetic field, magnetic field gradien
t, magnetic
susceptibility, electric current, voltage, pressure, mechanical displaceme
nt etc. with an
unprecedented sensitivity. The systems based on SQUID sensors offer a
wide bandwidth
(from near dc to hundreds of kHz), wide dynamic range (>100dB) and a
n intrinsically
quantitative response. The unprecedented sensitivity of the SQUID sensor togethe
r with the
use of superconducting pickup loops (used as input circuits) enables one to real
ize practical
measuring instruments for the measurement of extremely weak magnetic si
gnals with a
high sensitivity.
2.1 Principle of operation
The SQUID is basically a superconducting sensor, which operates below
the
superconducting transition temperature (T
c
) of the superconducting materials used for the
fabrication of the device. The basic phenomena governing the operation
of SQUID devices
are flux quantization in superconducting loops and the Josephson effect
(fig.1.). While
27
detailed descriptions are available in the literature (J. Clarke, 1993;
H. Koch, 1989), a brief
description of the working principle of the SQUID sensor is included
here to make this
chapter self-contained.
Fig. 1. (a) Flux quantization. (b) Non hysteretic I-V characteristic of a resist
ively shunted
Josephson junction
Flux quantization refers to the fact that the total flux linked with a supercond
ucting loop is
always constrained to be an integral multiple of a flux quantum (u
0
).
0 tot ext
LJ n u =u + = u (1)
where u
ext
is the externally applied magnetic flux, L is the self inductance o
f the
superconducting loop, J is the screening current induced in the superc
onducting loop
because of the application of the external magnetic flux and n is an
integer. The Josephson
effect refers to the ability of two weakly coupled superconductors to sustain at
zero voltage
a supercurrent associated with the transport of Cooper pairs, whose magnitude de
pends on
the phase difference between the two superconductors.
0
sin I I o = (2)
where I
0
is the maximum current the junction can sustain without developing a
ny voltage
and is known as critical current of the Josephson junction and o is
the phase difference
between the two weakly coupled superconductors. When two superconductors
are
separated by a very thin oxide barrier (tunnel junction), the establis
hment of tunneling
assisted phase coherence leads to Josephson effect; I-V characteristic
of such a tunnel
junction shows hysteretic behaviour due to non-negligible value of junc
tion capacitance.
This hysteresis is, however, undesirable and can be eliminated by shunting the j
unction with
appropriate on-chip thin film resistor to provide sufficient damping of the phas
e dynamics.
A DC SQUID consists of two such non-hysteretic Josephson junctions con
nected by a
superconducting loop.
To describe the operation of the dc SQUID, we assume that the bias current I
b
is swept from
zero to a value above the critical current (2I
0
) of the two junctions. An external magnetic flux
varying slowly in time is applied perpendicular to the plane of the loop. When t
he external
applied magnetic flux is zero (or nu
0
, n is an integer), there is no screening current induced
in the superconducting loop and the bias current I
b
simply divides equally between the two
junctions assuming the SQUID to be symmetric. When external magnetic f
lux u
ext
is
applied, the requirement of flux quantization generates a screening cur
rent J = -(u
ext
-
nu
0
)/L
s
, where L
s
is the inductance of the SQUID loop and n is an integer which makes the
J
u
ext
(a)
I
V
(b)
I
0
28
value of nu
0
to be nearest to the applied flux u
ext
. The screening current induced in the
SQUID loop adds to the bias current flowing through junction 1 and s
ubtracts from that
flowing through junction 2. When the junction 1 reaches its critical
current I
0
= I
b
/2+J, the
current flowing in junction 2 is I
b
/2-J (that is, I
0
-2J) and the total current flowing in the
SQUID is 2I
0
-2J. At this point the SQUID switches to the non-zero voltage state.
When the
applied flux is increased to u
0
/2, the screening current J reaches a value of u
0
/2L
s
and the
critical current falls from 2I
0
to 2I
0
(u
0
/L
s
) as shown in fig.2 (b). When the flux u
ext
is
increased further the SQUID makes a transition from the flux state n = 0 to n =
1, J changes
its sign and reaches zero again when u
ext
becomes equal to u
0
. At this point, the critical
current of the SQUID is restored to its maximum value of 2I
0
. In this way the critical current
oscillates as a function of u
ext
. If we bias the SQUID with a dc current greater than the
critical current of the two Josephson junctions, the voltage developed
across the SQUID
oscillates with a period of u
0
when the input magnetic flux steadily increases. Thus, the
SQUID produces output voltage in response to a small input flux ou (
<< u
0
), and is
effectively a flux to voltage transducer. The voltage swing oV produced at the o
utput of the
SQUID when the flux changes from nu
0
to (n +1/2)u
0
is known as the modulation depth of
the SQUID. The modulation depth of a typical low T
c
DC SQUID based on Nb/AlO
x
/Nb
Josephson junctions is ~ 20 to 30 V. The modulation depth oV is maximum for bias
currents
a little above the maximum critical current of the SQUID, and during operation t
he SQUID
is tuned to the bias current at which the modulation depth is a maximum.
Fig. 2. (a) dc SQUID (b) I-V characteristic of the dc SQUID with the applicatio
n of input
magnetic flux.
3. Flux Locked Loop (FLL) readout electronics
The periodic output voltage of the SQUID allows the SQUID to be operated in a sm
all signal
mode around the optimum working point but the linearity of response i
s limited to flux
range much less than u
0
/2. The small signal readout can only be used when the amplitude
of variation of magnetic flux signal is limited to the linear range around the
working point
(<u
0
/4). However, in most of the applications, the signal flux, which is
required to be
measured, varies from a fraction of a flux quantum to several hundred
s of flux quanta.
u
ext
= nu
0
u
ext
=(n+) u
0
I
I
b
V
J = u
0
/L
s
2I
0 2I
0
2J
u
ext
u
ext
oV
u0
u0
(b)
u
ext
I
b
I
b
/2+J I
b
/2-J
J V
Input
coil
(a)
29
Therefore, the measurement system based on the SQUID sensor should be
designed to
provide a wide dynamic range as required in any application. In order
to linearise the
periodic output voltage, the SQUID should be operated in a feedback loop as a nu
ll detector
of magnetic flux; the voltage at the output of the readout electronic
s will then be
proportional to the input signal flux. In order to suppress the 1/f noise and dc
drifts in the
preamplifier the signal of interest is shifted to frequencies well abo
ve the threshold of 1/f
noise by using high frequency flux modulation scheme. The flux modulat
ion scheme is
illustrated in fig.3. In this scheme, the signal flux, ou
sig
which is to be measured is
modulated by a high frequency carrier flux u
m
(t). The sinusoidal modulation flux, u
m
(t) of
frequency f
m
with a peak-to-peak value of nearly u
0
/2 is applied to the SQUID. When there
is no applied signal flux or the applied input flux is nu
0
, the SQUID produces output
voltage with a frequency twice the frequency of the modulation flux u
m
(t) and there is no
component at the modulation frequency present in this output. When this voltage
output is
fed to a lock-in detector referenced to f
m
, the output of the lock-in detector is zero. On the
other hand, when the applied signal flux is (n+1/4) u
0
, the SQUID output voltage has a
component at frequency f
m
, which is in-phase with the carrier frequency and the output of
the lock-in detector is a maximum. Similarly, when the signal flux is (n-1/4) u
0
, the SQUID
output voltage has a component at frequency f
m
, which is out of phase with the carrier
frequency and the output of the lock-in detector has a maximum negati
ve value. Thus, as
one increases the flux from nu
0
to (n+1/4) u
0
, the output from the lock-in detector
referenced to f
m
steadily increases from zero to a maximum positive value; if instead the flux
is decreased from nu
0
to (n-1/4) u
0
, the output from the lock-in detector referenced to f
m
decreases from zero to a maximum negative value. The lock-in output is integrate
d and fed
back to the same coil as that used for flux modulation via a feedba
ck resistor. The signal
flux ou
sig
applied to the SQUID produces an output of -ou
sig
from the feedback loop to
maintain a constant flux in the SQUID, while producing an output volt
age across the
feedback resistor, which is proportional to ou
sig
.
Fig. 3. Schematic representation of flux modulation scheme when the signal flux
is at (a)
nu
0
, (b) (n+1/4) u
0
and (c) (n-1/4) u
0
. (d) Linearized output as the signal flux varies from (n-
1/4) u
0
to
(n+1/4) u
0
.
(b)
(c)
(a)
u
m
u0
usig
u0
u
m
usig
u0
u
m
u
sig
(d)
V
u0/4
usig
-u0/4
30
The schematic diagram of the FLL readout electronics is shown in fig.4. In order
to achieve a
low system noise, it is necessary to match the output impedance of t
he SQUID device and
input impedance of the preamplifier. LT1028 (Linear Technology) was cho
sen to construct
the preamplifier. The spectral density of the voltage noise (e
n
) and current noise (i
n
) of the
preamplifier at a frequency of 100 kHz are specified by the manufactu
rer to be about 0.9
nV/\Hz and 1 pA/\Hz respectively. The optimum input impedance (e
n
/i
n
) of the
preamplifier is, therefore, about 900 O. Since the dynamic resistance
of the SQUID at its
optimum bias point is about 1 O, one requires a coupling circuit wit
h an impedance
transformation of about 900 for effective signal extraction from the SQUID. In a
ddition, the
bandwidth of the coupling circuit should be as large as possible to
extract the 100 kHz
modulation signal without attenuation. For this, a room temperature ste
p up transformer
with a turns ratio of 30 has been fabricated and used as an impedance matching c
ircuit. The
room temperature transformer consists of 10 turns of 24 SWG copper wi
re as primary coil
and 300 turns of 28 SWG copper wire as secondary coil wound on a toroidal ferrit
e core. The
inner and outer diameters of the toroidal core are 10 mm and 18 mm respectively.
The step
up transformer is housed in the preamplifier box and the preamplifier is mounted
at the top
of SQUID insert. The SQUID is biased with an optimum dc bias current, I
b
to get maximum
voltage modulation from the SQUID sensor. The magnetic flux, which is to be meas
ured (in
actual measurements) is applied to the input coil of the SQUID, which
is inductively
coupled to the SQUID loop via the mutual inductance, M
i
o \L
i
L
s
, where L
i
is the self
inductance of the input coil and L
s
is the self inductance of the SQUID loop. The signal flux
is modulated by a 100 kHz sinusoidal flux whose peak-to-peak amplitude is less t
han u
0
/2.
The modulated output voltage from the SQUID is stepped up by the imp
edance matching
transformer and further amplified by a two stage amplifier with suffic
ient gain and fed to
the signal input channel of the analog multiplier. The modulated output is phase
sensitively
detected with respect to the reference signal supplied from the same
100 kHz oscillator to
Fig. 4. Schematic diagram of the flux locked loop electronics with flux modulati
on scheme.
I
b
Ref
SQUID
@ 4.2K
Step up
Transformer
Amplifier
100 kHz, u
0
p-p
Multiplier
Oscillator
Phase shifter
Integrator
I/P
O/P
M
f
M
i
Offset
L
s
R
f
R
f
switch
Volt, V
f
31
the reference input channel of the analog multiplier. The output of t
he analog multiplier is
integrated and fed back as a current to the feedback coil in order
to counterbalance the
signal flux applied to the SQUID. The voltage, V
f
developed across the feedback resistor R
f
is
proportional to the input flux and the transfer function of the system (cV
FLL
/cu
sig
) is given
by (R
f
/M
f
), where R
f
is the feedback resistance and M
f
is the mutual inductance between the
feedback coil and SQUID (J. Clarke et al., 1976). When the feedback
switch is closed, the
feedback flux at the SQUID cancels the input flux and the system uses the SQUID
as a null
detector of magnetic flux. The voltage developed across the feedback resistor is
proportional
to the applied input flux and the time variation of this voltage is an exact rep
lica of the time
variation of the input signal flux coupled to the SQUID. The system
gain can be varied by
simply varying the value of the feedback resistor. The figures of merit of the S
QUID system
with the FLL readout electronics are the flux noise, slew rate and b
andwidth; these values
are listed in table 1 for the system developed in our laboratory.
Flux noise
(u
0
/\Hz)
Slew rate (u
0
/s) Bandwidth (kHz)
10 5x10
5
10
Table 1. Benchmarks of the SQUID system
4. Important issues in building SQUID based measuring systems
There are some important issues to be considered while designing and building me
asuring
instruments based on SQUID sensors. Since the sensing area of the SQU
ID is very small
(typically 10
-2
mm
2
), direct sensing of the signal flux leads to poor field resolution. The field
resolution can be improved to some extent by increasing the effective
area of the SQUID
sensors square washer (J.M. Jaycox & M.B. Ketchen, 1981). A better way to increas
e the field
resolution, however, involves the use of superconducting pickup loop wi
th larger area in
the form of a magnetometer or a gradiometer connected to the on-chip integrated
multi-turn
input coil which is magnetically tightly coupled to the SQUID. The noise energy
coupled to
the SQUID is minimum only when L
i
= L
p
, where L
i
and L
p
are the inductances of the input
coil and the pickup loop respectively. For a typical low-T
c
dc SQUID and radius of the
pickup loop r
p
~ 10 mm, one can expect the white magnetic field noise of the ord
er of 1
fT/\Hz. Such sensitivities can only be utilized fully inside a superconducting s
hield or in a
magnetically shielded room. In unshielded environments, fluctuations in
the earths
magnetic field, local fields at power line frequency and disturbances
due to strong sources
such as rotating machinery will dominate the applied magnetic field to
the SQUID. For
sensitive measurements using SQUID sensors, like the measurement of bio
magnetic fields
and high resolution measurement of magnetic susceptibility, the use of
gradiometers for
rejection of distant sources of magnetic noise is almost inevitable. F
ig.5(b) shows the
principle of a first derivative, axial gradiometer consisting of two p
ickup loops of equal
turns areas connected in series; the two loops are wound in opposition and are s
eparated by
the baseline, b. A uniform magnetic field B
z
ideally induces zero net screening current into
the two loops, and hence couples no net flux to the SQUID. A gradient cB
z
/cz, on the other
hand, induces a net screening current in the gradiometer, which flows through th
e input coil
and a corresponding flux is inductively coupled to the SQUID. Fig.5(c)
shows the
configuration of the second derivative axial gradiometer consisting of
two first derivative
gradiometers wound in opposition. For uniform magnetic field and first
order field
32
gradient, the net screening current induced in the second order gradio
meter is zero. The
second derivative of the magnetic field c
2
B
z
/cz
2
induces screening current in such a
gradiometer and couples a net flux to the SQUID.
Fig. 5. Pickup loop configuration with different geometries connected to the inp
ut coil of the
SQUID (a) magnetometer, (b) first order gradiometer and (c) second order gradiom
eter
The SQUID based measuring system has to be designed based on the nat
ure of the signal
source of interest. In some cases the signal source may be exposed to high tempe
rature, high
magnetic fields like magnetization measurements and in some other cases
like
nondestructive testing, biomagnetism, geophysics, SQUID microscope etc. the sign
al source
may have to be kept at room temperature and is often outside the cryostat.
5. SQUID based NDE system
The SQUID based NDE system designed and constructed in our laboratory
consists of a
SQUID probe with superconducting pickup loop in the form of first order axial gr
adiometer
housed in a nonmagnetic liquid helium cryostat, stepper motor driven precision X
Y scanner
with non-magnetic platform for the sample movement and the data acquis
ition module to
acquire the SQUID output signal with respect to the positional coordin
ates of the sample
which is under investigation.
5.1 SQUID probe
The SQUID probe comprises of a SQUID sensor and a superconducting fir
st order
gradiometer connected to the input coil of the SQUID. The SQUID senso
r is enclosed by a
superconducting magnetic shield while the superconducting gradiometer is
exposed to
detect the magnetic signal of interest. The local source of interest
generates a much larger
field gradient at detector (superconducting pickup loop) than that gene
rated by a more
distant noise source.
L
i
L
p
M
i
z
B V c o
B
z
(a)
L
i
L
p
) / (
2 2
z B V
z
c c o
b
B
z
b
(c)
M
i
L
i
L
p
b
) / ( z B V
z
c c o
B
z
(b)
M
i
33
A holder to mount the SQUID sensor together with a former to wind the pickup loo
p of the
gradiometer has been fabricated as a single piece and is connected at one end of
the 13 mm
diameter thin walled stainless steel (SS) tube designed to be insertib
le in a liquid helium
cryostat. The electrical leads in the form of twisted pairs as requir
ed for the SQUID sensor
are routed through the thin walled SS tube and terminated at the top
with appropriate
electrical feedthroughs. A first order gradiometer made of superconducting NbTi
wire with
diameter of 0.1 mm is wound over the former and is inductively coupl
ed to the SQUID
device. The gradiometer consists of two loops of 4-mm diameter wound
in opposition and
separated by a baseline of 40 mm. This configuration enables discrimination agai
nst distant
sources of magnetic noise which produce equal and opposite response fr
om the two loops
constituting the gradiometer. The SQUID sensor is shielded with a lead
cylinder with an
inner diameter of 16 mm and a length of about 120 mm.
The SQUID probe is housed in a FRP liquid helium cryostat having a capacity of 1
1.5 litres
of liquid helium. The cryostat has a low boil-off rate of under 2.2
litres per day and is
designed to have a minimum warm-to-cold distance of 6 mm. The cryostat is equipp
ed with
a top loading clear access of 25 mm diameter for the insertion of SQUID probe. T
he cryostat
was constructed with fiberglass-reinforced epoxy, which has been found
to have suitable
structural and thermal properties without introducing a high level of
magnetic noise and
distorting noise fields. The setup was calibrated by measuring the sys
tem response to the
magnetic field produced by a large circular coil and the calibration constant wa
s inferred to
be 20 nT/cm per flux quantum coupled to the SQUID. Fig.6. shows the cross sectio
nal view
of the SQUID probe housed in a liquid helium cryostat.
Fig. 6. Cross sectional view of the SQUID probe housed in liquid helium cryostat
. 1. LHe
cryostat, 2. Radiation shields, 3. Vacuum space, 4. SQUID, 5. Pickup loop
LHe
Pre amplifier
1
2
3
4
5
34
5.2 XY scanner
One of the most important requirements for SQUID based NDE system is
the scanner to
move the specimen in order to scan it under the stationary liquid he
lium cryostat. In most
SQUID based NDE systems, the specimen is moved under a rigidly fixed stationary
SQUID
system. Commercial general purpose scanners are not suitable for the S
QUID NDE
applications because of the use of magnetic materials in their constru
ction and the
consequent generation of heavy electromagnetic noise. Therefore a steppe
r motor driven
nonmagnetic precision XY scanner has been specially designed and fabric
ated for SQUID
NDE system operating in our laboratory to enable precision movement of the sampl
e.
The XY scanner has been designed for the scanning of flat plate samp
les with a positional
accuracy of 0.025 mm and repeatability of 0.1 mm. The detailed specif
ications of the XY
scanner have been listed in Table 2. The major components of the XY scanner are
computer
controlled XY stage, supporting platform which moves smoothly over a f
rictionless table
and a non-metallic and non-magnetic sample holder. The whole assembly
has been
mounted on a single frame. Necessary vibration isolation has been provided to mi
nimize the
transmission of floor vibrations. Care has been taken to isolate the SQUID syste
m from the
magnetic noise generated by the high torque stepper motors. The details of the
XY scanner
are discussed below.
Sample Flat plates
Scanning area
Step size
Speed
300 mm x 300 mm
0.25 mm to 5 mm
1 to 50 mm/sec
Table 2. Specifications of X-Y scanner
5.2.1 XY scanner: Computer controlled XY stage
The main component of the XY scanner is a computer controlled stepper
motor driven XY
stage. For flat samples the XY table is activated to scan the sample in either X
or Y directions.
The XY table has a traverse of 300 mm in each direction and is driven by two ste
pper motors
(model KML092F13, Warner SLOSYN). Each stepper motor is connected to a micro-ste
pping
driver (SS2000 MD808, Warner SLO-SYN) controlled by a programmable step
per motor
controller (model SS2000PCi, SLO-SYN). The motion is achieved by driving the lea
d screws
through the stepper motors to achieve the desired position of the tab
le. Each axis of the
stage is equipped with proximity switches which set the travel end-lim
its of the stage and
also define the home position. The home position is a reference point
from which all
distances are measured, and is always approached from the positive end of travel
limit with
a preset velocity and acceleration, to ensure the reproducibility of s
cans to better than 0.1
mm. In addition, there are safety limit switches at both ends for over travel pr
otection.
5.2.2 Supporting platform
Since the SQUID system is extremely sensitive to any magnetic flux changes, it i
s necessary
to keep this system far away from the stepper motors. The high frequ
ency current pulses
applied to the high torque stepper motors generate undesirable magnetic field no
ise, which
could disturb the locking of the sensitive SQUID system. To avoid such magnetic
noise, the
35
stepper motors have been located 2.5 m away from the SQUID system. T
he sample holder
has been mounted at one end of the supporting platform which is made
of non-magnetic
and non-metallic materials. Since the platform has a large length, two
pairs of roller tables
have been provided. The central one is made of a stainless steel case with brass
ball bearings
and the other one close to the sample holder is made of a nylon case with glass
ball bearings.
These rollers smoothly move over the frictionless tables. The supportin
g frame is made of
fiber glass square pipes. The fasteners used in the region close to
the sample holder are
made of nylon.
5.2.3 Sample holder
The sample holder is fabricated using non-metallic and non-magnetic mat
erials such as
fiberglass and polypropylene. Care has been taken to avoid the use of
any magnetic
components around the sample holder up to a radial distance of about 500 mm.
5.2.4 Shielding
Even though the stepper motors are kept far away from the SQUID syst
em, it was found
that the SQUID picks the magnetic field noise associated with the ste
pper motors causing
the SQUID system to frequently unlock. The noise level has been reduced to some
extent by
covering the stepper motors by two layers of - metal and by wrappin
g all the cables
coming from the control panel to XY stage by a flat copper braid. The flat coppe
r braid has
been properly grounded with the main power ground. The electrical leads coming f
rom the
SQUID to the preamplifier mounted at the top of the liquid helium cr
yostat and the leads
that connect the preamplifier to the flux locked loop electronics modu
le kept in the
instrument rack have also been shielded and the shield is properly grounded.
5.3 Data acquisition system
Fig.7 shows the schematic diagram of the SQUID based NDE measurement
system. The
specimen or the source of magnetic signal is mounted over the sample
holder of the XY
scanner. Whenever there is a change of magnetic flux in the vicinity
of the pick-up loop, a
screening current is induced in the pickup loop, which is inductively coupled to
the SQUID.
The SQUID in conjunction with the flux locked loop readout electronics generates
a voltage
output which is proportional to the magnetic field gradient, (cB
z
/cz) over the baseline of the
pick-up loop. The oscilloscope is used to display the necessary signals for tuni
ng the SQUID
readout electronics by setting optimum bias current, amplitude of modul
ation flux, etc. A
complete data acquisition system has been developed based on Visual Ba
sic. The user can
select the initial and the final positions of the sample coordinates,
step size and scanning
speed for each scan axis. The software selects grid points within the selected a
rea depending
upon the specified step size. The sample is scanned under the SQUID
system and the
SQUID output is recorded in the computer as a function of position c
oordinates. For every
specified step size, the intelligent micro-stepping drive unit sends TT
L pulse to the data
logger (model 34970A, Agilent), which stores the data in its buffer whenever a t
rigger pulse
is received from the drive unit; the first data is collected by a s
oftware trigger. After each
line scan (say y-axis) the sample is moved one step with the specifi
ed step size
perpendicular to y-axis (say x-axis). During this time the data is tr
ansferred from the data
36
logger to the computer and stored for further analysis. This wait tim
e is user selectable
before the start of scan. In addition to this, a linear scale has been provided
along each scan
axis and the actual position coordinates displayed in a digital readou
t. Provision has been
made for manual operation of each scan axis, and for selection of ho
me position for each
scan axis. Fig.8 shows the photograph of the SQUID based NDE system with XY scan
ner.
Fig. 7. Block diagram of the SQUID based NDE system
Fig. 8. Photograph of the SQUID based NDE system
LH
XY Scanner
Pickup loop
SQUID
Preamplifier
XY controller
(x,y coordinates)
FLL electronics Oscilloscope
Agilent
Data logger
PC
Cryostat
TTL
37
To check the performance of the system a steel ball weighing 2.2 mg
has been magnetized
by applying a magnetic field of about 35 Oe for a preset time of a
bout 5 minutes. The
magnetized steel ball has been scanned under the SQUID system to meas
ure the remanent
magnetization. After scanning, the steel ball has been removed and the
background noise
has been recorded. It may be noted that even though the sample platf
orm is made of non-
magnetic material, the Y table which moves over the X table in the XY scanner is
constructed
by using lead screws and linear guides which are magnetic. The contri
bution of the
background has been eliminated by subtracting the background from the signal. Fi
g 9 shows
the recorded SQUID output voltage for the magnetized steel ball while i
t is scanned under
the SQUID system. The amplitude of the magnetic anomaly associated with the 2.2
mg steel
ball is about 80 millivolts while the width of the magnetic anomaly arises on ac
count of the
stand-off distance of about 10 mm between the steel ball and the sensing loop. I
t may also be
noted that based on the noise floor, one can estimate that the SQUID based NDE s
ystem can
detect the presence of a steel particle weighing as low as 10g at signal-to-noise
ratio unity.
Fig. 9. Recorded SQUID output voltage for the steel ball scanning under the SQUI
D. (a) Steel
ball with background noise (b) background noise and (c) steel ball alone
6. Utilization of SQUID based NDE system
This SQUID system has been used for the measurement of subsurface def
ects by inducing
eddy currents in conducting materials at relatively low frequencies and
also for the
measurement of extremely low content of o-ferrite in the 316L(N) stain
less steel weldment
specimens which are subjected to low cycle fatigue at high temperatures.
6.1 SQUID based eddy current NDE
Conventional eddy current testing is the popular NDE technique for det
ecting subsurface
defects in conducting structures like aluminum, stainless steel etc. In
this technique, eddy
currents are excited in a conductor using a magnetic field varying in
time at an excitation
frequency f.
38
A suitable coil carrying AC current is used to induce eddy currents
in the specimen by
producing a time varying magnetic field. The physical basis for the eddy current
induction
in conducting material is contained in the Faradays law
B
E
t
c
V =
c
which can be rewritten

as
H
J
t
o
c
V =
c
(3)
where is the magnetic permeability and o is the electrical conductivity of the m
aterial. The
time varying applied magnetic field will induce a current distribution inside th
e conductor
in accordance with equation 3 and these currents are referred to as eddy current
s.
If the applied magnetic field varies sinusoidally with respect to time, H = H
0
e
i
e
t
, then eq. (1)
becomes
J i H eo V = (4)
In qualitative terms, the curl of the eddy currents is out of phase
with the applied time
varying magnetic field, is proportional to the frequency of the applie
d magnetic field and
opposes the changes in the applied magnetic field with time. In conve
ntional eddy current
NDE, sinusoidal current with the frequency ranging from 10 kHz to 10
MHz is passed
through an excitation coil having hundreds of turns and the voltage induced acro
ss the coil
is measured to determine the impedance of the coil. The presence of
a conductor near the
coil affects its impedance and the changes in the coil impedance are
measured which are
associated with changes in some characteristics of the conductor (varia
tion in local
thickness, conductivity or permeability due to the presence of defects). In the
case of SQUID
based eddy current NDE the change of the magnetic field due to the presence of a
defect is
directly measured by the SQUID. The defects perturb the flow pattern of the indu
ced eddy
currents in the conductor and hence they manifest as localized magneti
c anomalies when
the conducting plate is scanned under the SQUID sensor. The amplitude
and shape of the
magnetic anomaly corresponding to a defect depend on the amplitude of
the eddy current
flowing in the immediate vicinity of the defect and also on the defect character
istics such as
size, orientation etc. The use of higher excitation frequencies tends to increas
e the amplitude
of the eddy currents induced on the surface; however, the strength of
the eddy current
induced exponentially decreases with depth d measured from the surface as e
-d/
o
,
where o is
the skin depth defined as
1 2
1
f
o
t o
| |
=
|
\ .
(5)
Here f is the frequency of the excitation current used to induce ed
dy current in the
conducting material. In the conventional eddy current testing, excitation freque
ncies higher
than 10 kHz are generally used to increase the voltage induced in th
e pick-up loop;
therefore, the depth of the defect detection is limited by the skin
depth (o) of the material,
which is rather low at such high frequencies. If the excitation frequency is red
uced in order
39
to increase the skin depth, the induced voltage in the pick-up loop
also decreases, often
reaching the system noise floor, and thus severely limiting the depth of defect
detection. The
system based on the use of SQUID as a detector promises better signa
l-to-noise ratio at
lower operating frequencies. Since the sensitivity of the SQUID device is indep
endent of the
operating frequency from near dc to several kHz (white noise regime), it is poss
ible to use a
SQUID based system for the detection of deep subsurface defects, which are not d
etectable
in the conventional eddy current testing owing to skin depth limitatio
ns. In the SQUID
based system, pickup loop can be wound using superconducting wire; unlike normal
metal
pickup loops for which induced voltage is proportional to the rate of
change of flux,
superconducting pickup loop directly senses the change in magnetic flux and henc
e its use
is crucial in retaining the requisite low frequency sensitivity. Severa
l groups have
successfully utilized this feature to detect subsurface defects not detectable b
y conventional
eddy current techniques (W.G. Jenks et al., 1991; P. Chen et al., 2002; H. Weins
tock, 1991; M.
Mck et al., 2005; U. Klein et al., 1997).
The excitation coil used to induce eddy currents in the SQUID based
eddy current NDE is
also often different from that used in the conventional eddy current
methods. In
conventional testing, the excitation coil is in the form circular coil and in so
me cases a set of
differential coils are used. To enhance the eddy current in the speci
men, it is required to
apply a large time varying magnetic field. If we use circular excit
ation coil for the SQUID
based eddy current NDE system the direct magnetic field coupled to th
e SQUID is very
large and the output voltage of the flux locked loop readout electron
ics may reach
saturation levels. Under these conditions, the changes of magnetic field due to
the presence
of a defect cannot be detected by the SQUID. To overcome this, a double D excitati
on coil
at room temperature, placed just below the cryostat, is often used to excite edd
y currents in
the specimen at room temperature. This double D configuration for the e
xcitation coil
enables realization of a wide dynamic range since the direct coupling of magneti
c flux to the
pick-up loop is a minimum when the axis of the excitation coil coinc
ides with that of the
pick-up loop. The schematic view of the double D excitation coil and t
he location of the
superconducting pick-up loop are shown in fig.10. The central axes of the excita
tion coil and
the pick-up loop are adjusted to coincide and held stationary during the measure
ment. The
current in the double D coil flows through the central straight segment
and then around
the perimeter of the semicircular loops. Thus the field through one D
section of the coil is
always 180 degrees out of phase with the field through the other D
section of the coil. The
specimen, which is to be evaluated, is kept at room temperature at a
stand-off distance of
about 14 mm from the bottom loop of the superconducting gradiometer l
ocated inside the
cryostat and is scanned under the stationary liquid helium cryostat. E
ddy currents were
excited in the specimen under investigation using the double D coil with 120 turns
and an
overall diameter of 40 mm carrying a current of 150 mA(rms). Changes in the indu
ced eddy
current flow associated with the presence of a defect manifest as chan
ges of the flux signal
detected by the SQUID device as the specimen is scanned. The SQUID, in turn, pro
duces an
output voltage corresponding to these changes in magnetic flux. This o
utput is phase
sensitively detected by a lock-in-amplifier whose reference channel is
fed by the oscillator,
which is used to supply the necessary excitation current to the double-D coil. T
he schematic
diagram of the experimental set-up for the SQUID based eddy current N
DE is shown
in fig.11.
40
Fig. 10. Schematic view of the double D excitation coil and pickup loop arrangemen
t
Fig. 11. The SQUID based eddy current NDE system

Both the real (in-phase with the excitation current) and imaginary (90 out-of-pha
se with the
excitation current) components of the SQUID output voltage were measured using t
he lock-
in-amplifier and were simultaneously recorded by a computer with respect to the
positional
coordinates of the defect as the plate was scanned under the stationa
ry cryostat using a
precision X-Y scanner.
Double D
excitation coil
Pickup loop
L
i
) / ( z B V
z
c c o
M
i
Agilent Data
logger
Preamplifier
Reference
Input
LHe
XY Scanner
Pickup loop
SQUID
XY controller
(x,y coordinates)
FLL electronics Oscilloscope
PC
Cryostat
TTL pulse
Lock-in amplifier
Oscillator with
power amplifier
Output
Excitation
current
Defect
41
6.2 Remnant magnetization measurements
Fatigue damage and effect of thermal ageing are the most important problems for
a number
of high temperature components in power plants particularly in welded regions. A
ustenitic
stainless steels are the preferred materials for sodium cooled fast br
eeder reactors as they
possess excellent high temperature mechanical properties for nuclear rea
ctor service. A
nitrogen-added low carbon version of SS 316 austenitic stainless steel
(316 L(N)) has been
chosen for the high temperature components of Prototype Fast Breeder R
eactor (PFBR) in
India. A major consideration during the welding of 316 L(N) stainless steel is i
ts resistance to
hot cracking. The presence of an optimum amount of delta ferrite in
the austenitic weld
metal is desirable to prevent hot cracking in the weldment. However the o-ferrit
e structure
is highly unstable during high temperature service and transforms to c
arbides and brittle
intermetallic phases, e.g. sigma phase. Small quantities of these phase
s would cause large
variations in the mechanical properties and influence corrosion behaviou
r of the material.
Hence the knowledge of these phases such as type, amount etc is essential in des
igning the
optimal operational parameters to achieve the desired lifetime for weld
ed stainless steel
components. Since the o-ferrite is a magnetic phase, one can evaluate the o-ferr
ite content by
measuring remanent magnetization of the weldment sample. Since the SQUID device
has an
extremely high sensitivity for magnetic flux signals, one can detect an extremel
y low content
of o-ferrite in the weldment samples through remanent magnetization mea
surements. The
experimental set up for the remanent magnetization measurements is similar to th
at shown
in fig.11 except that the double Dexcitation coil is not used. In this case, the S
QUID output
is directly read by data logger and recorded by the computer with respect to the
coordinates
of the sample as it is scanned under the stationary cryostat.
7. Results of experimental investigations
We have carried out two different kinds of studies such as detection
of artificially
engineered subsurface flaws in aluminum plates by using the SQUID base
d eddy current
technique and NDE of fatigue cycled stainless steel weldment specimens
to detect the o-
ferrite content through remanent magnetization measurement by the SQUID system.
7.1 Detection of subsurface flaws in aluminum plates
The SQUID based NDE system has been initially tested with an aluminum
plate with a
length of 300 mm, width of 100 mm, and thickness of 10 mm with art
ificially engineered
defects, as shown in fig. 12 (a). Two rectangular defects have been
engineered with a
separation of 150 mm. One defect has a length of 50 mm, width of 1
mm, and height of 1
mm, and another defect has a length of 50 mm, width of 1 mm, and
height of 0.5 mm and
the magnetic anomalies corresponding to these two defects were studied
as part of initial
exploratory studies on subsurface defect characterization. Eddy currents
were excited at a
relatively low frequency of about 200 Hz using double D excitation coil
. The aluminum
plate with defects on its bottom surface was scanned under the SQUID
probe and the
changes of magnetic field associated with the defects were recorded wi
th respect to the
positional coordinates. Fig. 12 (b) shows the magnetic anomalies associ
ated with these
defects recorded using a SQUID. The amplitude of the magnetic anomaly correspond
ing to
a defect having a height of 1 mm was 2.3 nT/cm. Since the noise floor of the sys
tem is few
42
tens of pT/cm, it is estimated that even a localized loss of conductor thickness
as low as 0.1
mm can be detected at a depth of 9 mm at signal to noise ratio un
ity. The potential of the
system for detection of subsurface defects using low-frequency eddy cur
rent excitation is
evident from this data.
Fig. 12. (a) Aluminum plate with artificially engineered defects. (b) Magnetic a
nomalies
associated with the defects buried at a depth of 9 mm below the surface recorded
by the
SQUID.
Subsequently detailed experimental studies have been carried out for th
e determination of
optimum eddy current excitation frequencies for the defects located at
different depths
below the top surface of an aluminum plate. This optimum excitation f
requency
corresponds to the frequency at which the amplitude of the magnetic a
nomaly associated
with a defect at a certain depth reaches a maximum value. The optimu
m excitation
frequency has been theoretically investigated by Sikora et al (R. Siko
ra et al., 2003)
and by
Baskaran and Janawadkar (R.Baskaran and M.P. Janawadkar, 2004)
using the analytical
solution (C.V.Dodd & W.E. Deeds, 1968) to the distribution of eddy cu
rrents excited by a
circular coil in a defect free semi-infinite conductor. It may, howeve
r, be noted that the
results of these theoretical investigations cannot be directly compared with the
experimental
results reported here since in the present studies, a non-axisymmetric
double-D coil was
used as an excitation coil to induce eddy currents in a conductor st
ack of finite thickness
with a defect embedded within the stack.
A set of aluminum plates with a length of 300 mm, width of 200 mm
and with different
thicknesses ranging between 2mm to 12 mm in steps of 2 mm were fabricated for th
is study.
An artificial defect is engineered in a separate 1 mm thick aluminum
plate with a defect
length of 60 mm and width of 0.75 mm to simulate a localized loss of conductor v
olume in
the plate. The length and width of the plate carrying the defect are identical
to those of the
plates without defects. SQUID based eddy current NDE measurements have
been carried
1 mm 0.5 mm
12(a)
10 mm
12(b)

43
out for the stack of aluminum plates with the defect plate located at different
depths ranging
from 2 mm to 14 mm (in steps of 2 mm) while keeping the total thickness of the c
onductor
stack to have a constant value of 15 mm throughout the experimental investigatio
ns. As an
example, to simulate the defect located at a depth of 2 mm below the top surface
, the defect
plate was sandwiched between a 2 mm thick aluminum plate at the top and a 12 mm
thick
aluminum plate at the bottom. An interchange of the top and bottom plates shifts
the defect
location to 12 mm below the top surface. In this way, the location of the defect
was shifted to
different depths from 2 mm to 14 mm below the surface. A photograph
of the aluminum
plates used in this study in order to vary the defect depth is show
n in fig. 13. The stack of
plates was scanned under the SQUID in a direction such that the defe
ct length is
perpendicular to the scanning direction. To avoid edge effects, the sc
anning region was
limited to a total length of 120 mm (60 mm on each side of the de
fect). For each scan, both
the real and imaginary components of the SQUID output voltage were measured usin
g the
lock-in-amplifier and were recorded with respect to the positional coordinates o
f the defect.
For each defect depth, the experiments were repeated at different exci
tation frequencies
ranging between 33Hz and 853Hz in steps of 20 Hz.
Fig. 13. Aluminum plates used in the study (plate with the defect is also shown)
.
As an illustration of the magnetic anomaly detected by the SQUID system, fig. 14
shows the
measurements corresponding to defect located at a depth of 14 mm belo
w the top surface
when the eddy currents are excited at the frequency of 33 Hz, 103 H
z and 243 Hz. At the
lower excitation frequency of 33 Hz, the skin depth is high and the
resultant amplitude of
the change of magnetic field gradient due to the defect is 0.657 nT/
cm. At the higher
excitation frequency of 103 Hz, the skin depth is lower but the indu
ced eddy current
amplitude is higher and thus the detected magnetic anomaly has a higher amplit
ude of 1.8
nT/cm. When the excitation frequency is further increased to 243 Hz,
the exponential
attenuation of the eddy current amplitude at the location of the defe
ct more than
compensates for the small linear increase resulting from the use of h
igher frequency of
excitation; the amplitude of the detected magnetic anomaly therefore de
creases to 1.356
nT/cm. In this case, as shown in fig.14 (b) one can also clearly see the signal
changes due to
the small variations in lift-off as the sample moves relative to the gradiometri
c pickup loop.
At the optimum excitation frequency of 103 Hz, the signal to noise ratio is evid
ently better.
It may also be noted that the phase of the magnetic anomaly changes
systematically when
the excitation frequency is varied as shown in fig.14. The change in the resulta
nt amplitude
of the magnetic anomaly corresponding to a defect located at a depth
of 14 mm below the
surface when eddy currents are excited at different frequencies is shown in fig.
14(d). At low
frequencies, the eddy current induced in the conductor stack has low amplitude a
nd hence
the amplitude of the magnetic anomaly associated with the defect is a
lso low. As the
44
excitation frequency increases, the amplitude of the eddy current induc
ed on the surface
increases monotonically; however, owing to the skin effect, the amplitu
de of the eddy
currents at a depth of 14 mm increases initially, reaches a maximum
at the optimum
excitation frequency of 103 Hz and decreases steadily thereafter. At higher fre
quencies, the
eddy current tends to be concentrated at the upper surface of the ma
terial due to the skin
effect. Therefore the amplitude of the magnetic anomaly due to the presence of t
he defect at
a depth of 14 mm decreases when excitation frequencies higher than 103 Hz are us
ed. In the
present experimental investigations, the excitation frequency was varied from 33
Hz to 253
Hz in steps of 10 Hz and the optimum frequency was determined to be 103 Hz for t
he defect
located at a depth of 14 mm below the surface. The experiments were
repeated for defects
located at different depths of 12 mm, 10 mm, 8 mm, 6 mm, 4 mm and
2 mm from the top
surface and a comprehensive characterization of the magnetic anomalies was carri
ed out as
a function of defect depth and frequency. Fig.15 shows the magnetic a
nomaly detected by
the SQUID system as the sample with defect located 10 mm below the
surface is scanned.
These plots show the variations in the magnetic anomalies resulting fr
om the use of
different excitation frequencies on the same scale. Fig. 16 shows the changes in
the resultant
amplitudes of the magnetic anomalies corresponding to defects located a
t different depths
from the surface when eddy currents are excited at different frequenci
es. The excitation
frequencies were varied in steps of 10 Hz for the defects at greater depths and
20 Hz for the
defects at lower depths. The experimental parameters such as standoff distance,
excitation
Fig. 14. (a),(b) and (c) show the changes in the magnetic anomaly corresponding
to the
defect buried at a depth of 14 mm below the surface with eddy current excitation
at a
frequency of 33 Hz, 243Hz and 103 Hz respectively. (d) The resultant amplitude o
f the
magnetic anomaly due to the defect at different excitation frequencies.
45
Fig. 15. The magnetic anomalies detected by the SQUID system for different frequ
encies
when the defect was located 10 mm below the surface.
Fig. 16. The resultant amplitude of the magnetic anomalies corresponding to defe
cts located
at different depths when excited at different frequencies
coil current, flux locked loop electronics gain, scanning speed, step size, time
constant of the
lock in amplifier etc were kept constant for the entire series of experiments. T
he frequency at
which the amplitude of the measured magnetic anomaly reaches a maximum value is
taken
to be the optimum frequency of excitation for defect located at a particular dep
th. As a result
of these detailed experimental investigations, the optimum excitation frequencie
s have been
estimated for the defects located at different depths from the surface and are l
isted in table 3.
46
Depth (mm) 2 4 6 8 10 12 14
f
opt
(Hz) 453 313 233 183 143 123 103
Table 3. Optimum excitation frequencies for the defect at different depths
As shown in fig.17, the square root of the optimum excitation frequen
cy is found to be
inversely proportional to the defect depth. For the defects at lower depths from
the surface,
the signal to noise ratio is excellent for a wide range of frequencies; for the
defects at greater
depths from the surface, however, the range of frequencies which can
be used for defect
detection becomes narrow due to the poor signal to noise ratio and t
he skin effect of the
conducting materials. It may also be noted that such defects can be detected on
ly by using
the SQUID based system for nondestructive evaluation.
Fig. 17. The variation of the optimum excitation frequency with defect depth
7.2 Remanent magnetization measurements on fatigue cycled stainless steel
weldments
The 316L(N) stainless steel weldment specimens were prepared by welding
316L(N) base
metal with 316N electrodes by a manual metal-arc welding process. Weld
ing was carried
out on a 25 mm thick plate with a double-V configuration with an included angle
of 70
o
(Fig.
18). Welds were radio-graphed and only sound joints were taken for th
e fabrication of the
specimens. The low cycle fatigue (LCF) tests were conducted at a strain amplitud
e of 0.6%
using an Instron servohydraulic fatigue testing machine under total axi
al strain control
mode at 600
o
C. A strain rate of 3 x 10
-3
s
-1
was employed for the test. The magnetic flux
signal arising from the remanent magnetization of the sample was coupl
ed to the SQUID
device via a superconducting flux-transformer. As the sample was scanne
d under the
cryostat, SQUID output revealed a characteristic magnetic anomaly when
the center of the
gauge length of the sample passed under the pickup loop indicating th
e presence of a
magnetic phase at the location of the weld. To show the repeatability
of the XY scanner a
virgin weldment sample was scanned under the SQUID system for measuring the rema
nent
magnetization. The sample was placed at room temperature; about 25 mm below the
pickup
loop and scanned using the XY scanner. The scanning started far away from the sa
mple and
as the sample approaches the pick up loop, SQUID output increased and
it subsequently
decreased as the sample moved away from the pick up loop. The experiment has als
o been
repeated by rotating the sample by 90
o
, 180
o
and 270
o
about its central axis. As the sample
was rotated about its axis, depending on the orientation of the magnetization of
the sample
47
relative to the axis of the pick-up loop, the shape of the magnetic anomaly chan
ged from bell
shaped to dipolar The magnetic flux profile measured near the center
of the weldment
sample for different angles of rotation is shown in Fig. 19. It may be noted tha
t the center of
the measured magnetic anomaly coincides for all scans showing excellent
repeatability of
the XY scanner. To show the system capability in measuring the virgin
sample which was
strongly magnetic and fractured sample which was magnetically very weak
, the two
samples were kept at a separation of 150 mm and scanned under the SQUID. Fig. 20
shows
the magnetic flux profile for the virgin and the fatigue fractured sa
mple. For the virgin
sample, the amplitude of the magnetic anomaly was measured to be ~1.7
8 u
0
, whereas for
the sample which was subjected to fatigue testing at 600
o
C until fracture, the amplitude of
the magnetic anomaly was measured to be as small as ~0.05 u
0
under identical
measurement conditions. To complete the SQUID based measurements, measurement of
o -
ferrite was also carried out using a magnagage along the 25 mm gauge
length of both the
virgin and LCF tested samples. Figs. 21 (a) and (b) show the o - ferrite distrib
ution profiles in
the virgin and LCF tested specimens respectively. Measurements were mad
e in two
Fig. 18. Weld pad geometry used for the weld joint specimens.
Fig. 19. Magnetic line profiles perpendicular to the gauge length of the weldmen
t samples
for different angles of rotation.
48
Fig. 20. Comparison of the magnetic flux profiles for the virgin and the fractur
ed samples.
Fig. 21. (a) Delta ferrite distribution profile in the virgin sample. (b) Delta
ferrite distribution
profile in the LCF tested sample.
different orientations 90
o
apart from the central axis. The o - ferrite was seen to peak around
the mid-gage portion consisting of the joint proper. A considerable reduction in
the amount
of o - ferrite after the LCF cycling could be seen. The experiments
clearly show that the
magnetic phase present in the virgin samples transforms to a non-magne
tic phase (o or
carbides, etc.) during the fatigue deformation at 600
0
C. The transformation of o - ferrite to
brittle o - phase has been known to influence the cracking behavior and thereby
the fatigue
life in the weld metals and weld joints of 316 and 316 L(N) stainless steels (M.
Valsan et al.,
1995, 2002; A. Nagesha et al., 1999). The amount of transformation of
the o - ferrite into
brittle phases has also been found to be a strong function of the temperature an
d frequency
of testing (A. Nagesha et al., 1999). The results reported here demon
strate the potential of
SQUID based measurements in tracking the magnetic-nonmagnetic (ex.o-o) t
ransformation
in stainless steel welds when subjected to high temperature fatigue lo
ading. A series of
experiments has been carried out in the stainless steel weldment specimen
s to evaluate the
49
transformation of magnetic phase to nonmagnetic phases at different lev
els of fatigue
deformation. In this study, a single weld joint was selected for rema
nent magnetization
measurements in the virgin state and subsequently after every 50 fatig
ue cycles. At every
stage prior to fatigue cycling and remanent magnetization measurement, the weld
joint was
properly demagnetized by subjecting it to low-frequency alternating magn
etic field to
eliminate the influence of the past history and then remagnetized by
applying a preset dc
magnetizing field (35 Oe) for a preset time (300 s) before the measu
rement of remanent
magnetization using the SQUID-based setup commenced. The peak value of
demagnetization field was kept slightly higher than the magnetizing fie
ld. To evaluate the
relative changes in the magnetic content of the weldment specimen, the
parameters of the
measurement setup such as stand-off distance, FLL gain, etc., were mai
ntained constant
throughout the whole series of measurements. The virgin weldment specim
en gave a
maximum SQUID signal of 15 u
0
, which decreased rapidly to 6.67 u
0
when the sample was
subjected to LCF for 50 cycles. Thereafter, no significant changes in
the maximum SQUID
signal could be observed up to 150 cycles. However, a marked decrease in the SQU
ID signal
was noticed when the sample was subjected to LCF for 200 cycles acco
mpanied by the
initiation of a crack at the boundary of the weldment specimen. Micro
cracks were seen at
the boundary of the weldment specimen when the specimen was examined
through a
microscope. The magnetic profile of the weldment specimen scanned under
the SQUID
probe is shown in fig. 22 in the virgin state as well as after sub
jecting the specimen to
different levels of fatigue deformation. Fig. 23 shows the variation in the maxi
mum SQUID
signal as a function of the number of cycles of fatigue loading and
portrays the
transformation of magnetic o-ferrite to nonmagnetic phases when the weldment spe
cimen is
subjected to LCF at 600
0
C. It may be noted that although the initial signals were large,
SQUID-based measurements were intended to look for small magnetic anoma
lies, if any,
which could be correlated with the residual life of the weldment spec
imen. Further
experiments are underway to clarify these issues.
Fig. 22. Magnetic flux profile for the virgin and fatigue cycled weldment specim
en
50
0 100 200 300
0
5
10
15
20
Number of cycles
S
Q
U
I
D
o
u
t
p
u
t
(
u
0
)
0
100
200
300
400
F
i
e
l
d
g
r
a
d
i
e
n
t
(
n
T
/
c
m
)
Fig. 23. Change of SQUID output vs number of fatigue cycles for the weldment spe
cimen
subjected to fatigue deformation
8. Conclusions
It is evident that the high magnetic field sensitivity of the SQUID
sensors at relatively low
frequencies can be usefully harnessed for a number of applications in
nondestructive
evaluation. In this chapter, we have attempted to illustrate the wide spectrum o
f possibilities
by describing only a few representative studies and the reader is dir
ected to the vast
published literature in this emerging area. With the advent of HTS SQUID sensors
operating
at liquid nitrogen temperatures, SQUID based NDE is poised to be incr
easingly used in
critical applications where the requirement of high sensitivity, not at
tainable by any other
sensor technology, overrides the cost of cryogenic operating temperatures.
9. References
Bain, R.J.P; Donaldson, G.B.; Evanson, S. & Hayward, G. (1985) SQUID
gradiometric
detection of defects in ferromagnetic structures. In SQUID 85 - Superco
nducting
Quantum Interference Devices and their Applications, (eds.) Hahlbohm, H.
D. &
Lubbig, H., pp. 841-846, ISBN 0899251447, Walter de Gruyter & Co., Berlin, New
York
Bain, R.J.P.; Donaldson, G.B.; Evanson, S. & Hayward, G. (1987). Desig
n and operation of
SQUID-based planar gradiometers for non-destructive testing of ferromagne
tic
plates. IEEE Trans. Magn. Vol. 23, No. 2, pp. 473-476, ISSN: 0018-9464
Baskaran, R. & Janawadkar, M.P. (2004). Universal dependence of optimum
frequency on
defect depth for SQUID based eddy current evaluation. Proc. National
Seminar on
NDE, Pune, India, December 2004
Chen,P.; Chen, L.; Li, J. & Ong, C.K. (2002). Novel excitation coils
for non-destructive
evaluation of non-magnetic metallic structures by high-T
c
dc SQUID.
Superconductor Science and Technology, Vol. 15, No.6, pp. 855-858, ISSN 0953-204
8
51
Clarke, J.; Goubau, W.M. & Ketchen, M.B. (1976). Tunnel junction d.c
SQUIDs: fabrication,
operation and performance. J. Low Temp. Phys., Vol. 25, No. 1-2, pp.
99-144, ISSN
0022-2291
Clarke,J. (1993). SQUIDs: theory and practice. in The New Superconducti
ng Electronics
(eds) Weinstock, H. & Ralston, R. W., Kluwer, Dordrecht, pp.123180, ISB
N 0-
7923-2515-X
Dodd, C.V. & Deeds, W.E. (1968). Analytical Solutions to Eddy-Current
Probe-Coil
Problems, J. App. Phys., Vol. 39, pp. 2829-2838, ISSN 0021-8979.
Donaldson, G.B.; Cochran, A. & McKirdy, D.A. (1996) in: H. Weinstock
(Ed.), SQUID
sensors: Fundamentals, Fabrication and Applications, Kluwer, Dordrecht, pp
. 599-
628, ISBN 0-7923-4350-6
Evanson, S.; Bain, R.J.P.; Donaldson, G.B.; Stirling, G. & Hayward, G.
(1989) A comparison
of the performance of planar and conventional second-order gradiometers coupled
to a SQUID for the NDT of steel plates, IEEE Trans. Magn. Vol. 25, No. 2, pp. 12
00-
1203, ISSN: 0018-9464
Hans Koch, H. (1989) in Sensors A Comprehensive Survey (eds Boll,
R. & Overshott, K. J.),
VCH Publishers, NY, Vol. 5, pp. 381445, ISBN 978-3-527-62058-6
Jaycox, J. M. & Ketchen, M. B. (1981) Planar coupling scheme for ultra low noise
DC SQUIDs
IEEE Trans. Magn., Vol. 17, pp. 401403, ISSN 0018-9464
Jenks, W.G.; Sadeghi, S.S.H. & Wikswo, J.P. (1997) SQUIDs for nondestr
uctive evaluation.
J.Phys.D. Vol. 30, pp. 293-323, ISSN: 0022-3727
Klein,U.; Walker, M.E.; Carr, C.; McKirdy, D.M.; Pegrum, C.M.; Donaldso
n, G.B.; Cochran,
A. & Nakane, H. (1997). Integrated low-temperature superconductor SQUID
gradiometers for nondestructive evaluation. IEEE Trans. Appl. Supercond.,
Vol. 7,
No. 2, pp. 3037-3039, ISSN 1051-8223
Krause, H.-J. & Kreutzbruck, M.v. (2002) Recent developments in SQUID
NDE, Physica C,
Vol. 368, No. 1, pp. 70 79, ISSN 0921-4534
Krieger, J.; Krause, H.-J.; Gampe, U. & Sawade, G.(1999) . Magnetic f
ield measurements on
bridges and development of a mobile SQUID system. in: Proc. Intl. Con
f. on NDE
Tech. Aging Infrastr., pp. 229-239, Newport Beach, USA, (March 1999)
Lang, M.; Johnson,J.; Screiber, J.; Dobmann, G.; Bassler, H.J.; Eifler,D.; Ehrli
ch, R. & Gampe,
U. (2000). Cyclic deformation behaviour of AISI 321 austenitic steel a
nd its
characterization by means of HTC-SQUID. Nuclear Engineering and Design, Vol. 198
,
No.1, pp. 185-191, ISSN 0029-5493
Mck, M.; Korn, M.; Welzel, C.; Grawunder, S. & Schlz, F. (2005). Nonde
structive
evaluation of various materials using a SQUID-based eddy-current system.
IEEE
Trans.Appl.Supercond. Vol. 15, No. 2, pp. 733-736, ISSN 1051-8223
Nagesha, A.; Valsan, M.; Bhanu Sankara Rao, K. & Mannan. S.L.(1999) S
train rate effect on
the low cycle fatigue behaviour of type 316L(N) SS base metal and 31
6 SS weld
metal, Proceedings of the international welding conference IWC99, Vol. 2
,
February 1517, 1999, New Delhi, pp. 696-703
Sawade, G.; Straub, J.; Krause, H.J.; H. Bousack, Neudert, G. & Ehrli
ch, R. (1995). Signal
analysis methods for remote magnetic examination of pre-stressed element
s. in:
Schckert, G. & Wiggenhauser, H. (eds) Proc.Intl. Sympos.on NDT in Civil Eng. (ND
T-
CE), vol. II, DGZfP, Berlin, pp. 1077-1084
52
Sikora,R.; Chady, T.; Gratkowski, S.; Komorowski, M. & Stawicki, K. (2
003). Eddy Current
Testing of Thick Aluminum Plates with Hidden Cracks. Review of progres
s in
Quantitative nondestructive evaluation, Vol. 22, AIP Conference Proceedings, Vol
. 657,
pp. 427-434, ISBN 0-7354-0117-9, Bellingham, Washington, July 2002
Tavrin, Y.; Siegel, M. & Hinken, J.-H. (1999). Standard method for de
tection of magnetic
defects in aircraft engine discs using a HTS SQUID gradiometer. IEEE Trans. Appl
.
Supercond. Vol. 9, No.2, pp. 3809-3812, ISSN 1051-8223
Valsan, M.; Sundararaman, D.; Bhanu Sankara Rao, K. & Mannan, S.L. (1
995). High
Temperature Low Cycle Fatigue of Steels and their Welds, Metall. Mater
. Trans.,
Vol. 26A, pp. 1207-1219, ISSN 1073 - 5623
Valsan, M.; Nagesha, A.; Bhanu Sankara Rao, K. & Mannan, S.L. (2002).
High temperature
low cycle fatigue and creep-fatigue interaction behaviour of 316 and 3
16(N) weld
metals and their weld joints. Trans Ind Inst Met. Vol. 55, No. 5 pp
. 341-348, ISSN
0019-493X
Weinstock. H. & Nisenoff, M. (1985) Nondestructive evaluation of metallic struct
ures using
a SQUID gradiometer, SQUID 85, Proc. 3rd Int. Conf. On Superconducting Quantum
Interference Devices and their Applications (eds) Hahlbohm, H.D. & Lubbi
g, H.
(Berlin: deGruyter) pp. 843847 ISBN 0899251447
Weinstock, H. & Nisenoff, M. (1986) Defect detection with a SQUID magnetometer R
eview of
Progress in QNDE 6 (ed) Thompson, D.O. & Chimenti, D. (New York: Plenum) 669-
704
Weinstock, H. (1991). A review of SQUID magnetometry applied to nondes
tructive
evaluation. IEEE Trans. Mag., Vol. 27, No. 2, (March 1991), pp. 3231-3
236, ISSN
0018-9464
Weinstock, H. (Ed.), (1995) SQUID sensors: Fundamentals, Fabrication and
Applications,
Kluwer, ISBN 0-7923-4350-6, Dordrecht
3
Applications of Current Technologies for
Nondestructive Testing of Dental Biomaterials
Youssef S. Al Jabbari
1,2,*
and Spiros Zinelis
3,4
1
Director, Dental Biomaterials Research and Development Chair, School of Dentistr
y,
King Saud University, Riyadh,
2
Prosthetic Dental Sciences Department, School of Dentistry,
King Saud University, Riyadh,
3
Department of Biomaterials, School of Dentistry, University of Athens, Athens
4
Dental Biomaterials Research and Development Chair, School of Dentistry,
King Saud University, Riyadh
1,2,4
Saudi Arabia
3
Greece
1. Introduction
Dental biomaterials consist of a wide range of synthetic products that are norma
lly used to
restore patients oral health, function and aesthetic appearance. Generall
y, dental
biomaterials are classified on the basis of their atomic bonding into
metallic, ceramic,
polymer and composite materials. In addition to this classification, bi
omaterials can be
classified according to their interactions with the surrounding oral tissues (Ma
no et al. 2004).
Based on the tissue responses, biomaterials can be divided into three
different categories
commonly known as bioinert, bioactive and bioresorbable materials.
Bioinert materials are materials that have minimal interactions with t
he surrounding
tissues, such as partially stabilized zirconia, alumina, pure titanium
and some of its
alloys, some grades of stainless steel and ultra high molecular weight polyethy
lene.
Bioactive materials are materials that start to interact positively wi
th the surrounding
hard and sometimes soft tissues after placement inside a biological bo
dy. Synthetic
hydroxyapatite [Ca
10
(PO
4
)
6
(OH)
2
] and bioglass are the best representative examples of
commonly used bioactive materials.
Bioresorbable materials are materials that begin to progressively disso
lve and slowly
become replaced by the adjacent biological tissues after placement or implantati
on inside
a biological body. Polylactic-polyglycolic acid copolymers and tricalcium
phosphate
[Ca
3
(PO
4
)
2
] are two examples of the most commonly used bioresorbable materials.
Further to these previous classifications of dental biomaterials, it is
important to mention
that the final products normally placed in the oral cavity are produc
ed in two different
ways. The first involves industrial line production, wherein thousands
of identical dental
*
Corresponding Author
54
devices are manufactured, such as dental implants, endodontic files and
orthodontic
wires/brackets. The second is the production of custom-made devices that are pro
duced in
dental laboratories, such as crowns, fixed partial dentures, removable
partial dentures,
complete dentures and orthodontic devices (Figure 1).
As with any industrial application, the structural integrity of dental devices h
as to be tested
to preserve the reliability of function and avoid premature failures.
This testing starts with
quality control aiming to discard any defective items, while the end products sh
ould comply
with safety regulations and materials specifications (i.e. ISO standards
). Information for
assuring proper quality is acquired through destructive and non-destruct
ive testing. These
techniques are easily distinguished from each other, as the latter lea
ves the tested devices
undamaged and suitable for additional testing and/or permanent usage in
patients. Non-
destructive testing (NDT) involves a broad spectrum of analytical techn
iques, including
macroscopic/microscopic visual inspection, eddy current testing, radiograp
hy, X-ray
computed tomography (XCT), resonance frequency analysis (RFA) and many
others. NDT
aims to qualitatively and/or quantitatively characterize the tested devi
ces by detecting,
Fig. 1. a) Representative examples of industrially-made dental devices: 1) ortho

dontic
archwire made of Ni-Ti alloy, 2) endodontic file made from Ni-Ti alloy, 3) Mini-
orthodontic
implant made of cp-Ti and 4) dental implant made of Ti. b) Industrially-made ort
hodontic
brackets fabricated from stainless steel alloy. c) Custom-made ceramo-metal crow
ns.
d) Custom made removable partial denture made of Cr-Co alloy and acrylic resin.
c d
Applications of Current Technologies for Nondestructive Testing of Dental Biomat
erials
55
locating and sizing any external/internal structural defects. The acquir
ed information is
then assessed to evaluate whether the tested devices meet acceptable c
riteria or whether
they should be rejected. Taking advantage of the NDT capabilities, the
quality of the end
products is increased while the level of reliability is enhanced. The
application of these
sophisticated NDT methods can easily be conducted industrially to produ
ce acceptable
dental devices. However, conducting extensive systematic quality control for cus
tom-made
dental devices is restricted owing to 1) the tremendous increase in relative cos
ts, 2) the need
for availability of specifically designed testing equipments in dental offices a
nd laboratories
and 3) the need for well-trained dentists and dental technicians to operate thes
e equipments.
Besides all these factors, if proper NDT is applied routinely to custom-made den
tal devices,
it will dramatically increase the total cost of already expensive dental service
s.
Although there are a number of limiting factors for routine application of NDT i
n dentistry,
NDT has a variety of important applications in the dental field. The
best example is X-ray
radiography, which is used for the detection of pores that are typica
lly located in thin
regions of cast custom-made dental devices (i.e. clasp shoulders of a
cast metal removable
partial denture framework). The early detection of pores at these crit
ical regions and stage
will allow the dentist and/or dental technician to correct the problem
before the final
delivery of the prostheses to the patients. The existence of such por
es will lead to
catastrophic premature failure in the oral cavity. In research applicat
ions, NDT contributes
significantly to the determination of adverse effects of the oral envi
ronment on dental
biomaterials/devices and vice versa. In addition, NDT is very helpful
in studies of the
interaction mechanisms between the oral environment and dental biomaterials as a
result of
intraoral aging. A variety of NDT methods can be used to fully characterize dent
al devices
preoperatively (as received) and after retrieval from the oral cavity
by tracking the
occurrence of any changes during intraoral aging. Stereomicroscopy, variable pre
ssure SEM
(VPSEM) without the need for a conductive coating and X-ray microanalysis have b
een used
to monitor the morphological and elemental alterations after intraoral
aging of dental
devices. In addition, computed X-ray micro-tomography (micro-XCT) has be
en used to
correlate the locations and sizes of internal defects found preoperativ
ely in all ceramic
bridges with the fracture origin under clinical conditions. Such inform
ation acquired from
micro-XCT regarding the defect locations will provide tremendous insight
s into the
underlying mechanism of intraoral degradation. Other examples for applications o
f NDT in
dentistry are 1) noninvasive and in situ testing of intraoral implant
stability utilizing RFA
and 2) detection of hidden or sub-surface caries (tooth decay) by uti
lizing X-radiography
and laser technology measuring laser fluorescence within the tooth structure.
The aim of this chapter is to present and discuss the applications of current te
chnologies for
NDT of dental biomaterials covering a wide range of NDT techniques and their imp
acts on
daily dental practice, noninvasive diagnosis and dental biomaterials research.
2. Applications of NDT for quality assurance purposes during dental
treatment
Commonly, NDT techniques are performed by dentists and/or dental technicians to
ensure
the proper quality of custom-made dental devices. For example, they ca
rry out routine
checks by the naked eye or using a stereoscope for the marginal inte
grity of dental
restorations (crowns) for precision before the restorations are delivered
permanently to the
patients. However, it is important to mention that precision in dentistry is a s
ubjective term

56
and varies from one dentist to another. This means that proper precis
ion and acceptability
are dependent on a dentists personal experience and skills. Therefore,
this type of
testing/inspection is optional and in many situations is performed with
out specific
requirements to accept or reject custom-made dental devices (e.g. crown
s). Internal defects
in cast and welded metallic custom-made devices are not uncommon and
X-ray testing is
performed for early detection of such defects. This section will prese
nt the dental
applications of stereoscopic and X-ray examinations and testing.
2.1 Stereomicroscopic examination and testing
Stereomicroscopic examination is performed routinely by dental technicians and/o
r dentists to
evaluate the quality of recently fabricated custom-made dental devices (e.g. cro
wns). Detection
of any problems at this stage will allow for proper correction before
the final prosthesis
insertion in the patients mouth. A recent study found that implant ret
aining screws
deteriorate over a period of time when they are used to hold a dental prosthesis
in a patients
mouth (Al Jabbari et. al, 2007a). Therefore, the authors encouraged de
ntists who provide
extensive implant treatment in their practice to equip their offices with a ster
eomicroscope to
enable regular evaluation of the quality of the tiny retaining implant
screws at follow-up
appointments. A low power stereomicroscope was found to be a powerful and useful
tool for
evaluating the quality of the external structures/surfaces of tiny dental device
s (e.g. prosthetic
retaining screws) (Al Jabbari et. al, 2007a). As illustrated in Figure 2, the st
ereomicroscope was
able to clearly reveal the damage and deterioration of an implant scr
ew head and threads,
whereas it is almost impossible for a dentist to observe such damage and deterio
ration under
the naked eye. The great advantage of performing this type of NDT in dental offi
ces is that it
will allow dentists to replace any severely damaged or deteriorated retaining sc
rews with new
ones. Failure to detect such deterioration and damage may lead in the
future to a more
complicated and expensive dental treatment for patients (Al Jabbari et. al, 2007
b).
(a)
Fig. 2. Threaded segment (a) and slotted head segment (b) & (c) of new and retri
eved
implant retaining screws examined under stereomicroscope. Examination of these t
iny
dental implant devices reveals significant threads deterioration/thinning black a
rrows
and screw head damage (c) when compared to the intact threads profile red arrows and
normal slotted head (b).
(b)
(c)
erials
57
As mentioned previously, dental technicians routinely evaluate the quali
ty of their
fabricated devices. For example, they use a stereomicroscope to examine
devices made by
casting. If they detect any external large casting porosity, they will
correct and repair the
problem by filling the pores with solder materials or they may need
to remake the whole
casted device before they pass it for clinical application. In addition, they re
gularly evaluate
the passivity of the fit of their cast prosthesis superstructure utilizing a ste
reomicroscope. If
they detect any misfit between the cast superstructure and the support
ing teeth and/or
implants, they will correct the problem using recasting, soldering, wel
ding and electro-
discharging machining techniques (Contreras et al. 2002; Ntasi et al.
2010; Romero et al.
2000; Zinelis 2007).
2.2 X-ray testing
Many dental devices (such as crowns, and fixed and removable partial
dentures) are
traditionally manufactured by casting dental alloys. Recently, the prepa
ration of metal-free
crowns and bridges has introduced the concept of casting ceramic mater
ials. However, the
mechanical stability and biocompatibility of dental devices depend on the proper
ties of the
materials and on the accuracy of the manufacturing process. Unfortunate
ly, the dental
casting procedure unavoidably leads to the development of pores in dental cast f
rameworks
owing to gas entrapment or shrinkage, which may adversely affect the quality and
efficacy
of dental devices. The development of undesirable porosity is dependent on vario
us factors
and is a common complication for precious, semiprecious and base conte
mporary dental
alloys (Elliopoulos et al. 2004; Li et al. 2010; Neto et al. 2003; Ucar et al. 2
011; Zinelis 2000).
This may negatively affect the long-term mechanical stability. For example, whe
n there are
cast external porosities, corrosion resistance decreases because of crev
ice formation and
plaque accumulation in the oral cavity.
In industrial applications, internal voids can be readily investigated
by employing X-ray
examination, and the same technique has been adopted in dental practic
e as a non-
destructive method for the same purposes. Pores can be easily distinguished as d
ark regions
on radiographs, thereby providing significant information for the size,
location and
distribution of imperfections. The same methodology can be readily appl
ied to identified
internal voids in dental cast frameworks (Dharmar et al. 1993; Eisenbu
rger et al. 2002;
Eisenburger et al. 1998; Elarbi et al. 1985; Mattila 1964; Wictorin e
t al. 1979). However, the
visibility of voids and the picture quality depend on the combination
of the material to be
tested and the analytical conditions applied for X-ray testing (Eisenburger & Ad
dy 2002).
Going back to the principle of this technique, it must be noted that
the attenuation of a
narrow beam of monoenergetic photons with specific E and intensity I
o
passing through a
homogeneous material of thickness t is determined by Lamberts low:
I=I
o
e
-(, , )*t
(1)
where I is the photon intensity that goes out to the sample, and i
s the linear attenuation
coefficient that depends on the material density, atomic number Z and
beam energy.
Therefore, the X-ray absorption depends on the atomic number, the density of the
material
and the energy of radiation. Accordingly, different dental alloys can be penetra
ted by X-rays
to different extents during testing. As stated above, various precious, semiprec
ious and base
58
metal alloys are used in the dental field that have different element
al, mechanical and
physical properties (Roberts et al. 2009; Wataha 2002). Table 1 shows
some representative
types of dental alloys with densities ranging from 4.51 g/cm
3
for Ti to 19.3 g/cm
3
for pure
Au. The vast majority of dental frameworks are produced by casting of
precious and base
metal alloys (Roberts et al. 2009) with the exception of pure gold for use in de
ntistry, which
employs an electroforming technique (Vence 1997). As can be expected f
rom equation (1)
and the density values of the base metal alloys in Table 1, lower l
evels of X-ray absorption
facilitate X-ray penetration. Low absorption coefficients and a high energy beam
are needed
to penetrate the thick metallic parts of cast dental frameworks. Figur
e 3 demonstrates the
attenuation coefficients of some of the dental alloys presented in Table 1. As c
an be seen in
Figure 3, the absorption coefficients decrease with increasing beam ene
rgy. This is not the
case for pure Au and Au-based dental alloys where the absorption coef
ficient rises at the
energy of 80 kV because the K absorption edge limits the penetration
of Au and Au-based
dental alloys to 0.6 mm (Eisenburger & Tschernitschek 1998). Non-precious and ba
se dental
alloys such as Ni-Cr, Co-Cr and commercially pure Ti (cpTi) can be p
enetrated by up to
several millimeters depending on the accelerating voltage and exposure
time (Eisenburger
& Addy 2002; Eisenburger & Tschernitschek 1998).
Brand name Mass content (%) Type
Density
(gr/cm
3
)
Electroforming* Au: 99.9 Pure Au 19.3
IPS d.SIGN 98
Au:85.9, Pt:12.1, Zn:1.5, In<1.0
Ir<1.0, Fe<1.0, Mn<1.0, Ta<1.0,
High Au 18.9
Degudent U94*
Au:76.0,Pt:9.6,Pd:8.9,Ag:1.2,Cu:0.3
Sn:0.8,In:1.5,Ir:0.1,Re:0.2,Ta:0.2.
High Au 17.9
Degulor M*
Au:70.0,Ag:13.5,Au:8.8,Pt:4.,
Pd:2.0, Zn:1.2, Ir: 0.1
High Au 15.5
Degupal G*
Pd:77.3, Ag:7.2, Au:4.5, Sn:4.0,
Ru: 0.5; Ga: 6.0; Ge: 0.5
Pd based 11.7
4 all
Ni:61.5, Cr:25.7, Mo:11.0,
Si:1.5, Mn<1.0, Al<1.0
Ni-Cr 8.4
IPS d.SIGN 30
Co:60.2,Cr:30.1,Ga:3.9,Nb:3.2
Si:0.9, Mo:0.6, Fe:0.5, B:0.3, Li<1.0
Co-Cr 7.8
Ti*
Ti:balance,< 0.2,< 0.06,
C<0.08, < 0.013, Fe < 0.25
cp Ti
(gradeII)
4.5
Table 1. Brand names, elemental compositions, alloy types and densities of some
representative dental alloys used for the production of metalloceramic crowns an
d bridges.
The alloys are sorted in descending order of the density. The attenuation coeffi
cients of the
alloys indicated by asterisks are shown in Figure 3.
As examples, Figure 4 shows dental devices casted from grade II cpTi
and analyzed by X-
rays at 70-kV tube voltage, 8-mA beam current and 0.32-s exposure time. In Figur
e 3a, it can
be seen that large pores are located at the connector areas of a ca
st framework for a fixed
partial denture. Similar sized pores can be seen in cast rectangular
specimens used for the
determination of metalloceramic bonding strength values (Figure 4b). The spheric
al shape of
these pores is a strong indication of gas entrapment, which is a typi
cally reported problem
erials
59
when cpTi and Ti alloys are cast (Elliopoulos et al. 2004; Zinelis 2
000). Figures 4c and 4d
show spherical pore formation on a cast framework for a removable partial dentur
e located
critically at the shoulder of a circlet clasp, which is considered to
be a common region for
such pore formation.
Fig. 3. Mass attenuation coefficient of three dental alloys (Eisenburger & Tsche
rnitschek
1998). The same values for pure Au, Ni and Ti are presented for comparison purpo
ses
(Hubbell et al. 1996).
The main advantage of radiographic X-ray testing is the ability to pr
ovide quick valuable
information about the quality of the metal framework casting by reveal
ing the locations,
numbers and sizes of internal defects that are not possible to detect
by visual inspections.
The occurrence of internal defects/porosity at critical regions such as the clas
p shoulders of
cast removable partial denture frameworks will lead to premature clinic
al failures.
Therefore, early detection will allow proper correction of the problem prior to
final delivery
of a prosthetic device to a patient.
Previous studies have reported that fractures and premature failures of
removable partial
denture frameworks range from 16% to 19% owing to the occurrence of internal por
osity in
cast Co-Cr frameworks (Dharmar et al. 1993; Elarbi et al. 1985; Wicto
rin et al. 1979). In
addition, they revealed that the frequencies of the common locations o
f these defects were
22% to 73% at major connectors, 5% to 43% at clasps/clasp shoulders and 6% to 8%
at minor
connectors (Dharmar et al. 1993; Elarbi et al. 1985). Therefore, and because of
the high rates
of occurrence of such defects, dentists and/or dental technicians are encouraged
to perform
radiographic X-ray inspections at early stages of removable partial den
ture fabrication.
60
Early detection of these casting defects and imperfections may allow e
asy repair or simple
remaking of the metal framework. On the other hand, late detection may mandate r
emaking
of the entire removable prosthesis, which is considered to be a costly choice to
be performed
by dentists, to avoid intraoral premature failures of a removable prosthesis.
Fig. 4. X-ray images of dental casting. a) A dental cast framework for fixed par
tial denture,
b) cast specimens for testing metalo-ceramic bond strength c) a cast framework f
or
removable partial denture and d) High magnification of the highlighted region in
(c) and it
shows internal porosity occurrence (small arrow) at the clasp shoulder.
Another valuable dental application of radiographic nondestructive X-ray
testing is in
retrieval analysis studies. Conventional dental X-ray units can be read
ily used for X-ray
analysis of dental alloys, except for pure Au and Au-based alloys whe
re the penetration
depth is limited to approximately 0.6 mm and thus increasing acceleration voltag
e to 120 kV
is not possible with conventional dental X-ray machines. Microfocus X-r
ay systems (CRX
1000/CRX 2000; CR Technology Inc., Aliso Veijio, CA, USA) have been u
sed to detect the
occurrence of internal defects in prosthetic retaining implant screws m
ade of gold-based
alloys (Al Jabbari et al, 2007a). For NDT, utilization of a microfocu
s X-ray machine is
possible and very valuable for evaluating various tiny dental devices
made from precious,
semiprecious and non-precious alloys. However, this type of machine is known to
be mainly
useful for dental biomaterials research purposes.
erials
61
3. Applications of NDT for research purposes in the dental biomaterials field
Besides the valuable applications of NDT in quality assurance of dental devices,
NDT makes
significant contributions and has important applications in the dental biomateri
als research
field. Normally, dental biomaterials in the oral cavity are exposed to
various aggressive
conditions. When biomaterials are placed intraorally in the form of fi
lling materials or
prosthetic appliances, they go through a variety of degradation mechanisms such
as fatigue,
wear, corrosion and discoloration. Therefore, studies of the degradation
mechanisms of
biomaterials are important toward the design of new biomaterials with
increased efficacy
and longevity.
Laboratory or in vitro testing is widely applied to dental biomaterial
s to determine the
properties of the materials. However, in vitro testing cannot provide any reliab
le information
that can predict the in vivo behavior of biomaterials because the con
ditions of the oral
environment cannot be simulated experimentally in research laboratories.
Accordingly,
NDT can be effectively used over a long period of time to monitor the occurrence
of changes
in a specific dental biomaterial or device resulting from intraoral aging. Gener
ally, NDT of a
retrieved specific dental biomaterial or device that has been placed in a patien
ts mouth for a
reasonable time and comparison with the properties of a new (unused) b
iomaterial/device
will provide more useful and significant information about the degradat
ion mechanism
resulting from long-term use in vivo. Under this general concept, a variety of N
DT methods
and techniques have been developed such as micro-XCT, VPSEM-EDS, optical profilo
metry
and X-ray diffraction (XRD).
The main two limiting factors in conducting such research protocols ar
e ethical and cost
reasons. For example, a permanently placed filling dental biomaterial c
annot be retrieved
from the mouth for only research purposes. Retrieval of a biomaterial and replac
ement with
a new one will subject the tooth to additional unnecessary clinical p
rocedures. Multiple
clinical procedures will cause repeated insults to a specific tooth th
at may lead to pulpal
irritation and/or necrosis. Needless to say, that is ethically unaccept
able. Similarly, a
successful dental appliance or device cannot be retrieved from the mouth solely
to conduct
NDT before it fails, since making a new one is costly for the patie
nt and there is no
guarantee that the newly fabricated device will be as successful as the first on
e.
There are two additional obstacles that may also make the performance of NDT in
retrieved
dental biomaterials infeasible. First, it requires frequent patient foll
ow-ups before the
biomaterials and/or devices are retrieved. Unfortunately, not all patien
ts will accept many
follow-ups for research purposes only. Second, many retrieved dental bi
omaterials and/or
devices are unsuitable for the conduction of certain types of NDT. Fo
r example, XRD
analysis requires flat surfaces of a few square millimeters in dimension, a requ
irement that
is hard to fulfill in dental devices.
3.1 Micro-XCT
Currently, computed tomography is extensively used in the medical field
for
diagnostic/treatment purposes. During the last two decades, new bench-to
p models have
been introduced for use in the characterization of materials that employ similar
principles to
medical computed tomography. The only difference is that the bench-top
models have an
isotopic resolution capable of reaching a few tens of nanometers. Cont
rary to medical
62
computed tomography machines, the specimens tested with bench-top models
can be
rotated while the detector and X-ray source are fixed within the mach
ine. The micro-XCT
scanning produces hundreds of horizontal slices for a tested specimen, which are
then used
to reconstruct the entire specimen. Reconstruction of a specimen is ac
complished by two
reconstruction algorithms commonly known as iterative and filtered back
projection
methods. In addition, computer software can be utilized in the develop
ment of three-
dimensional models, pseudocoloring and quantitative determination of geometrical
features
of an irregular dental biomaterial device.
Figure 5 shows a good example of NDT utilizing a micro-XCT analysis
of a fixed partial
denture (FPD). It reveals the importance of this tool for nondestructi
vely analyzing the
internal structure of the whole ceramic FPD. The FPD was analyzed pri
or to permanent
cementation of the prosthesis in the patients mouth. The analysis reveal
ed that the joining
procedure of the three different parts of the alumina core was not done properly
because of
the entrapment of large voids at and within the bulk of the cementing material (
Figure 5d).
Unfortunately, after final insertion of the FPD in the patients mouth,
it did fail at the
connector area after being in service for a short period of time. Therefore, it
can be said that
micro-XCT is a powerful tool for evaluating the quality of industrially and/or c
ustom-made
dental devices and for failure analysis of dental biomaterials.
3.2 SEM-VPSEM-EPMA
Scanning electron microscopy (SEM) combined with electron probe microana
lysis (EPMA)
is considered to be a powerful analytical tool for providing morpholog
ical and elemental
information about tested samples at low (6) and high (150,000) magnifica
tions. SEM is
able to bridge the gap between optical stereomicroscopy and transmissio
n electron
microscopy. Recent advancements in SEM manufacturing technology can prov
ide imaging
of non-conductive specimens (low-vacuum SEM) and samples at 99% relativ
e humidity
(environmental SEM). These new operating modes are also known as VPSEM
and confer
tremendous capabilities to SEM. Additional information regarding the operation p
rinciples
and applications of these new SEM models can be found in relevant pr
evious reports
(Bergmans et al. 2005; Danilatos; Danilatos 1991; Danilatos 1993; Danilatos 1994
; Danilatos et
al.; Kodaka et al. 1992).
In dentistry, brazing is the main joining technique for making metalli
c orthodontic
appliances. A space maintainer is an example of the most commonly use
d orthodontic
appliances. This appliance is made of two stainless steel tooth bands
joined by a stainless
steel orthodontic wire. The orthodontic wire and the two bands are jo
ined by brazing
utilizing low fusing silver brazing alloys (Figure 6). Figure 7 shows
a high magnification
SEM photomicrograph of an orthodontic space maintainer appliance, reveal
ing that the
soldered area joins the stainless steel bands to the orthodontic wire.
NDT utilizing X-ray
EDS analysis was performed at that area at two different times (befor
e and after dental
treatment). The purpose of the NDT and the analysis was to determine
the effects of long-
term use in vivo on the Ag-based brazing alloy. Small porosities (ind
icated by arrows in
Figure 7) were used to identify the area for X-ray EDS analysis. The two spectra
obtained at
the two different times are shown in Figure 8, and reveal significant decreases
in the Cu and
Zn composition after intraoral aging.
erials
63
Fig. 5. Micro XCT analysis of a fixed partial denture (FPD) that was fabricated
from all
ceramic materials. The core of the FPD was made from alumina that was veneered w
ith
dental porcelain. a) FPD X-ray image before the FPD was cemented in the patients
mouth
showing (A) the core alumina, (B) the veneering porcelain and (C) the connector
area joining
the three unites of the FPD. It is easy to distinguish between the three parts b
ecause the
differences in X-ray absorption. b) A pseudocoloring reconstruction of a perpend
icular cross
section for the FPD showing the alumina core, the veneering porcelain and the ma
terial used
for cementation at the connector area. Pores are easily identified as white circ
le areas in the
regions of porcelain layer and at the core porcelain interface. Red line located
on the small
attached image indicates the cross sectional plane while the bar indicates the a
bsorption
scale. c) 3D image of the reconstructed structure after digital processing helpf
ul in total
volume determination. d) Alternative 3D image showing the distribution of intern
al pores in
dental porcelain and the occurrence of big voids near and within the cement laye
r.
64
Fig. 6. An orthodontic device known as space maintainer made of two bands and a
wire
with two soldered joints (arrows). Bands and wires were manufactured from stainl
ess steel
whereas the soldering alloy is Ag-based alloy containing Cu, Zn and Sn.
Fig. 7. Secondary Electron Images (SEI) of a joint area between the stainless st
eel band (A)
and the orthodontic wire (C) soldered with Ag-based soldering alloy (B). (a) As-
received
appliance from the dental laboratory and before it was placed in the patients mou
th. (b) The
retrieved appliance from the patients mouth after it serviced for approximately 1
4 months
of treatment period. External surface porosities (arrows) were used as a referen
ce for
locating exact area used for x-ray EDS analysis.
erials
65
Fig. 8. EDS x-ray spectrum for the Ag-based alloy solder before and after intrao
ral aging.
Note the decrease in Cu, Zn after long-term use in vivo.
Figures 8 and 9 provide additional information regarding the surface s
tructure of the Ag-
based soldering alloy. Both figures confirm that the significant decrea
ses in Cu and Zn are
appended to the dissolution of the low atomic contrast second phase,
which is enriched in
Cu and Zn. A possible explanation is that dental biomaterials with multiple phas
e structures
might be prone to galvanic corrosion. Of course, the presence of the stainless s
teel bands and
wire might have an additional effect on this phenomenon. However, this
is only an
assumption and the verification of galvanic corrosion requires further
extended research.
The important contribution of this NDT and X-ray EDS analysis is sign
ificant because it
confirms other previous findings that Ag-based alloys are prone to cor
rosion and ionic
release (Grimsdottir et al. 1992; Locci et al. 2000a; Mockers et al. 2002; Staff
olani et al. 1999).
These findings might be a reason for mucosal irritation, which was re
ported in a previous
study (Bishara 1995). The significant release and dissolution of Cu and
Zn during intraoral
aging must be taken seriously because Cu ions have toxic effects on the human bo
dy (Locci
et al. 2000b; Vannet et al. 2007; Wataha et al. 2002).
4. Dental applications of NDT as non-invasive diagnostic methods
Radiographic X-rays are used routinely in dental offices to non-invasiv
ely diagnose hard
dental tissue diseases or to detect dental caries (tooth decay). Howev
er, in recent years, new
technologies have been developed and introduced into the dental field for use as
non-invasive
diagnostic tools. The best two examples are RFA (Meredith et al. 1997
) and fluorescence
measurements (Jablonski-Momeni et al. 2011; Lussi et al. 2003; Rodrigues et al.
2010).
66
Fig. 9. Backscattered Electron Images after etching of Ag-based alloy solder bef
ore (a) and
after (b) intraoral aging. The Ag based alloy has two different microstructure p
hases with
dispersion of lower atomic contrast phase in the matrix. This second phase has b
een
completely diluted leaving behind a random distribution of surface craters. (Ori
ginal
magnification: 1000X).
4.1 RFA
The main dental application of RFA as a diagnostic method is to quan
tify dental implant
stability after surgical implant placement in the human jaw bone. Norm
ally, when dental
implants are placed in healthy human jaws, they will form and establish strong s
table bond
with the surrounding bone tissues after a period of several months and this phen
omenon is
known as Osseointegration (Albrektsson et al. 1986). It has been suggested in se
veral studies
that the stiffness of the boneimplant interface can be assessed by RFA
(Valderrama et al.
2007; Sakoh J et al. 2006; Alsaadi G et al. 2007). Therefore, clinic
ians have been advised to
utilize RFA to determine the strength and adequacy of the established
bond
(Osseointegration) before they restore implants with dental prostheses.
A commonly used
device for this purpose is the Osstell Mentor device (Integration Diag
nostics, Goteborg,
Sweden) (Figure 10a). As shown in Figure 10, part of the Osstell Mentor device c
omprises L-
shaped transducers. These transducers will record all the information as an impl
ant stability
quotient (ISQ), which is a function of the boneimplant stiffness (N/m)
and the marginal
bone height. The ISQ is a dimensionless quantity, and larger values indicate gre
ater levels of
interfacial boneimplant stiffness (meaning a higher established osseointeg
ration with
greater stability).
Besides the aforementioned beneficial diagnostic applications of RFA, it
has been utilized
extensively in research studies. Traditionally, research studies have ev
aluated the
established bond between an implant and bone by histomorphometric tests
. However, the
main disadvantage of these tests is that they are destructive (Meredith N. 1998)
. Therefore, it
has been suggested that RFA can be used periodically to evaluate the
established bond
(osseointegration occurrence) between an implant and bone without sacrificing th
e object in
an in vivo study (Figure 10b) (Meredith N. 1998; Huang HM. et al. 2003).
erials
67
(a) (b)
Fig. 10. (a) The Osstell device utilized normally for RFA. (b) Illustrating con
duction of RFA
in an animal study. It is important that the transducer be placed in same positi
on each time a
measurement is taken for a specific dental implant. Different ISQ values could b
e obtained
by positioning the transducer in different directions.
4.2 Fluorescence measurements
Dental caries are also known as tooth decay and result from demineral
ization of inorganic
components of the outer layer (enamel) of the tooth structure. Released bacteria
l lactic acid
will normally lead to tooth enamel demineralization. Detection of tooth caries a
t early stages
is crucial because it only requires a non-costly simple treatment. How
ever, the early stages
of dental caries may not be easily detected by the naked eye during
routine dental
examinations. Therefore, fluorescence measurements have been recommended
for early
detection and diagnosis of enamel demineralization (Jablonski-Momeni et al.
2011; Lussi et
al. 2003; Rodrigues et al. 2010). The structure of healthy and sound
tooth enamel is
characterized by a low baseline fluorescence level, while demineralized and infe
cted enamel
will have an increased fluorescence level. In addition, the fluorescence level i
ncreases as the
caries process advances (Lussi et al. 2001). A recently developed device for det
ecting dental
caries at its early stages based on fluorescence measurements is the
DIAGNOdent device
(KaVo, Biberach, Germany). The DIAGNOdent device emits red light at 655 nm and d
etects
bacterial metabolites in the demineralized tooth structure (Lussi et al
. 2003; Lussi et al.
2006a; Lussi et al. 2006b). The DIAGNOdent device then classifies the
tested regions
according to the calibrated fluorescence intensity as follows: scores 013 = no c
aries, scores
1420 = early (incipient) enamel caries and scores above 20 = advanced
dentine caries
(Figure 10a) (Lussi & Hellwig 2006b).
Another recently developed fluorescence camera, VistaProof (Drr Dental, B
ietigheim-
Bissingen, Germany), is used for early diagnosis of dental caries. The VistaProo
f emits blue
light at 405 nm (Jablonski-Momeni et al. 2011) and records fluorescenc
e from the probed
tooth surfaces in the form of digital images (Rodrigues et al. 2008).
Intact tooth structures
show green fluorescent images, while infected and demineralized tooth s
tructures show
68
blue-violet fluorescent images. Infected areas with increased numbers of
bacteria and
bacterial byproducts show red fluorescent images (Figure 11b). The digital softw
are utilized
by the VistaProof quantifies the color components and provides scoring
outcomes ranging
from 04 that indicate the penetration depth of the dental caries withi
n the tooth structure
(Figure 11b). The scoring outcomes are good diagnostic values for the
presence or absence
and the severity of dental caries. The values are used as follows: 00.9 = sound a
nd healthy
tooth structure; 0.91.5 = initial (incipient) enamel caries; 1.52.0 = deep enamel
caries; 2.0
2.5 = dentine caries; and above 2.5 = deep dentine caries. It is im
portant to mention that,
despite the reported reliable applications of the DIAGOdent and VistaPr
oof for non-
invasive diagnosis of dental caries, dentists routinely verify their fi
ndings by taking
radiographic X-rays to diagnose the occurrence of dental caries.
Fig. 11. Occlusal surface of a molar tooth with dental caries. The infected area
s were
diagnosed by DIAGNOdent (a) and VistaProof (b). Numbers in both images provide
valuable information regarding the extent and penetration depth of dental caries
within the
tooth structure.
5. Conclusions
NDT plays very important roles in dentistry, and in the dental biomaterials rese
arch field in
particular. However, the application of NDT on a daily basis for dental diagnosi
s purposes
and for assuring adequate therapeutic quality is limited, mainly becaus
e of the significant
increases in relevant time and cost. Luckily, NDT along with its various applica
tions in the
dental biomaterials research field is performed routinely by scientific
researchers.
Consequently, this NDT has led to noticeable enhancements of the quality, perfor
mance and
biocompatibility of dental biomaterials that are placed daily into patients mouth
s.
6. Acknowledgments
The author s would like to thank Prof. George Eliades, Dr. Triantaffi
los Papadopoulos, Dr.
Abdulaziz Al-Rasheed, Dr. Anil Sukumaran, Dr. Maysa Al-Marshood, Peter Ts
akiridis, Dr.
Alerxis Tagmatarxis and Maria Michalaki for their help in providing some of the
pictures for
this manuscript.
erials
69
7. References
Al Jabbari YS., Fournelle R., Ziebert G., Toth J., Iacopino AM. (2007a). Mechani
cal behavior
and Failure analysis of prosthetic retaining screws after long-term use in vivo.
Part
I: Charactrization of adhesive wear and structure of retaining screws.
Journal of
Prosthodontics, Vol.17, No.3, pp.168-180, ISNN 1532-849X
Al Jabbari YS., Fournelle R., Ziebert G., Toth J., Iacopino AM. (2007b). Mechani
cal behavior
and Failure analysis of prosthetic retaining screws after long-term use in vivo.
Part
III: Preload and fracture tensile load testing. Journal of Prosthodonti
cs, Vol.17, No.3,
pp.192-200, ISNN 1532-849X
Alsaadi G., Quirynen M., Michiels K., Jacobs R., van Steenberghe D. (2007). A bi
omechanical
assessment of the relation between the oral implant stability at inser
tion and
subjective bone quality assessment. Journal of Clinical Periodontology,
Vol.34, No.4,
pp. 359-366, ISNN 0303-6979
Bergmans L., Moisiadis P., Van Meerbeek B., Quirynen M., Lambrechts P.
(2005).
Microscopic observation of bacteria: review highlighting the use of env
ironmental
SEM. International Endodontic Journal, Vol.38, No.11, pp. 775-88, ISNN 0143-288
5
Bishara S. E. (1995). Oral Lesions Caused by an Orthodontic Retainer
- a Case-Report.
American Journal of Orthodontics and Dentofacial Orthopedics, Vol.108, No.2, pp
. 115-
117, ISNN 0889-5406
Contreras E. F. R., Henriques G. E. P., Giolo S. R., Nobilo M. A.
A. (2002). Fit of cast
commercially pure titanium and Ti-6Al-4V alloy crowns before and after marginal
refinement by electrical discharge machining. Journal of Prosthetic Dentistry, V
ol.88,
No.5, pp. 467-472, ISNN 0022-3913
Danilatos G. D. (1991). Review and Outline of Environmental Sem at Pr
esent. Journal of
Microscopy-Oxford, Vol.162, pp. 391-402, ISNN 0022-2720
Danilatos G. D. (1993). Introduction to the Esem Instrument. Microscopy
Research and
Technique, Vol.25, No.5-6, pp. 354-361, ISNN 1059-910X
Danilatos G. D. (1994). Environmental Scanning Electron-Microscopy and M
icroanalysis.
Mikrochimica Acta, Vol.114, pp. 143-155, ISNN 0026-3672
Dharmar S., Rathnasamy R. J., Swaminathan T. N. (1993). Radiographic and metallo
graphic
evaluation of porosity defects and grain structure of cast chromium co
balt
removable partial dentures. Journal of Prosthetic Dentistry, Vol.69, No.4, pp.
369-73,
ISNN 0022-3913
Eisenburger M., Addy M. (2002). Radiological examination of dental castings -- a
review of
the method and comparisons of the equipment. Journal of Oral Rehabilitation, Vol
.29,
No.7, pp. 609-14, ISNN 0305-182X
Eisenburger M., Tschernitschek H. (1998). Radiographic inspection of den
tal castings.
Clinical Oral Investigations, Vol.2, No.1, pp. 11-4, ISNN 1432-6981
Elarbi E. A., Ismail Y. H., Azarbal M., Saini T. S. (1985). Radiographic detecti
on of porosities
in removable partial denture castings. Journal of Prosthetic Dentistry,
Vol.54, No.5,
pp. 674-7, ISNN 0022-3913
Elliopoulos D., Zinelis S., Papadopoulos T. (2004). Porosity of cpTi c
asting with four
different casting machines. Journal of Prosthetic Dentistry, Vol.92, No.4, pp.
377-381,
ISNN 0022-3913
70
Grimsdottir M. R., Gjerdet N. R., Henstenpettersen A. (1992). Compositi
on and Invitro
Corrosion of Orthodontic Appliances. American Journal of Orthodontics an
d
Dentofacial Orthopedics, Vol.101, No.6, pp. 525-532, ISNN 0889-5406
Hubbell J., Seltzer S. (1996) NIST X-ray Attenuation Databases. In NIST ed. NIST
Databases.
Huang HM., Chiu CL., Yeh CY., Lin CT., Lin LH., Lee SY (2003). Early detection o
f implant
healing process using resonance frequency analysis. Clinical Oral Implants Resea
rch,
Vol.14, No.4, pp. 437-443, ISNN 1600-0501.
Jablonski-Momeni A., Schipper H. M., Rosen S. M. et al. (2011). Perfo
rmance of a
fluorescence camera for detection of occlusal caries in vitro. Odontolo
gy, Vol.99,
No.1, pp. 55-61, ISNN 1618-1247
Kodaka T., Toko T., Debari K., Hisamitsu H., Ohmori A., Kawata S. (1
992). Application of
the environmental SEM in human dentin bleached with hydrogen peroxide
in
vitro. Journal of electron microscopy (Tokyo), Vol.41, No.5, pp. 381-6, ISNN 00
22-0744
Li D., Baba N., Brantley W. A., Alapati S. B., Heshmati R. H., Daeh
n G. S. (2010). Study of
Pd-Ag dental alloys: examination of effect of casting porosity on fati
gue behavior
and microstructural analysis. Journal of Materials Science-Materials in
Medicine,
Vol.21, No.10, pp. 2723-2731, ISNN 0957-4530
Locci P., Lilli C., Marinucci L. et al. (2000a). In vitro cytotoxic
effects of orthodontic
appliances. Journal of Biomedical Materials Research, Vol.53, No.5, pp. 560-567
, ISNN
0021-9304
Locci P., Marinucci L., Lilli C. et al. (2000b). Biocompatibility of alloys used
in orthodontics
evaluated by cell culture tests. Journal of Biomedical Materials Research, Vol.5
1, No.4,
pp. 561-568, ISNN 0021-9304
Lussi A., Hack A., Hug I., Heckenberger H., Megert B., Stich H. (200
6a). Detection of
approximal caries with a new laser fluorescence device. Caries Research
, Vol.40,
No.2, pp. 97-103, ISNN 0008-6568
Lussi A., Hellwig E. (2006b). Performance of a new laser fluorescence
device for the
detection of occlusal caries in vitro. Journal of Dentistry, Vol.34, No.7, pp.
467-471,
ISNN 0300-5712
Lussi A., Longbottom C., Gygax M., Braig F. (2003). The influence of
professional cleaning
and drying of occlusal surfaces on DIAGNOdent readings. Journal of Den
tal
Research, Vol.82, pp. B64-B64, ISNN 0022-0345
Lussi A., Megert B., Longbottom C., Reich E., Francescut P. (2001). Clinical per
formance of a
laser fluorescence device for detection of occlusal caries lesions. European Jou
rnal of
Oral Sciences, Vol.109, No.1, pp. 14-19, ISNN 0909-8836
Mano J. F., Sousa R. A., Boesel L. F., Neves N. M., Reis R. L. (2004). Bloinert,
biodegradable
and injectable polymeric matrix composites for hard tissue replacement: state of
the
art and recent developments. Composites Science and Technology, Vol.64,
No.6,
pp. 789-817, ISNN 0266-3538
Mattila K. (1964). A Roentgenological Study of Internal Defects in Chrome-Cobalt
Implants
and Partial Dentures. Acta odontologica Scandinavica, Vol.22, pp. 215-28, ISNN 0
001-
6357
Meredith N. (1998). Assessment of implant stability as a prognostic de
terminant.
International Journal of Prosthodontics, Vol.11, No.5, pp.491-501, ISNN 0893-217
4
Meredith N., Book K., Friberg B., Jemt T., Sennerby L. (1997). Resona
nce frequency
measurements of implant stability in vivo. A cross-sectional and longitudinal st
udy
erials
71
of resonance frequency measurements on implants in the edentulous and partially
dentate maxilla. Clinical Oral Implants Research, Vol.8, No.3, pp. 226-33, ISNN
0905-
7161
Mockers O., Deroze D., Camps J. (2002). Cytotoxicity of orthodontic ba
nds, brackets and
archwires in vitro. Dental Materials, Vol.18, No.4, pp. 311-317, ISNN 0109-5641
Neto H. G., Candido M. S. M., Junior A. L. R., Garcia P. P. N. S. (2003). Analys
is of depth of
the microporosity in a nickel-chromium system alloy - effects of elect
rolytic,
chemical and sandblasting etching. Journal of Oral Rehabilitation, Vol.3
0, No.5, pp.
556-558, ISNN 0305-182X
Ntasi A., Mueller W. D., Eliades G., Zinelis S. (2010). The effect o
f Electro Discharge
Machining (EDM) on the corrosion resistance of dental alloys. Dental M
aterials,
Vol.26, No.12, pp. E237-E245, ISNN 0109-5641
Roberts H. W., Berzins D. W., Moore B. K., Charlton D. G. (2009). M
etal-ceramic alloys in
dentistry: a review. Journal of Prosthodontics, Vol.18, No.2, pp. 188
-94, ISNN 1532-
849X
Rodrigues J. A., Hug I., Diniz M. B., Lussi A. (2008). Performance of
fluorescence methods,
radiographic examination and ICDAS II on occlusal surfaces in vitro. C
aries
Research, Vol.42, No.4, pp. 297-304, ISNN 0008-6568
Rodrigues J., Diniz M., Hug I., Cordeiro R., Lussi A. (2010). Relatio
nship between
DIAGNOdent values and sealant penetration depth on occlusal fissures. C
linical
Oral Investigations, Vol.14, No.6, pp. 707-711, ISNN 1432-6981
Romero G. G., Engelmeier R., Powers J. M., Canterbury A. A. (2000).
Accuracy of three
corrective techniques for implant bar fabrication. Journal of Prosthetic
Dentistry,
Vol.84, No.6, pp. 602-607, ISNN 0022-3913
Sakoh J., Wahlmann U., Stender E., Nat R., Al-Nawas B., Wagner W (2006). Primary
stability
of a conical implant and a hybrid, cylindric screw-type implant in vi
tro.
International Journal of Oral and Maxillofacial Implants, Vol.21, No.4,
pp. 560-566,
ISNN 0882-2786
Siegwart, R. (2001). Indirect Manipulation of a Sphere on a Flat Disk
Using Force
Information. International Journal of Advanced Robotic Systems, Vol.6, N
o.4,
(December 2009), pp. 12-16, ISSN 1729-8806
Staffolani N., Damiani F., Lilli C. et al. (1999). Ion release from
orthodontic appliances.
Journal of Dentistry, Vol.27, No.6, pp. 449-454, ISNN 0300-5712
Ucar Y., Brantley W. A., Johnston W. M., Dasgupta T. (2011). Mechanic
al Properties,
Fracture Surface Characterization, and Microstructural Analysis of Six N
oble
Dental Casting Alloys. Journal of Prosthetic Dentistry, Vol.105, No.6,
pp. 394-402,
ISNN 0022-3913
Valderrama P., Oates TW., Jones AA., Simpson J., Schoolfield JD., Coch
ran DL. (2007).
Evaluation of two different resonance frequency devices to detect implant stabil
ity:
A clinical trial. Journal of Periodontology, Vol.78, No.2, pp.262-272. ISNN 0022
-3492
Vannet B. V., Hanssens J. L., Wehrbein H. (2007). The use of three-dimensional o
ral mucosa
cell cultures to assess the toxicity of soldered and welded wires. European Jour
nal of
Orthodontics, Vol.29, No.1, pp. 60-66, ISNN 0141-5387
Vence B. S. (1997). Electroforming technology for galvanoceramic restora
tions. Journal of
Prosthetic Dentistry, Vol.77, No.4, pp. 444-9, ISNN 0022-3913
72
Wataha J. C. (2002). Alloys for prosthodontic restorations. J Prosthet Dent, Vol
.87, No.4, pp.
351-63, ISNN 0022-3913
Wataha J. C., Lockwood P. E., Schedle A., Noda M., Bouillaguet S. (2002). Ag, Cu
, Hg and Ni
ions alter the metabolism of human monocytes during extended low-dose
exposures. Journal of Oral Rehabilitation, Vol.29, No.2, pp. 133-139, ISNN 0305
-182X
Wictorin L., Julin P., Mollersten L. (1979). Roentgenological detection
of casting defects in
cobalt-chromium alloy frameworks. Journal of Oral Rehabilitation, Vol.6,
No.2, pp.
137-46, ISNN 0305-182X
Zinelis S. (2000). Effect of pressure of helium, argon, krypton, and
xenon on the porosity,
microstructure, and mechanical properties of commercially pure titanium castings
.
Journal of Prosthetic Dentistry, Vol.84, No.5, pp. 575-582, ISNN 0022-3913
Zinelis S. (2007). Surface and elemental alterations of dental alloys
induced by electro
discharge machining (EDM). Dental Materials, Vol.23, No.5, pp. 601-607
, ISNN
0109-5641
4
Neutron Radiography
Nares Chankow
Department of Nuclear Engineering, Faculty of Engineering
Chulalongkorn University
Thailand
1. Introduction
A few years after the discovery of neutron by James Chadwick in 1932
, H. Kallman and E.
Kuhn started their work on neutron radiography in Germany using neutro
ns from a small
neutron generator. Due to the second World War, their first publicatio
n was delayed until
1947. However, the first report on neutron radiography was published b
y Peters in 1946, a
year before Kallman and Kuhns. After research reactors were available, in 1956 Th
ewlis and
Derbyshire in UK demonstrated that much better neutron radiographic ima
ges could be
obtained by using intense thermal neutron beam from the reactor. Speci
fic applications of
neutron radiography were then started and expanded rapidly particularly
where research
reactors were available.
The radiographic technique was originally based on metallic neutron converter sc
reen/film
assembly. Neutron converter screen and film were gradually improved unt
il early 1990s
when computer technology became powerful and was available at low cost
. Non-film
neutron radiography was then possible to be used for routine inspection of speci
mens. After
2005, imaging plate specially designed for neutron radiography was avai
lable and could
provide image quality comparable to the best image quality obtained fr
om the gadolinium
foil/film assembly with relative speed approximately 40 times faster. Neverthele
ss, neutron
radiography has not been widely employed for routine inspection of spe
cimen in industry
like x-ray and gamma-ray radiography due to two main reasons. Firstly,
excellent image
quality still needs neutrons from nuclear reactor. Secondly, neutron ra
diography is only
well-known among the academic but not industrial people. It is actuall
y excellent for
inspecting parts containing light elements in materials even when they
are covered or
enveloped by heavy elements. Nowadays, neutrons from small neutron gene
rator and
californium-252 source can give neutron intensity sufficient for modern
image recording
system such as the neutron imaging plate and the light-emitting neutro
n converter
screen/digital camera assembly.
2. Principle of neutron radiography
Neutrons are fundamental particles which are bound together with proton
s within the
atomic nucleus. Neutron is electrically neutral and has mass of nearly the same
as a proton
i.e. about 1 u. Once a neutron is emitted from the nucleus it becomes free neutr
on which is
not stable. It decays to a proton and an electron with a half-life of 12 minutes
.
74
Neutron radiography requires parallel beam or divergent beam of low en
ergy neutrons
having intensity in the range of only 10
4
10
6
neutrons/cm
2
-s to avoid formation of
significant amount of long-lived radioactive isotope from neutron absorp
tion within the
specimen. The transmitted neutrons will then interact with neutron conv
erter screen to
generate particles or light photons which can be recorded by film or
any other recording
media. Free neutrons emitted from all sources are fast neutrons while neutron ra
diography
prefers low energy neutrons. To reduce neutron energy, neutron sources
are normally
surrounded by large volume of hydrogeneous material such as water, pol
yethylene,
transformer oil and paraffin. Neutron collimator is designed to bring
low energy neutron
beam to the test specimen. As illustrated in Figure 2, attenuation co
efficient of gamma-ray
increases with increasing of the atomic number of element while attenuation coef
ficients of
neutron are high for light elements like hydrogen(H), lithium (Li) and
boron(B) as well as
some heavy elements such as gadolinium (Gd), cadmium(Cd) and dysprosium
(Dy). In
contrast, lead (Pb) has very high attenuation coefficient for gamma-ray
but very low for
neutron. Neutron radiography therefore can make parts containing light
elements; such as
polymer, plastic, rubber, chemical; visible even when they are covered
or enveloped by
heavy elements.
Neutrons may interact with matter in one or more of the following reactions.
i. Elastic scattering: (n, n) reaction
Neutron collides with the atomic nucleus, then loses its kinetic energ
y. It should be noted
that neutron loses less kinetic energy when it collides with a heavy
nucleus. In contrast, it
loses more kinetic energy when collides with a light nucleus. Hydrogen(
1
H) is therefore the
most effective neutron moderator because it is the lightest nucleus hav
ing mass almost the
same as neutron (1u). Elastic scattering is most important in productio
n of low energy or
slow neutrons from fast neutrons emitted from the source for neutron ra
diography. Water,
paraffin and polyethylene are common neutron moderators. In fact, hydro
gen-2 (
2
H, so
called deuterium) is the best neutron moderator due to its extremely
low neutron
absorption probability. Heavy water (D
2
O) has neutron absorption cross section only about
1/500 that of light water (H
2
O) but heavy water is very costly.
Fig. 1. Major components of typical neutron radiography system
Neutron Radiography
75
Fig. 2. Mass attenuation coefficients for thermal neutrons ( ) and gamma-rays as a
function
of atomic number of elements (reproduced from [3] with some modifications)
ii. Inelastic scattering: (n, n) or (n, n) reaction
Similar to elastic scattering but when a neutron collides with the at
omic nucleus it has
enough kinetic energy to raise the nucleus into its excited state. After colli
sion, the nucleus
will give off gamma-ray(s) in returning to its ground state. Even ine
lastic scattering also
reduces energy of fast neutron but this is not preferable in neutron
radiography because it
increases in contamination of gamma-rays to the system.
iii. Neutron capture: (n, ) reaction
A neutron can be absorbed by the atomic nucleus to form new nucleus
with an additional
neutron resulting in increasing of mass number by 1. For example, whe
n cobalt-59 (
59
Co)
captures a neutron will become radioactive cobalt-60 (
60
Co). The new nucleus mostly
becomes radioactive and decays to beta-particle followed by emission of gamma-ra
y. Some
of them are not radioactive such as
2
H,
114
Cd,
156
Gd, and
158
Gd. A few of them only decay by
beta-particle without emission of gamma-ray such as
32
P. This reaction plays a vital role in
neutron radiography when metallic foil screen is used to convert neutr
ons into beta-
particles and gamma-rays.
iv. Charged particle emission: (n, p) and (n, ) reactions
Most of charged particle emission occurs by fast neutrons except for
the two important (n,
) reactions of lithium-6 (
6
Li) and boron-10 (
10
B). The
6
Li(n, )
3
and
10
B(n, )
7
Li reactions
play important roles in neutron detection and shielding. In neutron radiography,
these two
reactions are mainly employed to convert neutrons to alpha particles or
to light. The (n, p)
reaction is not important in neutron radiography but it may be useful when solid
state track
detector is selected as the image recorder.
76
v. Neutron producing reaction: (n, 2n) and (n, 3n) reactions
These reactions occur only with fast neutrons which require a threshol
d energy to trigger.
They may be useful in neutron radiography particularly when utilizing
14-MeV neutrons
produced from a neutron generator. By inserting blocks of heavy metal
like lead (Pb) or
uranium (U) in neutron moderator, low energy neutron intensity can be
increased by a
factor of 2 3 or higher from (n, 2n) and (n, 3n) reactions.
vi. Fission: (n, f) reaction
Fission reaction is well-known for energy production in nuclear power
plant and neutron
production in nuclear research reactor. A heavy nucleus like uranium-235 (
235
U), plutonium-
239(
239
Pu) undergoes fission after absorption of neutron. The nucleus splits
into 2 nuclei of
mass approximately one-half of the original nucleus with emission of 2 3 neutron
s. When
uranium (U) is used to increase neutron intensity by the above (n, 2n) and (n, 3
n) reactions,
fission reaction also contributes additional neutrons to the system. Th
e degree of
contribution depends on the ratio of uranium-235 to uranium-238 in uranium.
Attenuation of neutron by the specimen depends on thickness and the a
ttenuation
coefficient similar to gamma-ray as follows.
I
t
= I
0
exp (-t) (1)
Where
I
t
is the transmitted neutron intensity (n/cm
2
-s)
I
0
is the incident neutron intensity (n/cm
2
-s)
t is the specimen thickness (cm)
is the macroscopic cross section (cm
-1
)
Fig. 3. Attenuation of neutrons by specimen
Neutron Radiography
77
is equivalent to the linear attenuation coefficient of gamma-ray () and is the ch
aracteristic
of elements in the specimen. and are the product of atom density of elements con
tained
in the specimen (in atoms/ cm
3
) and their effective microscopic cross sections () to the
reactions of interest (in cm
2
). is the effective cross section, not the actual physical cross
section of the nucleus. It indicates probability of occurrence for each neutron
interaction. For
examples,
(n, )
indicates the probability of (n, ) reaction and

indicates the probability of
scattering reaction which combines elastic (n, n) and inelastic (n, n) scattering
cross sections.
and of pure elements and common compounds or mixtures (such as water, heavy wate
r
and concrete) can be found in literatures. Figure 4 illustrates perc
entage of 0.0253 eV
neutron transmission through different kinds of material having thickness of 1 c
m.
Element or Molecular
Atomic
number
Nominal Density
(g/cm
3
)

, barns
a
, cm
-1
Boron 5 2.3 759 97.23
Carbon(graphite) 6 1.60 3.410
-3
2.72810
-4
Heavy water
1.105 1.3310
-3
4.4210
-5
Water
1.0 0.664 0.02220
Aluminum 13 2.699 0.230 0.01386
Iron 26 7.87 2.55 0.2164
Copper 29 8.96 3.79 0.3219
Silver 47 10.49 63.6 3.725
Gadolinium 64 7.95 49000 1492
Dysprosium 66 8.56 930 29.50
Gold 79 19.32 98.8 5.836
Lead 82 11.34 0.170 5.60310
-3
Table 1. Microscopic and macroscopic cross sections of some elements and compoun
ds [4]
Fig. 4. Comparison of 0.0253 eV neutron and 0.5 MeV gamma-ray transmission throu
gh
materials having thickness of 1 cm
78
3. Neutron sources
Neutron sources for neutron radiography can be divided into 3 groups.
These are
radioisotope source, electronic source and nuclear reactor.
i. Radioisotope neutron source
Nowadays, two radioisotope sources are appropriate and available for neutron rad
iography
i.e. americium-241/beryllium (
241
Am/Be) and californium-252 (
252
Cf).
241
Am/Be produces
neutron from (, n) reaction by bombardment of beryllium (Be) nucleus with alpha p
articles
from
241
Am. The average neutron energy and the neutron emission rate are approximately
4.5 MeV and 2.2 x 10
6
neutrons/second per 1 curie (Ci) of
241
Am with a half-life of 432 years.
241
Am/Be can be available up to several tens curies of
241
Am.
252
Cf emits neutrons from
spontaneous fission with average neutron energy of 2 MeV and the emission rate o
f 4.3 x 10
9
neutrons/second per curie or 2.3 x 10
6
neutrons/second per microgram of
252
Cf.
252
Cf has a
half-life of 2.6 years and is the best radioisotope source for neutro
n radiography due to its
extremely high neutron output, low average emitted neutron energy and small size
.
ii. Electronic neutron source
Particle accelerator and neutron generator are neutron emitting sources
produced by
nuclear reactions. Particles are accelerated to a sufficient energy and
brought to hit target
nuclei to produce neutrons. Compact neutron generators are now availabl
e for field use
with neutron emission rate of 10
9
to 10
12
neutrons per second. The reactions below are
commonly used to produce neutrons.
1
D
2
+
1
D
2
0
n
1
+
2
He
3
+ 3.28 MeV, so called DD Reaction
1
T
3
+
1
D
2
0
n
1
+
2
He
4
+ 17.6 MeV, so called DT Reaction
1
D
2
+
4
Be
9
0
n
1
+
5
B
10
+ 4.35 MeV
1
H
1
+
4
Be
9
0
n
1
+
5
B
9
- 1.85 MeV
Energy of fast neurons produced from the above reactions is monoenergetic and de
pends on
the incoming particle i.e.
1
H and
2
D. The DT reaction is a well-known fusion reaction for
generating 14 MeV neutrons.
iii. Nuclear reactor
Nuclear reactor generally produces neutrons from fission reaction of ur
anium-235 (
235
U).
The fission neutron energy is in the range of 0 10 MeV with the m
ost probable and the
average energy of 0.7 and 2 MeV respectively. Fission reactions take
place in the nuclear
reactor fuel rods which are surrounded by neutron moderator. The moder
ator reduces the
neutron energy to thermal energy or slow neutron. Due to its high slow neutron i
ntensity in
the reactor core of about 10
12
10
14
neutrons/cm
2
per second, good collimation of neutron
beam can be easily obtained to give excellent image quality for neutron radiogra
phy.
Maximum neutron flux in moderator is a function of neutron emission r
ate from neutron
source and neutron energy. Thermalization factor, as shown in Table 3,
is the ratio of the
neutron emission rate (in neutrons per second, s
-1
) to the maximum neutron flux (in
Neutron Radiography
79
neutrons per second per square centimeter, cm
-2
s
-1
) in moderator. Neutron flux in water
moderator per a neutron emitted from the neutron source at any distances can be
obtained
from Figures 6 and 7. For example, the thermalization factor of
252
Cf obtained from Table 3 is
100. The neutron emission rate of
252
Cf is 2.3 x 10
6
neutrons per microgram. If a 500 mg
252
Cf
is used, the maximum flux in water can be calculated from 500 x 2.3 x 10
6
/100 = 1.15 x 10
7
cm
-2
s
-1
. From a graph in Figure 6 for
252
Cf, the maximum flux is at 1 cm distance from the
source which indicates the neutron flux of about 1 x 10
-2
per a neutron emission from
252
Cf.
Thus, the maximum neutron flux can be calculated from (500 x 2.3 x 10
6
)
x (1 x 10
-2
) = 1.15 x
10
7
cm
-2
s
-1
. Neutron flux at other distances can also be obtained from Figure 6. It should
be
noted that the thermalization factor increases with increasing emitted neutron
energy from
the source.
As mentioned earlier, neutron radiography requires low energy neutrons.
The lower
neutron energy gives better image contrast. Fast neutron or high energ
y neutrons emitted
from the source are slowed down by moderator such as water to produc
e slow or low
energy neutrons. The slow neutron energy in moderator is dependent of
moderator
temperature and the energy distribution follows Maxwellians for gas mole
cules and
particles. The slow neutron is therefore called thermal neutron. Cold neu
trons can be
produced by cooling the moderator/collimator down, such as with liquid helium, t
o obtain
better image contrast. The cadmium ratio of cold neutrons is indicated
in Table 2 where
infinity () means there is no epicadmium neutron (energy > 0.5 eV) in the beam.
Source Comments
Radioisotope
constant neutron output, low cost, maintenance free, no operating
cost, low neutron flux/long exposure time, acceptable image
quality, mobile unit is possible
Accelerator
moderate cost, moderate operating and maintenance cost, medium
neutron flux/medium exposure time, good image quality, mobile
unit is possible for small neutron generator
Nuclear Reactor
constant neutron output, high cost, high maintenance and operating
cost, high neutron flux/short exposure time, excellent image
quality, mobile unit is impossible
Table 2. Neutron sources and their general characteristics

80
Source
Emitted neutron
energy (MeV)
Normal thermal neutron
flux (cm
-2
s)
Therma-
lization
factor
L/D ratio Cd ratio
Range Mean in moderator at specimen
Radioisotope
-
241
Am/Be
(50 Ci)
-
252
Cf (1 mg)
0 10
0 - 10
4.5
2.0
10
8
10
9
10
3
- 10
4
10
3
10
5
400
100
30-50
50-100
5-20
5-20
Accelerator
-
2
D(d, n)
3
He
-
3
T(d, n)
4
He
-
9
Be(p, n)
9
B
-
9
Be(d, n)
10
B
2.7
14.1
1.15
3.96
2.7
14.1
1.15
3.96
10
8
10
9
10
8
10
9
10
8
10
9
10
8
10
9
10
4
10
5
10
4
10
5
10
4
10
5
10
4
10
5
200
600
50
100
20-30
20-30
20-30
20-30
5-20
5-20
5-20
5-20
Nuclear Reactor 0-10 2.0 10
11
- 10
14
10
6
10
8
100 100-300 100-
Table 3. Neutron sources and their technical characteristics [2, 6]
Fig. 5. Energy spectra of neutrons from neutron sources (reproduced from [5])
Neutron Radiography
81
Fig. 6. Neutron flux in water per source neutron emitted from radioisotope neut
ron
sources [6]
82
Fig. 7. Neutron flux in water per source neutron emitted from neutron producing
accelerators [6]
Neutron Radiography
83
4. Neutron collimators
Neutrons in moderator are scattered in all directions which are not suitable for
radiography.
Neutron collimator is a structure designed to extract slow neutron beam from the
moderator
to the specimen. Ideally, parallel neutron beam is preferred because i
t gives best image
sharpness. If this is the case, Soller or multitube collimator is use
d. However, divergent
collimator is easier to construct and gives good image sharpness depen
ding on the
geometrical parameters as will be discussed later.
i. Soller or multitube collimator: This collimator is constructed with
neutron absorbing
material; such as boron, cadmium and gadolinium; as illustrated in Fig
ure 8 so as to
bring parallel neutron beam to the test specimen. Neutrons can only g
et into the
collimator from one end which is in the moderator then get out to t
he other end.
Neutrons those are not travel in parallel with the collimator axis will hit the
side of the
tube or plate and are then absorbed allowing only neutrons travelling
in parallel with
the tube axis to reach the test specimen. This type of collimator is applicable
to nuclear
reactor where input neutron intensity to the collimator is high. The drawbacks a
re that
the pattern of parallel plates or tubes may be seen on the image and it is more
costly to
construct in comparison to the divergent collimator.
ii. Divergent collimator: Divergent collimator is designed in the way
that neutrons are
allowed to get into the collimator only through a small hole from one end then d
iverge
at the other end. The collimator is lined with neutron absorber to a
bsorb unwanted
scattered neutrons. It is easy to construct and can be used with non
-reactor neutron
source like radioisotope and accelerator where slow neutron input is l
ow. The
drawback is that image sharpness may not be as good as the Soller collimator. Fo
r low
neutron intensity as in radioisotope system, neutron output at the spe
cimen position
can still be increased by making part of the collimator on the input
or source side free
from neutron absorber as shown in Figure 10. Neutrons can thus enter
the collimator
through this part resulting in increasing of neutron intensity. From e
xperience with
241
Am/Be and
252
Cf sources, neutron intensity can be increased approximately by 10 -
60 % and the cadmium ratio can also be increased from about 5 to 20.
In doing so, the
image contrast is significantly improved while the image sharpness is a little p
oorer.
Fig. 8. Multitube neutron collimator
84
Fig. 9. Divergent neutron collimator allowing neutrons to get into the collimato
r only
through the hole of diameter D
Fig. 10. Divergent neutron collimator with part of the source side contains no n
eutron
absorber allowing more neutrons to get into the collimator
Neutron Radiography
85
5. Neutron radiographic techniques
After neutrons pass through the specimen they interact with the converter screen
to produce
radioisotope, alpha particle or light which can be recorded by film,
imaging plate, optical
camera or video camera. Image recording medium must be selected to ma
tch with the
particles or light emitted from the neutron converter screen so as to
obtain the maximum
efficiency. The neutron converter screen/image recording device assembl
ies commonly
used in neutron radiography are described below.
i. Metallic foil screen/film: Metallic foil with high neutron cross section is
employed to convert
slow neutrons to beta-particles, gamma-rays and/or conversion electron while ind
ustrial
x-ray film is normally used as the image recorder. Gadolinium (Gd) fo
il is the best
metallic screen for neutron radiography in terms of having extremely h
igh neutron
absorption cross section, giving the best image resolution and not becoming radi
oisotope
after neutron absorption.
155
Gd and
157
Gd are found 14.9 and 15.7 percent of natural Gd
isotopes with neutron absorption cross sections of 61,000 and 254,000 barns re
spectively.
155
Gd and
157
Gd absorb neutrons then become
156
Gd and
158
Gd correspondingly which are
not radioactive. Prompt captured gamma-rays emitted during neutron absor
ption can
cause film blackening. More importantly, prompt gamma-rays may hit atom
ic electrons
resulting in ejection of electrons from the atoms (so called conversion
electron) which
are more effective to cause film blackening. It should be noted that
less than a few
percentage of gamma-ray photons cause film blackening. Electrons and beta-partic
les are
preferred because they interact with film much more than gamma-rays.
Film may be replaced by imaging plate (IP) which has more than 10 times faster s
peed than
the x-ray film. Gd foil/x-ray film requires relatively high neutron ex
posure thus it is not
possible to carry out neutron radiography with low neutron flux system using rad
ioisotope.
About 5 years ago, Fuji started to produce neutron imaging plate by
adding Gd into the
imaging plate which can give the image quality comparable to that fro
m the Gd foil/x-ray
film assembly with approximately 50 times reduction of neutron exposure
. It is therefore
possible to be used with low neutron flux system.
Other metallic foil screens can also be used (as listed in Table 4) but the imag
e quality is not
as good as that obtained from Gd. This is mainly because low energy electrons em
itted from
Gd have very short ranges resulting in much better image sharpness. In case of h
aving large
gamma-ray contamination in the neutron beam and specimen containing gam
ma-ray
emitting radioisotopes, dysprosium (Dy) is often used. To avoid gamma-ray exposu
re to x-
ray film, the transfer method must be applied by exposing only the D
y screen with
transmitted neutrons from the specimen. During exposure, radioisotopes
165m
Dy and
165
Dy
are formed with half-lives of 1.26 minutes and 2.3 hours respectively.
The Dy foil is then
removed from the neutron beam and placed in close contact with an x-ray film to
produce a
latent image. The film density or film darkness is corresponding to t
he activity of Dy
radioisotopes formed in each part of the Dy foil.
Formation of radioisotope from neutron irradiation follows the equation below.
A = n (1 e
-
) (2)
86
Where A is the radioactivity of radioisotope formed in disintegration per secon
d (dps) after
completion of neutron irradiation. n is the number of original stable isotope a
toms. is the
neutron absorption cross section of the original stable isotope in cm
2
. is the neutron flux in
cm
-2
s
-1
. is the decay constant of the radioisotope formed in s
-1
and T is the irradiation time
in second (s).
The decay constant () can be obtained from :
= 0.693/T
1/2
(3)
where T
1/2
is half-life of the radioisotope. Form equation (1), more than 96 %
of the
maximum radioactivity can be obtained if the irradiation time is greater than 5
times of the
half-life.
Fig. 11. Illustration of direct exposure method
Fig. 12. Illustration of two steps of indirect or transfer exposure method
Neutron Radiography
87
Material
Mode of
Production of
active Isotope
Cross-
section
(barns)
Half-life
Major
particle
emitted
Direct or
Transfer
technique
Lithium Li
6
(n, )
3
935 Stable Direct
Boron B
10
(n, )Li
7
3,837 Stable Direct
Rhodium Rh
103
(n, )Rh
104
144 43 s

Diect
Rh
103
(n, n)Rh
103m
57 min x-ray
Rh
103
(n, )Rh
104m
11 4.4 min

Cadmium Cd
113
(n, )Cd
114
20,000 Stable Direct
Indium In
115
(n, )In
116
45 14 s

n e
In
115
(n, )In
116m
154 54 min

Samarium Sm
149
(n, )
m
150
41,500 Stable Direct
Sm
152
(n, )
m
153
210 46.7 h

Gadolinium Gd
155
(n, )Gd
156
58,000 Stable e
-
Direct
Gd
157
(n, )Gd
158
240,000 Stable e
-
Dysprosium Dy
164
(n, )Dy
165
800 2.3 h

n e
Dy
164
(n, )Dy
165
2,000 1.26 min

Table 4. Characteristics of Some Possible Neutron Radiography Converter Material
s [2, 7]
Radioisotope decays exponentially according to its half-life. If A
0
is the radioactivity of the
radioisotope after completion of neutron irradiation, the radioactivity
at any time t can be
calculated from
A
t
= A
0
t
e
(3)
For Gd foil, no radioisotope is formed during neutron irradiation. Emi
ssion of prompt
gamma-rays and conversion electrons follows neutron absorption by
155
Gd and
157
Gd at the
rate of n per second. In case of Dy foil,
165m
Dy and
165
Dy are formed with the radioactivity
following equation (2). After removal from the neutron facility,
165m
Dy and
165
Dy will decay
with half-lives of 1.26 minutes and 2.3 hours respectively. Film is
exposed to emitted
radiation while placing in close contact with the radioactive foil. Bu
ild-up and decay of a
radioisotope is illustrated graphically in Figure 13.
ii. Light emitting screen/film: Light-emitting screen is a mixture of
scintillator or phosphor
with lithium-6 (
6
Li) and/or boron-10 (
10
B). Neutrons interact with
6
Li or
10
B to produce
alpha-particles via (n, ) reaction. Light is then emitted from energy
loss of alpha-
particles in scintillator or phosphor. Light sensitive film, digital camera or v
ideo camera
can be used to record image. This makes real-time and near real-time
radiography
possible. The most common light-emitting screen is NE426 available fr
om NE
Technology which is composed of ZnS(Ag) scintillator and boron compound
.
Gadolinium oxysulfide (terbium) [Gd
2
O
2
S (Tb), GOS] and lithium loaded glass
scintillator are also common in neutron radiography. GOS itself is a
scintillator.
Conversion electrons as well as low energy prompt gamma-rays emitted f
rom
interaction of neutrons with Gd cause light emission. Glass scintillato
r is sensitive to

88
charged particles such as alpha- and beta-particles. Lithium is added
into the glass
scintillator so that alpha-particle will be emitted from
6
Li(n, )
3
reaction resulting in
emission of light.
Fig. 13. Graphical illustration of build-up and decay of radioactivity of a radi
oisotopein
neutron converter screen using indirect or transfer method
Fig. 14. Real-time or near real-time neutron imaging system using light-emitting
neutron
converter screen
Neutron Radiography
89
Light-emitting screen offers highest speed but gives poorest image sharpness com
paring to
other screen/film assemblies. This is the only type of screen that ca
n be used with low
neutron flux system using radioisotope neutron source. From experience, photogra
phic film
is more suitable with the light-emitting screen than industrial x-ray
film by the following
two main reasons. Firstly, photographic film is less sensitive to gamm
a-ray. As a result, it
gives better image contrast particularly when neutron beam is contamina
ted by large
fraction of gamma-rays. Secondly, photographic film is cheaper and easily availa
ble.
iii. Alpha-emitting screen/track-etch film: Alpha-emitting screen is ma
de of lithium and/or
boron compound. Particles emitted from
6
Li(n, )
3
and
10
B(n, )
7
Li reactions interact
with track-etch film (or so called solid state track detector, SSTD)) to produce d
amage
tracks along their trajectories. The detector is later put into hot c
hemical solution to
enlarge or etch the damage tracks. After etching, the damage tracks can
be made
visible under an optical microscope with a magnification of x 100 up.
Radiation dose
and/or neutron intensity can be evaluated by counting number of tracks per unit
area.
The area where track density is so large becomes translucent while th
e area with low
track density is more transparent. The degree of translucence depends on track d
ensity
resulting in formation of visible image on the film. However, contrast
of the image is
poor while sharpness is comparable to the Gd foil/x-ray film assembly.
Methods for
viewing the image is needed to improve image contrast such as reprinting the ima
ge on
a high contrast film. It has been reported that the simplest method is to scan i
mage on
the track-etch film using a desktop scanner [8]. Track-etch film is not sensitiv
e to light,
beta-particle and gamma-ray. The alpha-emitting screen/track-etch film as
sembly can
therefore be used to radiograph radioactive specimens by the direct me
thod. No
darkroom is needed for film processing. Kodak LR115 Type II, CA80-15
Type II and
CN85 Type B have been widely used during the past two decades. They
are cellulose
nitrate films coated with lithium metaborate (Li
2
B
4
O
7
). The optimum etching condition
is 10 - 40 % sodium hydroxide (NaOH) at 60 C for a duration of 30 40 minutes. Lat
er,
BE-10 screen of 93 % enriched boron-10 in boron carbide (B
4
C) form manufactured by
Kodak became available and has been widely used since then due to its highest ne
utron
conversion efficiency. Kodak LR115, CA8015 and CN85 cellulose nitrate f
ilm without
lithium metaborate are used with the BE-10 screen. CR39 plastic or po
ly(alyl diglycol)
carbonate is also available and is used extensively for alpha detection due to i
ts higher
track registration efficiency. The CR39/BE-10 assembly will probably become the
most
common in track-etch neutron radiography.
6. Applications of neutron radiography: Facilities and sample images
Neutron radiography has been employed for non-destructive testing of specimens.
Parts of
test specimen containing light elements; such as rubber, plastic, chemi
cals; can be made
visible even when they are covered or enveloped by heavy elements. Nuclear react
or gives
the best thermal or cold neutron beam for neutron radiography as can
be seen in Tables 3
and 6. The cadmium ratio is normally greater than 10 and can be as
high as 300 or even
infinity if required. The L/D ratio is always greater than 100 which indicate ex
cellent image
sharpness. Nuclear research reactor generally provides excellent beam po
rts for neutron
experiments including for neutron radiography. All neutron converter scr
een/image
90
recorder assemblies mentioned above can be employed for inspection of specimens
but the
exposure times vary considerably. Neutron exposure for some converter s
creen/film
assemblies can be estimated by using the curves in Figure 15.
Converter Screen Emitted

Particles
Recording Medium Comment
Metallic foil screen
- Gd foil
electrons Industrial x-ray film Best image quality, needs
high neutron exposure
- Dy foil
Beta-particles,
Gamma-rays
Industrial x-ray film Best for transfer method, good
image quality, needs high
exposure
Light-emitting screen
- NE426
Light
Photographic film,
Industrial x-ray film,
Optical or video
camera
Fastest speed, needs low
exposure, acceptable image
quality, allows real-time or
near real-time imaging
- GOS
-
6
Li loaded
glass
scintillator
Alpha-emitting screen
- Li
2
B
4
O
7
Alpha-particles
Track-etch film e.g.
CR39, cellulose
nitrate
Good image quality, needs
high exposure, low contrast,
requires image viewing
technique, does not require a
darkroom for film loading
and processing
- BE-10
Table 5. Common neutron converter screens and image recording media

Neutron Radiography
91
Fig. 15. Thermal neutron exposure required film for radiography [9]
For example, total thermal neutrons per square centimeter required for
Gd metallic foil
screen/film and NE-426 light emitting screen/film assemblies to make a
density of 1.5 on
film are approximately 5.5 x 10
8
and 5 x 10
6
respectively. If neutron flux at the specimen
position is 10
6
cm
-2
s
-1
, the exposure time needed for the two screens are 550 and 5 seconds
respectively.
When a 1 mg (1000 g) Cf-252 is used as in Table 6, the maximum neutron flux in wa
ter will
be 1000 g x (2.3 x 10
6
s
-1
g
-1
)/100 = 2.3 x 10
7
cm
-2
s
-1
. The neutron flux at the specimen
position for a circular cross-section, divergent collimator with an L/D
of 12 (as in Table 6)
can be calculated from:

exit
=
ouce
(D/L)
2
/16 (4)
exit
= 2.3 x 10
7
(1/12)
2
/16 = 9.98 x 10
3
10
4
cm
-2
s
-1
Where
ouce
and

exit
are neutron fluxes at the source side and the specimen position
respectively. The neutron flux obtained from calculation agrees with th
e value in Table 6.
Thus, the exposure time required for the Gd metallic foil screen/film
and NE-426 light
emitting screen/film assemblies to make a density of 2.0 on film will
be 5.5 x 10
4
and 500
seconds respectively. It is therefore impossible to use the Gd foil screen/film
assembly with
Cf-252 source. However, the neutron flux can still be increased by le
aving part of the
collimator on the source side without neutron absorber. The neutron fl
ux will then be
increased by a factor of (1 + 2a/L), where a is the length of the collimator wi
thout neutron
absorber. For example, if the length of the total collimator (L) is 30 cm and th
e part without
neutron absorber (a) is 10 cm, the neutron flux is increased by a factor of [1 +
(2 x 10)/30] =
92
1.67 or 67 %. The exposure time will then be reduced from 500 secon
ds to 300 seconds. In
doing so, the cadmium ratio is increased from about 5 to 15 -20 res
ulting in significant
improvement in image contrast but the image sharpness is gradually red
uced [10]. Use of
the second neutron converter screen can also decrease the exposure time by a fac
tor of up to
2.2 as shown in Table 7. During the past decades, the image recordin
g devices have been
rapidly improved in speed as well as graininess including film, imagin
g plate (IP), digital
optical camera, digital video camera, CCD and CMOS chips. The new dev
ices allow
radiographers to perform non-film neutron radiography with neutron generator and
Cf-252
neutron sources. The Fuji neutron imaging plate offers speed several t
en times faster than
that of the Gd/film assembly with comparable image quality [11-13]. Th
e light-emitting
screen coupled with a digital camera with light sensitivity from ISO
1600 and time
integration mode makes non-film neutron radiography by Cf-252 possible.
An image
intensifier or a microchannel plate (MCP) is useful for real-time or
near real-time neutron
imaging in low flux system. Examples of neutron radiographic images taken from d
ifferent
neutron facilities and by different techniques are illustrated in Figures 18 to
23.
Source
Collimator Typical beam characteristics
Position
Base flux
(cm
-2
s
-1
)
Intensity
(cm
-2
s
-1
)
L/D
ratio
Cd
ratio
Multi-purpose research
reactor
Radial 10
14
10
8
250 2-5
Tangential 10
13
10
7
250 10-50
Cold source 210
11
10
6
100
Radiography reactor Radial 10
12
10
6
250 2-5
Tangential 410
11
210
6
100 10-50
Be(d, n); 3 MeV, 400 A Radial 310
9
210
5
33 5-20
Be(, n); 5.5 MeV, 100 A Radial 410
8
810
4
18 5-20
T(d, n)+ U; 120 keV, 7 mA Radial 10
8
210
4
18 5-20
252
Cf ; 5 mg + sub-critical
Reactor assembly
Radial 310
9
210
5
18 5-20
252
Cf ; 1 mg Radial 210
7
10
4
12 5-20
Table 6. Examples of common neutron radiography facilities [2, 6]
Neutron Radiography
93
Converter Foil
Technique
Foil Thickness (m)
Relative Speed
Front Back
Rh/Gd Direct 250 50 5.3
Rh/Rh Direct 250 250 4.7
Gd/Gd Direct 25 50 3.7
In/In Direct 500 750 3.7
Dy/Dy Direct 150 250 3.7
Cd/Cd Direct 250 500 3.3
Ag/Ag Direct 450 450 2.7
Dy Direct - 250 2.5
Gd Direct - 25 2.4
Cd Direct - 250 2.2
Rh Direct - 250 2.1
In Direct 500 - 1.7
Dy Transfer 250 - 16.4
In Transfer 50 - 11.2
Remark : The relative speed for the direct and transfer methods are not comparab
le
Table 7. Relative Speed of Neutron Converter Screens [2]
Fig. 16. An example of Cf-252 based neutron radiography system
94
Fig. 17. An example of neutron generator based neutron radiography system
Fig. 18. A neutron radiograph of pistol bullets [14] (research reactor, Gd foil
/film technique)
Neutron Radiography
95
Neutron(Gd/film) Neutron (imaging plate) X-Ray
Neutron(Gd/film) Neutron (imaging plate) X-Ray

Fig. 19. Neutron radiographs with a floppy disk drive (above) and RS-232 connect
ors
(below) using neutron beam from a TRIGA Mark III research reactor in comp
arison with
x-ray radiograph [12]
Neutron (imaging plate) X-Ray
Fig. 20. Neutron radiograph of a Buddha statue using neutron beam from a TRIGA M
ark III
research reactor in comparison with x-ray radiograph [13] (Clay can be clearly
seen in the
middle part of the body in the neutron radiograph)
96
Neutron (imaging plate) X-Ray
Fig. 21. Neutron radiograph of an ancient lacquerware using neutron beam from a
TRIGA
Mark III research reactor in comparison of x-ray radiograph [13] (Pattern of em
broidered
bamboo thread can be seen clearer in the neutron radiograph)
X-Ray Neutron
Fig. 22. Neutron radiograph of an RS-232 connector using low intensity neutrons
from Cf-
252 and NE426 light-emitting screen/photographic film [15]
Neutron Radiography
97
(a) (b)
Fig. 23. Neutron radiograph (b) of a hard disk drive using neutrons from a resea
rch reactor
in comparison with x-ray radiograph (a) [16] (The neutron radiograph from the t
rack-etch
film was scanned by using a desktop scanner with a shiny polished metal sheet us
ed as the
light reflecting surface.)
7. Quality control of neutron radiographic image
Quality of neutron radiograph is affected by various factors not only
the L/D ratio and the
cadmium ratio as mentioned previously but also the gamma-ray content,
the geometric
unsharpness, type of converter screen, type of image recording medium and film p
rocessing.
The image sharpness is improved with increasing of the L/D ratio while the image
contrast is
improved with increasing of the cadmium ratio. The gamma-ray content in neutron
beam will
deteriorate the image contrast. The other factors affect the neutron r
adiographs in the same
way as in x-ray and gamma-ray radiography. The ASTM Beam Purity Indicator (BPI)
and the
ASTM Sensitivity Indicator (SI) are common neutron beam quality indicators as il
lustrated in
Figures 24 and 25 respectively. Department of Nuclear Engineering of C
hulalongkorn
University also developed a neutron beam quality indicator, so called C
U-NIQI, as
illustrated in Figure 26. The CU-NIQI is used with a 3 mm thick lead (Pb) plate
to determine
gamma-ray content. The quality indicators are based on the same princi
ples. Teflon and
polyethylene are hydrogeneous materials used for indicating proportion o
f slow neutrons to
fast neutrons. Cadmium and boron are good neutron absorbers used for indicating
proportion
of neutrons of energy below 0.5 eV to beyond 0.5 eV. Lead strip or
wire is the indicator for
gamma-ray content. Film density readings at the positions corresponding
to those materials
can be used to evaluate quality of the neutron beam.
98
Fig. 24. The ASTM Beam Purity Indicator [2]
Fig. 25. The ASTM Sensitivity Indicator [2]
Neutron Radiography
99
Diameter of the central through hole = 3 mm, diameter of hole no. 1 - 4 = 2 mm.
Depth of hole no. 1, 2, 3 and 4 are 1, 2, 3 and 4 mm respectively.
Fig. 26. The CU-NIQI Neutron Beam Purity Indicator developed by the Department o
f
Nuclear Engineering of Chulalongkorn University
8. Methods for determining neutron exposure
In practice there is no standard procedure for constructing an expos
ure curve for neutron
radiography as in x-ray and gamma-ray radiography. For specific applica
tion, the
radiographer can first begin with trial and error to choose the best
exposure. Without
knowing the specimen composition and thickness to determine the neutron attenuat
ion, the
exposure cannot be obtained. The best method, so far, is to measure
transmitted neutron
intensity by using a small neutron detector at a few positions behind
the specimen before
placing the screen/film assembly. The transmitted neutron intensity is
inversely
Fig. 27. Illustration of a method for determining the neutron exposure by measur
ing
transmitted neutron intensity with a small neutron detector
100
proportional to the exposure. In doing so, the radiographer do not ne
ed to perform decay
and distance corrections. This method can also be applied in x-ray an
d gamma-ray
radiography by changing the neutron detector to x-ray/gamma-ray detector.
9. Acknowledgements
The author would like to express his deepest gratitude to Mr. JateCha
n Channuie and Mr.
Kittiwin Iaemsumang for their assistance in the preparation of the figures and t
he tables in
this chapter. Thanks are also extended to my eldest daughter, Miss Ka
triya Chankow, and
my wife, Mrs. Julie Chankow, for their times spent in correcting the English tra
nslation.
10. References
[1] J.P. Barton. Neutron Radiography An Overview, Practical Applications
of Neutron
Radiography and Gauging. ASTM STP 586, American Society of Testing and
Materials, 1976, p. 5 - 19.
[2] Commission of the European Communities. Neutron Radiography Handbook. Edite
d by P.
von der Hardt and H. Rttger, D. Reidel Publishing Company, Dordrecht, 1981.
[3] D.A. Garrett and H. Berger. The Technological development of Neutron Radiogr
aphy,
Atomic Energy Review. Vol. 15(2), 1977, p.123 - 142.
[4] J.R. Lamarsh. Introduction to Nuclear Engineering. 2
nd
Ed., Addison-Wesley, New York, 1983.
[5] A.A. Harms. Physical Processes and Mathematical Methods in Neutron Radiograp
hy,
Atomic Energy Review. Vol. 15(2), 1977, p.143 - 168.
[6] M.R. Hawkesworth. Neutron Radiography : Equipment and Methods, Atomic
Energy
Review. Vol. 15(2), 1977, p.169 - 220.
[7] H. Berger. Detection System for Neutron Radiography, Practical Applications o
f Neutron
Radiography and Gauging. ASTM STP 586, American Society of Testing and
Materials, 1976, p. 35 - 57.
[8] N. Chankow. A Simple Method for Viewing Track-Etch Neutron Radiographic Imag
es
Using a Scanner. Proceedings of the 2
nd
International Meeting on Neutron Radiography
System Design and Characterization, Yokosuka, Nov. 12-18, 1995.
[9] Y. Suzuki. et al. Development of Imaging Converter, Proceedings of
the 3
rd
World
Conference on Neutron Radiography, Osaka, 1989.
[10] S. Jaiyen. , Development of A Prototype for Low-Flux Thermal Neutron Radiogr
aphy
System Using Cf-252, Master Thesis, Department of Nuclear Technology,
Chulalongkorn University, 2002.
[11] S. Wonglee. Neutron radiography using Neutron Imaging Plate, Master
Thesis,
Department of Nuclear Technology, Chulalongkorn University, 2007.
[12] N. Chankow, S. Punnachaiya and S. Wonglee. Neutron radiography usin
g Neutron
Imaging Plate, Applied Radiation and Radioisotopes. Vol.68 (4), 2010, p. 662-665.
[13] R. Funklin. Application of Neutron radiography using Neutron Imaging

Plate for
Inspection of Ancient Objects, Master Thesis, Department of Nuclear Technology,
Chulalongkorn University, 2009.
[14] P. Suksawang. A Study of Neutron radiography, Master Thesis, Departm
ent of
Nuclear Technology, Chulalongkorn University, 1981.
[15] P. Orachorn. Cf-252 Based Neutron radiography using Neutron Imaging Plate,
Master
Thesis, Department of Nuclear Technology, Chulalongkorn University, 2009.
[16] W. Ratanathongchai. Neutron radiography using BE-10 Neutron Converte
r Screen,
Master Thesis, Department of Nuclear Technology, Chulalongkorn University, 1997.
5
Flaw Simulation in Product Radiographs
Qian Huang and Yuan Wu
South China University of Technology
China
1. Introduction
X-ray inspection machines have been widely used in modern industry to explore th
e interior
of products, and to control product quality. With the advancement of
automatic product
inspection, we need lots of flaw radiographs to validate the systems sensitivity
and to help
tune the automatic inspection parameters. Flaw simulation in product radiographs
has been
receiving an increasing attention for over two decades.
Before implementing the program to inspect the work pieces, a large n
umber of sample
images are needed to tune the algorithm, examine its performance, and ensure its
accuracy.
Product radiographs with defects from the production line are the best
, however, they are
often not available in sufficient quantities or variety. Simulation o
f casting defects is an
alternative approach to deal with this problem.
Recently, the CAD model method, the Monte Carlo method and the genera
tive image
method have become the three main approaches for flaw simulation.
1. The CAD model method enables simulations to be produced for comple
x three
dimensional (3D) casting objects. Ray-tracing, together with calculation
of the X-ray
attenuation, is the basis of this model to produce 3D casting defect simulations
.
2. Monte Carlo simulation is a method for iteratively evaluating a de
terministic model
using sets of random numbers as inputs. The creation of a proper physical model
is the
most critical task for the simulation.
3. Generative image model for flaw simulation of the product radiograp
hs has been
developed from the idea of the superimposition technique, and is based
on defect
analysis.
In this chapter, we will discuss flaw simulation in product radiograph
s. The chapter is
organized as follows. Section 2 introduces the CAD models for simulati
on, the three main
parts of CAD models will be discussed including the X-ray source, the
geometric and
material properties of the objects, and the imaging process. Section 3
presents the
applications of the Monte Carlo method for X-ray image simulation, a brief intro
duction of
the Monte Carlo method and how to combine it in the simulation application are d
iscussed.
Section 4 concludes with the generative image model for flaw simulation in radio
graphs, the
generative models and simulation results are outlined, the authors them
selves have
produced significant results in this field. Section 5 concludes and
offers suggestions for
further research. Finally, a reference list is provided for further reading.
2. CAD Models for simulation
The CAD model simulates the entire X-ray imaging process. There are t
hree main
components of an X-ray penetration model, the X-ray generator, the interaction b
etween the
X-ray and the object and the imaging process (Tillack et al., 2000). In CAD simu
lation, the X-
ray source, the geometric and material properties of objects as well
as the imaging process
are simulated independently. For product flaw simulation, the CAD model
can be used in
the two approaches, which are for a whole product with defects or flaws embedded
, or for
the flaws only but using post-processing for its applications.
2.1 CAD Models for radiographs
The X-ray beam comes from the X-ray source, penetrates the object, at
tenuates and finally
projects into the image detector, see Fig. 1. So each pixel ( , ) u v
in the simulated image can
be projected back though the modelled object to the X-ray beam source
. Since it is very
difficult to make the X-ray source from the x-ray tube into a single
point light source, this
can be seen from Fig. 1. The source, which refers to the area of the emitter apert
ure should
be very small, to minimise the blurring of the edges of the work piece in a radi
ograph or on
the film, the enlarged circle in Fig. 1.
Fig. 1. The X-ray imaging system with the problems of the source and object edge
s in the
radiograph
The X-ray attenuation law is the basis of the simulation process (Haken &Wolf, 1
996):
=
0
cxp (px) (1)
where
0
is the incident radiation intensity, is the transmitted intensity, x is the thi
ckness
of the object, and p is the energy attenuation associated with the material. The
gray value of
the simulated image can be computed as follows:
I = A
0
(E)0cxp _p
()x
_ +B (2)
Flaw Simulation in Product Radiographs 103
where
0
(E) is the incident radiation intensity with energy E, is the solid
angle that
corresponds to the pixel observed from the source point, p
(E) represents the attenuation of

material i at energy E, and x
is the total path length through material i. Fig.2 is an example

of a work-piece radiograph simulated from a CAD model and attenuation data.
Fig. 2. A three-dimensional (3D) model of a workpiece and its generated radiogra
ph
Some flaws, such as cavities, can be simulated as nil X-ray absorptio
n and others, such as
incrusted material, with a different attenuation coefficient p
d
. We rewrite the equation (1) as
following:
=
0
cxp(p(x J)) cxp(p
d
J) = (x)cxp(J(p p
d
)) (3)
Where d is the thickness of the flaw. If the flaw is a cavity with
zero absorption, p
d
= u
(Mackenzie & Totten, 2006). Fig. 3 shows the small cavities in a workpiece on th
e left and on
the right the simulated X-ray image compared with its real radiograph in the mid
dle.
Fig. 3. The 3D image of a work-piece at left volume and its radiograph in the mi
ddle and the
simulated image at right
In this technique, the entire object, including its defects, will be
built up by independent
CAD software. It offers excellent flexibility for setting objects and
flaws, but it is not
convenient and user friendly. The object itself may be complicated. It is time c
onsuming to
build up such a complicated model, moreover, in which every point sho
uld be calculated
according to the attenuation law to form a simulated radiograph. What we need is
to show
flaws where they may appear on a real product. It is possibly easier
to build up the flaw
model only and superimpose the simulated flaws onto a real X-ray radiograph.
2.2 CAD models for flaw only
There is a serious disadvantage in the first approach. Building up an
entire CAD model of
the object and computing each pixel of the simulated image make the
enormous
computation. The second approach can be regarded as an improvement of
the first one. In
this technique, a 3D modelled flaw is projected and superimposed onto a real X-r
ay image.
The gray value displayed from the captured CCD image can be expressed
as a linear
function of the transmitted radiation:
I(x) = A(x) + B (4)
where x is the thickness of the object. If there is a cavity with
thickness d, according to
equation (3), the equation will be changed into:
I(x-u) = A(x)exp(u) + B (5)
According to equation (4), A(x) can be substituted by I(x) B:
I(x J) = (I(x) B)exp(pJ) +B (6)
In an X-ray image, the brightest point I
max
is obtained when the thickness is zero and the
darkest point I
min
is obtained when the thickness is a maximum.
I
mux
= A
0
+ B (7)
I
mn
= A
0
exp(px
mux
) + B (8)
With equation (6) and (7), B can be estimated as following:
B = I
mux

I
mcx
-I
min
1-cxp(-x
mcx
)
(9)
In this technique, a 3D flaw model, the material absorption coefficien
t p, and a real X-ray
image are needed. Then, parameter B is estimated according to equation
(9). The 3D flaw
model with thickness d is projected and superimposed onto the real X-
ray image using
equation (6). The new gray value of a pixel, including the projected
3D flaw, comes from
equation (6). All we need are the absorption coefficient and the flaw
thickness (Mery, 2001).
The original gray value provides us part of the information needed. The work req
uired in this
approach is much less than in the first approach. But in this approach, we do
n t consider the
solid angle 0 of the X-ray beam that is projected onto the pixel, which generates
an error.
Fig. 4. Very simple simulated casting defects compared with the real ones
2.3 Coordinate system
The CAD model and the X-ray source use virtual 3D world coordinates, but the ima
ge uses
2D digital image coordinates. We need to transform between these two
systems. 3D
coordinates can be denoted as:
H = |X Z]
1
(10)
We assume that the digital image has the horizontal and vertical scale factors K
u
and K
. The
focus distance is f. To maintain the projected system, the projected
plane must be placed
within the focus distance f. If the digital image has size (x
max
, y
max
) pixels, the projected plane
must have the physical size of (x
mux
K
u
, y
mux
K
)millimetres. It means that the projected

plane samples every 1 K
u
millimetres horizontally and 1 K
millimetres vertically during

the simulation process. A pixel m=(x, y) in the simulated image can be converted
into
H = j
0
0
+x
K
u
v
0
-
K
[
1
(11)
in the 3D coordinate system while (u
0
, I
0
) is the origin of the projected plane.
1/Kv
1
U
K
Fig. 5. The coordinate system.
To simulate the flaw there are 4 steps needed. First, capture a real X-ray image
, and take the
parameters f, K
u
and K
v
, (x
max
, y
max
) ; Second, build a 3D flaw model as a 3D mesh; Third,
position the 3D flaw in the virtual coordinate system and simulate th
e X-ray capture
process, obtaining a depth map matrix W(x,y) which stores the length
travelled within the
component of each ray that penetrated the 3D flaw. Finally, compute t
he new gray value
with the depth matrix using equation (6). A single ray is emitted from every pix
el of the 2D
image to the X-ray source. The distance travelled by the ray inside
the flaw will be
calculated and stored in the depth matrix (Hahn & Mery, 2003).
2.4 Discussion of CAD model simulation
First, each intersection alternates the state of the ray from outside to inside.
The final depth
is the sum of all internal traces and the attenuation calculation of the X ray w
ill be difficult
for irregular product shape.
Second, the computation is enormous. Every pixel of the simulated imag
e needs to emit a
ray to penetrate the 3D model to the source. For simplification, we
assume that the X-ray
comes out from a point source. But actually, the X-ray source is not
an ideal point source
and it has a scatter effect. Another cause that extends the computati
on is the shape of the
projected plane. The projected plane is by definition a plane with the central p
ixel closer to
the X-ray source than the corner pixels. The intensity of the X-rays is stronger
in the middle.
We know a perfect detector would be a spherical one which is far mo
re complicated in
coordinate computation.
Third, the CAD model is not user friendly. A three-dimensional CAD mo
del is needed as
seen from Fig.2, but building up such a CAD model is a time consuming job. Altho
ugh the
second approach doesnt need an entire object model, the 3D flaw model
is needed.
However there are no prepared parameters for flaws normally. Considerin
g the stochastic
shape of the most defects, generating 3D CAD flaw model, which have to be create
d before
all other calculation and projection, is a key technique, because usin
g a fixed CAD model
lacks randomization of flaw models.
Finally, there is a lack of description language and criteria for man-made 3D fl
aws. Its hard
to tell whether man-made flaws are like or unlike real flaws.
3. Monte Carlo methods and their application in X-ray image simulation
3.1 Monte Carlo methods
A brief introduction of Monte Carlo methods is presented to better understand th
e concept
and the four steps that Monte Carlo methods usually follow. An example of estima
ting is
summaiiseu anu the applications aie uiscusseu.
3.1.1 Introduction
The Monte Carlo method was developed in the 1940s by John von Neuman
n, Stanislaw
Ulam and Nicholas Metropolis, while they were working on nuclear weapo
n projects
(Manhattan Project) in the Los Alamos National Laboratory. It was named in homag
e to the
Monte Carlo Casino, a famous casino, where Ulam s uncle would often g
amble away his
money (Metropolis, 1987).
Monte Carlo methods are a class of computational algorithms that rely on repeate
d random
sampling to compute their results. Monte Carlo methods are often used
in simulating
physical and mathematical systems. These methods are most suited to ca
lculation by a
computer and tend to be used when it is infeasible to compute an ex
act result with a
deterministic algorithm.
Monte Carlo methods are especially useful for simulating systems with
many coupled
degrees of freedom, such as fluids, disordered materials, strongly coup
led solids, and
cellular structures. They are used to model phenomena with significant
uncertainty in
inputs, such as the calculation of risk in business. They are widely used in mat
hematics, for
example to evaluate multidimensional definite integrals with complicated
boundary
conditions. When Monte Carlo simulations have been applied in space ex
ploration and oil
exploration, their predictions of failures, cost overruns and schedule overruns
are routinely
better than human intuition or alternative "soft" methods.
3.1.2 Steps
Monte Carlo methods vary in different situations, but tend to follow
these particular and
simple steps:
1. Define a domain of possible inputs.
2. Generate inputs randomly from a probability distribution over the domain.
3. Perform a deterministic computation on the inputs.
4. Aggregate the results.
For example, consider a circle inscribed in a unit square. Given that the circle
and the square
have a ratio of areas that is /4, the value of can be approximated
using a Monte Carlo
method (Kalos &Whitlock, 2008):
1. Draw a square on the ground, and then inscribe a circle within it.
2. Uniformly scatter some objects of uniform size (grains of rice or sand) over
the square.
3. Count the number of objects inside the circle and the total number of object
s.
4. The ratio of the two counts is an estimate of the ratio of the
two areas, which is /4.
Multiply the result by 4 to estimate .
In this procedure the domain of inputs is the square that circumscrib
es our circle. We
generate random inputs by scattering grains over the square then perform a compu
tation on
each input (test whether it falls within the circle). Finally, we aggregate the
results to obtain
our final result, the approximation of .
To get an accurate approximation for this procedure should have two
other common
properties of Monte Carlo methods. First, the inputs should truly be
random. If grains are
purposefully dropped into only the centre of the circle, they will no
t be uniformly
distributed, and so our approximation will be poor. Second, there should be a la
rge number
of inputs. The approximation will generally be poor if only a few gr
ains are randomly
dropped into the whole square. On average, the approximation improves as more gr
ains are
dropped.
3.1.3 Applications
Monte Carlo methods are especially useful for simulating phenomena with
significant
uncertainty in inputs and systems with a large number of coupled degr
ees of freedom.
Areas of application include: physical sciences, engineering, computational biol
ogy, applied
statistics, games, design and visuals, finance and business, telecommunications,
etc.
3.2 Monte Carlo and computer simulation
Computer simulation had developed rapidly just after the invention of
the computer and
the first large-scale deployment is to model the process of nuclear d
etonation. In this
simulation, the scientists used a Monte Carlo algorithm.
3.2.1 Monte Carlo and computer simulation
A computer simulation is a computer program that attempts to simulate an abstrac
t model
of a particular system. Computer simulations have become a useful part
of mathematical
modelling of many natural systems in physics (computational physics), a
strophysics,
chemistry and biology, human systems in economics, psychology, social s
cience, and
engineering. Simulations can be used to explore and gain new insights into new t
echnology,
and to estimate the performance of systems too complex for analytical solutions.
Computer simulation developed hand-in-hand with the rapid growth of the
computer,
following its first large-scale deployment during the Manhattan Project
in World War II to
model the process of nuclear detonation. It was a simulation of 12 h
ard spheres using a
Monte Carlo algorithm. Computer simulation is often used as an adjunct
to, or substitute
for, modelling systems for which simple closed form analytic solutions
are not possible.
There are many types of computer simulations; the common feature they
all share is the
attempt to generate a sample of representative scenarios for a model
in which a complete
enumeration of all possible states of the model would be prohibitive or impossib
le.
3.2.2 Monte Carlo simulation software
Many software systems and computer platforms can perform Monte Carlo simulation
tasks,
Matlab and C++ are the most well-known. Some specially-designed softwar
e systems for
Monte Carlo simulation are very popular with researchers. Two of them will be in
troduced
below.
1. EGSnrc
The EGSnrc system (EGSnrc, URL) is a package for the Monte Carlo simu
lation of coupled
electron-photon transport. Its current energy range of applicability is consider
ed to be 1 keV
to 10 GeV. The EGS acronym stands for Electron Gamma Shower, and EGS
nrc is an
extended and improved version of the EGS4 package originally developed
at Stanford
Linear Accelerator Center (SLAC). In particular, it incorporates significant imp
rovements in
the implementation of the condensed history technique for the simulatio
n of charged
particle transport and better low energy cross sections.
2. GoldSim
GoldSim is the premier Monte Carlo simulation software solution for dy
namically
modelling complex systems in business, engineering and science (GoldSim, URL). G
oldSim
supports decision and risk analysis by simulating future performance wh
ile quantitatively
representing the uncertainty and risks inherent in all complex systems.
Organizations
worldwide use GoldSim simulation software to evaluate and compare alter
native designs,
plans and policies in order to minimize risks and make better decisio
ns in an uncertain
world.
3.3 X-ray images simulation by Monte Carlo methods
As the X-ray imaging system must be modelled by Monte Carlo simulatio
n, the physical
principles of an X-ray imaging system will be introduced firstly. Two publicatio
ns with the
corresponding models and the simulation results are discussed.
3.3.1 Principles of the X-ray imaging system
Typical X-ray imaging systems consist of a point X-ray source and an
image recording
device (detector). The object to be imaged is placed between the X-ra
y source and the
detector. The images are produced by local reduction in the intensity of the pri
mary X-rays
produced by the X-ray source reaching the detector through the interna
l structure of an
object in their path. These local intensity reductions correspond to t
he distribution of
absorption (attenuation) in the object under study. The special amount of attenu
ation by the
object depends, in turn, on its thickness, its chemical composition, a
nd the X-ray energy.
Besides the source and detector, other auxiliary devices exist in a real imaging
system. The
schematic diagram is illustrated in Fig. 6.
Fig. 6. The schematic depiction of an X-ray imaging system.
3.3.2 Direct imaging simulation model
U. Bottigli et al. developed a voxel-based Monte Carlo simulation model
of X-ray imaging.
They try to get the X-ray images from this simulation model directly. Two main s
teps can be
summarised as below:
1. Modelling
The energy spectrum, polarization and profile of the incident beam can be define
d so that X-
ray imaging systems can be modelled. Photoelectric absorption, fluorescent emiss
ion, elastic
and inelastic scattering are included in the model. The detailed model and the c
omplicated
mathematical descriptions can be found in the literature (Bottigli et al., 2004)
.
2. Simulation
The Monte Carlo method is employed in the course of simulation. The core of the
simulation
is a fast routine for the calculation of the path lengths of the photon trajecto
ry intersections
with the grid voxels. The selected simulation result can be seen in Fig. 7 below
.
In this experiment, only the absorption is considered, i.e. the photon
s from scattering or
fluorescence emission are discarded. This type of simulations is useful
for testing the
correctness and the speed of the procedure for numerical integration of the abso
rption along
an arbitrary path inside a grid. The CPU time for the simulation of one projecti
on image was
about 3 seconds using a PC based on a PENTIUM IV 1800MHz processor.
(a) Experimental image
(b) Simulated image
Fig. 7. The simulation result of the voxel-based Monte Carlo simulation model. (
a) is the
experimental image, and (b) is the simulated image.
3.3.3 Scattered image simulation model
Different from the direct imaging model, the scattered image model foc
uses on simulating
the scattered image. The final image can be obtained by subtracting t
he scattered image
from the real image taken during the inspection. Frank Sukowski et al
. provided a simple
description of this model in (Sukowski & Uhlmann, 2011).
A lot of effects affect the image quality in radioscopy systems. By optimizing t
he throughput
of the inspection system it is of essential interest to suppress all
effects reducing the image
quality. One effect e.g. is the scattered X-ray radiation from inside
the specimen during
inspection which hits the detector and reduces the contrast and sharpness of the
projection.
This effect leads to reduced possibility of detection of small defects
. With the Monte Carlo
simulation is it possible to simulate the scattering effects in the s
pecimen and also the
distribution of the scattered radiation on the detector. If we know the intensit
y distribution
of the scattered radiation from the specimen on the detector, it can
be subtracted from the
real image taken during the inspection. With this operation it is possible to ge
t images of the
specimen with nearly zero intensity of scattered radiation resulting in
better contrast and
higher sharpness of the image. The author believes that simulation is the only w
ay to get a
realistic and not approximated intensity distribution of scattered radia
tion. In Fig. 8 the
simulated projection of a step wedge and the simulated intensity distr
ibution of the
scattered radiation are shown.
3.4 Discussion about Monte Carlo simulation method
Monte Carlo simulation is a method for iteratively evaluating a determ
inistic model using
sets of random numbers as inputs. This method is often used when the
model is complex,
nonlinear, or involves more than just a couple uncertain parameters. A
n X-ray based
Fig. 8. The scattered image simulation. The left half is the simulated projectio
n of a step
wedge, and the right half is the simulated scattered image.
imaging system is a very complex system including lots of factors which can affe
ct the final
image. To simulate this kind of complex system with a Monte Carlo me
thod, to build a
proper model, is difficult and important. In this research area there
is no readily available
model that can be followed, even a tiny difference of hardware or device positio
n can lead a
very different model, and the objects under inspection can also affect
this model. Another
problem is the computing cost of simulation, which may be extremely high due to
the model
complexity. However, the Monte Carlo method is a very powerful tool t
o solve complex
simulation problems, and its application in X-ray imaging systems is still being
explored.
4. Generative image for flaw simulation in product radiographs
Defect superimposition is another technique that attempts to simulate c
asting defects. It
differs from the CAD approach in the process of defect creation using 2D image t
echnology,
to superimpose the simulated casting defects onto real radioscopic imag
es (Filbert et al.,
1987). It needs neither complex 3D software packages nor a model of
the casting specimen
under test, furthermore, it offers a radioscopic image of a real prod
uct with a range of
possible defects to test, validate and measure the accuracy of differe
nt radiograph analysis
procedures, or for tutorial and training purposes.
4.1 Image generation technology
The image generation technology for flaw simulation of product radiogra
phs is developed
from the idea of the defect superimposition technique, and is based on defect an
alysis. The
descriptions of defects or flaws defined by different standard organizations for
specimens or
products use a language and high level semantics which can be underst
ood by engineers
and workers in the field. Though many sample images are used as stan
dards and are
available for demonstration, they cannot include all because almost any
defects in the
products are distinguished greatly from others in their shapes, grey c
ontrasts, salience of
edges, size of areas, girth of counters etc. The defects are recogniz
ed from a product X-ray
image normally depending on an understanding of the descriptions or de
finitions of the
flaw but also rely on the calculation of their image parameters. We can generate
flaw images
according to the image semantic interpretation rather than depend heavi
ly on the system
parameters of the X-ray imaging or CAD models which provide the geometrical para
meters
of products.
The flaw generation process, as shown in Fig. 9, consists of five steps. The fir
st step is defect
image collecting. Though the definition for one kind of defect is uniform or uni
que, images
have great variety, i.e. polymorphism in the defect formation. From a
certain number of
defect samples, the qualitative visual features are extracted and analysed. In t
he second step,
which is a key step called seed featureor main high frequency feature creation pro
cess, the
defect images are formed or grown from it. Simultaneously, the fuzzy image repre
sentations
for each type of feature are expected to apply to the image generati
on after obtaining the
qualitative features and statistical analysis, fuzzy membership is neede
d and is applied to
these image features in the third step. The fourth step is generating low freque
ncy features
in the regions of the creative defect, to subject the defect appearan
ces to their semantic
definitions. The last step is image synthesizing or superimposing the
created flaw image
onto real product radiographies.
Defect image
collecting
Image
synthesizing or
superimposing
High & low
frequency
features
generation
Fuzzy image
representations
Qualitative visual
features
Fig. 9. Phases of flaw image generation for product radiographs
4.2 Defect image collecting
As X-ray radiation is passed through the material under test, and a
detector senses the
radiation intensity attenuated by the material, a defect in the material modifie
s the expected
radiation received by the sensor. The contrast in the X-ray image bet
ween a flaw and a
defect-free area of the specimen is distinctive. In an X-ray image we can see th
at the defects,
such as voids, cracks, or bubbles, show up as bright areas compared to their bac
kground, i.e.
the imaging of defect-free material.
Defects are defined according to their appearances, as many different kinds are
described by
the natural language along with some radiograph images, as we can see
in Fig.10. The
images shown in Fig. 10 are chosen from the reference radiographs pre
sented by The
American Society for Testing and Materials (ASTM) E155 (Hoff, 1999). T
hese are standard
radiographs of castings produced to help the radiographer make appropriate asses
sments of
the defects in casting components.
As discussed in detail in the NDT Resource Center (Iowa State Univers
ity, 1996), the
characteristics of casting defects, the discontinuities produced by gas porosity
or blow holes,
sand and dross inclusions, various kinds of shrinkages, are summarized
in radiography-
NDT course material (Iowa State University, 1996) are quoted here:
(a) (b) (c) (d)
(e) (f) (g)

Fig. 10. The sample patterns of the American Society for Testing and Materials (
ASTM). (a)
gas porosity or blow holes (b) shrinkage of needle cavity(c) shrinkage of round
cavity (d)
sponge shrinkage (e) sand inclusions or dross (f) filamentary shrinkage (g) dend
ritic
shrinkage
1. Gas porosity or blow holes are usually smooth-walled rounded cavities of a sp
herical,
elongated or flattened shape caused by accumulated gas or air which is trapped b
y the
metal. The gas or air will be trapped as the molten metal begins to
solidify while the
sprue is not high enough to provide the necessary heat transfer needed to force
the gas
or air out of the mold.
2. Sand inclusions and dross are nonmetallic oxides, which appear on
the radiograph as
irregular, dark blotches. These come from chipped parts of the mold o
r core walls
and/or from oxides which have not been skimmed off prior to the intr
oduction of the
metal into the mold gates.
3. Shrinkage has various forms of discontinuity occurring in all cases as the m
olten metal
shrinks as it solidifies in all parts of the final casting. Shrinkage in its var
ious forms can
be recognized by a number of characteristics on the radiographs. There are at le
ast four
types of shrinkage: (a) cavity; (b) dendritic; (c) filamentary; and (d) sponge t
ypes. Some
documents designate these types only by number to avoid misunderstandin
g. Their
characteristics are summarized in the following:
(a) Cavity shrinkage appears as areas with distinct jagged boundaries.
It may be produced
when metal solidifies between two original streams of melt coming from opposite
directions
to join a common front. Cavity shrinkage usually occurs at a time when the melt
has almost
reached solidification temperature and there is no source of supplement
ary liquid to feed
possible cavities.
(b) Dendritic shrinkage is a distribution of very fine lines or small
elongated cavities that
may vary in density and are usually unconnected.
(c) Filamentary shrinkage usually occurs as a continuous structure of
connected lines or
branches of variable length, width and density, or occasionally as a network.
(d) Sponge shrinkage shows itself as areas of lacy texture with diffu
se outlines, generally
toward the mid-thickness of heavier casting sections. Sponge shrinkage may be de
ndritic or
filamentary. Filamentary sponge shrinkage appears more blurred as it is p
rojected through
the relatively thick coating between the discontinuities and the film surface.
The X-ray image is usually captured with a frame-grabber and stored in a matrix.
The size of
the image matrix corresponds to the resolution of the image. The gray value betw
een 0 and
255 associated with the image brightness denote from the one hundred
percent black to
white in an image. Tuned by the voltage and current for the X-ray tube, the gray
intensity of
a product and its defects will vary significantly, that means the contrast betwe
en the defect
and its background will change, the examples shown in Fig. 11, in which the tube
current is
1.5mA but the tube voltages are varied from 110KV to 105KV and 75KV
in each from
left to right.
Fig. 11. The radiographs with different tube voltage for the same specimen, the
tube voltage
used from left to right are separately 110KV, 105KV and 75KV and all with tube c
urrent
1.5mA
From the pictures in Fig. 11 we can see that not only the contrast of the defect
s against their
backgrounds vary in every image but the brightness of the defects als
o changes. Although
using high voltage, the defect in the left image is the darkest comp
ared with the defects in
the other two images with relatively low voltages, and the contrast around the d
efect in the
left image associated with its background is very low.
It is necessary to collect the sample defect radiographs and classify
them into different
semantic categories and if possible, get their radiographs with different voltag
es.
4.3 Visual feature analysis and nonparametric sampling
Deduced from example-based technique for image generation, the nonparametric sam
pling
and modelling that relies on the qualitative visual feature analysis c
ontributes to set up
nonparametric models for both main construction of defects, their textu
re synthesis and
other low frequency information. The introduction of nonparametric sampl
ing and defect
image reconstruction leads to an intricate feature analysis problem. He
re we mention two
different classical nonparametric sampling and modelling approaches, one
is a stochastic
defect counter based on a hierarchical template model and the other i
s a skeletonization-
based model.

4.3.1 Counter based hierarchical template model
Most casting defects are located on the inner part of the work piece
s, as seen in Fig. 12,
where the two diagrams on the left are typical of actual radiographs with defect
s generated
from the production line, the images sizes are 540 540 pixels, and the
others are the
diagrams of what is seen when the porosity defects are cut from the
actual images and
magnified to display with image sizes of 200 200 pixels. The intensit
ies of casting defects
in the radiographs are usually lighter than the parts around them. If
the work pieces have
different structures and shapes, the defects with the same size often appear as
different grey
values on different work pieces. This is a result of the different absorption of
the X-rays by
the metal due to the different thicknesses. The energies required for X-ray pene
tration of the
thicker structures are significantly higher, and the grey values of th
e casting defects are
strongly affected by the corresponding backgrounds. This is illustrated
in Fig. 12, showing
casting defects in casting products. Though it is not easy to classif
y them into type one or
type two defects, all of them are single large and discontinuous. The aim of thi
s paper is to
generate simulated defects similar to these real ones.
Fig. 12. The sample defects in real casting pieces
The cavities, as shown in Fig. 13 (a) and (b), are both single large casting de
fects, however,
they are considered separately as blow hole and shrinkage defects, and they diff
er greatly in
appearance and volumetric measurement from the porosity defect shown in Fig. 13
(c).
(a) (b) (c)

Fig. 13. Sample defects in real casting pieces
By careful inspection of the visual features of about 2000 images of casting def
ects, we have
found that stochastic shapes with closed defect contours are the most
prominent features.
Based on the principle of Active Qualitative Vision (Aloimonos, 1990), an initia
l shape needs
to be designed to provide a starting point followed by the adjustment of the fol
lowing image
parameters during the simulation: grey level value, shape, size, orientation, sh
ading effects,
and contrast with background. The random shape template is designed as three emb
edded
nesting stencil-plates, as illustrated in Fig. 14, in which the shapes
of each template can be
given randomly but their sizes are limited so that every smaller one
of three shapes fits
entirely inside the larger one for the appropriate grey value adjustme
nt. The analysis and
the measurement of the profiles of the largest cavity defects, from a
ll of our casting defect
images, leads to a size for the profile of the largest stencil-plate of 78x46 pi
xels (1cm=31.889
pixels), which approximates to the size of the single defect profiles that appea
r normally in
casting pieces.
From Fig. 14 and from the physics of the x-ray process it is easy to find that t
he middle of
the cavities is normally brighter than their edges. For convenience the higher g
ray value is
set to the middle of the entire shape; and also there is a general
requirement for the
template set that their areas vary, with the smaller being totally en
closed by the larger.
With the proposed defect creation algorithm, the stencil-plates are not
limited in shape,
and changes in their formation will not significantly influence the ou
tcome of the
generated defects. Furthermore, the templates used are binary images. The white
parts in
the templates are the shapes used in the defect simulation. Though th
ese white parts are
not the exact final shape of the simulated defects, they approximate
to the final shape
having some of the features of real casting defects with smooth, continuous, and
irregular
contours.
Fig. 14. The sample defects in real casting pieces
We use one of shapes in Fig.14 to design the superposing operation to locate the
simulated
defect onto a specified region of the original X-ray image; this superposing ope
ration can be
used by all stencil-plate superposition, the details can be found in refere
nce paper (Huang,
et al., 2009).
4.3.2 Skeletonization based model
The skeleton of some dendritic flaws such as filamentary shrinkage and
dendritic
shrinkage in castings or cracks in any materials can be the main feature or so c
alled seed
feature to let simulated flaws grow from it. Skeletonization is a proc
ess for reducing
foreground regions in a binary image to a skeletal remnant that large
ly preserves the
extent and connectivity of the original region while throwing away mos
t of the original
foreground pixels(Gonzalez, 2008). It can be a key step when extracting the nonp
arametric
features from a sample defect. To see how this works, imagine that the foregroun
d regions
in the input binary image are made of some uniform slow-burning mater
ial. Light fires
simultaneously at all points along the boundary of this region and wa
tch the fire move
into the interior. At points where the fire travelling from two diffe
rent boundaries meets
itself, the fire will extinguish itself and the points at which this happens for
m the so called
quench line. This line is the skeleton. Under this definition it is cl
ear that thinning
produces a sort of skeleton.
The dash line in set A denotes its skeleton S(A), as shown in Fig. 15,the notion
of a skeleton.
Fig. 15. Set A and its skeleton denoted by dashed lines
The skeleton of set A can be expressed in terms of erosions and openings as foll
owing:
S(A) = _S
k
(A)
k
k=0
(12)
With
S
k
(A) = (A kB) (A kB) B (13)
Where B is a structuring element, and (A kB) indicates k successive erosions of A
K times,
(A kB) = ( (A B) B) ) B (14)
and K is the last iterative step before A erodes to an empty set. In other words
,
K = mox {k|(A kB) ] (15)
The formulation states that S(A) can be obtained as the union of the skeleton su
bsets S
k
(A).
Also, it can be shown that A can be reconstructed from these subsets by using th
e equation
A = (S
k
(A) kB)
K
k=0
(16)
Where ( ( ) )
k
S A kB denotes k successive dilations of S
k
(A);that is
(S
k
(A) kB) = (( (S
k
(A) kB) B) ) B (17)
Another term medial axis transform (MAT) is often used interchangeably
with
skeletonization but they are not quite the same. The skeleton is simp
ly a binary image
showing the simple skeleton, while MAT is a graylevel image where e
ach point on the
skeleton has an intensity which represents its distance to a boundary
in the original object,
shown in the Fig. 16.
Fig. 16. Skeletonization of a binary image, from left to right: the original bin
ary image,
normal skeletons, the skeleton image formed by the medial axis transformation
The medial axis of an object is the set of all points having more than one close
st point on the
object s boundary. In mathematics the closure of the medial axis is k
nown as the cut locus.
In 2D, the medial axis of a plane curve S is the locus of the centres of circles
that are tangent
to curve S in two or more points, where all such circles are contained in S. (It
follows that the
medial axis itself is contained in S.) The medial axis of a simple
polygon is a tree whose
leaves are the vertices of the polygon, and whose edges are either straight segm
ents or arcs
of parabolas. The medial axis together with the associated radius function of th
e maximally
inscribed discs is called the medial axis transform , which is a com
plete shape descriptor,
meaning that it can be used to reconstruct the shape of the original domain.
According to American Society for Testing and Materials (ASTM), Filamen
tary shrinkage
usually occurs as a continuous structure of connected lines or branche
s of variable length,
width and density, or occasionally as a network. From Fig.17 we can
see that, the most
obvious feature is the continuous tree structure, wider in the trunk
and thinner in the
branches. In grayscale, the parts closer to the center of the structure are dark
er. The outside
part is lighter and the branches are between the centre and the outs
ide part. Because of its
structure and features, the skeleton description is one of the best. In the appl
ication, several
layers of skeletons are needed, as showing in Fig.17, the simple and the complic
ate skeletons
for the left dendritic shrinkage are as Fig. 17(c) and (d), which wi
ll be used to represent
different parts of the defects, some like trunks and some branches.
(a) (b) (c) (d)
Fig. 17. Skeletonization of a real dendritic shrinking for flaw simulation: (a)
a radiograph of
a workpiece, (b) zoomed dendritic shrinking, (c), (d) the skeleton images based
on binarized
images of different thresholds
Fig. 18. Randomly created skeleton for crack generation and simulation
Apart from sample defects, skeletons can also be created randomly such
as for cracks or
filamentary shrinkages, and so on. Please refer Fig. 18.
4.4 Visual feature analysis and fuzzy representation
Since the qualitative visual features are fuzzy, so a fuzzy representa
tion of them is
appropriate. Set the image grayscale range [1, Ng], is defined as the membership
function
of the gray value in a specified image area, it makes the gray valu
e mapped to [0, 1]. On a
view, like a factor image with piexls point to a single point of am
biguity in the original
image, its corresponding values indicate the membership degree that the
image point
belongs to the object region. All the pixels in the image form a si
ngle dot fuzzy matrix
expressed with (I).
The following eight fuzzy geometric features impact on the accuracy and speed of
operation
with more good compromise effects. The image size is M N, the (I) is denoted by ,
the
membership definitions of these fuzzy geometric features and the classification
functions for
the object structures are as follow:
1. Area:
o(p) = p (18)
Area is the weighted sum of the pixels in an object image range.
2. Circumference:
p(p) = |p(i, ]) (i, ] +1)| + |p(i, ]) p(i +1, ])|
N-1
]=1
M
=1
N-1
]=1
M
=1
(19)
Circumference pixel horizontal distance and vertical distance weighted a
nd is used to
measure the undulations of the image area.
3. Firmness
comp(p) =
u()
p
2
()
(20)
Firmness reflects the degree of fuzzy compact region
4. Height and width
Height
(p) = mox
]
{p(i, ])]
M
=1
(21)
Width
w(p) = mox
]
{p(i, ])]
N
]=1
(22)
Height is the total number of the largest membership of the rows of
the image, and width
isthe total number of the largest membership of the columns of the i
mage. The height and
width can be used, respectively, to reflect the image of certain area
s in the overall vertical
and horizontal directionfeatures.
5. Length and width
Length
l(p) = mox
]
| p(i, ])
M
| (23)
Width
b(p) = mox

| p(i, ])
N
]
| (24)
Length is the total number of the largest membership of the rows of the image, w
idth is the
total number of the largest membership of the columns of the image.
The use of the X-ray
image can be described in certain regions in the vertical and horizon
tal distribution of the
extreme characteristics of the membership.
6. Area coverage
I0AC(p) =
u()
I()b()
(25)
The characteristic quantity is the fuzzy area defined by the length and width of
the moment
shaped area ratio, length and width defined by the area of image certain areas o
f the largest
possible area of fuzzy.
4.5 Generation of the low frequency information
No matter how the generative flaw image is rendered from a creative
counter or from a
simple skeleton, the counters or skeletons normally are the crucial ch
aracteristics of most
inner defects of the materials and are used mainly to produce the si
milar perceivable
structures with the original flaw. While some types of flaws are not
obvious in their high
frequency information, but present as apparent textural features such a
s shrinkages of
needle cavity and small round cavity as well as sponge shrinkage show
ing in Fig. 19
(Hoff, 1999).
Fig. 19. Defect texture, from left to right: shrinkages of needle cavity and sma
ll round cavity
as well as sponge shrinkage
4.5.1 Texture generation
Efros and Leung (1999) have proposed a nonparametric model to synthesi
ze an output
image from a given metric sample. Their approach works surprisingly well for a w
ide range
of textures. Their algorithm consists of the following steps:
1. The output image is initialized with a random noise whose histogra
m is equalized with
respect to the histogram of the input sample.
2. For each pixel in the output image, in scan-line order, the alrea
dy-synthesized values in
neighbourhood of current pixel z of specific size N(z) is considered and is comp
ared with all
possible neighbourhoods from the input sample. The value of the input
pixel x with the
most similar N(x) is then assigned to z.
Efros & Leung s algorithm is significantly useful for some kinds of defect simul
ation. After,
initialize a synthesized texture with a 3x3 pixel "seed" from the sou
rce defect texture, for
every unfilled pixel which borders some filled pixels, find a set of
patches in the source
image that most resemble the unfilled pixel s filled neighbors. Choose
one of those patches
at random and assign to the unfilled pixel a grey value from the center of the c
hosen patch,
and do it repeatedly. The only crucial parameter is how large a neighbourhood yo
u consider
when searching for similar patches. As the "window size" increases, th
e resulting textures
capture larger structures of the source image. The finer details of this approac
h can be found
in the Efros paper. The other problem which arises from radiograph sim
ulation is to
merge together the generated flaws with the workpiece or other parts o
f the background
image. In a similar manner to the heat diffusion equation, the image
diffusion technique is
adopted, which is discussed below.
4.5.2 Imagediffusion technique
Embedding simulated flaws into the objects radiographs, which is called augmented
reality
in the simulation research field, is not an avoidable step and is important both
in simulation
application and the technology of image superimposing. Take the X-ray
penumbra into
account, we know that the edges of the defects are blurred and sometimes fused w
ith their
background, furthermore, the grey values of the defect vary when the background
changes,
constrained by the law of X-ray attenuation and simultaneity, the grey
values of the work-
piece and its defect vary enormously in the large range of tube potential.
Similar to the effects of the heat diffusion equation, image diffusion
is suitable for edge
preserving blur and can be used widely in a flaw superimposing into a real radio
graph, the
apparent visual characteristics of blurred edges by diffusion can be seen from F
ig. 20.
Fig. 20. The original panda image with the diffused ones on their right
To an initial image u(x,y,0) change to be image u(x,y,t) after a tim
e t, the anisotropic
diffusion first proposed by Perona and Malik (1990) is denoted as:
(x,,t)
t
= Ji:|g(x, y, t)7u(x, y, t)] (26)
where u(x, y, t) is the gradient of the image and g(x, y, t) is diffusion coeffic
ient, which can
be isotropic or anisotropic and depends on different applications. Teboul et al.
(1998) further
discussed the conditions of the coefficient g(x, y, t) and three con
straints are given as
follows:
1. lim
|vI|0
g(x) = Zu < Z <
2. lim
|vI|
g(x) = u
3. g(x) is a strictly decreasing function of |7I(x)|.
There are many research contributions to the diffusion coefficient func
tion, such as Gauss
Smoothing Diffusion(Catte, et al. 1992), Morphological Diffusion(Segall&
Acton, 1997),
Degeneration Diffusion Mode(You, et al. 1995)Statistical Diffusion Mode(Black, e
t al. 1998),
Curvature Diffusion Mode(Alvarez, et al. 1992), but the anisotropic diffusion mo
dels are ill-
posed, it is difficult to determine the convergence condition. In the flaw simul
ation, we still
rely on the qualitative visual feature analysis of real radiographs.
4.6 Synthesizing and generating radiographs
The flaws generated in Fig. 21, follow a similar procedure though their seed feat
ures differ
from greatly from each other. After analysis of sample images and gro
uping them into
different defect categories, extract their qualitative visual features;
choosing the seed
feature and decide the flaw rendering approach such as the template me
thod or the
skeletonization method or texture synthesizing. Based on statistical ana
lysis, calculate
thefuzzy image representations for each type of feature, which are exp
ected to be relevant
for the image generation, fuzzy membership as needed is applied to th
ese image features
during the image generation process. As the seed feature grows, simult
aneously, the low
frequency features in the regions of the creative defect should be taken out to
ensure that the
defect appearances reflect their semantic definitions. Finally, image superimpos
ing onto real
product radiographs is also an important step that contributes to the
success of the
simulation application. The super impositions rely on the location as well as th
e associated
gray values, which generally are implanted in the simulated flaws according to t
he contrast
of the defect against its background.
Fig. 21. Comparison simulated flaws with real defects on different products, at
the third row
the original is on the left and the simulated is on the right
5. Conclusion
X-ray defect simulation is associated with many uncertainties in terms
of shape,
inconsistancy of grey values and the blurring of simulated defect edge
s against the
background. The approaches discussed in this section have shown how to
simulate and
create various kinds of defects including stochastic large discontinuity
defects and regular
shape defects such as circular and elliptical spots, which can be use
d to build a sample
library for testing and tuning automatic inspection programs. The new defect sup
erposition
methods using nesting stencil-plates, texture generation as well as skeletonizat
ion have been
shown to be able to simulate the normal range of casting defects. Th
ese new methods are
not only capable of simulating the casting defects, but also able to
be used to simulate
irregular objects in different kinds of X-ray images. Apart from the
crucial high frequency
image features that were adopted to render the structures of the flaw
s, the anisotropic
diffusion method has received more attention as nearly all the objects
appearing in
radiographs have shading effects, which could be visually represented b
y diffusion
technology.
Compared with the CAD model and the Monte Carlo method, the advantage
s and
characteristics of image generation method, that is the augmented reali
ty simulation
method, can be summarized as: (1) It has more randomness and flexibil
ity and provides a
better match for the shape variability of the defects: (2) The seed features are n
ot limited in
shape or size, and changes in their form will not significantly influ
ence the simulated
defects. (3) Many and varied defects can be created by the user and added to a r
eal casting
piece image. The program is easy to use and, by choosing and tuning the paramete
rs, it can
be used for the majority of casting pieces without any advanced measurements on
the work
piece. (4) The program works directly on the radiographs, and the location for t
he simulated
defect can be chosen randomly by the user in any X-ray image. (5) S
ince no part of the
programme is based on a CAD platform, there is no need of a CAD software package
or any
other database, the defect simulation function can be organized as a
program module and
integrated into the casting automatic inspection system to share the stored data
.
6. References
Aloimonos, J. (1990). Purposive and Qualitative Active Vision, Proceedin
gs of 10th
International Conference on Pattern Recognition (ICPR), pp. 346-361
Alvarez, L. et al. (1992). Image selective smoothing and edge detectio
n by nonlinear
diffusion II. SIAM Journal of Numerical Analysis, (1992), Vol.29, No.3, pp. 845
-866
Black, M. et al. (1998). Robust anisotropic diffusion, IEEE Transaction
s on Image Processing,
(1998), Vol. 7, No.3, pp. 421-432, ISSN: 1057-7149
Bottigli et al. (2004). Voxel-based Monte Carlo Simulation of X-ray Im
aging and
Spectroscopy Experiments, Spectrochimica Acta Part B: Atomic Spectroscopy
, (2004),
Vol.59, No.10-11, pp. 1747-1754
Catte, F. et al.(1992). Image selective smoothing and edge detection by nonlinea
r Diffusion,
SIAM Journal of Numerical Analysis, et al. (1992), Vol.29, pp. 182-193
Efros, A. & Leung, K. (1999). Texture Synthesis by Non-parametric Sampling, Proc
eedings of
IEEE International Conference on Computer Vision, pp. 1033-1038, Corfu,
Greece,
September 1999
EGSnrc (URL). Available from http://irs.inms.nrc.ca/software/egsnrc/
Filbert, D.; Klatte, R.; Heinrich, W. & Purschke,, M. (1987). Computer
Aided Inspection of
Castings. Proceedings of the IEEE-IAS Annual Meeting, pp. 1087-1095, Atlanta, US
A
GoldSim(URL). Available from http://www.goldsim.com/Home/
Gonzalez, C. & Woods, E. (2008). Digital Image Processing, Prentice Ha
ll, ISBN: 0201180758,
0130946508
Haken, H. &Wolf, H. (1996). The Physics of Atoms and Quanta: Introduction to Exp
eriments and
Theory. Berlin, Heidelberg: Springer, 5th edition, 1996
Hoff, A. (1999). Reference Radiographs, In: The American Society for N
ondestructive testing,
03.08.2011, Available from
http://www.asnt.org/publications/materialseval/basics/oct99basics/oct99basics.
htm
Iowa State University. (1996), Radiograph Interpretation Castings, In : NDT reso
urce center,
04.08.2011, Available from
http://www.ndt-ed.org/EducationResources/CommunityCollege/Radiography/
TechCalibrations/RadiographInterp_Castings.htm
Kalos, H. & Whitlock, A. (2008). Monte Carlo Methods, Wiley-VCH, ISBN: 978-3527
407606
Mackenzie, D.& Totten, G. (2006). Anyalytical Characterization of Alumin
um, Steel, And
Superalloys, CRC Press, Taylor & Francis Group, (2006). pp. 718-721, I
SBN:
9780824758431
Mery, D. (2001). Flaw Simulation in Castings inspection by radioscopy, INSIGHT,
Journal of
the British Institute of Non-Destructive Testing, (2001), Vol.43, No.10, pp.664-
668
Metropolis, N. (1987). The Beginning of the Monte Carlo Method, Los A
lamos Science (1987
Special Issue dedicated to Stanislaw Ulam), pp. 125130
Perona, P. & Malik, J. (1990). Scale-space and edge detection using a
nisotropic diffusion.
IEEE Transactions Pattern Anal. Mach. Intel., (1990), Vol.12, No.7, pp. 629639, I
SSN:
0162-8828
Segall, A. & Acton, T. (1997). Morphological anisotropic diffusion, Pro
ceedings of IEEE
international conference on image processing, (1997), No.1, pp. 356-361
Sukowski, F. & Uhlmann, N. (2011). Applications of Monte Carlo Method
in Science and
Engineering, Shaul Mordechai (Ed.), InTech, ISBN: 978-953-307-691-1
Teboul, S. et al. (1998). M. Barlaud. Variational approach for edge-preserving r
egularization
using coupled PDEs. IEEE transactions on Image Processing, (1998), Vol
7, No.3, pp.
387-397, ISSN: 1057-7149
Tillack, G. et al. (2000). X-ray modeling for industrial applications,
NDT & E International,
Vol.33, No.7, (2000), pp. 481-488
You,Y. et al. (1995). On ill-posed anisotropic diffusion models. Procee
dings of International
Conference on Image Processing, 1995, No.2, pp. 268-271
You,Y. et al.(1996). Behavioral analysis of anisotropic diffusion in im
age processing. IEEE
Transactions on Image Processing. (1996), Vol.5, No.11, pp. 1539-1553, ISSN: 105
7-7149
6
Study of Metallic Dislocations by
Methods of Nondestructive Evaluation
Using Eddy Currents
Bettaieb Laroussi, Kokabi Hamid and Poloujadoff Michel
Universit Pierre et Marie Curie (UPMC), Laboratoire dElectronique et
Electromagntisme (L2E), Paris
France
1. Introduction
This chapter summarizes a work which we have conducted over the perio
d 2005-2010
[Bettaieb et al., 2008; Bettaieb, 2009]. The aim was to better understand the No
n Destructive
Evaluation (NDE) method based on the use of eddy currents. A very simp
le example, well
adapted for beginners, was the evaluation of a homogenous rectangular aluminum p
late of
constant thickness, in case of a calibrated crack. Such a choice had
a great advantage: it
allowed the development of simple theoretical analysis, and therefore a
very good
understanding of the physics of the problem.
Naturally, theoretical results must be checked experimentally. To this
end, we built a
Helmholtz arrangement of two circular coils, to create a convenient uniform alte
rnating field
within the plate. Then, to measure the reaction field of the plate, we had to us
e a sensor. We
used alternately a SQUID and a Hall sensor. The choice between those is not obvi
ous, as will
be shown later. Indeed, the SQUID is about 1uuu times more sensitive than the Ha
ll sensor,
but its sensitive part is more remote than the Hall sensor, and this may reduce
significantly
its advantage. We found this comparison extremely interesting [Bettaieb et al.,
2010].
Indeed, if we cut a rectangular plate, the cut being parallel to one
side and normal to the
excitation field, it seems natural that the plate is equivalent to a
homogeneous one.
Therefore, the cut cannot be detected. Just in case, we made the exp
eriment and there was
an unexpected signal!! We rapidly realized the reason of this result: the cut th
rough the plate
did not interrupt any current, but the cutting tool had damaged the
microstructure of the
aluminum over a small part of each half plate. Therefore, none of th
e two half parts had
remained homogeneous. We realized that the microstructure of a small p
art of the half
plates had been modified. In fact, the two halves of the original pl
ate were equivalent to a
homogeneous plate, except along the cut. Therefore, we did not detect the cut,
but the metallic
dislocations around it. Some more theoretical investigations allowed to
evaluate the
importance of the change in the electrical resistivity associated with
the dislocations, and
their extension. Then, we started to investigate the dislocations due t
o hammer shocks and
to mechanical flexions.
This chapter is a report of all our efforts. And we hope that other
studies, by other
investigators, will extend these results.
2. Theoretical study of a defect free rectangular aluminium plate
We consider the case of a homogeneous rectangular aluminum plate (resi
stivity
p = S.82 1u
-8
. m). Its dimensions are o = 11umm, b = 1uumm, c = Smm in the x, y, z
directions respectively (figure 1). This plate is immersed in an alternative uni
form magnetic
induction field B
parallel to the 0y direction, varying at the 18uEz frequency, with

an
amplitude of u.1S 1u
-4
I if we use a SQUID and 2.4S 1u
-4
I if we use a Hall probe
sensor.
Fig. 1. Defect free rectangular aluminium plate under consideration
This excitation frequency has been chosen because it is one of those
which are the less
present in the general environment in and around our laboratory.
We shall study the induced current density using the more appropriate
simplifying
assumptions, in order to develop a good physical understanding of the phenomena.
In a first step, we assume that all the induced currents lines are straight line
s parallel to the
0x direction from one side to the other (figure 1). If this is so,
the external field is easy to
evaluate by Biot and Savart law. An interesting remark is that the f
ield of the induced
currents, inside the plate, is quite much smaller than the excitation field.
In a second step, we improve the evaluation of the induced current l
ines in a plane obcJ
parallel to the plane x0z (figure 2).
In such planes, the value of B
is given under the form of a double Fourier expansion

B(x, z) = o
mn
cos [
mnx
u
cos (
nnz
c
)
m,n
(1)
and the values of the amplitude of the components ]
x
and ]
z
of the current density are
Study of Metallic Dislocations by Methods of Nondestructive Evaluation Using Edd
y Currents 129
]
x
= -
o
p
16uc
2
n
B
1
n
sIn[
mn
2
sIn (
nn
2
)
(mnu)
2
+(nnc)
2
sin [
mnx
u
cos (
nnz
c
)
m,n
(2)
]
z
=
o
p
16cu
2
n
B
1
m
sIn[
mn
2
sIn (
nn
2
)
(mnu)
2
+(nnc)
2
cos [
mnx
u
sin (
nnz
c
)
m,n
(3)
respectively. It may be noted that ]
x
and ]
z
are the derivative of a current function:
u(x, z) =
1
p
(])(
4uc
n
)
2
B
mn
sIn[
mn
2
sIn (
nn
2
)
(mnu)
2
+(nnc)
2
cos [
mnx
u
cos (
nnz
c
)
m,n
(4)
so that ]
x
= -
0(x,z)
x
and ]
z
=
0(x,z)
z
.
Fig. 2. Rectangular aluminium plate under consideration showing the cross sectio
n plane
obcJ for the evaluation of the induced currents lines
The important point is to understand that, in the case of figure 2 (o = 11umm, c
= Smm), the
current density is essentially parallel to the 0x axis, and that this
justifies the above first
approximation: the current lines are parallel to 0x (figure 3).
Fig. 3. Lines of induced currents for a defect free rectangular aluminium plate
(o = 11umm,
c = Smm), the current density is essentially parallel to the 0x axis
3. Theoretical study of a rectangular aluminum plate with a calibrated crack
Consider now the above plate, where we have made a calibrated crack, as shown in
figure 4.
Fig. 4. Rectangular aluminium plate with a calibrated crack having a width (w =

b
2
) and a
depth (Jcptb)
We analyze the induced currents produced by the alternating excitation field B
, considering
that the plate is equivalent to 3 sane plates of widths b
1
, b
2
= w and b
3
(b
1
+b
2
+b
3
= 1uumm) and of thickness c, c - Jcptb and c respectively. The above analysis
stands for each of those three plates; the numerical analysis is straight forwar
d, and is given
below (figures 8 and 9).
4. Experimental set-up
In our laboratory, we have established an experimental apparatus based on Helmho
ltz coils
arrangement for excitation, and fixed magnetic sensors (SQUID and Hall
probe) (figure 5).
The structure has been made mainly of Plexiglas and other non magneti
c materials (brass,
copper, Al, wood) in order to avoid any magnetic noise. The samples a
re moved
underneath the sensor by means of an x-y scanning stage. The stage and the data
acquisition
are controlled with a LABVIEW program which is optimized to automate
the entire
measuring process. The sensor (SQUID or Hall probe) measures the magne
tic field
perpendicular to the excitation field and to the scanning direction. A
lock-in amplifier has
been used to achieve a synchronous detection.
The two versions of the equipment are shown in figure 6. The sensor
is located in a fixed
position. In the case of the SQUID, mechanical distance between upper side of th
e plate and
the Dewar is u.Smm, and the sensor itself is 12.Smm higher. In the
case of the Hall probe
(right part of the figure), the total distance between the plate and
the active part of the
sensor is 1mm.
The difference between the two sensors appears immediately. Indeed, the
excitation field,
when the SQUID is in use, is limited to u.1S 1u
-4
I to avoid the saturation. When the Hall
y Currents 131
probe is in use, this limit reaches 2.4S 1u
-4
I with our generator; but it might be higher
with better equipment. In both cases, the frequency is 18uEz. The sens
itivity of the SQUID
measuring sensor is 4.2SII, and the sensitivity of the Hall sensor is SmII.
Fig. 5. High T
c
SQUID NDE instrumentation set-up
Fig. 6. The two versions of the experiment set-up: on the left is a SQUID sensor
and on the
right a Hall probe
5. Experimental verification
Figure 7 shows the plate with a calibrated crack immersed in the uni
form excitation field
described above (18uEz, u.1S 1u
-4
I if we use a SQUID and 2.4S 1u
-4
I if we use a Hall
probe sensor).
Fig. 7. Rectangular aluminium plate with a calibrated crack immersed in a unifor
m
excitation field created by Helmholtz coils
The vertical component of the resulting induction field, measured between A and
A
is given
at a vertical distance J of 1mm (measured with the Hall probe), or at a vertical
distance J of
12.8mm (measured with the SQUID). It is clear that the computed values and the m
easured
values tally in two very different cases (figures 8).
(A)
Fig. 8.A. Computed (a) and measured (b) values of B
z
between A and A
in the case of a
rectangular aluminium plate with a calibrated crack (width w = 1mm, Jcptb = 1mm)
:
measurement with a Hall probe at J = 1mm above the plate
-20 -10 0 10 20
-6x10
-7
-3x10
-7
0
3x10
-7
6x10
-7
B
z
(
T
e
s
l
a
)
y(mm)
(a)
(b)
y Currents 133
(B)
Fig. 8.B. Computed (a) and measured (b) values of B
z
between A and A
in the case of a
rectangular aluminium plate with a calibrated crack (width w = 1mm, Jcptb = 1mm)
:
measurement with a SQUID at J = 12.8mm above the plate
Figure 9 represents the case of a defect free rectangular aluminum pl
ate. The vertical
component B
z
of the resulting induction field, measured between A and A
is given at a
vertical distance J of 1mm (measured with the Hall probe), and at a
vertical distance J of
12.8mm (measured with the SQUID). Note that the difference between the
oretical values
and experimental values is so small that they cannot be distinguished.
(A)
-20 -10 0 10 20
-1,5x10
-8
-1,0x10
-8
-5,0x10
-9
0,0
5,0x10
-9
1,0x10
-8
1,5x10
-8
B
z
(
T
e
s
l
a
)
y(mm)
(a)
(b)
-20 -10 0 10 20
-4,0x10
-5
-2,0x10
-5
0,0
2,0x10
-5
4,0x10
-5
B
z
(
T
e
s
l
a
)
y(mm)
(a)
(b)
-2, 1
-7
-4, 1
-7
-,
-2, 1
-7
4, 1
-7

(B)
Fig. 9. Computed (a) and measured (b) values of B
z
between A and A
in the case of a defect

free rectangular aluminium plate: (A) measurement with a Hall probe at J = 1mm
above
the plate, (B) measurement with a SQUID at J = 12.8mm above the plate
6. Case of a crack with zero width showing metallic dislocations
We endeavored to repeat the same kind of experiment as above, but wi
th a slot of zero
width (b
2
= w = umm, Jcptb = c) (figure 10).
Fig. 10. Rectangular aluminium plate with a slot of zero width under considerati
on
We expected to find, along the line AA
, a variation of the vertical component of the induced

field depicted by curve (a) of figure 11, because all current lines
are parallel to the 0x
direction. In other words, we did not expect to detect the cut. We were extremel
y surprised
to find the variation depicted by curve (b) of the same figure. In fact, we had
considered that
the two halves of the plate were equivalent to a homogeneous plate. What we negl
ected was
that, when you cut a metallic plate, the cutting tool may damage the metallic mi
crostructure
-20 -10 0 10 20
-1x10
-8
0
1x10
-8
2x10
-8
B
z
(
T
e
s
l
a
)
y(mm)
(a)
(b)
y Currents 135
of the plate, at least over small depth. Therefore, the intuitive representation
of the situation
is shown in figure 12. In this figure, the aluminum parts are themse
lves divided into two
parts: one where the resistivity is normal (p = S.82 1u
-8
. m), and one where the
resistivity is higher: p
> p.
Fig. 11. Measured values of B
z
, with a Hall probe at J = 1mm above the plate, between A
and A
: (a) in the case of a defect free rectangular aluminium plate, (b) in the case
of a
rectangular plate with a slot of zero width
In figure 12, the thickness of the higher resistivity zone is considered
w
2
at the edge of each
half plate, so the total damaged edge area width is w
i
.
Fig. 12. Representation of the two half plates laid near each other, and of two
assumed
fatigued zones of resistivity p
i
Therefore, the next problem is to determine the values of w
and p
which account
conveniently for the experimental results. This is a simple classical optimizati
on problem. In
our case, we pick at random one arbitrary value of w
and p
. This allows to evaluate the z

component of the induced field. It is then necessary to define a glo
bal error e(w
, p
)
between the observed field, and the field corresponding to w
and p
. Then we modify either

w
or p
to decrease e and repeat this process until e is as small as possible. This is

developed
in [Poloujadoff et al., 1994] and in the reference [Bettaieb et al., 2010].
This led us to determine a best value of p
p = S.82 1u
-8
. m = 1.S p) and of w
(w
=
Sum) which provided a theoretical curve of B
z
shown in figure 13 (a).
Fig. 13. (a) Theoretical curve with optimized values of w
= Sum and p
= 1.S p, (b)
experimental curve established with a Hall sensor at d=1mm above the aluminium p
late in
the case of a zero width crack
Since the publication of this reference, Dr. Denis GRATIAS suggested a further v
erification
of this approach. This consisted in annealing the two half plates, th
en reputing the
measurements. The annealing cycle is shown in figure 14.
Fig. 14. Annealing cycle of the two half plates after cutting a plate
Then, we place again the two half plates in the same uniform magnetic excitation
field. The
curve of the vertical component of the induction field still shows a
variation of w
and p
,
but much smaller than previously; this show that annealing has been very effecti
ve, but has not
been long enough (figure 15).
-30 -20 -10 0 10 20 30
-4x10
-7
-2x10
-7
0
2x10
-7
4x10
-7
B
z
(
T
e
s
l
a
)
y(mm)
(a)
(b)
0 100 200 300 400 500 600 700
0
100
200
300
400
T
e
m
p
e
r
a
t
u
r
e
(
C
)
Temps (min)
Sb
19umin
S8umin
y Currents 137
Fig. 15. Experimental curve established with a Hall sensor at d=1mm above the al
uminium
plate in the case of a zero width crack before and after annealing
7. Metallic dislocations created by a shock
The above study of dislocations can be carried on by creating small
damaged zones by a
direct or indirect shock with a hammer (figure 16). We have already reported a p
reliminary
study of this phenomenon [Bettaieb et al., 2010], and we report some
more recent progress
below.
Fig. 16. (a) Creation of an impact zone by stroking a steel ball, (b) impact upo
n the square
aluminium plate, (c) zoom of the shock print zone with modelling dimensions (D,
d)
-30 -20 -10 0 10 20 30
-8,0x10
-7
-4,0x10
-7
0,0
4,0x10
-7
8,0x10
-7
1,2x10
-6
B
z
(
T
e
s
l
a
)
y(mm)
before annealing
after annealing

In our latest experiments, we stroke a steel ball in the middle of
a square aluminum plate
with a hammer. The dimensions of the damaged zones depend naturally o
n the radius of
the steel ball and on the violence of the hammer shocks. We have co
nsidered only three
cases; the depth J and the diameter of some shock prints are given in the table
I.
Dimeter depth J
Impact 1 4mm 0.11mm
Impact 2 3mm 0.09mm
Impact 3 2mm 0.07mm
Table 1. Dimensions of the damaged zones
Consider the first impact, produced near the middle of the square pla
te. The Aluminum
plate is laid upon the stage which has been already described in the figure 6. T
he excitation
field being again at 18uEz with an amplitude of u.1S 1u
-4
I if we use a SQUID and
2.4S 1u
-4
I if we use a Hall probe sensor. The stage is moved as described i
n 4, the
variation of the vertical induction field component, measured by the Hall probe
is shown in
figure 17 (A). This proves that the impact may be easily detected, s
ince the perturbation
signal varies between _1u
-5
I. It is also possible to scan a square domain
(-1umm x, y +1umm) around the center of the impact and to represent it
as in
figure 17 (B).
(A)
(B)
Fig. 17. (A) Measured values of B
z
along the 0y axis for x = umm, (B) matrix representation
of the values of B
z
across a square domain (-1umm x, y +1umm). Measurement with a
Hall probe at d=1mm above the square aluminium plate ( = 4mm, J = u.11mm)
-10 -5 0 5 10
-1x10
-5
-1x10
-5
-5x10
-6
0
5x10
-6
1x10
-5
B
z
(
T
e
s
l
a
)
y(mm)
y Currents 139
Considering the second impact, we found similar results, but the signa
l varies within a
smaller interval (_u.6 1u
-5
I) (see figures 18 (A) and 18 (B)). As for the third impact, it is
not really detectable.
(A)
(B)
Fig. 18. (A) Measured values of B
z
along the 0y axis for x = umm, (B) matrix representation
of the values of B
z
across a square domain (-1umm x, y +1umm). Measurement with a
Hall probe at d=1mm above the square aluminium plate ( = Smm, J = u.u9mm)
How may we interpret these results?
It is extremely important to consider that these results may be assig
ned to two different
causes: the change of the shape of the aluminum square plate and/or a modificati
on of the
metallic microstructure inducing a change of the electric resistivity. To clarif
y this question,
we have assumed that the effects of these two causes can be separate
d. Unfortunately, a
rigorous mathematical analysis of such a structure would involve a cyl
indrical geometry
and a uniform field, resulting in a very complicated 3D analysis. For
this reason, we made
the following approximations which happen to give a sufficient feeling of what h
appens.
This means that we first replaced the portion of a sphere, in figure
16 (c) by an empty
parallelepiped with a square basis of side w = and height b = J chosen to have t
he same
volume than the portion of the half sphere (figure 19 (A)). This mod
el accounts for the
observed results (figure 19 (B)).
-10 -5 0 5 10
-8x10
-6
-4x10
-6
0
4x10
-6
8x10
-6
B
z
(
T
e
s
l
a
)
y(mm)
(A)
Fig. 19.A. Square aluminium plate with an empty space of same volume that the on
e in
figure 16 (c)
(B)
Fig. 19.B. Theoretical values of B
z
corresponding to the figure 19 (A)
(A)
(B)
Fig. 20. (A) Square aluminium plate with an inner portion of resistivity p
> p, (B) theoretical

values of B
z
corresponding to the figure 20 (A).
-20 -10 0 10 20
-1,5x10
-5
-1,0x10
-5
-5,0x10
-6
0,0
5,0x10
-6
1,0x10
-5
1,5x10
-5
B
z
(
T
e
s
l
a
)
y(mm)
-20 -10 0 10 20
-3x10
-5
-2x10
-5
-1x10
-5
0
1x10
-5
2x10
-5
3x10
-5
B
z
(
T
e
s
l
a
)
y(mm)
y Currents 141
In a second step, we assumed that the deformation of the plate was
negligible, but the
metallic microstructure has been changed over a depth as large as 1mm
. We have
considered that in this region the resistivity has been changed by 10
% (figure 20 (A)). This
model also accounts for the experimental results (figure 20 (B)).
Therefore, this experiment and the corresponding models show that the effect of
the shock is
partially explained by a deformation of the square aluminum plate and
partially by
dislocations created in the metal over same depth. We have proved tha
t the effects of the
dislocations are much more important [Bettaieb, 2009].
8. Metallic dislocations created by flexions
8.1 Experimental study
Consider a long aluminum strip (150mm) shown in figure 21 which has been bent by
hand,
so as to create a fatigue zone in the middle. In the already quoted previous pap
er [Bettaieb
et al., 2010], we had shown that the fatigue of the metallic bent p
art could be detected. To
this end, we had placed the strip in a uniform alternating horizontal
field B
, and we have
scanned its upper surface with a Hall probe. However, this first publ
ication was limited to
the feasibility of the detection.
Fig. 21. Fatigued strip, the length of the damaged (bent) zone is approximately
equal to 3mm
Since that time, we have greatly improved our experimental process, and have dev
eloped an
inverse problem method for a better mathematical analysis of the results.
Our first effort has been to flatten the long strip carefully before measurement
, as shown by
figures 22 (figure 22 (A) is a side photo, and figure 22 (B) an upside down view
of the strip
during flattening). As a result, the signal obtained along the sensor
track is symmetrical
about its center (figure 23 (A)).
(A)
(B)
Fig. 22. Bent long strip being flattened in a shaper to eliminate the geometrica
l deformation
before eddy current measurement (A: side view; B: upper view)
y Currents 143
(A)
(B)
Fig. 23. (A) Values of B
z
induced by the eddy currents within a flattened fatigued long strip
(measurement with a Hall sensor at J = 1mm above the strip), (B) Corresponding v
alues of
the estimated local resistivity along the damaged zone using the inverse solutio
n
8.2 Theoretical evaluation of resistivity along the strip
The principle of the inverse problem is well known. The relationship between the
induction
field and the induced currents in the long strip is linear, the coef
ficients being the
conductivity. Simultaneously, the relationship between the strip induced
current and the
outside induced induction field is also linear. Therefore, if the exci
ting induction field and
the induced induction field are known, the resistivity can be determin
ed [Bettaieb et al.,
2010]. It is what has been done to estimate the resistivity (figure 23 (B)) alon
g the strip. This
latter result clearly shows the effect of the fatigue where the strip has been b
ent.
9. Conclusion
In this very original chapter, we have shown that a methodology gener
ally used to detect
cracks can be also used to measure the effects of a mechanical fatig
ue created by cutting
tools, impact shocks or flexions. The natural continuation should be to study qu
antitatively
-20 -10 0 10 20
-6x10
-5
-3x10
-5
0
3x10
-5
6x10
-5
9x10
-5
B
z
(
T
e
s
l
a
)
y(mm)
-20 -10 0 10 20
6,0x10
-8
6,5x10
-8
7,0x10
-8
7,5x10
-8
8,0x10
-8
R
s
i
s
t
i
v
i
t
(
O
h
m
.
m
)
y(mm)
most of the microstructure properties which influence the resistivity o
f the alloy. These
include irradiations defects and phase transitions. We would certainly
welcome
collaboration offers from colleagues.
10. Acknowledgment
We would like to thank warmly Dr. Denis GRATIAS from ONERA/CNRS for several usef
ul
advices and discussions notably related to the nature of dislocations. He also s
uggested the
check of our ideas by the annealing procedure.
11. References
Bettaieb, L. (2009). Contributions to Non Destructive Evaluation of ama
gnetic metlas using
eddy currents. Phd Thesis at University of Pierre and Marie Curie (UPMC), in Fre
nch,
(September 2009).
Bettaieb, L.; Kokabi, H.; Poloujadoff, M.; Sentz, A. & Krause, H. J.
(2008). Non Destructive
Testing (NDT) With High Tc RF SQUID. Journal of Physics: Conference S
erie 97,
(March 2008), 6pp.
Bettaieb, L.; Kokabi, H.; Poloujadoff, M.; Sentz, A.; Krause, H. J. &
Coillot, C. (2010).
Comparison of the use of SQUID an Hall effect sensors in NDE. Materi
als
Evaluation, Vol.68, No.5, (2010), pp. 535-54.
Bettaieb, L.; Kokabi, H.; Poloujadoff, M.; Sentz, A. & Tcharkhtchi, C. (2010).
Fatigue and/or
crack detection in NDE. Nondestructive Testing and Evaluation, Vol.25,
No.1,
( March 2010), pp. 13-24.
Bettaieb, L.; Kokabi, H.; Poloujadoff, M.; Sentz, A. & Krause, H. J.
(2009). Analysis of some
Nondestructive Evaluation Experiments Using Eddy Currents. Research in
Nondestructive Evaluation, Vol.20, No.3, (July 2009), pp. 159-177.
Poloujadoff, M.; Christaki, E. & Bergman, C. (1994). Univariant search:
An opportunity to
identify and solve conflict problems in optimization. IEEE trans. Energ
y Conver.,
EC(9), (December 1994), pp. 659-664.
7
Magnetic Adaptive Testing
Ivan Tom
1
and Gbor Vrtesy
2
1
Institute of Physics, Praha,
2
Research Institute for Technical Physics and Materials Science, Budapest
1
Czech Republic
2
Hungary
1. Introduction
Tests of impending material degradation of engineering systems due to
their industrial
service make an indispensable part of any modern technological processes. Destru
ctive tests
are extremely important. Their considerable advantage consists in their
straightforwardness.
In great majority of cases they directly test the very property, whic
h is at stake. For instance,
the limiting endurable stress before the material yields is directly measured by
a mechanical
loading test, or the number of bending cycles before the system breaks is direct
ly counted to
find the fatigue limits, and so on. Besides, the destructive inspection
s are regularly used in
situations of any materialized failure, when knowledge of the final condition o
f the material
at the moment of the breakdown helps to avoid any next malfunction of the same o
r similar
systems under equivalent circumstances. However, the destructive tests c
annot be used on
the finished or half-finished parts in process of their industrial pro
duction, simply because
those parts after having been destructively tested cannot be used fo
r their primary
purpose any more. The only possible application of destructive tests i
n industry is to
examine destructively every n
th
produced piece only, which is not sufficient for reliability of
goods presently required.
Nondestructive tests, do not suffer by those problems. Causing no harm, they can
be used at
each produced object, and they can be periodically applied even on sy
stems in service.
Judging by the vivid interest in the presently observed improvement of
traditional
nondestructive tests and in development of some recently discovered one
s, the
nondestructive evaluation of material objects attracts currently attention perha
ps even more
than the destructive assessment does. Evaluation of nondestructive tests
keeps the user
informed about the actual condition of the system, and should ensure
avoidance of any
failure before it ever happens. There are numerous methods of nondestructive tes
ts, based on
optical, acoustic, electrical, magnetic and other properties of the mat
erials, which can be
correlated with the watched quality of the whole system. The necessity
of the unambiguous
correlation between the nondestructively measured physical property and
the guarded
property of the system at stake is an unreservedly required claim. It
calls for the
nondestructive tests to be examined and re-examined to their one hundr
ed percent reliability
before they can be really applied in crucial cases. Multi-parametric o
utput of a
nondestructive testing method is therefore an extremely valuable and we
lcomed property,
146
which enables to check and to cross-check validity of the necessary correlation
and ensures
reliability of single measurement with respect to the guarded physical q
uality of the tested
system. The presently described method of Magnetic Adaptive Testing (MAT) is jus
t one of
the few multi-parametric nondestructive tests available.
Many industrial systems have construction parts, which are fabricated f
rom the most
common ferromagnetic construction materials, such as steel or cast iron
. Nondestructive
tests of their structural-mechanical quality can be very suitably based
on magnetic
measurements, because processes of magnetization of ferromagnetic materia
ls provide
detailed reflection of the material microstructure and its degradation.
The physical processes, which take place in the course of magnetizatio
n of ferromagnetic
materials by an applied magnetic field, are well known. They are clas
sified as reversible
(mostly continuous changes of direction of the magnetization vector wit
hin magnetic
domains) and irreversible (mostly discontinuous changes of volumes of magnetic d
omains,
caused by jumps of domain walls from a position to a next one). These processes,
the latter
ones in particular, are closely correlated with structure of the sampl
e, with its uniformity,
texture, material defects, internal and external stresses and even with the samp
le shape. This
can be disturbing, if we want to measure magnetic properties of the
pure material.
However, if we are interested just in the material structure, this co
rrelation serves as a
starting point of all magnetic techniques used for magnetic nondestruct
ive tests of any
ferromagnetic material. For detailed survey of such magnetic methods see e.g. pa
pers (Jiles,
1988, 2001, Blitz, 1991, Devine, 1992) and many references to actual
measurements
quoted there.
Magnetic hysteresis methods belong to the most successful indirect ways
of structural
investigation of ferromagnetic construction materials. They are mostly based on
detection of
the material structural variations via variation of the traditional par
ameters of the major,
saturation hysteresis loop, such as coercive field, H
C
, remanent magnetic induction, B
R
,
maximum permeability,
MAX
, and a few others. At first, these magnetic parameters are
experimentally correlated with independently measured real structural /
mechanical
characteristics of the samples, and then from measurement of the former ones the
latter can
be determined. The point is that the magnetic parameters are measured non-destru
ctively and
with less difficulty than the real structural / mechanical characterist
ics, which in most cases
can be learned destructively only. The few traditionally employed magnetic param
eters are
actually special points or slopes on the magnetic major hysteresis loo
p. These traditional
parameters are very well suited for characterization of magnetic proper
ties of ferromagnetic
materials, but they were never optimized for magnetic reflection of mo
dified structural
properties of the measured samples. Besides, they are by no means the
only available
magnetic indicators of various nonmagnetic modifications of ferromagnetic
materials.
Correlation between alternative magnetic parameters and concrete structural chan
ges of the
studied material, can possibly be even better adapted to each specific task.
Structural aspects of ferromagnetic samples cover material non-uniformiti
es such as local
mechanical stresses, clusters of dislocations, grains, cavities, inclusio
ns and many others.
Their presence, distribution and magnitude control details of the magne
tization processes,
and there is no reason to expect that each type of structural defect affects all
regions of each
magnetization process with the same intensity. Measurement of a suitably chosen
magnetic
147
variable, e.g. magnetic induction, B, or differential permeability, , of
differently degraded
samples of the same material, while changing the applied field in a
systematic way,
confirms the fact. Analysis of a large family of minor hysteresis loops is a goo
d example of such
a systematic investigation of magnetization processes. Besides, as known from th
e Preisach
model of hysteresis (Bertotti, 1998), such volume of data is able to
make a complete
magnetic picture of the sample.
A novel method, Magnetic Adaptive Testing (MAT), for magnetic nondestructive ins
pection
of ferromagnetic construction materials (Tom, 2004) is introduced in this
chapter. The
method is based on systematic measurement of minor magnetic hysteresis
loops. The
essential difference between material testing by the traditional hysteresis- and
by the MAT-
approach is shown in Fig. 1 schematically. The left hand part of the
figure represents the
traditional measurement of the single major (saturation) hysteresis loop
. The major loop is
measured for each of the degraded samples and the material degradation can be de
scribed
through variation of values of any of the few major loop parameters, e.g. H
C
, B
R
, ... as functions
of an independent degradation variable, c, (e.g. as functions of mecha
nical tension). The
right hand part of the figure depicts schematically volume of the mea
sured data for MAT.
The large family of minor hysteresis loops is measured for each of t
he degraded samples
and the material degradation can be then described through variation of values o
f any of the
point (and/or slope) on any of the minor loops, i.e. B(F
i
, A
j
) (and/or (F
i
, A
j
)), as functions of an
independent degradation variable, c, (e.g. as functions of mechanical tension).
B(F
i
, A
j
)
= field coordinates
MAX
B
M
B
R
H
C
A
r
b
i
t
r
a
r
y
d
e
s
c
r
i
p
t
o
r
s
B
(
F
i
,
A
j
)
M
a
n
y
a
v
a
i
l
a
b
l
e
p
a
r
a
m
e
t
e
r
s
sweeping field, F
F
e
w
a
v
a
i
l
a
b
l
e
p
a
r
a
m
e
t
e
r
s
T
r
a
d
i
t
i
o
n
a
l
d
e
s
c
r
i
p
t
o
r
s
H
C
,
B
M
,
B
R
,
M
A
X
.
.
.
sweeping field, F
loop amplitude, A
Fig. 1. Schematic comparison of the traditional magnetic hysteresis testing (lef
t) and
Magnetic Adaptive Testing (right). The traditional testing uses only a few param
eters of the
major loop for description of the material. Magnetic Adaptive Testing has the ch
oice to pick
up the best from many available parameters indexed by the field coordinates.
The method of Magnetic Adaptive Testing utilizes systematic measurement of large
families
of minor hysteresis loops, from minimum amplitudes up to possibly the maximum (m
ajor)
ones on degraded ferromagnetic samples/objects. From the large volume o
f the recorded
data, those are applied for evaluation of the degradation, which refle
ct the material
degradation in the most sensitive or otherwise the most convenient way. Such bes
t adapted
for the investigated case data are used as the MAT-parameter(s) and its / their
dependence
on an independent variable accompanying the inspected degradation is re
ferred to as the
148
MAT degradation function(s). The following text is dedicated to description of b
asic principles
of MAT, to specific advantages and disadvantages of the method and to a number o
f cases
when MAT was successfully applied.
2. Description of the method
The basic features and processes of use of the method of Magnetic Ad
aptive Testing are
probably most instructive and easy to understand, if they are demonstr
ated on a typical
concrete example of MAT application. Therefore, throughout Section 2, p
roperties are
described of samples manufactured from low carbon steel, which was degraded (pla
stically
deformed) by previous application of 7 different magnitudes of mechanic
al tension. The
seven steel samples are ring-shaped with the magnetizing and the pick-
up coils wound on
each of them. All the figures presented in Section 2 refer to this example.
The magnetic induction method appears to be the easiest way of the s
ystematic
measurement for MAT. A Permeameter described in Fig. 2 makes a practical example
of the
possible experimental arrangement. The driving coil wound on the magnet
ically closed
sample gets a triangular waveform current with stepwise increasing amplitudes an
d with a
fixed slope magnitude in all the triangles (see Fig. 2, right). This
produces a triangular
variation of the magnetizing field with time, t, and a voltage signal
, U, is induced in the
pick-up coil for each k
th
sample:
U(dF/dt, F, A
j
, c
k
) = K*cB(dF/dt, F, A
j
, c
k
)/ct = K*(dF/dt, F, A
j
, c
k
)* dF/dt, (1)
where K is a constant determined by geometry of the sample and by t
he experimental
arrangement. As long as F=F(t) sweeps linearly with time i.e. ,dF/dt, is (the sa
me) constant
for measurement at each of the samples, Eq.(1) states, that the measu
red signal is simply
proportional to the differential permeability, , of the measured magnetic circuit
, as it varies
with the applied field, F, within each minor loop amplitude, A
j
, for each k
th
measured
sample. If correct results without influence of any previous remanence should be
obtained,
it is evident that each sample has to be thoroughly demagnetized before it is me
asured.
Pick-up
coil
Driving
coil
Voltage
amplifier
Resistor
Signal
amplifier
Sample
Power
amplifier
Function
generator
A
/
D
I
n
/
O
u
t
c
a
r
d
N
o
t
e
b
o
o
k
c
o
m
p
u
t
e
r
121.5 122.0 122.5 123.0 123.5 124.0
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
Step, AI
F
M
a
g
n
e
t
i
z
i
n
g
c
u
r
r
e
n
t
,
I
F
[
A
]
Time, t [a.u.]
Fig. 2. Left: Block-scheme of the Permeameter. The driving coil produces triangu
lar
variations of the applied magnetic field. The signal coil picks-up the induced v
oltage
proportional to differential permeability of the sample. This scheme shows arran
gement
when a magnetically closed sample is measured and both the coils are wound direc
tly on
the sample. Right: Triangular variation of the magnetizing current with time.
149
The Permeameter works under control of a notebook PC, which sends the
steering
information to the function generator and collects the measured data. An input/o
utput data
acquisition card accomplishes the measurement. The computer registers ac
tually two data
files for each measured family of the minor -shaped loops. The first on
e contains detailed
information about all the pre-selected parameters of the demagnetization
and of the
measurement. The other file holds the course of the voltage signal, U, induced i
n the pick-up
coil as a function of time, t, and of the magnetizing current, I
F
, and/or field, F. A typical
example of one family of the -shaped loops (the reference, unstrained sample with
c
0
=0%)
is shown in Fig. 3a, and Fig. 3b presents the seven families of all the seven ri
ng samples of
the steel degraded by the previously applied mechanical tension to the strain va
lues c
k
= 0%,
1.7%, 3.5%, 5.8%, 7.8%, 9.8%, 17.9%, respectively. Evidently it is a lot of data
and our task is
to compare them and to find the most suitable ones for description o
f the investigated
material degradation.
-2000 -1000 0 1000 2000
-15
-10
-5
0
5
10
15
(a)
c = 0%
P
i
c
k
-
u
p
c
o
i
l
s
i
g
n
a
l
,
U
~
[
m
V
]
Magnetizing field, F, [A/m]
-2000 -1000 0 1000 2000
-15
-10
-5
0
5
10
15
(b)
c = 0%
c = 1.7%
c = 3.5%
c = 5.8%
c = 7.8%
c = 9.8%
c = 17.9%
P
i
c
k
-
u
p
c
o
i
l
s
i
g
n
a
l
,
U
~
[
m
V
]
Magnetizing field, F, [A/m]
Fig. 3. Examples of families of the -shaped loops vs. magnetizing field, F. (a) M
easured on
the unstrained c
0
=0% low carbon steel sample. The positive and negative parts of the
signal correspond to the increasing and decreasing parts of the triangular wavef
orm of the
current (field), respectively. (b) Measured on all the seven ring samples, manuf
actured from
low carbon steel, which was degraded by mechanical tension down to strain values
c
k
= 0%,
1.7%, 3.5%, 5.8%, 7.8%, 9.8%, 17.9%. The curves are plotted in dependence on the
magnetizing field, F, within the limits 2000 A/m.
Instead of keeping the signal and the magnetizing field in shapes of
continuous time-
dependent functions, it is practical to interpolate the family of data for each
c
k
-sample into a
discrete square (i, j)-matrix, U(F
i
, A
j
, c
k
), with a suitably chosen step, AA = AF. (Because
dF/dt is a constant, identical for all measurements within one experiment, it is
not necessary
to write it explicitly as a variable of U.) MAT is a relative metho
d (practically all the
nondestructive methods are relative), and the most suitable information
about degradation
of the investigated material can be contained in variation of any element, of su
ch matrices as
a function of c, relative with respect to the corresponding element o
f the reference matrix,
U(F
i
,A
j
,c
0
). Therefore all U(F
i
, A
j
, c
k
) elements will be divided by the corresponding elements
U(F
i
,A
j
,c
0
) of the reference sample matrix and normalized elements of matrices
of relative
differential permeability (F
i
,A
j
,c
k
) = U(F
i
, A
j
, c
k
)/U(F
i
, A
j
, c
0
), and their proper sequences
150
(F
i
,A
j
,c) = U(F
i
, A
j
, c)/U(F
i
, A
j
, c
0
) (2)
as normalized -degradation functions of the inspected material will be
obtained. Throughout
the text of this chapter, all the degradation functions are considered
as normalized with
respect to the corresponding values of the reference sample.
Beside the -matrices and the corresponding -degradation functions, also m
atrices of the
integrated, B=}dF, or the differentiated,
F
=d/dF, and/or
A
=d/dA values can be
computed and used for definition of the B-degradation functions, B(F
i
,A
j
,c), of the
F
-
degradation functions,
F
(F
i
,A
j
,c), and/or of the
A
-degradation functions,
A
(F
i
,A
j
,c). In fact
the B-,
F
-, and
A
-degradation functions do not contain more or other information than the
-degradation functions, but their use is sometimes more helpful and they are occa
sionally
able to show certain material features with higher sensitivity or in
different relations. This
will be shown later in the application examples. (Note: As the
A
degradation functions are
actually not used in any of the examples here,
F
is simply denoted as thoughout this
chapter.) In some cases it turns out, that degradation functions of r
eciprocal values, such as
1/-degradation functions and the similar others, are more convenient than the dir
ect ones.
Application of the reciprocal degradation functions proves effective espec
ially in situations
when with the increasing parameter c the direct degradation functions approach k
ind of
a saturation.
An example of the -degradation functions and 1/-degradation functions of the low c
arbon
steel degraded by the mechanical tension is shown in Fig. 4.
0 5 10 15
0
1
2
3
1
2 3
4
5 6
7
A
B
C
D
E F
G
a
b
c
d
e
f
g 1
2 3
4
5 6
7
A
B
C
D E
F
G
a
b
c
d
e
f
g
1
2 3
4
5
6
7
A
B
C
D
E
F
G a
b
c
d
e
f
g
1
2
3
4
5 6 7
A
B
C D E F G
a
b
c
d
e
f g
1
2
3 4 5 6 7
A
B
C
D
E F G
a
b
c
d
e f
g
1
2
3
4
5 6 7
A
B
C D E F G
a
b
c
d
e f g
1
2
3
4
5 6
7
A
B
C
D
E F
G
a
b
c
d
e f
g
1
2
3
4
5
6
7
A
B
C
D E
F
G
a
b
c
d
e
f
g
1
2
3
4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3 4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3
4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e f
g
1
2
3 4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3
4
5 6
7
A
B
C D
E
F
G
a
b
c
d
e
f
g
1
2
3
4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3
4 5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3
4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
1
2
3
4
5
6
7
A
B
C
D
E
F
G
a
b
c
d
e
f
g
(a)
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
F0A200
Fn200A400
F0A400
F200A400
Fn400A600
Fn200A600
F0A600
F200A600
F400A600
Fn600A800
....
0 5 10 15
0
2
4
6
8
1
2 3 4 5 6 7
A
B C D E F G
a
b c
d e f g 1
2 3 4 5 6 7
A
B C D E F G
a
b c d e f g
1
2 3 4 5 6 7
A
B C D
E F G a
b
c
d
e f g
1
2
3
4
5 6
7
A
B
C
D
E
F
G
a
b
c
d
e
f g
1
2
3
4
5
6
7
A
B
C
D
E
F G
a
b c d e f g
1
2
3
4
5 6
7
A
B
C
D
E
F
G
a
b
c
d
e
f g
1
2 3 4 5 6 7
A
B C D E F G
a
b
c
d
e f g
1
2
3 4 5 6 7
A
B
C D E F G
a
b
c d e f g
1
2
3
4 5
6 7
A
B C D E
F G
a
b
c d e
f g
1
2 3 4 5 6 7
A
B
C D E F G
a
b
c d e f g
1
2
3 4 5 6 7
A
B C D E F G
a
b c d e f g
1
2 3 4 5
6 7
A
B
C D E
F G
a
b
c d e
f g
1
2
3 4 5 6 7
A
B C D E F G
a
b
c d e f g
1
2
3
4 5 6 7
A
B
C
D E F G
a
b
c
d e f g
1
2
3
4 5 6 7
A
B
C
D E F G
a
b
c
d e
f g
1
2
3 4 5 6 7
A
B
C D E F G
a
b c d e f g
1
2
3
4 5
6 7
A
B
C
D E
F G
a
b
c
d e
f g
(b)
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
F0A200
Fn200A400
F0A400
F200A400
Fn400A600
Fn200A600
F0A600
F200A600
F400A600
Fn600A800
....
Fig. 4. The -degradation functions (a) and 1/-degradation functions (b) of the low
carbon
steel degraded by the mechanical tension. The functions were calculated for the
step AA =
AF = 200A/m and they are plotted up to the maximum amplitude A
j
= 3600 A/m.
Number of the degradation functions obtained from the MAT measurement
depends on
magnitude of the maximum minor loop amplitude, A
j
, up to which the measurement is
done, and on choice of the step value AA = AF which is used for c
omputation of the
interpolated data matrices. The maximum amplitude A
j
= 3600 A/m and the step AA = AF =
200 A/m were applied in the presented example of investigation of the
degraded steel.
Under these conditions there are altogether 289 exploitable -degradation function
s (all the
151
unreliable boundary (F
i
=A
j
) degradation functions were excluded from the total number
of
n=0
E
n=17
(2n+1)=324) and the same number of 1/-degradation functions, as they ar
e
plotted in Fig. 4.
Once the degradation functions are computed, the next task is to find
the optimum
degradation function(s) for the most sensitive and enough robust descri
ption of the
investigated material degradation. A 3D-plot of sensitivity of the degradation f
unctions can
substantially help to choose the optimum one(s). The - and 1/-sensitivit
y maps
corresponding to the degradation functions of Fig. 4 are plotted in Fig. 5. Slop
e of the linear
regression of each degradation function is here defined as the function sensitiv
ity. Thus the
sensitivity map is a 3D-graph of these slope values plotted against the degradat
ion functions
field coordinates (F
i
, A
j
). As it follows from the presented sensitivity maps, the most
sensitive -degradation functions are those with field coordinates around
(F
i
=-1400 A/m,
A
j
=3600 A/m), and/or (F
i
=2200 A/m, A
j
=3600 A/m), whereas the most sensitive 1/-
degradation functions evidently have the F-coordinate equal to F
i
= 400 A/m and the
optimum amplitudes start from A
j
= 600 A/m (the most sensitive 1/-degradation function)
and the sensitivity decreases (but keeps rather high anyway) for large
r minor loop
amplitudes. The most sensitive - and 1/-degradation functions are plotted in Fig.
6.
-3000 -2000 -1000 0 1000 2000 3000
-3000
-2000
-1000
0
1000
2000
3000
(a)
-sensitivity map
Magnetizing field, F [A/m]
M
i
n
o
r
l
o
o
p
a
m
p
l
i
t
u
d
e
,
A
[
A
/
m
]
-0.02800
-0.01400
0
0.01400
0.02800
0.04200
0.05600
0.07000
0.08400
-3000 -2000 -1000 0 1000 2000 3000
-3000
-2000
-1000
0
1000
2000
3000
(b)
1/ -sensitivity map
Magnetizing field, F [A/m]
M
i
n
o
r
l
o
o
p
a
m
p
l
i
t
u
d
e
,
A
[
A
/
m
]
-0.02500
0.03000
0.08500
0.1400
0.1950
0.2500
0.3050
0.3600
Fig. 5. The -sensitivity map (a) and 1/-sensitivity map (b) of the low carbon stee
l degraded
by the mechanical tension. The maps were calculated for the step AA = AF = 200 A
/m and
they are plotted up to the maximum amplitude A
j
= 3600 A/m.
0 5 10 15
2
4
6
8
1
/
a
n
d
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
1/ F=400, A=600
1/ F=400, A=1400
1/ F=400, A=1800
1/ F=400, A=3600
F=-1400,A=3600
F=2200, A=3600
Fig. 6. The most sensitive degradation functions for plastic deformation of a st
eel for the step
AA = AF = 200 A/m. The top sensitive is the 1/-degradation function with field pa
rameters
F
i
= 400 A/m, A
j
= 600 A/m. The most sensitive -degradation functions are more flat.
152
3. Methodological hints
During measurement of the induced signal in the example presented in
previous Section,
samples shaped like thin rings were used, and the magnetizing and the
pick-up coils were
wound directly on them. Relatively small field was able to magnetize
the material
substantially, as there was no demagnetization effect, and homogeneous
magnetization of the
sample material was rightfully assumed. Signal measured on such magneti
cally closed
samples is proportional to differential permeability of the samples material, an
d magnetizing
field, F, inside the samples can be easily calculated as
F = N I
F
/L [A/m, 1, A, m], (3)
where N is number of turns of the magnetizing coil, I
F
is the magnetizing current and L is
circumferential length of the magnetic circuit (e.g. of the sample ring).
If the measured samples are magnetically open, the situation is more complicated
. Long and
narrow specimens can be also successfully magnetized by coils placed around their
body, as
the demagnetization of such shapes is still acceptable for the MAT measurement.
However,
demagnetization field comes here into picture and the total inside field cannot
be calculated
as easily as suggested by Eq.(3). And what to do with short stout o
r large flat shapes? The
magnetic circuit of such samples can be artificially closed, and the
samples can be
magnetized and measured with the aid of a magnetically soft yoke. The yoke can
be either
passive with the coils wound around body of the sample, or the yoke can be activ
e (we speak
then about an active inspection head) with the magnetizing coil wound around the
bow of the
yoke and the pick-up coil around one or both legs of the yoke.
Problems may appear in cases of rough or uneven surfaces of samples,
where quality of
magnetic contact between the sample and the yoke fluctuates from sampl
e to sample. This
introduces fluctuation of quality of the magnetic circuit and therefore
also of the pick-up
signal, which can be mistaken for variation of the samples material.
The most rigorous
solution of such situation is then simultaneous measurement of the tan
gential field on the
sample surface (Stupakov, 2006, Stupakov et al., 2006, Perevertov, 2009), which
substantially
complicates the experiment, however. Another approach, which is able to solve th
e problem
satisfactorily in majority of situations is to insert a non-magnetic spacer betw
een the sample
/ yoke touching surfaces. The spacer makes the fluctuation relatively
smaller and allows
successful MAT measurement even on rough surfaces of magnetically open
samples, see
(Tom et al., 2012) and Section 3.5. However, fortunately, in great majority of inv
estigated
samples the different pieces of sample series have more-or-less similar
quality of surface,
which ensures reliable and repeatable measurements, even if the surface
is rough,
considering that MAT is a relative measurement, and degradation functions are no
rmalized
with the corresponding value of the reference sample.
In MAT measurements with open samples and attached yokes we deal with
non-uniform
magnetization. As a consequence, the magnetizing field inside the sampl
e cannot be easily
calculated like (3), we cannot use the magnetizing field coordinates (
F
i
, A
j
), but we use the
magnetizing current coordinates (I
Fi
, I
Aj
) instead. Besides, with the non-uniform magnetic
circuits we cannot speak about the signal U to be proportional to th
e differential
permeability of the material, but evidently we deal with an effective differenti
al permeability

153
of the existing circuit instead. As MAT is a relative method, the current coordi
nates are also
well applicable for identification of the mutually corresponding magneti
c states of the
samples to be related and compared.
Measurements on closed samples and on open ones, either in a solenoid or helped
by attached
yokes, evidently yield quantitatively different data. If measured on id
entical materials, are
they comparable? Do they show similar trends? Next two Sections of this part giv
e answers to
such questions and describe MAT measurements performed on equivalent series of c
losed and
open samples (Section 3.1) and on the same series measured in a solenoid and wit
h the aid of
yokes (Section 3.2). The remaining Sections of this part present sever
al other points of the
methodological discussion, namely the potential of multi-parametric chara
cter of MAT
(Section 3.3), the influence on MAT sensitivity of speed of magnetizat
ion (Section 3.4) and of
nonmagnetic spacers (Section 3.5), and eventually the influence of samp
le material
temperature on the MAT results (Section 3.6). The last Section of this part, Sec
tion 3.7, refers to
several traditional hysteresis and Barkhausen measurements of degraded s
amples and
compares their results with those obtained by MAT.
3.1 Magnetically closed and open samples
Structural degradation of samples manufactured from compressed steel are here st
udied by
MAT and their optimal 1/-degradation functions are compared with their rolling re
duction,
and with their independently measured Vickers hardness (HV). The material is low
-carbon
steel with 0.16 wt.% of C, 0.20 wt.% of Si, 0.44 wt.% of Mn, with
the rest of Fe, and the
samples were produced from steel plates cold rolled down to 0, 5, 10, 20 and 40%
reduction.
Three sample geometries are used, namely picture-frames, rectangular pla
tes and
rectangular bars. Dimensions and shapes of the samples are shown in Fig. 7.
Fig. 7. Dimensions and shapes of the frames, of the plates and of the bar-sample
s.
MAT measurements were performed on all the samples. The frame samples
were
equipped with identical magnetizing and pick-up coils each. All plate-
and bar-samples
were measured by the same inspection head, directly attached (no non-m
agnetic spacers
were used for the polished surfaces) on a surface of each sample. Th
e optimum 1/-
degradation functions were taken for each series of the samples. Repre
sentative sets of
such optimal 1/-degradation functions vs. rolling reduction and vs. the Vickers h
ardness
are plotted in Fig. 8 for the frame-, for the plate- and for the bar-shaped samp
les. In each
measurement series, all the degradation functions are normalized by dat
a of the virgin
(not rolled) sample of the relevant shape. As seen from Fig. 8, cons
ecutive series of the
optimal MAT degradation functions well reflect the material modification
s, regardless of
the actual sample shapes. The results on series of differently shaped
samples agree with
154
each other qualitatively, and the frame-shaped magnetically closed sampl
es show the
highest sensitivity, as expected. For making the measurements also quan
titatively
comparable and repeatable, it is necessary to perform the tests always
under the same
experimental conditions (speed of magnetization, steps of increasing current
, etc.), to use
the same measuring head (and/or the same coils), and evidently also t
he same shape of
samples (see also Vrtesy et al., 2008a).
0 10 20 30 40
0
2
4
6
8
10
12
(a)
Bar
Plate
Frame
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Rolling reduction [%]
140 160 180 200
0
2
4
6
8
10
12
(b)
Bar
Plate
Frame
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Vickers hardness
Fig. 8. Optimal 1/-degradation functions for the three series of differently shap
ed samples.
(a) Dependence on the rolling reduction. (b) Dependence on Vickers hardness.
3.2 Open samples between yokes and in a long solenoid
Low carbon commercial steel CSN 12050 was plastically elongated in an
Instron loading
machine up to final strain values 0, 0.1, 0.2, 0.9, 1.5, 2.3, 3.1, 4.0, 7.0 and
10.0%. A single series
of identically shaped long flat samples (115x10x3 mm
3
) was manufactured from the
degraded steel pieces and it is here studied by MAT in two different ways, both
applicable
for these magnetically open samples. In one measurement series the sam
ples were
magnetized by a short coil, placed around middle of each sample and
the magnetic circuit
was artificially closed from both sides of the sample by two symmetri
cally positioned
passive yokes (no non-magnetic spacers). The arrangement is shown in F
ig. 9a, while Fig.
10a shows pick-up coil signals of the investigated samples. In the other measure
ment series
the same samples were magnetized as open, in a long solenoid. This s
olenoid, with one of
the samples inside, is shown in Fig. 9b. Fig. 10b demonstrates the p
ick-up coil signals
originating from this experimental arrangement.
Dramatic quantitative differences can be observed between signals of Figs. 10a a
nd b, even
though the double-peak character of the signal is qualitatively preserv
ed in both. And, in
spite of difference between the directly measured signals, character of the MAT
degradation
functions vs. plastic strain is again qualitatively the same. It is sh
own in Fig. 11, where the
optimal 1/-degradation functions of the discussed two cases are compared
. Evidently,
sensitivity of measurements is significantly higher if the samples are
at least artificially
closed by the yokes, but even in the worse case of the open long thin samples in
a solenoid
the material degradation is well measurable.
155
(a) (b)
Fig. 9. The arrangements, used for the measurements. (a): The samples are magnet
ized by
the short coil placed around middle of each sample and a pair of soft passive yo
kes closes
the magnetic circuit from bottom and top. (The 2 kg weight is used only for pres
sing the
yokes better on the sample surfaces.) (b): The samples are magnetized inside the
long
solenoid.
-1.0 -0.5 0.0 0.5 1.0
-100
-50
0
50
100
(a)
S
i
g
n
a
l
o
f
t
h
e
p
i
c
k
-
u
p
c
o
i
l
[
m
V
]
Magnetizing current, I
F
[A]
S1
S3
S5
S7
S10
-1.0 -0.5 0.0 0.5 1.0
-80
-40
0
40
80
(b) S
i
g
n
a
l
o
f
t
h
e
p
i
c
k
-
u
p
c
o
i
l
[
m
V
]
F
[A]
S1
S3
S5
S7
S10
Fig. 10. Several pick-up coil signals of the investigated samples. (S1 is the re
ference, not
deformed sample, S10 is the sample with 10 % strain). (a): Samples with artifici
ally closed
magnetic circuit by a pair of passive yokes. (b): Open samples in the solenoid.
Remark:
Scales in the (a) and (b) figures are not mutually comparable as neither the mag
netizing
solenoids nor the pick-up coils are the same in the two experimental arrangement
s.
3.3 Full use of the multi-parametrical character of MAT
Magnetic Adaptive Testing offers a great number of degradation function
s. Usually those
degradation functions are used as calibration curves for assessment of
unknown samples,
which are monotonous and most sensitive to the investigated material d
egradation. However,
in some cases, magnetic reflection of the degraded material yields onl
y non-monotonous
degradation functions, see e.g. study of low cycle fatigue in (Devine et al., 19
92, Tom et al.,
2010). A single non-monotonous degradation function does not allow to
decide whether a
measured unknown sample belongs to the ascending or to the descending part of it
. It will
156
be shown in this Section, that the multi-parametric behavior of MAT can solve th
is question
and that a suitable combination of two or more non-monotonous degradat
ion functions
obtained from a single MAT measurement makes such a decision possible, see also
(Vrtesy
& Tom, 2012a).
0 2 4 6 8 10
2
4
6
(a)
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Strain (%)
0 2 4 6 8 10
1,0
1,2
1,4
1,6
1,8
(b)
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Strain (%)
Fig. 11. MAT degradation functions vs. the plastic strain obtained on the tensil
e stressed
steel samples. (a): Samples were artificially closed by a double yoke. (b): Open
samples in
the solenoid.
As an illustration of such a solution again the series of long flat samples (115
x10x3 mm
3
) is
presented, which were manufactured from the commercial steel CSN 12050,
plastically
deformed by tensile stress up to the ten strain values 0, 0.1, 0.2, 0.9, 1.5, 2.
3, 3.1, 4.0, 7.0 and
10.0%. The same samples were used in Section 3.2 already for demonstration of eq
uivalency
between results of two different ways of measurement. Data from the arrangement
with two
passive yokes (see Fig. 9a) are described here.
The non-monotonous character of the 1/-degradation functions is shown in Fig. 11,
and the
same is true about the 1/-degradation functions, which are presented in Fig. 12. T
he most
sensitive 1/-degradation functions come from the area of coordinates arou
nd I
F
=40 mA,
I
A
=150 mA values (see the red area in the 1/-sensitivity map plotted in
Fig. 13), but as
shown in Fig. 12a, they are non-monotonous at the range of large strains. If 1/-de
gradation
functions from other areas of the sensitivity map are plotted, all of
them show the non-
monotonous shape as well. However, not all of them have the peaks in
the same range of
strains as the most sensitive ones do. An example of such a 1/-degrada
tion function is
plotted in Fig. 12b. The function in Fig. 12b comes from the area around I
F
=400 mA, I
A
=600
mA (see the yellow area at the right-hand-side in the 1/-sensitivity ma
p in Fig. 13.), it is
substantially less sensitive than the optimal 1/-degradation function of
Fig. 12a, but it is
nicely monotonous at large strains (and non-monotonous at low strains, where the
sensitive
function does not have any problem). Evidently, the sensitive 1/-degradat
ion function of
Fig. 12a can be used as the main calibration function, but if an unknown sample
falls into its
non-monotonous part, the auxiliary 1/ -degradation function of Fig. 12b
answers the
question about whether the unknown sample belongs to the ascending or to the des
cending
part of the Fig. 12a-curve.
157
0 2 4 6 8 10
0
5
10
15
20
25
(a)
I
F
= 40 mA, I
A
= 150 mA
T
h
e
m
o
s
t
s
e
n
s
i
t
i
v
e
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Strain (%)
0 2 4 6 8 10
0,5
1,0
1,5
2,0
2,5
(b)
I
F
= 400 mA, I
A
= 600 mA
T
h
e
a
u
x
i
l
i
a
r
y
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Strain (%)
Fig. 12. (a): The optimum 1/-degradation function from the (I
F
=40 mA, I
A
=150 mA) area.
(b): The auxiliary 1/-degradation function from the (I
F
=400 mA, I
A
=600 mA) area.
-200 0 200 400
0
200
400
600
800
F
[mA]
M
i
n
o
r
l
o
o
p
a
m
p
l
i
t
u
d
e
,
I
A
[
m
A
]
Fig. 13. The 1/-sensitivity map of the steel degraded by tensile stress. Crossing
lines
indicate the areas from where degradation functions of Fig. 12 were taken.
3.4 Speed of magnetization
The most frequent way of MAT measurement proceeds by the electromagnet
ic induction
method in a Permeameter similar to the scheme in Fig. 2 under the condition ,dF/
dt,=const.
This condition guarantees that the magnetization processes in the samples contin
ue always
at the constant speed, which is desirable. As according to Eq.(1) the
measured signal is
directly proportional to dF/dt, this field-slope is set high enough with the int
ention of getting
a sufficient signal-to-noise ratio. At the same time, however, it is
advisable to set it low
enough in order to minimize eddy currents and other dynamic effects, whi
ch can adversely
influence shape of the signal and of the degradation functions via their implici
t dependence
on the value of dF/dt. In this Section influence of the rate of cha
nge of the magnetization
processes on sensitivity of degradation functions in Magnetic Adaptive
Testing will be
investigated, see also (Tom et al., 2009).
The samples employed for this investigation were the same series of c
ircular rings of low
carbon steel plastically deformed by uniaxial tension, which was introd
uced as an
illustrative example in Section 2. Arrangement of the experiment coinci
ded with Fig. 2, so
158
that the varying magnetization of each sample was achieved by applicat
ion of a time-
dependent magnetic field, F(t), due to triangular waveform current, I
F
(t), in the magnetizing
coil, with step-wise increasing field-amplitudes, A
j
, corresponding to the step-wise
increasing current-amplitudes, I
Aj
. The rate of change in all the triangles was constant (but
for its sign), namely dI
F
/dt=const. for the current-slope and/or dF/dt=const. for the
magnetizing field-slope, in each measured family of the minor loops an
d in any
measurements mutually compared. Three measurements of the same series of samples
were
carried out. The applied current-slopes and the corresponding field-slop
es in the three
measurements were dI
F
/dt = 0.5 A/s, 4 A/s, 32 A/s, and dF/dt = 0.8 kA/m/s, 6.4 kA/m/s,
51.2 kA/m/s, respectively.
The sensitivity maps of the - and of the 1/-degradation functions compu
ted for the
measurement with the lowest field-slope dF/dt = 0.8 kA/m/s were plotted in Fig.
5 for the
field-coordinates -A
j
s F
i
s +A
j
, 0 < A
j
s 3.6 kA/m, with the step AA = AF = 0.2 kA/m. The -
and of the 1/-sensitivity maps for dF/dt = 6.4 kA/m/s, and 51.2 kA/m/
s were
qualitatively similar to those in Fig. 5. Three interesting regions of the field
-coordinates (F
i
,
A
j
) can be seen in each of the sensitivity maps, with slight shifts of the extreme
regions due
to different field-slopes. The white regions in Fig. 5a indicate the most sensit
ive increasing -
degradation functions, the black areas indicate the most sensitive decr
easing -degradation
functions. For the 1/-degradation functions (see Fig. 5b) it naturally
is just reversed. The
most sensitive degradation functions for the field-slope dF/dt = 0.8 kA/m/s are
plotted in
Section 2. in Fig. 6. The same degradation functions, together with t
heir counterparts for
higher values of the field slope, namely for dF/dt = 6.4 kA/m/s, and
51.2 kA/m/s are
drawn in Fig. 14 in the present Section.
As the curves in Fig. 14 show, there is very little influence of the
varied magnetizing field-
slope from 0.8 till 51.2 kA/m/s (i.e. 64 times larger speed of magnetization) on
sensitivity of
the best -degradation functions within their regions of monotonous increase (i.e.
in the two
white areas of the sensitivity map of Fig. 5a). However, the best degr
adation functions
from the white region of Fig. 5b, (i.e. those with a monotonous increa
se of the 1/-
degradation functions) are influenced substantially as it is seen on F
ig. 14c. As the
degradation functions are created by the signals ratio shown in Eq. (
2), explanation of this
behavior can be found through a closer look at the recorded induced voltage sign
als both for
the degraded samples, c
k
, and for the normalizing reference sample, c
0
, as was discussed in
details in (Tom et al., 2009).
There is no universal advice as for what magnetizing field-slope to chose and wh
at region of
degradation functions to use for the most successful MAT nondestructive tests in
individual
cases. As the name of the method suggests, it is recommended to adapt optimally
the choice
both of the degradation functions field-coordinates region and the magn
etizing field-slope
as to obtain the satisfactory signal-to-noise ratio and the best degra
dation function
sensitivity at the same time. The concrete adaptation is completely go
verned by properties
of the investigated material degradation and by the level of noise an
d the available rate of
change of the magnetizing field of the used measuring technique.
However, generally speaking it can be stated, that as long as the MAT degradatio
n functions
are picked-up from localities of the field coordinates where the local
differential
permeability is high (frequently close to the maximum permeability at the used m
inor loop),
159
then the (usually reciprocal) degradation functions possess very high sen
sitivity, but level
of their sensitivity is strongly field-slope-dependent. On the other ha
nd, if the MAT
degradation functions are chosen from localities of the field coordinates where
the material
is closer to saturation and the local differential permeability is low
, sensitivity of such
degradation functions is not extremely high, but it is little dependent on the a
pplied magnetizing
field-slope.
0 5 10 15 20
0
1
2
3
4
5
6
(a)
O
p
t
i
m
u
m
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
F
i
=-1.4kA/m, A
j
=3kA/m, slope 0.8kA/m/s
F
i
=-1.4kA/m, A
j
F
i
=-1.4kA/m, A
j
0 5 10 15 20
0
1
2
3
4
5
6
(b)
O
p
t
i
m
u
m
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
F
i
=2.2kA/m, A
j
F
i
=2.2kA/m, A
j
F
i
=2.2kA/m, A
j
0 5 10 15 20
0
1
2
3
4
5
6
(c)
O
p
t
i
m
u
m
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain, c [%]
F
i
=0.4kA/m, A
j
F
i
=0.4kA/m, A
j
F
i
=0.4kA/m, A
j
Fig. 14. The steepest degradation functions measured at the three magnetizing fi
eld-slopes
dF/dt = 0.8 kA/m/s (), 6.4 kA/m/s (O), and 51.2 kA/m/s (A), at the three importan
t
coordinate regions of the sensitivity maps. For easy comparison all the degradat
ion
functions are determined at the minor loop with amplitude A
j
= 3kA/m: (F
i
,A
j
,c) at F
i
= -1.4
kA/m (a), (F
i
,A
j
,c) at F
i
= +2.2 kA/m (b), and 1/(F
i
,A
j
,c) at F
i
= +0.4 kA/m (c).
3.5 Non-magnetic spacers
Magnetic measurement of flat samples, which cannot be magnetized (and/o
r closed
magnetically) in any better way than by an attached magnetizing/sensing soft yok
e, set into
direct contact with the sample surface, suffers often by fluctuation of quality
of the magnetic
sample/yoke contact. This is a well-known problem, with unpolished surfaces in p
articular,
which can be improved by using as large yoke as possible and/or by application o
f a spacer
between the yoke and the sample. However, size of the sample frequently does not
allow a
very large yoke to be used and even a thin nonmagnetic spacer decrea
ses and distorts the
measured signal substantially, so that measurement of basic magnetic pa
rameters of the
sample material in this way is very difficult, if not impossible.
However, spacers are quite applicable for magnetic structuroscopy, i.e. f
or magnetic
measurement of relative structural changes of ferromagnetic construction
materials,
especially if the measurement is carried out by a method analyzing the measured
signal like
it is done e.g. in Magnetic Adaptive Testing. The following measuremen
t explains and
illustrates use of spacers, see also (Tom et al., 2012).
Samples of gradually increasing brittleness were prepared from ferromagn
etic steel
15Ch2MFA in the shape of rectangular prisms 10 x 10 x 30 mm
3
. Material for the samples
was embrittled in the same way as described in Section 4.3.2. Three samples of e
ach grade of
brittleness (of the same tempering temperature T
T
of their preparation, see Section 4.3.2)
were used for the reported measurement. Quality of surfaces of the sa
mples corresponded
to their ordinary machining (milling) and grooves from the milling or
even scratches were
visible on some of them. No polishing of the surfaces was implemented, some surf
aces were
evidently worse than others.
160
The measurement was done inductively on the magnetically open samples
using a pair of
passive yokes. Magnetization of the samples was carried out by a short driving c
oil, which
was each of the samples before the measurement inserted into. The ind
uced voltage signal
was recorded from a pick-up coil, placed co-axially with the driving one. In ord
er to be able
to magnetize the short thick samples by the small driving coil appreciably, two
soft yokes of
laminated Si-Fe sheets (cross sections of the legs 10x20 mm
2
, maximum height in the bow 30
mm) were used to short-cut the magnetic flux during magnetization of
the samples. The
yokes were attached to surfaces of the samples either directly, or ov
er thin spacers (paper
stickers) glued to each contact face of the two yokes. Thickness of the applicab
le spacers was
from zero up to 0.9 mm for the starting examination, spacers with th
ickness 0.08 mm and
0.23 mm were chosen as optimal for the reported measurements. Arrangem
ent of the
sample, yokes and the driving and pick-up coils are shown in Fig. 15 (left) and
photo of the
sample holder can be seen in Fig. 15 (right), showing a sample, the driving coil
(the pick-up
coil is covered by the driving one), the yokes, and a system of levers insuring
a reproducible
pressure of the yokes on each sample.
Yoke
Spacer
Pick-up
coil
Driving
coil
Voltage
amplifier
Resistor
Signal
amplifier
Sample
Power
amplifier
Function
generator
A
/
D
I
n
/
O
u
t
c
a
r
d
N
o
t
e
b
o
o
k
c
o
m
p
u
t
e
r
Fig. 15. Left: Sketch of the Permeameter with a magnetically open sample, coils,
yokes and
spacers. Right: The sample holder with the coils and the attachable yokes. The p
rismatic
sample is placed into the coil system (by shifting it in the metal tray rightwar
ds down to the
brass stop) and the black yokes are pressed against the sample by the system of
levers with
the handle on the right in the picture.
The speed of change of the magnetizing current in its linear region
was 1 A/s for the
measurements without spacers and with spacers 0.08 mm thick, and it was
3 A/s for the
measurement with the spacers 0.23 mm thick. (Increase of the speed of change of
current
in the latter case as compared with the former ones was used in o
rder to make the
signal conveniently larger.) Also the maximum amplitudes and the consta
nt steps were
different for different measurements, their applied values can be approximately
read from
Figs. 16a,b,c. Their choice was rather arbitrary, guided by the intention
to get the sample
appreciably magnetized and to fill up the range between the minimum a
nd maximum
amplitudes of the magnetizing current with a representative number of minor loop
s. The
voltage signals measured in the pick-up coil were proportional to the effective
differential
permeability of the used magnetic circuits and their characteristic sha
pes can be seen in
Fig. 16.
161
-1.0 -0.5 0.0 0.5 1.0
-80
-40
0
40
80 spacer 0 mm
(a)
S
i
g
n
a
l
~
[
m
V
]
F
[A]
T
T
= 690
0
C
T
T
= 670
0
C
T
T
= 660
0
C
T
T
= 650
0
C
T
T
= 640
0
C
T
T
= 630
0
C
T
T
= 620
0
C
-2 -1 0 1 2
-40
-20
0
20
40
spacer 0.08 mm
(b)
F
[A]
-4 -2 0 2 4
-20
-10
0
10
20 spacer 0.23 mm
(c)
F
[A]
Fig. 16. (a): Measured signals from the most ductile (T
T
= 690
0
C) till the most brittle (T
T
=
620
0
C) samples, with the yokes without any spacer.
(b): Signals of the same samples measured with non-magnetic spacers 0.08 mm thic
k, glued
on the yokes. Notice the range of the vertical axis is twice smaller and of the
horizontal axis
is twice larger than in figure (a).
(c): Signals of the same samples measured with non-magnetic spacers 0.23 mm thic
k, glued
on the yokes. Notice the range of the vertical axis is four times smaller and of
the horizontal
axis four times larger than in figure (a).
Decision about optimum thickness of spacers can be found empirically,
following
measurement of peak values of signals of good and bad samples at a conv
enient (the
same for all spacers) speed of change of the magnetizing field, using
a series of available
spacers. Fig. 17 shows result of such a measurement on two most duct
ile (T
T
= 690
0
C)
samples, one with the best surfaces and the other with the worst ones
. As it can be seen in
Fig. 17, magnitudes of the peak values of the two samples approach e
ach other with
increasing spacer thickness. Spacers with thickness 0.08 mm and 0.23 m
m were chosen as
optimum ones for the illustrative measurement.
0.0 0.2 0.4 0.6 0.8 1.0
5
10
15
20
S
i
g
n
a
l
p
e
a
k
[
a
.
u
.
]
Spacer thickness [mm]
BEST SURFACE
WORST SURFACE
Fig. 17. Peak values of signals of two most ductile (T

T
= 690
0
C) samples, one with the best
surfaces and the other with the worst ones. The two curves approach each other w
ith
increasing spacer thickness.
162
The optimum -degradation functions and the
F
-degradation functions for the
measurement without spacer and with spacers 0.08 mm and 0.23 mm thick
are plotted in
Fig. 18.
-700 -680 -660 -640 -620
1.0
1.5
2.0
(a)
B
e
s
t
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
(-1)*Tempering temperature [
0
C]
SPACER=0
SPACER=0.08mm
SPACER=0.23mm
-700 -680 -660 -640 -620
1
2
3
4
(b)
B
e
s
t
F
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
0
C]
SPACER=0
SPACER=0.08mm
SPACER=0.23mm
-700 -680 -660 -640 -620
5
10
15
(c)
B
e
s
t
F
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
0
C]
SPACER=0
SPACER=0.08mm
SPACER=0.23mm
Fig. 18. (a): The best -degradation functions measured without any spacer (O) on
the yoke
faces, with spacers 0.08 mm (V) and with spacers 0.23 mm ().
(b): The best
F
-degradation functions measured without any spacer (O) on the yoke faces,
with spacers 0.08 mm (V) and with spacers 0.23 mm (). Notice the vertical scale i
s doubled
in comparison with figure (a).
(c): The same curves as in figure (b), just the vertical scale is four times lar
ger than in (b).
Influence of the non-magnetic spacers inserted between the sample and
the magnetically
soft yokes is clearly seen in Figs. 16a,b,c: The increased distance b
etween surface of the
sample and faces of the yokes increases demagnetization within the magnetic circ
uits. This
means a decrease of the magnetizing field in the sample with spacers at the use
of the same
magnetizing current as for the circuit without the spacers. However, t
he decreased
magnetizing field fluctuates less. Beside the control of fluctuation of
the magnetizing field,
presence of well-defined spacers (their thickness should be found empir
ically by trial-and-
error approach) modifies shapes of the signals measured on the series of degrade
d samples.
Generally, with application of a spacer, substantially decreased scatter
of the experimental
points accompanied by a small decrease of sensitivity of the degradati
on functions can be
expected. Such a regular behavior is presented in Figs. 18, for the - and the
F
-degradation
functions measured with thin spacers (0 and 0.08 mm) in particular. A substantia
lly thicker
spacer (in our case 0.23 mm), however, is able even to modify shape
of the signals
considerably (see Figs. 16c), which can result in a qualitatively new shape of t
he degradation
functions with quantitatively increased sensitivity (see Fig. 18c).
Use of the spacers for testing of material structural degradation can
be recommended as a
simpler alternative to more sophisticated and more demanding measurement
with a direct
on-line recording of the magnetic field at the sample surface. Especiall
y, when it is
measured by a series of field sensors and the resulting field value
is obtained by
extrapolation of their readings down to the sample surface, see e.g.
(Stupakov, 2006,
Perevertov, 2009), the outcome can be very reliable and allows compensation of f
luctuating
quality of the sample-yoke magnetic contact by determination of the immediate va
lue of the
magnetizing field inside the samples. Such measurements are even able
of a reasonable
determination of absolute magnetic parameters of the samples, which evi
dently cannot be
done with the spacers only.
163
Successful use of the non-magnetic spacers can be applied for relative
indication of the
samples micro-structural conditions. The spacers modify shapes of the measured si
gnals, they
substantially reduce scatter of experimental points accompanied by a sl
ight decrease of the
degradation functions sensitivity. Sometimes, thick spacers in particular
, are able to modify
shapes of the measured signals qualitatively and to bring about consid
erable increase of
sensitivity, especially in the degradation functions computed from the
signal derivatives.
However, spacers hardly can be employed (without a simultaneous measure
ment of the
sample surface field) for determination of purely magnetic parameters of the sam
ples.
3.6 Temperature dependence
An important feature of any NDT technique is the permissible range of
temperatures within
which the method can be successfully applied. For magnetic methods, th
e tested objects are
mostly ferromagnetic iron-based construction materials, namely various kinds of
steel or cast
iron. These materials are electrical conductors, their electric conducti
vity is substantially
reduced not far above room temperature already, and this is the reason
, why e.g. the widely
successful eddy current inspections loose above room temperature a grea
t portion of their
efficiency. Not so the magnetic methods. Magnetic properties of iron-ba
sed materials do not
change around the room temperature significantly yet, as they are stil
l pretty far from their
Curie point. Therefore, magnetic methods and relative magnetic NDT meth
ods in particular,
give results rather independent of temperature, if the tests are carried out wit
hin the working
temperatures of most industrial objects, let us say up to about 200
0
C (see Vrtesy et al. 2010a).
Such behavior of MAT is demonstrated on a series of 5 ductile cast
iron specimens,
solidified with different cooling rates, which in its turn lead to five differen
t microstructures
with different values of Brinell hardness (see Vrtesy et al. 2010b for
details). MAT
measurements were performed on each specimen at two different temperatu
res (20
0
C and
180
0
C) by application of an inspection head. Different rates of change (sl
opes) of the
magnetizing current were used. The specimens were heated by a hot plate and temp
erature
of surface of the specimens was measured by a thermocouple. The obtai
ned optimally
chosen -degradation functions are shown in Fig. 19. As it can be seen
there, linear
150 160 170 180
1
2
3
4
T=20
0
C, slope 0.27 A/s
T=180
0
C, slope 0.27 A/s
T=180
0
C, slope 2.70 A/s
T=180
0
C, slope 15.0 A/s
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Brinell hardness
Fig. 19. The -degradation functions vs. the Brinell hardness measured at 20
0
C and at 180
0
C
temperatures. Different rates of change of magnetizing current (slopes) were use
d in the
latter case.
164
correlation was found between the nondestructively determined optimal -de
gradation
functions and the Brinell hardness of the investigated material, and this correl
ation does not
depend on the temperature as long as slope of the magnetizing current is kept lo
w enough
(see the red and blue lines in Fig. 19). If higher slopes are used,
the influence on the -
degradation functions is negligible at room temperature: the -degradation functio
ns at 20
0
C at
rates 2.7 and 15.0 A/s coincide with the red line (not shown in Fig. 19, but kin
dly compare
with Figs. 14a,b). At temperature 180
0
C another behavior is seen: the -degradation
functions at slopes 2.7 and 15.0 A/s increased their sensitivity with increase o
f the slope (see
the orange and green lines).
3.7 Comparison of MAT descriptors with traditional magnetic parameters
As mentioned in Section 2, Magnetic Adaptive Testing is a magnetic hy
steresis method,
which in contrast to the traditional hysteresis ones makes use of data taken not
only from
the saturation-to-saturation major hysteresis loops of a series of degr
aded ferromagnetic
materials, but collects information from systematically measured families of min
or hysteresis
loops. Such a large system of data is supposed to be able to completely characte
rize (see e.g.
Preisach, 1935, Mayergoyz, 1991, Bertotti, 1998) each of the measured material,
and therefore
it is natural to expect it to contain also those special magnetic da
ta, which vary with
degradation of the investigated material with top sensitivity. The top-
sensitive data can be
those, utilized by the traditional hysteresis approach, but experience
shows that the top
sensitive data are usually different from the traditional ones.
For making a comparison between results of MAT and of a number of o
ther widely used
nondestructive magnetic methods, measurements are considered on two series of di
fferently
shaped samples of different materials plastically deformed by two different ways
. The first
was a series of TRIP steel magnetically open flat long bar samples (
the same series as in
Section 4.2.2) plastically elongated by uniaxial tension. The results o
f this first series are
plotted in Fig. 20a. The other series were low carbon steel magnetically closed
frame shaped
samples (the same series as in Section 3.1) plastically compressed by cold rolli
ng. The results
of this second series are plotted in Fig. 20b. Each of the series w
as measured by several
magnetic methods. For making the comparison possible, the results of e
ach measurement
were normalized by the corresponding value of the reference (not strained) sampl
e. Fig. 20
shows data of the two applied ways of plastic deformation as they ar
e reflected by the
traditional major loop parameters (H
C
and, 1/
MAX
and hysteresis losses, W), by Barkhausen
noise measurement (1/RMS), and by the top sensitive MAT degradation fun
ctions (1/ and
1/). It shows that in both these illustrative cases the traditional hysteresis par
ameters, and
also the Barkhausen data result in qualitatively the same correlations between t
he investigated
magnetic characteristics and the material degradation. It is the first important
conclusion of
these comparative measurements, which shows in general the usefulness and equiva
lency of
magnetic measurements.
On the other hand, this comparison clearly demonstrates that properly
chosen MAT
degradation functions the 1/ -degradation functions in particular reflec
t the material
degradation (at least that due to plastic deformation, as shown in ou
r example) with
substantially higher sensitivity. The MAT-degradation functions are typic
ally much more
165
sensitive than the traditional parameters. The evident reason for this
statement is the fact, that
the MAT parameters are optimized just for the investigated material an
d for the studied
way of degradation. It is worth of mentioning that similarly good results were a
chieved on
other materials and other ways of degradation, as well.
0 10 20 30
0
2
4
6
8
10
(a)
R
e
l
a
t
i
v
e
c
h
a
n
g
e
o
f
m
a
g
n
e
t
i
c
i
n
d
i
c
a
t
o
r
s
Strain due to plastic deformation by uniaxial tension [%]
1/ - MAT
1/ - MAT
1/RMS - Barkhausen
1/
MAX
- major loop
H
C
- major loop
0 10 20 30 40
0
5
10
15
20
(b)
R
e
l
a
t
i
v
e
c
h
a
n
g
e
o
f
m
a
g
n
e
t
i
c
i
n
d
i
c
a
t
o
r
s
Strain due to plastic deformation by cold rolling [%]
1/ - MAT
1/ - MAT
1/
MAX
- major loop
1/RMS - Barkhausen
Hc - major loop
W - major loop
Fig. 20. Comparison of the available MAT most sensitive degradation functions pi
cked up
from the families of minor loops, with other available magnetic parameters. (a):
TRIP steel
open samples elongated by uniaxial tension. (b): Low carbon closed samples compr
essed by cold
rolling.
Higher sensitivity, however, is not the single advantage of the MAT approach as
compared
with the traditional hysteresis and other magnetic ways of NDT. Anothe
r good point for
MAT is its multi-parametric output, which allows not only re-check and
cross-check the
measured results, but in cases of non-monotonous dependencies it is of
ten possible to use
one set of data for clarification of another one, as it is shown e.g. in Section
3.3 or in (Vrtesy
& Tom, 2012). Another, experimentally very friendly feature of MAT is the fact, th
at there
is no need of saturation of the measured samples. This is a very im
portant advantage
especially with magnetically open samples, where to reach saturation is
usually an
extremely difficult task, and the traditional methods are often forced to use mi
nor instead of
major loops anyway and just to assume the differences from the real
major loops are not
substantial (which is often justified, however).
Speaking about the magnetically open samples, one more interesting prop
erty must be
mentioned, namely the unrestricted application of non-magnetic spacers f
or MAT
measurement with magnetizing/sensing inspection heads or with magnetic-fl
ux-short-
cutting yokes. The point is, that for purposes and needs of MAT actu
ally any magnetic
value, which can be parametrized by its field- and/or current-coordinates and wh
ich varies
in a definite way with the investigated degradation of the material,
is acceptable and
applicable for indication of the studied degradation. Use of non-magnet
ic spacers brings
substantial simplification of attitude and easiness of evaluation of te
sting, especially on
samples with not ideal surfaces. See Section 3.5 and (Tom et al., 2012).
Last but not least, systematic measurement of the family of minor loo
ps, with the constant
rate of change of the sample magnetization in particular, makes it po
ssible to use special
ways of extrapolation for rather precise determination of quasi-static
magnitudes of magnetic
166
parameters of the material even on magnetically open samples. This sim
ple idea is
thoroughly described by Uk in his recent paper (Uk, 2010).
4. Examples of application
4.1 Elastic deformation
Application of MAT on elastic deformation of steel samples was investi
gated only to a
limited extent, so far. There is still plenty of space for getting e
xperience in this field. An
unpublished example is shown of a preliminary investigation of the -degradation f
unctions
reflecting uniaxial tension applied on a sample of Mn-Si steel (CSN 13240: Fe, 0
.33-0.41%C,
1.1-1.4%Si, 1.1-1.4%Mn). Fig. 21 shows sketch of a magnetically closed s
ample shape, which
made it possible to clamp the sample by its top and bottom pads int
o jaws of a loading
machine, and to apply equal tensional stress to the two legs of the
sample window. Stress-
strain loading curve of the sample is shown in Fig. 22a. The magneti
zing and the pick-up
coils were wound on the legs and magnetization of the material was t
hus looped in the
closed path around the sample window.
Fig. 21. Magnetically closed steel sample for testing influence of the actively
applied
tensional stress on the shape of the MAT degradation functions. The red rectangl
es show
positions of the magnetizing and pick-up coils on legs of the sample window.
The result of the MAT measurement and evaluation is given in Fig. 22
b. The complicated
shape of the -degradation function reflects presence of at least three effects in
fluencing the
magnetic results: appearance of magnetic anisotropy with the pressure-in
duced easy axis
along direction of the positive pressure (pulling), appearance of magne
tic anisotropy with
the pressure-induced reorientation of the grains in the polycrystalline
material (probably
reorientation of the material crystalline grains so that they get more
and more oriented by
their (111) axis along the direction of the pull; this should very p
robably induce an
anisotropy with a hard axis along direction of the pull, and appearan
ce of dislocations and
variation of the dislocation distribution.
4.2 Plastic deformation
4.2.1 Plastic deformation of austenitic (paramagnetic) steel
4.2.1.1 Cold rolled austenitic steel
The aim of the example presented in this Section is to show applicat
ion of MAT for
characterization of cold rolled austenitic stainless steel specimens (Vrt
esy et al., 2007).
Titanium stabilized austenitic stainless steel, 18/8 type, was studied.
The specimens were
annealed at 1100
0
C for 1 hour. Then they were quenched in water in order to prevent
any
carbide precipitation, and to achieve homogeneous austenitic structure a
s the starting
material. The as-prepared stainless steel specimens were cold-rolled at
room temperature
down to about 60% strain. The compressive plastic deformation of the material in
creased its
167
hardness and at the same time the originally paramagnetic austenite sp
ecimens became
more and more ferromagnetic, due to appearance of bcc o-martensite. Vic
kers hardness,
HV, was measured destructively and nondestructive MAT measurements were
performed
on each specimen by attaching a magnetizing/sensing inspection head on
surface of the
stripe-shaped samples.
0.0 0.1 0.2 0.3 0.4
0
200
400
600
800
1000
(a)
S
t
r
e
s
s
[
M
P
a
]
Strain
0 200 400 600 800 1000
0.4
0.6
0.8
1.0
1.2
1.4
1.6
(b) O
p
t
i
m
u
m
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Stress [MPa]
Fig. 22. (a): Stress-strain loading curve of the sample of the CSN 13240 steel,
(b) The most
sensitive -degradation function (F
i
=280, A
j
=560 A/m) as it depends on the applied stress.
Experimental points on the two curves show stress values at which the sample was
magnetized and measured.
Values of the measured Vickers hardness, HV, together with the most s
ensitive -
degradation function are shown in Fig. 23 in dependence on the plastic deformati
on strain.
Evidently the MAT results offer substantially higher sensitivity and reliability
of indication
of the state of the stainless steel after the deformation than the traditionally
used destructive
hardness measurements.
4.2.1.2 Austenitic steel strained by tensile stress
Austenitic stainless steel SUS316L investigated in the present example
was also plastically
deformed as that in the previous Section 4.2.1.1, however this time not by compr
ession, but
by a tensile stress instead. Flat samples with original dimensions 100x70x3mm
3
were loaded
up to the 50% strain. In contrast to the compressed samples, the tensile deforma
tion did not
introduce such a large percentage of the ferromagnetic phase into the
deformed samples
and neither the signals measured by an attached inspection head nor t
he -degradation
functions gave much hope for sensitive enough magnetic indication of t
he strain values.
However, the B-degradation functions (and also the -degradation functions
not shown
here) gave very acceptable results. The magnetic indication of the str
ain was substantially
more efficient (up to about 30%) if the samples were magnetized along
the direction of the
material prolongation than normal to it (up to a few % only), as it
can be seen in Fig. 24.
However, it is evident, that MAT was applicable even in such an unpromising case
and that
it was also able to reflect anisotropy induced into the material by the stress.
For more details
see (Vrtesy et al., 2011).
168
30 40 50 60
0
5
10
15
30 40 50 60
1.0
1.1
1.2
HV
Strain [%]
O
p
t
i
m
a
l
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
H
V
(
n
o
r
m
a
l
i
z
e
d
)
Strain [%]

HV
Fig. 23. The optimally chosen -degradation function and the Vickers hardness, HV,
vs.
plastic strain of the samples. The inset shows the same values of HV but with mo
re detailed
vertical scale than that of the large plot.
0 20 40
1.0
1.2
1.4
B
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Strain [%]
Parallel
Perpendicular
Fig. 24. The optimal B-degradation functions vs. residual strain of stainless st
eel for parallel
and perpendicular magnetization with respect to the elongation.
4.2.2 Plastic deformation of ferromagnetic steel
Plastic deformation of ferromagnetic steel was investigated by MAT many
times, always
with excellent results. Papers (Stupakov & Tom, 2006, Tom et al., 2006,
Tom et al.,
2009a, 2009b, Vrtesy et al., 2008c) and others can be quoted here, an
d results of a typical
measurement on low carbon steel plastically deformed by uniaxial tension were us
ed for the
description of the MAT method in Section 2 and for investigation of the influenc
e of the rate
of change of the magnetizing field on sensitivity of the MAT degradat
ion functions in
Section 3.4. The results described in those two Sections are connected with the
same material
and similar experiments. MAT description of plastic deformation was als
o studied by
measurements performed on magnetically open and closed samples in Secti
on 3.1. At this
place also a concise information about MAT study of a series of Tran
sformation Induced
Plasticity (TRIP) steel samples as of a specific material is added.
169
Transformation Induced Plasticity steels are modern materials for car i
ndustry with an
excellent combination of ductility and high tensile strength. Typical mechanical
parameters
of TRIP materials are the upper yield strength about 540 MPa, the fr
acture strength about
750 MPa, and the fracture strain up to 30%. Our samples of TRIP 700
type steel with the
original dimensions 150x20x1.18 mm
3
were loaded by tensile stress along their length (the
production rolling direction of the material) and they were magnetized
also along this
direction by an attached inspection head in the MAT tests.
Eighteen samples were deformed to their relaxed strain values from 0%
up to 28.33%.
Vickers hardness (HV) measurements showed a significant increase of har
dness due to the
samples elongation. MAT experiments revealed a sensitive linear correspo
ndence between
the optimal - and
-degradation functions and HV (see Fig. 25) (and/or the strain not
presented here), and also a typical drop of the peak signal value (si
milar to that in Fig. 3b)
between the reference (not deformed) sample and the very first deformed (with st
rain 1.7%
only) sample. The scatter of points observed in Fig. 25 is evidently
due to the untreated
industrial surface of the commercial material together with application
of the attached
inspection head for the MAT measurement. No non-magnetic spacers (by which the s
catter
of points would certainly be reduced) were used in those experiments yet.
220 240 260 280 300 320 340
0
2
4
6
8
10
O
p
t
i
m
a
l
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Vickers hardness, HV
Fig. 25. The linear dependence of the optimal
-degradation function and Vickers hardness
of the TRIP-700 steel plastically deformed by tensile loading.
4.3 Steel embrittlement
Decrease of ductility accompanied by increase of brittleness makes stee
l construction
elements less reliable and more fault-prone. Number of processes exists
, which can induce
such a (usually adverse) change. Among examples of so far tested targ
ets of MAT it is
considered expedient to mention especially measurements of embrittled st
eel aiming at
possible application of the Magnetic Adaptive Testing as a nondestructi
ve method for
checking the state of ductility/brittleness of materials used as steel coats of
pressure vessels
in nuclear reactors. The operational conditions of nuclear pressure vess
el steel, that lead to
such quality degradation include long-term bombardment by neutrons combi
ned with
intense material ageing in the elevated working temperature. As experim
enting with
170
irradiated material requires protected environment of hot laboratories, fi
rst investigation
of applicability of potential methods is usually performed on steel sa
mples, which
underwent simulated not radioactive treatment by artificial thermal agei
ng and artificial
mechanical or thermal embrittlement.
Abilities of MAT to reflect structural modifications of thermally aged
steel (Vrtesy et al.,
2008b), of steel embrittled by cold rolling (Vrtesy et al., 2008a, Tom
et al., 2009a, 2010)
and also of steel embrittled by special thermal processing (Tom et al.,
2011) were
investigated. Only after these experiments with the not radioactive simulated ma
terials, the
measurements proceeded to irradiated pressure vessel steel in a hot la
boratory. Concise
information about these experiments is presented in the following text
in Section 4.3.1
(embrittlement by thermal ageing), Section 4.3.2 (embrittlement by thermal proce
ssing) and
Section 4.3.3 (embrittlement by neutron irradiation). Successful MAT tes
ts of several kinds
of steel embrittled by cold rolling was described in Sections 3.1, 3.7 and 4.2.1
.1.
4.3.1 Steel embrittlement by thermal ageing
Cu precipitates are known to be formed in the Fe metal matrix by th
ermal ageing of Fe-1
wt.% Cu samples at 500
0
C and these precipitates have equally large impact on mechanical
properties of this material as they have in the real pressure vessel steel. Two
series of this Fe-
Cu alloy were annealed at 850
0
C for 5 hours, followed by water quenching. Then one part of
the samples was cold-rolled down to 10% deformation and isothermally aged at 500
0
C for 0,
50, 500 and 5000 min. The other part of samples was aged similarly,
but without the
previous cold rolling (0% deformation). Vickers hardness, HV, of the deformed ma
terial was
measured with the standard Vickers indentation technique. Then 2 mm th
ick ring-shaped
samples were prepared having 18 mm outer and 12 mm inner diameter, r
espectively. The
samples were equipped with a magnetizing coil and a pick-up coil each
. MAT
measurements were performed on those rings.
The most sensitive -degradation functions of the two sample series (with 10% and
with 0%
deformation) were determined and plotted together with the normalized V
ickers hardness
values in Fig. 26. The definite variation of mechanical hardness, HV,
of the aged material
testifies about the structural modifications, leading to observable chan
ge of the
ductile/brittle properties. However, as it is shown in Fig. 26, MAT r
eflects those changes
with substantially higher sensitivity than HV and in contrast to HV,
MAT is monotonous.
Details about the experiment and detailed discussion on the material changes can
be found
in (Vrtesy et al., 2008b).
4.3.2 Steel embrittlement by thermal processing
One of available technologies simulating radiation embrittlement of steel is a s
pecial thermal
processing described in (PHARE, 1995). At this place (see (Tom et al.,
2011) or details)
investigation by MAT is shortly presented of a large series of sample
s of the 15H2MFA
reactor steel (widely used in nuclear plants pressure vessels VVER 440
), which were
artificially embrittled in this way. The magnetic measurement was accom
panied by
destructive Charpy impact tests done on the same samples, so that the
resulting MAT
degradation functions could be plotted in direct dependence on the cri
tical ductile-brittle-
transition-temperature (DBTT) of the material. The samples were prepared
(30 pieces of
171
each embrittled grade) in the shape of rectangular prisms 10 x 10 x 55 mm
3
, with a V-notch
in the middle, applicable for the mechanical Charpy tests. However, be
fore they were
sacrificed for the mechanical destruction, they were nondestructively tested mag
netically by
MAT with the use of soft yokes short-cutting the magnetic flux and w
ith nonmagnetic
spacers (see Section 3.5 and Fig. 15) suppressing scatter of experimen
tal results. The most
sensitive degradation functions proved to be the 1/ - and degradation fu
nctions (as
usual) and the optimal ones are plotted in Fig. 27 as functions of the DBTT.
0 1000 2000 3000 4000 5000
1
2
3
4
5
6
7
0 % deformation - HV
10 % deformation - HV
0 % deformation - MAT
10 % deformation - MAT
O
p
t
i
m
a
l
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
N
o
r
m
a
l
i
z
e
d
V
i
c
k
e
r
s
h
a
r
d
n
e
s
s
Ageing time at 500
0
C [min]
Fig. 26. The most sensitive -degradation functions and the averaged normalized Vi
ckers
hardness values of the thermally aged Fe-Cu alloy in dependence on the ageing ti
me. The
blue curves (, ) describe the samples without previous deformation, the red ones (,
A)
belong to the samples compressed by 10% before start of the ageing.
MAT is shown in this Section to be sensitive enough to reflect variation of brit
tleness of the
nuclear pressure vessel steel. It is essential, however, to keep in mind that th
is test of MAT
on the steel, which was embrittled by thermal processing, reports abou
t ability of MAT to
reflect presence and density and size of defects, which were introduced into the
material just and
only by the applied thermal processing. The thermal processing shifted
DBTT and Vickers
hardness in this steel, but microstructure of this processed steel was certainly
different from
that of the same kind with DBTT shifted about the same (i.e. of the
same mechanical
brittleness), but embrittled by combination of the real fluence of neu
trons and thermal
ageing in a nuclear reactor. The material defects after different trea
tments are substantially
different even though the Charpy impact tests may show identical britt
leness. Magnetic Adaptive
Testing is called adaptive just for the reason that it is able to rea
ct selectively and
differently to different types of defects; it can be optimally adapted
to them. This is
considered to be one of the advantages of the method, and it was ma
ny times confirmed.
Therefore, the present investigation proved MAT to see defects causing m
odification of
brittleness by the thermal processing, similarly as it saw defects modif
ying brittleness of
low carbon steel by cold rolling as reported in (Tom et al., 2009a) o
r in (Takahashi et al.,
2006), and here in Sections 3.1 and 3.7. We can anticipate by analog
y MAT to be able to
see also defects caused by combination of neutron irradiation and thermal ageing
, but this is
all. In order to prove it, real experiments on irradiated/aged samples will have
to be carried
172
out. However, magnetic structuroscopy and its various methods were alread
y
successfully tested at least on some irradiated samples, see e.g. (Tak
ahashi et al., 2006,
Dobman et al., 2008, Vandenbossche, 2009, Gillemot & Pirfo Barroso, 20
10) and the next
Section shows, that also measurements by MAT on irradiated samples wer
e equally
successful.
-40 0 40 80 120 160
1
2
3
4
5
R
e
c
i
p
r
o
c
a
l
1
/
-
a
n
d
d
i
r
e
c
t
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Ductile-brittle-transition-temperature, DBTT, [
0
C]
1/ - MAT
- MAT
Fig. 27. Optimal reciprocal 1/ -degradation function and direct -degradation functi
on of
the 15H2MFA nuclear pressure vessel steel embrittled by the special thermal trea
tment.
Values of the DBTT were taken from destructive Charpy impact tests carried out o
n the
same samples.
4.3.3 Steel embrittlement by neutron irradiation
For studying influence of neutron irradiation, block samples with dimensions 10x
10x30 mm
3
were prepared from three types of nuclear reactor pressure vessel stee
l: JRQ is the
western, 15H2MFA is the eastern type reactor steel, 10ChMFT is the welding metal for
steel 15H2MFA.
The samples were irradiated by neutron fluence in the 1.58x10
19
1.19x10
20
n/cm
2
range
by E>1 MeV energy neutrons. (The value 1.19x10
20
n/cm
2
corresponds to the neutron
fluence, which a pressure vessel coat of a VVER 440 reactor obtains
in approximately 20
years.) Always two samples of each material were irradiated with the
same fluence, two
not-irradiated ones of each material were available as reference samples. For in
vestigation
of irradiated samples by MAT technique a specially designed sample hol
der was built,
very similar to that in Fig. 15. The samples were magnetized by a s
hort solenoid, placed
around the sample and the magnetic circuit was closed from one side
of the sample by a
passive soft magnetic yoke. A plastic dummy yoke was used from the o
ther side, just to
keep the sample fixed in the holder. Application of a single yoke ma
de it possible to
measure each sample independently four times from all four sides of the block. T
hin non-
magnetic spacers (0.08 mm) were placed between the sample and the yok
e, in order to
reduce fluctuation of quality of magnetic contact between the sample s
urface and the
yoke. Measurements were performed in a protected hot laboratory, the sam
ples were
never touched by naked hand.
173
The resulting optimal MAT degradation functions are plotted in Fig. 28
a for each of the
three different materials. The 1/ -degradation functions characterized the
material
modification the best, and very satisfactory correlation was found between the d
egradation
functions and the neutron fluence. The type of correlation is similar for all th
e investigated
materials: the 1/ -degradation functions monotonously increase with increasing amo
unt of
neutron irradiation. Fig. 28b presents how DBTT of the 15H2MFA steel
can be determined
from the 1/ -degradation function of Fig. 28a if a power law formula between the n
eutron
fluence and DBTT (corresponding to typical conditions of a VVER 440 reactor) is
applied for
recalculation. DBTT
0
is the ductile-brittle transition temperature of the 15H2MFA steel
before irradiation.
0 2 4 6 8 10 12
1.0
1.5
2.0
2.5
3.0
(a)
JRQ
15H2MFA
10ChMFT
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Neutron fluence, n/cm
2
(*10
19
)
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6
-60
-50
-40
-30
-20
-10
0
(b)
DBTT
0
= -54.5
0
C
D
B
T
T
,
[
0
C
]
1/ -degradation function
DBTT of 15H2MFA
Fig. 28. (a): The optimally chosen 1/ -degradation functions for the three investi
gated
steels. (b): DBTT of the 15H2MFA steel in dependence on the 1/ -degradation functi
on of
Fig. 28a.
In conclusion, the applicability of MAT was also demonstrated for insp
ection of neutron
irradiation embrittlement of nuclear reactor pressure vessel steels. It should b
e noted at the
same time, that geometry of the investigated samples was rather unfavo
rable for any
magnetic measurement. For next measurements, and for industrial application in p
articular,
another geometry of the samples is recommended. For instance long thin
rods (e.g. 30-50
mm length and 2-3 mm diameter) could be used as surveillance samples
for magnetic
nondestructive tests. Such samples should be measured contactless in a so
lenoid, and much
better results, not complicated by the severe problems with bad surfaces of samp
les having
spent long periods inside a nuclear reactor, can be expected in such case.
4.4 Low cycle fatigue
Here MAT response is described to varied structures of a series of c
ommercial steel CSN
12021 (used for high pressure pipelines), which were modified by appli
cation of low cycle
fatigue. Twelve large flat steel samples were fatigued by cycles with
the constant strain
value 0.25% and the constant strain rate of change 0.3%/s. The numbers of kilocy
cles (kc)
applied to different samples were 0, 2, ..., 20, 22 kc. The samples would be des
troyed near to
24 kc. After application of the relevant numbers of cycles, magnetical
ly closed samples
(rectangles 38x20x5 mm
3
with a hole C10 mm in the middle) were cut-out from the fatigued
middle parts of the plates. The magnetizing and pick-up coils were wo
und directly on the
rectangular samples through the middle hole. An optimum 1/-degradation f
unction is
presented in Fig. 29.
174
0 8 16 24
0.8
1.2
1.6
2.0
2.4
O
p
t
i
m
u
m
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Fatigued samples lifetime, c [kc]
Fig. 29. The most sensitive 1/-degradation function of the low cycle fatigued deg
raded
steel. The error bars show the 5% range, which corresponds to the scatter of magn
etic
properties measured on nominally identical samples of the used commercial steel.
Shape of the degradation functions (both of the 1/-ones plotted here a
nd of the -ones in
(Tom et al., 2010) confirm the expected profile published earlier in (L
o et al., 2000), but
suggesting perhaps more details and certainly more sensitivity. Magnetic
reflection of the
microstructure reveals radical change in the first 15% of the samples lifetime,
corresponding
probably to redistribution of dislocations into a volume configuration
of areas with higher
and lower dislocation densities, which create kind of a cellular pattern
. In the second
phase, up to about 70% of the lifetime, the situation is stabilized,
which is manifested by a
constant level of the degradation functions: the cellular structure of
dislocation bands
persists in a dynamic equilibrium with nucleation and annihilation and
with some
periodical shifting. During the last part of the samples lifetime, microscopical
displacements
of the material lead to appearance of micro-cracks, mostly at the samples surfac
e. High local
stresses at ends of the micro-cracks cause significant changes of the
magnetic degradation
functions again.
These last part variations of the degradation functions are the most
interesting/important
from the practical point of view. If indicated in time, they could s
erve as a warning of the
approaching end of lifetime of the sample, or better of an industrial
object. Shape of the
degradation function in Fig. 29, suggests possible existence of such a
warning. The limited
number of the individual degraded samples did not allow investigation
of this area in as
much details as necessary within the described experiment. Behavior of
the degradation
functions suggests possibly even more substantial variations than in (L
o et al., 2000). The
problem certainly deserves more attention and more experiments in futur
e, with focus on
the last part of the samples lifetime. Beneficial assistance of nonmag
netic spacers together
with use of inspection heads on magnetically open fatigued samples sho
uld yield
satisfactory results.
4.5 Decarburization of steel surface
If a steel object is exposed to normal air atmosphere at elevated te
mperature for an
appreciable period of time, unwanted softening of the steel surface by
decarburization can
appear. The surface is modified into a layer of practically pure iron the ferrit
e and fatigue
lifetime of such an object is dramatically shortened. Existence and al
so thickness of the
175
magnetically soft surface decarburized layer can be detected by the classical MA
T approach,
of course, as it can be found in (Skrbek et al., 2011). It will be shown in this
Section, however,
that - as suggested by (Perevertov et al., 2011) - in the case of
decarburized surfaces a
modified MAT line of attack is possible with advantage. It is based
on the idea that the
decarburized surface ferrite layer gets magnetically saturated in substa
ntially lower applied
fields than rest of the sample, and that value of the saturated magn
etic flux of each ferrite
layer should be proportional to its thickness. Principles and straightf
orward results of this
MAT attitude will be shortly described here, whereas its details can
be found in
(Tom et al., 2011).
The modified MAT tests were carried out by magnetization of ring samples of carb
on spring
steel 54SiCr6 (D/d/w = 60/54/3 mm, D outer diameter, d inner diameter, w axial w
idth).
All the samples were annealed at 800
0
C in air atmosphere for T = 0, 1, 4, 8, and 20 hours,
respectively. A magnetizing and a pick-up coil were wound on each of
the samples. The
measured voltage signals of each sample were numerically integrated, and the int
egral at a
minor loop amplitude A
j
, measured on a sample annealed in air for the time-period, T, was
denoted as the corresponding maximum magnetic flux u
T
(A
j
). Indicating u
0
(A
j
) such
maximum magnetic flux of the reference sample, differences (u
T
- u
0
)(A
j
) were plotted as
functions of the loop amplitudes, A
j
, in Fig. 30a.
These reduced magnetic fluxes, (u
T
- u
0
), of the decarburized surfaces were plotted for each
loop with amplitude A
j
in dependence on values of the optically measured thickness, t, of
the surface ferrite (t was measured independently destructively on auxi
liary samples
sections). It turned out, that the plots (u
T
- u
0
)(t) were very linear within surprisingly broad
range of A
j
. Fig. 30b gives an example of such a linear relation for the minor loop amplitu
de
A
j
= 1 kA/m, and similar linear fits were obtained in all the investigated range o
f the minor
loop amplitudes. Fig. 30c shows the correlation coefficient, R, of such linear f
its.
In conclusion it can be stated that modified MAT measurement of suita
ble minor hysteresis
loops of ferromagnetic steel objects can serve as a fast, convenient and nondest
ructive method
for determination of the surface decarburization level. Provided, of co
urse, that a reference
sample and at least one object of the same material, with an indepen
dently measured
appreciable thickness of the decarburized layer is available so that t
he necessary calibration
line can be prepared. Range of the appropriate minor loop magnetizing amplitudes
was shown
to be broad, the measurement simple and not requiring any sophisticated equipmen
t.
0.0 1.0 2.0 3.0
0
2
4
6
(a)
(u
1
- u
0
)
(u
4
- u
0
)
(u
8
- u
0
)
(u
20
- u
0
)
M
a
x
i
m
u
m
f
l
u
x
v
a
l
u
e
s
(
u
T
-
u
0
)
(
A
j
)
[
a
.
u
.
]
Minor loops amplitude, A
j
[kA/m]
0 100 200 300 400
0
2
4
6
(b)
M
a
x
i
m
u
m
r
e
d
u
c
e
d
f
l
u
x
a
t
A
j =
1
k
A
/
m
(
u
T
-
u
0
)
(
t
)
[
a
.
u
.
]
Thickness of ferrite, t [m]
(u
T
- u
0
)(t), A
j
= 1kA/m
The best linear fit
0.0 1.0 2.0 3.0

0.7
0.8
0.9
1.0
(c)
C
o
r
r
e
l
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
,
R
Minor loops amplitude, A
j
[kA/m]
Fig. 30. (a): The maximum magnetic flux values of the samples reduced by the max
imum
flux value of the reference sample, (u
T
- u
0
)(A
j
), at minor loops with amplitudes A
j
. (b): The
reduced maximum magnetic flux values, (u
T
- u
0
), of the air-annealed samples, at the minor
loop with amplitude A
j
=1 kA/m, as a function of thickness, t, of the surface decarburized
ferrite. (c): Correlation coefficient, R, of the linear regression fit with (u
T
- u
0
)(t), for the
used range of the minor loop amplitudes, A
j
.
176
4.6 Welding quality tests
Point welding is a frequently used technology in industry, and checking quality
of welding
is very important. Welding process produces serious phase transformation
in the welded
area of the material, which is accompanied by modification of magnetic character
istics. It is
shown in this Section how MAT can be applied for inspection of point welding of
car-body
sheets.
Thickness of the point-welded sheets was 1.2 mm, the material was cold
rolled low carbon
steel. Two series of welded points were produced by 2.4 bar pressure
welding and the
welding current was modified from 15.6 up to 21.8 kA. Diameter of the welded poi
nts was
around 7 mm. A MAT-checked row of welded points is shown at the bot
tom of Fig. 31a, a
destructively tested control series of welded points can be seen at the top of F
ig. 31a. Result
of the destructive tests confirmed, that welded points made by the welding curre
nt between
18.5 and 20 kA were proper. The points with lower current were not
welded strongly
enough, the points above 20 kA were over-welded (over-welding damaged the materi
al).
MAT measurements were performed on each welded point with a magnetizin
g yoke. The
result of the measurement is shown in Fig. 31b, where the optimally chosen 1/-deg
radation
function vs. the welding current can be seen. It is worth to mention
, that even a small
change of the welding current results in a measurable modification of the MAT de
gradation
functions at the beginning of the curve, and this becomes more pronou
nced in the critical
region from 18.5 to 20 kA. All the MAT degradation functions were no
rmalized by the
corresponding value of a reference spot (where no welding was taken).
More than 100%
difference was found between the MAT degradation functions corresponding to the
lowest
and to the highest energy of welding, which makes possible to charact
erize quality of the
welding quite easily.
16 18 20 22
1.0
1.5
2.0
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Welding current, I [kA]
Fig. 31. (a): Two rows of welded points, each produced by a different welding cu
rrent. The
upper row shows results of the destructive test. (b): The optimally chosen 1/-deg
radation
function vs. the welding current, measured on the bottom row of welded points sh
own in
figure (a).
Choice of the optimal degradation functions is not critical, the top
sensitivity area of 1/-
degradation functions in the sensitivity map is very large. For the n
ext possible industrial
application also tolerance towards of the results with respect to exac
t positioning of the
177
magnetizing yoke was investigated. The same welded spot was measured s
everal times
after each other, the position of the yoke with respect to the welded spot was m
odified, and
MAT evaluation was performed. The results are seen in Fig. 32, where
three characteristic
cases are shown. In the left case the welded spot was in the central area of the
yoke, in the
middle case only half area of the welded spot was under the yoke, and in the rig
ht case the
welded spot was completely outside. The figure shows how the 1/-degrada
tion functions
values were changing compared with the central position of the yoke.
Even the most
unfavorable case caused 3% difference in the MAT degradation function value only
.
0 1 2
1.00
1.04
1.08
Point welding
Magnetizing yoke
Position of the magnetizing yoke [a.u.]
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Fig. 32. Variation of the MAT 1/-degradation functions, if the measurement was pe
rformed
on the same welded spot at three characteristic positions of the yoke with respe
ct to the
welded spot, as indicated in the figure.
4.7 Search for flaws
Detection and evaluation of thickness reduction of pipes are very impo
rtant issues for
prediction of lifetime of pipelines in order to avoid severe accidents
(Ju, 2007). Recently
many nondestructive testing techniques have been used for the measureme
nt of pipe wall
thinning, but none of them gives satisfactory result in the case of
ferromagnetic carbon
steels, especially if the steel is thick or if it is layered. In this Section it
is shown how Magnetic
Adaptive Testing can be applied for inspection of wall thinning in la
yered ferromagnetic
materials. Experiments were performed on a system of ferromagnetic carb
on steel plates,
one of them containing an artificial slot (see also (Vrtesy et al., 2012b)).
Preliminary experiments verified, that samples with different total thic
kness give different
signals from the MAT inspection head. Evidently similar modifications of the sig
nal can be
expected if thickness of the sample or of one of the layered samples is modified
only locally.
It was simulated by three plates (180x170x3 mm
3
) made of a carbon steel, with one of them
containing a 9x2 mm
2
slot in the middle. Fig. 33 shows the case if all the three sheets are put
together. The samples were measured in such a way, that they were ma
gnetized by an
inspection head from the top side, while the slot was placed at the
bottom. The
measurements were performed on both double and triple configurations (Fig.
34), and
the head was attached in both parallel and perpendicular orientations
with respect to the
axis of the slot. The head was moved on the top sample surface alon
g a direction
perpendicular to the axis of the slot. Distance, x, of the center of the head fr
om the middle of
178
the slot was the independent parameter. In this case cross-section of
the magnetic yoke of
the inspection head was 10.5x16 mm
2
, the total outside length 31 mm, and the total outside
height of the bow 31 mm. The distance between the legs was 10 mm.
Fig. 33. Measurement of the triple sheet with perpendicular orientation of the y
oke.
Results of the measurements, performed on the double sheet system and
sketch of the
experimental arrangement are shown in Fig. 34a. The yoke was oriented
perpendicular to
the axis of the slot in this case. The optimally chosen -degradation
functions were
normalized by the value, measured at the largest distance from the center of the
slot (x = - 60
mm). A very similar result was obtained if the yoke was oriented parallel with t
he slot axis
and moved similarly as in the former case. The measurement was perfor
med in a triple
sheet configuration, too, with the results shown in Fig. 34b. The exp
eriments revealed that
Magnetic Adaptive Testing is a promising tool for nondestructive detect
ion of thinning of
steel slabs from the other side of the specimen, and that it is able to give goo
d results even
for a layered ferromagnetic material.
-80 -60 -40 -20 0 20 40 60 80
0.8
1.0
1.2
1.4
(a)
O
p
t
i
m
a
l
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Position of yoke s center from axis of the defect, x [mm]
-80 -60 -40 -20 0 20 40 60 80
0.8
1.0
1.2
1.4
(b)
O
p
t
i
m
a
l
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Position of yoke s center from axis of the defect, x [mm]
Fig. 34. Optimal MAT degradation functions in dependence on the inspection head
position
for perpendicular configuration in a (a) double sheet system, (b) triple sheet s
ystem.
179
4.8 Cast iron
4.8.1 Ductile cast iron
Special series of ductile cast iron samples was investigated (see also
(Vrtesy et al., 2010c))
Staircase-shaped samples were prepared, where different cooling rates resulted i
n different
structures. Correlations were studied of the nondestructively measured M
AT degradation
functions with the destructively determined Brinell hardness and also w
ith the electrical
conductivity of the samples. Four staircase-shaped block specimens with
5 steps each were
prepared. (One of these samples was used here earlier for temperature
dependence
investigations, see Section 3.6.) Chemical composition of the cast iron
samples is shown in
Table 1. The amount of ferrosilicon and of commercial pure iron in the melt was
controlled,
in order to get different carbon equivalents. The targeted carbon equivalent of
sample 1 was
set to a low value, in order to have cementite in the matrix. The
targeted chemical
compositions of samples 2, 3 and 4 were the same, but they were giv
en different heat
treatments, in order to have different matrices.
Sample C Si Mn P S Mg
1 3.05 2.041 0.166 0.049 0.016 0.024
2 3.58 2.592 0.127 0.065 0.016 0.025
3 3.46 2.575 0.128 0.066 0.015 0.025
4 3.43 2.523 0.130 0.064 0.016 0.026
Table 1. Chemical composition of the ductile cast iron samples (values in wt%).
These staircase-shaped block specimens were used for magnetic measuremen
t: each step
was measured. The samples were magnetized by a magnetizing inspection
head, attached
on the surface. The optimally chosen B-degradation functions of the sa
mples are shown in
Fig. 35 in dependence on the Brinell hardness and on the electric conductivity.
All the steps
of all the measured (four) samples are taken into account within this
figure. (Fig. 19 in
Section 3.6 shows the dependence of a -degradation function on the hardness for s
ample 4.)
The degradation function values of different samples were normalized by
the sample with
the lowest hardness.
150 200 250 300
1.0
1.5
2.0
2.5
(a)
B
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Brinell hardness, HB
16 18 20
1.0
1.5
2.0
2.5
(b)
2.5
2.0
1.5
1.0
B
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Conductivity [10
5
S/m]
sample 1
sample 2
sample 3
sample 4
Fig. 35. The most sensitive B-degradation functions in dependence on Brinell har
dness (a)
and on conductivity (b).
180
The Brinell hardness and the MAT degradation functions correspond to e
ach other very
well, linear correlation was found. It should be emphasized that the pr
esented relationship
was obtained when all the samples (with different chemical compositions
and preparation
conditions) were considered for the evaluation. (Considering the steps
within each single
sample, the scatter of points is significantly smaller, as shown in F
ig. 19). In any case the
reliability and reproducibility of the MAT degradation functions are go
od, as determined
from the sensitivity maps. A good correlation was also found between the MAT deg
radation
functions and conductivity, but in this case the samples with different chemical
composition
should be handled separately, as can be seen in Fig. 35b. The optima
lly chosen MAT
degradation functions of samples 2, 3 and 4, which have rather simila
r compositions, lie
more-or-less on a straight line as a function of conductivity. This is also vali
d for sample 1,
but with a different slope.
4.8.2 Flake graphite cast iron
Flake graphite cast iron materials with various graphite structures and
matrices were
prepared and their magnetic properties were measured and evaluated syst
ematically.
Correlation of the nondestructively measured MAT parameters has been st
udied with the
destructively determined Brinell hardness, the pearlite/ferrite ratio and
the graphite
morphology. Three flake graphite cast iron materials with different che
mical
compositions listed in Table 2 were prepared. Disks, prepared of these
materials were
subjected to two kinds of heat treatment: annealing (AN) to obtain a
ferrite-based matrix
and normalization (NR) to obtain a pearlite-based matrix. To treat the
se annealed and
normalized disks, they were heated at 1123 K in a furnace for one h
our and then either
cooled in the furnace or in air for AN and NR, respectively. Altoget
her 9 flake graphite
cast iron materials with various matrices and graphite shapes were thu
s produced. After
grinding the specimen surfaces, their Brinell hardness, HB, was measured confirm
ing that
the normalizing and annealing treatments were successful in producing t
he fully ferritic
and fully pearlitic matrices, respectively. Metallographic examination al
so revealed, that
the as-cast samples (samples without the heat treatments) had different
ly mixed
pearlite/ferrite matrices.
Material
Chemical composition (mass%)
C Si Mn P S Cr Ti
CE4.7 3.77 2.78 0.78 0.025 0.015 0.029 0.015
CE4.1 3.36 2.15 0.69 0.018 0.010 0.014 0.011
CE3.7 3.13 1.66 0.72 0.017 0.020 0.038 0.010
Table 2. Chemical composition of flake graphite cast iron samples (values in wt%
).
The samples were magnetized by an attached inspection head for the MA
T investigation
and degradation functions of all the investigated samples were evaluated. The op
timal 1/-
degradation functions are plotted against the Brinell hardness in Fig. 36.
181
90 100 110 120 130
1.0
1.2
1.4
1.6
1.8
annealed
as cast
normalized
Sample CE4.7
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Brinell hardness
100 120 140 160 180 200 220
1.0
1.2
1.4
1.6
1.8
2.0
2.2
annealed
as cast
normalized
Sample CE4.1
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
s
Brinell hardness
120 140 160 180 200 220
1.0
1.2
1.4
1.6
1.8
annealed
as cast
normalized
Sample CE3.7
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Brinell hardness
Fig. 36. Optimally chosen 1/ -degradation functions of the sample series CE4.7, C
E4.1 and
CE3.7 vs. Brinell hardness.
The applied annealing resulted in completely ferrite-based matrices (the 0% pear
lite/ferrite
ratio) and the normalization resulted in completely pearlite-based ones
(the 0%
pearlite/ferrite ratio). The pearlite/ferrite ratio of the as-cast sampl
es was different for
different sample series, and it was determined by metallographic examin
ation. The
pearlite/ferrite ratio and the average length of graphite flakes were
evaluated using an
image processing software. Correlation between the optimal 1/-degradation
function and
the pearlite/ferrite ratios is given in Fig. 37.
0 20 40 60 80 100
1.0
1.2
1.4
1.6
all samples annealed
normalized
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Pearlite/ferrite ratio
Fig. 37. Optimally chosen 1/-degradation function of all the samples vs. perlite/
ferrite
ratio. The annealed and the normalized samples are indicated.
Graphite area and graphite length characterize the graphite structure o
f the investigated
samples. MAT degradation functions were evaluated as functions of these quantiti
es for the
three as-cast samples, and the corresponding values were compared with
each other.
Correlation between the magnetic parameters and the graphite area can be seen in
Fig. 38a,
while correlation between the magnetic parameters and the graphite leng
th is plotted in
Fig. 38b.
Concluding the observations, closely linear correlation was found betwee
n the MAT
degradation functions and the Brinell hardness. It is valid if influence of the
heat treatments
was investigated within the series of samples with the same chemical
composition (as
shown in Fig. 36), but also if influence of chemical composition was studied for
different as-
cast samples, and also if even all the samples were considered within the same g
raph (these
182
last two cases are not shown here). This confirms the expectation tha
t magnetic hardening
correlates with the mechanical hardening of cast iron very well. This is why MAT
is also an
excellent tool for determining the pearlite/ferrite ratio. Degradation f
unctions of all 0%
pearlite/ferrite ratio samples are very close to each other, the diffe
rence is not larger than
the error of measurement, and the same holds for the normalized sampl
es (100%
pearlite/ferrite ratio case). The difference of about 50% in magnitude
of the shown 1/-
degradation function is more than enough for determination of variation
s between the
pearlite/ferrite ratios from 0% to 100% (see Fig. 37). The larger sca
tter in Fig. 37 of the
degradation functions evaluated on the as-cast samples can be probably
attributed to
inaccurate determination of the pearlite/ferrite ratio, which was rather
estimated from
microphotographs of their structure. MAT was also found to be a usefu
l tool for
measurement of parameters of graphite morphology. Very satisfactory line
ar correlation
between optimally chosen MAT degradation functions and both the graphite length
and the
graphite area was found on the as-cast samples in Fig.38. The correla
tion between MAT
degradation functions and the graphite morphology is evident even if a
ll the samples are
considered, however in this case the scatter of points is larger.
10 12 14 16 18
1.0
1.5
2.0
2.5
(a)
as cast samples
CE3.7
CE4.1
CE4.7
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Graphite area [%]
30 40 50 60 70
1.0
1.5
2.0
2.5
(b)
CE3.7
CE4.1
CE4.7 as cast samples
O
p
t
i
m
a
l
1
/
-
d
e
g
r
a
d
a
t
i
o
n
f
u
n
c
t
i
o
n
Graphite length [m]
Fig. 38. Optimally chosen 1/ -degradation functions of the as-cast samples vs. th
e graphite
area (a) and vs. the graphite length (b).
5. Conclusions
An original magnetic nondestructive method of Magnetic Adaptive Testing
is presented in
this chapter as a sensitive option to other ways of investigation of modificatio
n of properties
of ferromagnetic construction materials. Physical basis, principles and
ways of application
of the method are explained in the first and second Sections, suggestions with r
espect to its
efficient use are given in Section 3, and Section 4 is dedicated to
several examples of
measurements, where MAT was utilized, usually with very satisfactory results.
MAT belongs to the family of nondestructive magnetic hysteresis materia
l testing methods,
which generally investigate structural modifications of bulk of the tes
ted objects, and it is
shown to surpass sensitivity of the traditional hysteresis methods main
ly due to its specific
selection from the recorded large data-pool of that part of information, which i
s optimal with
respect to the investigated material and to the way of its degradation.
183
The most advantageous features of Magnetic Adaptive Testing were demonstrated th
rough
many successful measurements on many different materials, which underwen
t different
types of degradation (fatigue, annealing, plastic deformation, neutron i
rradiation, etc.).
MAT generally offers significantly larger sensitivity with respect to
any degradation
parameter than any of other magnetic nondestructive and probably even
of any existing
destructive method. A very important, advantageous feature of the MAT technique
is that it
does not require magnetic saturation of investigated samples. The most
appropriate MAT
descriptors of each type of degradation are chosen from a big data-po
ol, and this multi-
parametric choice frequently makes possible to characterize the material
degradation in a
more complex way, than other methods. It was also shown that the opt
imally chosen
degradation functions are reliable and repeatable. Magnetic Adaptive Tes
ting is a
comparative measurement, it does not offer any absolute magnetic quantities. Ins
pection of
any unknown sample is based on a previous teaching process.
The most serious drawback of this technique is that in great majority
of practical
applications the MAT inspection is required to operate on magnetically
open industrial
objects, where shape of the object (through the demagnetization factor) and cond
ition of its
surface can introduce uncertainty into the measurement. However, it was
shown that
qualitatively the same correlation exists between the MAT degradation f
unctions and the
degradation variables both for open and for closed magnetic circuits (
Section 3.1, and/or
3.2), and also that the surface problem can be treated rather successfully (Sect
ion 3.5).
Main directions promising to contribute to further development of the MAT applic
ation are
probably a next improvement of MAT inspection heads for contact measur
ement on
surfaces of industrially interesting objects, namely in finding a satis
factory elimination of
fluctuation of quality of magnetic contact between the inspection head
and surface of the
object, which is frequently uneven, rough and deteriorated. An importan
t step in this
direction was achieved by application of non-magnetic spacers, which wa
s described in
Section 3.5 and successfully applied in several latest measurements. Pe
rhaps a simple
measurement of a parameter proportional to the real value of the magn
etizing field inside
the object magnetized from its surface could do the trick.
As for the most promising next industrial application of MAT, the non
destructive
surveillance program for irradiated steel of nuclear pressure vessel re
actor steel (Section
4.3.3) should be pointed out, and quantitative measurement of steel surface deca
rburization
(Section 4.5). Probably also fatigue tests of periodically loaded objec
ts could give very well
applicable results, which, however, still require broad next experimental study,
substantially
more detailed than the first attempt mentioned in Section 4.4.
6. Acknowledgements
The authors if this chapter would like to express their gratitude to
all the colleagues, who
were co-authors of the quoted papers and also to those who took part at preparat
ion of the
presented experiments. Special thanks aim at the members of the Univer
sal Network for
Magnetic Nondestructive Evaluation, the annual workshops of which made
a very fruitful
and supporting platform for discussions and inspiration. The work was
financially
supported by the Czech Institute of Physics, v.v.i. of ASCR (project AVOZ 101005
20) and by
the Hungarian Research Institute for Technical Physics and Materials Science of
HAS, by the
184
Hungarian-Czech and by the Hungarian-Japanese Bilateral Intergovernmental
S&T
Cooperation projects, by the project No.101/09/1323 of the Grant Agency
of the Czech
Republic, and by the Hungarian Scientific Research Fund (projects T-0
35264, K-62466 and
A08 CK 80173).
7. References
Bertotti G. (1998) Hysteresis in Magnetism, Academic Press, San Diego
Blitz J. (1991) Electrical and Magnetic Methods of Nondestructive Testi
ng, Adam Hilger IOP
Publishing, Ltd, Bristol
Devine M.K. (1992) Magnetic detection of material properties (1992) J.
Min. Met. Mater.
(JOM), pp. 24-30
Devine M.K., Kaminski D.A., Sipahi L.B. Jiles D.C. (1992) Detection of
fatigue in structural
steels by magnetic property measurements, Journal of Materials Engineeri
ng and
Performance, Vol. 1, pp. 249-253
Dobmann G., Altpeter I., Kopp M., Rabung M., Hubschen G. (2008) ND-ma
terials
characterization of neutron induced embrittlement in German nuclear reac
tor
pressure vessel material by micromagnetic NDT techniques, Electromagnetic
Nondestructive Evaluation (XI), p.54, IOS Press, ISBN 978-1-58603-896-0
Gillemot F., Pirfo Barroso S. (2010) Possibilities and difficulties of
the NDE evaluation of
irradiation degradation, Proceedings of 8th International Conference on
Barkhausen
Noise and Micromagnetic Testing (ICBM8) (ISBN 978-952-67247-2-0)
Jiles D.C. (1988) Review of magnetic methods for nondestructive evaluat
ion, NDT
International Vol. 21, p. 311
Jiles D.C. (2001) Magnetic methods in nondestructive testing, In: Encyc
lopedia of Materials
Science and Technology, K.H.J. Buschow et al., Eds, p.6021, Elsevier Press, Oxfo
rd
Ju, Y. (2007) Remote measurement of pipe wall thinning by microwaves,
Advanced
Nondestructive Evaluation II, Volume 2 pp. 1128-1133
Lo C.C.H., Tang F., Biner S.B., Jiles D.C. (2000) Effects of fatigue-
induced changes in
microstructure and stress on domain structure and magnetic properties o
f Fe-C
alloys, J. Appl. Phys., Vol. 87, p. 6520
Mayergoyz I.D. (1991) Mathematical Models of Hysteresis, Springer-Verlag,
Berlin PHARE
PH2.01/95 Project Reactor Pressure Vessel 1995
Preisach F. (1935) Uber die magnetische Nachwirkung, Zeit. fur Physik Vol. 94 p
. 277
Perevertov O. (2009) Increase of precision of surface magnetic field m
easurements by
magnetic shielding, Meas. Sci. Technol., Vol. 20, p. 055107
Perevertov O., Stupakov O., Tom I., Skrbek B. (2011) Detection of sprin
g steel surface
decarburatization by magnetic hysteresis measurements, NDT&E International Vol.
44 pp. 490494
Skrbek B., Tom I., Kadlecov J., Ganev N. (2011) NDT Characterization of Decarburiza
tion
of Steel after Long-Time Annealing, Kovove Materialy Metallic Materials
, Vol. 49,
No.6, pp.401-409
Stupakov O. (2006) Investigation of applicability of extrapolation metho
d for sample field
determination in single-yoke measuring setup, J. Magn. Magn. Mater. Vol
. 307 pp.
279-287
Stupakov O., Tom I., Kadlecov J. (2006) Optimization of single-yoke magnetic testin
g by
surface fields measurement, J.Phys. D Vol. 39, pp. 248-254
185
Stupakov O., Tom I. (2006) Hysteresis minor loop analysis of plastically
deformed low-
carbon steel, NDT&E Int. Vol. 39 pp. 554-561
Takahashi S., Kobayashi S., Kikuchi H., Kamada Y. (2006) Relationship between me
chanical
and magnetic properties in cold rolled low carbon steel, J. Appl. Phy
s., Vol. 100,
p.113908
Takahashi S., Kikuchi H., Ara K., Ebine N., Kamada Y., Kobayashi S.,
Suzuki M. (2006) In
situ magnetic measurements under neutron radiation in Fe metal and low
carbon
steel, J. Appl. Phys. Vol. 100, p.023902
Tom I. (2004) Non-Destructive Magnetic Adaptive Testing of Ferromagnetic
Materials, J.
Magn .Magn. Mater. Vol. 268/1-2, pp. 178-185
Tom I., Stupakov O., Kadlecov J., Perevertov O. (2006) Magnetic Adaptive Testing lo
w
magnetization, high sensitivity assessment of material modifications, J.
Magn.
Magn. Mater. Vol. 304 pp.168-171
Tom I., Vrtesy G., Kobayashi S., Kadlecov J., Stupakov O. (2009a) Low-car
bon steel
samples deformed by cold rolling analysis by the magnetic adaptive te
sting, J.
Magn. Magn. Mater., Vol. 321 pp. 2670-2676
Tom I., Vrtesy G., Kadlecov J. (2009b) Influence of rate of change of m
agnetization
processes on sensitivity of Magnetic Adaptive Testing, J. Magn. Magn.
Mater. Vol.
321 pp. 1019-1024
Tom I., Vrtesy G., Kadlecov J. (2009c) Influence of rate of change of m
agnetization
processes on sensitivity of Magnetic Adaptive Testing, J.Magn. Magn. Ma
ter. Vol.
321 pp. 1019-1024
Tom I., Kadlecov J., Vrtesy G., Skrbek B. (2010) Investigation of Structural Modific
ations
in Ferromagnetic Materials by Magnetic Adaptive Testing, Proceedings of
8th
International Conference on Barkhausen Noise and Micromagnetic Testing (
ICBM8),
pp.153-162 (ISBN 978-952-67247-2-0)
Tom I., Kadlecov J., Konop R., Dvokov M. (2011) Magnetic nondestructive indication
of varied brittleness of 15Ch2MFA steel, Proceedings of 9th International Confer
ence on
Barkhausen Noise and Micromagnetic Testing (ICBM9) ISBN 978-952-67247-4-4
(paperback), ISBN 978-952-67247-5-1 (CD-ROM), pp. 55-63
Tom I., Kadlecov J., Vrtesy G. (2012) Measurement of flat samples with r
ough surfaces
by Magnetic Adaptive Testing, IEEE Trans. Magn., in press
Uk E. (2010) A new approach to the evaluation of magnetic parameters f
or non-
destructive inspection of steel degradation, Journal of Electrical Engineering,
Vol. 61.
pp. 100-103
Vandenbossche L. (2009) Magnetic Hysteretic Characterization of Ferromagn
etic Materials
with Objectives towards Non-Destructive Evaluation of Material Degradatio
n,
Gent University, Gent, Belgium, PhD Thesis
Vrtesy G., Tom I., Mszros I. (2007) Nondestructive indication of plastic deformation
of
cold-rolled stainless steel by magnetic adaptive testing, J. Magn. Magn.
Mater. Vol.
310 pp. 76-82
Vrtesy G., Tom I., Takahashi S., Kobayashi S., Kamada Y., Kikuchi H. (2008a) Inspec
tion
of steel degradation by Magnetic Adaptive Testing, NDT & E INTERNATION
AL,
Vol. 41. pp. 252-257
186
Vrtesy G., Tom I., Kobayashi S., Kamada Y. (2008b) Investigation of ther
mally aged
samples by Magnetic Adaptive Testing, Journal of Electrical Engineering,
Vol. 59.
pp.82-85
Vrtesy G., Tom I., Takahashi S., Kobayashi S, Kamada Y, Kikuchi H., (200
8c) Inspection
of steel degradation by Magnetic Adaptive Testing, NDT & E INTERNATION
AL,
Vol. 41. pp. 252-257
Vrtesy G., Uchimoto T. Tom I., Takagi T. (2010a) Temperature dependence
of magnetic
descriptors of Magnetic Adaptive Testing, IEEE Trans..Magn., Vol. 46. pp. 509-51
2
Vrtesy G., Uchimoto T. Tom I., Takagi T. (2010b) Nondestructive character
ization of
ductile cast iron by Magnetic Adaptive Testing, J. Magn. Magn. Mater. Vol. 322
pp.
3117-3121
Vrtesy G., Uchimoto T., Takagi T., Tom I. (2010c) Nondestructive inspecti
on of ductile
cast iron by measurement of minor magnetic hysteresis loops, Material
s Science
Forum, Vol. 659. pp. 355-360
Vrtesy G., Ueda S., Uchimoto T., Takagi T., Tom I., Vrtesy Z. (2011) Evaluation of P
lastic
Deformation in Steels by Magnetic Hysteresis Measurements, In: Electroma
gnetic
Nondestructive Evaluation (XIV), T. Chady et. al., Eds., pp. 371-378,
IOS Press,
Amsterdam
Vrtesy G., Tom I. (2012a) Complex characterization of degradation of ferr
omagnetic
materials by Magnetic Adaptive Testing, IEEE Trans. Magn., in press
Vrtesy G., Tom I. Uchimoto T., Takagi T. (2012b) Nondestructive investiga
tion of wall
thinning in layered ferromagnetic material by magnetic adaptive testing, NDT & E
INTERNATIONAL, Vol 47. pp. 51-55
Part 3
Concrete Nondestructive Testing Methods
8
Elastic Waves on Large
Concrete Surfaces for Assessment of
Deterioration and Repair Efficiency
D. G. Aggelis
1
, H. K. Chai
2
and T. Shiotani
3
1
Materials Science & Engineering Department, University of Ioannina,
2
Department of Civil Engineering, Faculty of Engineering, University of Malaya
3
Graduate School of Engineering, Kyoto University
1
Greece
2
Malaysia
3
Japan
1. Introduction
Most of societys infrastructure supporting certain sectors of human acti
vity is based on
cementitious materials. Bridges, highways, water intake facilities and o
ther structures are
made of concrete. These structures sustain external function loads, own
weight, as well as
deterioration by temperature cycles and attack of environmental agents during
their useful
life span. The number of civil infrastructures built more than 50 years ago may
be estimated
to several hundreds of thousands worldwide (Chai et al. 2010). The operational e
fficiency of
these structures is of primary importance for economic reasons but mostly for hu
man safety.
Concrete structures have ceased to be considered maintenance-free. They
should be
inspected in regular intervals, their damage level should be evaluated and when
necessary,
repair action should be applied. Maintenance or repair projects should
be based on
prioritization as to the importance of the structure and its damage s
tatus. Therefore,
economic, fast and reliable characterization schemes are highly demanded
. In most of the
currently available methods of assessment, elastic modulus of material
and strength
characteristics are the primary criteria of consideration. In order for
the information to be
obtained, it often becomes necessary that mechanical tests are conducte
d on samples
acquired from the target structure, through extraction of cores or oth
er similar exercises,
which turn out to be inflicting further damage to the already-degraded
target structure. In
addition, considering the nature of these exercises, which are usually
adopted at selected
locations, the assessment results are local and often not representativ
e for the overall
structure. Recently, extensive research has been reported in elastic wave-relate
d techniques
resulting in reliable assessment of the structural condition of actual
concrete materials and
structures. The characterization of concrete structures by Non Destructi
ve Testing (NDT)
and specifically stress wave methods produces mainly qualitative but very import
ant results
concerning damage. The surface layer of concrete suffers the most of
environmental
degradation as well as maximum stresses particularly by flexural loadin
g. It is reasonable
therefore, that considerable deterioration initiates from the surface. T
he deterioration is
often observed in the form of small-scale cracking or macroscopic surf
ace breaking cracks
which can propagate to the interior and accelerate the deterioration of the whol
e structure,
especially if the embedded metal reinforcement is exposed to various environment
al agents.
According to the results, proper repair action can be taken and therefore, the f
unctionality of
a structure can be extended. This is extremely important bearing in m
ind that the cost of
building new infrastructure is much higher compared to a focused repair project
aiming at
extending the service life of the structure for decades.
The most widely used feature in elastic wave NDT is pulse velocity, which is cor
related with
the damage degree. Despite its rough results, pulse velocity monitoring
of large concrete
structures is of great significance since it offers a general estimati
on of the condition and
enables proper repair action. Empirical correlations between pulse veloc
ity and damage or
strength have been exploited for a long time (Jones 1953, Kaplan 1959, Anderson
and Seals
1981, Popovics 2001). It is accepted that pulse velocity, even though not extrem
ely sensitive
to damage (reduced prior to final failure by a percentage of not mor
e than 20%), is very
indicative of the internal conditions of the material (Van Hauwaert et al. 1998,
Shiotani and
Aggelis 2009). It possesses essential characteristics making it the most widely
accepted and
used parameter in concrete NDT. One of these advantages is its direct relation t
o the elastic
constants through well known elasticity relations (Sansalone & Streett
1997, Naik et al.
2004). This way, elastic waves measurements enable the estimation of t
he modulus of
elasticity and consequently projection to the strength of the material
through empirical
relations (Kaplan 1960, Keating et al. 1989, Shah et al. 2000 , Monteiro et al.
1993, Philippidis
& Aggelis 2003).
Another essential feature of pulse velocity is its independence to the
propagation distance.
Pulse velocity is firmly fixed to the elastic constants and nominally its value
does not exhibit
changes with distance like other parameters, e.g. amplitude and frequen
cy which are
continuously decaying due to attenuation (Landis and Shah 1995, Owino
and Jacobs 1999,
Jacobs & Owino 2000). In addition to the well known use and interpre
tation of pulse
velocity, frequency and dispersion features have recently been studied with the
aim of more
accurate characterization concerning the damage content and characteristi
c size (Aggelis &
Philippidis 2004, Philippidis & Aggelis 2005, Punurai et al 2006, Aggelis & Shio
tani 2007a, In
et al. 2009, Chaix et al. 2006). Dispersion originates from inhomogene
ity and hence the
velocity dependence on frequency should be much more evident in damage
d than in
healthy media. Concrete is inhomogeneous by nature due to porosity and
aggregates and
exhibits moderate dispersive trends (Philippidis & Aggelis 2005, Punurai
et al. 2006).
Nevertheless, cracks due to size and severe impedance mis-match with the matrix
material,
are stronger scatterers of elastic waves and influence wave propagation
more effectively
than the inherent inhomogeneity of the material.
As aforementioned, concrete structures suffer deterioration in the form of distr
ibuted micro-
cracking as well as macroscopic surface breaking cracks. Cracking allow
s water and other
chemical agents to penetrate into the material, further deteriorating the struct
ure, oxidizing
the reinforcement and finally compromising its load bearing capacity (O
htsu & Tomoda
2008). It is desirable to mitigate these deteriorating actions or even strengthe
n the structure
by repair. Repair can take place in the form of cement injection eit
her targeted to seal
Elastic Waves on Large Concrete Surfaces
for Assessment of Deterioration and Repair Efficiency 191
specific macroscopic cracks (Aggelis & Shiotani 2007b, Thanoon et al. 2005, Issa
& Debs 2007,
Yokota & Takeuchi 2004), as well as a measure to reinforce the whole
surface applying
injection in a pattern of boreholes on the surface of the structure (Shiotani et
al. 2009, Shiotani
& Aggelis 2007). Concerning specific macro-cracks, their depth can be evaluated
by ultrasound
either by the travel time of the longitudinal wave refracted from the
tip of the crack, or the
amplitude of the surface Rayleigh wave that survives beneath the crack (Doyle &
Scala 1978,
Hevin et al. 1998, Liu et al. 2001, Pecorari 2001, Song et al. 2003, Zerwer et a
l. 2005, Aggelis et
al. 2009). The same measurements can be repeated after application of grouting a
gent in order
to estimate the efficiency of repair. However, although this can be d
one for specific surface
breaking cracks it would be not be practical for a large concrete su
rface that normally has
numerous cracks. Therefore, although specific cracks may be indicatively targete
d for detailed
measurements, it is also desirable to establish a way to evaluate the
general conditions of a
large area of the structure both before and after repair. The relative change of
wave parameter
values before and after maintenance actions offers a deterministic measure of re
pair efficiency
(Kase & Ross 2003, Aggelis & Shiotani 2007b, Shiotani et al. 2009, A
ggelis et al. 2009) which
otherwise should be based solely on empirical criteria. This way the effectivene
ss of the repair
can be quantified by the change of wave features, as elastic waves propagate in
long distances
through the material and gather information from different parts. In t
he present chapter,
distributed micro-cracking on large concrete surfaces is examined by elastic wav
es before and
after repair. Results are presented from an actual old and deteriorate
d structure, as well as a
large concrete block with simulated defects inside. It is exhibited th
at complementary use of
Rayleigh and longitudinal waves is appropriate for the evaluation of i
nitial condition and
repair efficiency, while combined use of wave velocity and major frequ
ency, as well as their
dependence (dispersion) enhances the evaluation of repair. This kind of dispersi
on effects are
attributed to inhomogeneity (Tsinopoulos et al. 2000) and are diminishe
d after repair.
Additionally, development in computer and electronic engineering enables
tomography
reconstruction of the interior characterized by the value of the propa
gation velocity of either
longitudinal or Rayleigh waves, which can be related directly to the mechanical
properties of
the material (Sassa 1988, Kepler et al. 2000, Kobayashi et al. 2007, Shiotani et
al. 2009, Aggelis et
al. 2011). For the full field measurement of a large concrete surface
, the newly developed
Rayleigh wave tomography methodology is discussed showing the potential
to successfully
map the subsurface defects, exploiting the connection between penetration depth
and applied
wavelength (Chai et al. 2010). The chapter comprehensively reviews the
current practice in
structural health monitoring of real structures using elastic waves, in
addition to discussing
new trends and features for more detailed assessment of the damage condition.
2. Longitudinal and Rayleigh waves
Wave propagation is strongly affected by the elastic properties and de
nsity of constituent
materials. Longitudinal wave velocity is given in Eq. (1):
( )
1
1 1 2
P
v
E
C
v v
| |
|
\ .
=
| |
+
|
\ .
(1)
where E is the elasticity modulus, is density, and is the Poissons ratio.
It has been correlated with strength (Kaplan 1959, Kaplan 1960, Kheder 1999, Qas
rawi 2000,
Naik et al. 2004) and damage (Van Hauwaert et al. 1998, Ono 1998, M
ikulic et al. 1999,
Aggelis & Shiotani 2008a, Shiotani & Aggelis 2009) of concrete materials, offeri
ng rough but
valuable estimations because the damage condition influences the mechani
cal properties
and, hence, the wave speed. Employing a number of sensors, the veloci
ty structure of the
material can be constructed and the internal condition can be visualiz
ed, highlighting the
existence of voids or cracks (Sassa 1988, Kepler et al. 2000, Kobayas
hi et 2007, Aggelis &
Shiotani 2007b, Shiotani et al. 2009). Furthermore, the use of Rayleig
h (or surface) waves
seems quite suitable for surface opening cracks investigation, since they propag
ate along the
surface of the structure. Additionally, they occupy higher percentage o
f energy than the
other types of waves. For example, as mentioned in (Graff 1975) a po
int source in a
homogeneous half space radiates 67% of its energy in the form of Ray
leigh waves, while
only 7% in compressional ones. Moreover, since they are essentially two-dimensio
nal, their
energy does not disperse as rapidly as the energy associated with thr
ee-dimensional
dilatational and shear waves. Specifically, their amplitude is inversely
proportional to the
square root of propagation distance while for a longitudinal wave the amplitude
is inversely
proportional to the distance (Owino & Jacobs 1999). This makes them more easily
detectable
than other kinds of waves, as will also be discussed in the present
text. The distinct
difference from longitudinal waves concerns the particle motion which is ellipti
cal with the
vertical component greater than the horizontal, as opposed to the unidi
rectional oscillation
of the particles in longitudinal waves (Graff 1975). The Rayleigh moti
on decreases
exponentially in amplitude away from the surface (Jian et al. 2006).
Practically, the
penetration depth of these waves is considered to be similar to their wavelength
(Sansalone
and Streett 1997, Jian et al. 2006). The velocity of surface Rayleigh waves, C
R
, which will be
also discussed in this chapter, is given by (2):
( )
( )
0.87 1.12
1 2 1
R
v E
C
v v
+
=
+ +
(2)
From equations (1) and (2) it is derived that by measuring both long
itudinal and Rayleigh
velocities, apart from the Youngs modulus, the Poissons ratio can be calculated as
well.
3. Experimental part
3.1 Repair
On the surface of the old concrete structure (dam) numerous cracks we
re observed, as is
typical due to long exposure to freezing-thawing cycles, and water att
ack. Repair was
applied by cement injection in three different ways. First, cement inj
ection was conducted
from the opening of the thicker cracks using syringes, see Fig. 1. As to thin cr
acks, they were
repaired by surface application of cement. Finally, cement was injected
in a pattern of
boreholes on the surface using a constant pressure, as described in (
Shiotani & Aggelis
2007). The actual result is that empty pockets in the structure created by crack
s or extensive
porosity were filled with cementitious material. This material is initi
ally liquid in order to
penetrate to the thin crack openings, but due its cementitious nature,
hydration reaction
transforms it into a stiff inclusion, sealing the crack sides. It results in con
siderable decrease
of permeability which is crucial for water intake facilities and resto
res some of the load
bearing capacity since voids are replaced by a stiff material reinforcing the st
ructures cross
section. Concerning initial crack depth evaluation, some of the cracks can be in
terrogated by
longitudinal or Rayleigh waves (Doyle et al. 1978, Liu et al. 2001,
Tsutsumi et al. 2005,
Aggelis et al. 2009). However, given the large number of cracks this
is not practical. A
measure of the quality of the entire surface must be obtained in a time-effectiv
e manner.
Fig. 1. (a) Multiple locations of injection on the surface, (b) detail of the in
jection point.
3.2 Wave measurements
In order to estimate the condition of the material elastic wave measu
rements were applied
before and after repair. In the specific case, three vertical arrays
of four sensors were
attached using a rectangular pattern to record the surface response, a
s seen in Fig. 2. All
three dimensions of the concrete block (dam pier) were much larger than the moni
tored area
and the wavelength, and therefore, no influence was expected to either
longitudinal or
Rayleigh waves. Excitation was conducted by a steel ball (35 mm in diameter) res
ulting in a
frequency peak of approximately 10 kHz (see Fig. 3(a)) and a longitud
inal wavelength of
approximately 400 mm while the Rayleigh wavelength is of the order of
200 mm. The
excitation was consecutively conducted near each sensor (which acted as a trigge
r) and the
rest acted as receivers. Therefore, several paths (all possible combina
tions between two
individual sensors) with different lengths were examined. The horizontal spacing
was 1.2 m,
the vertical 1.5 m, while the longest paths corresponding to the larg
est diagonals of the
whole monitored area were 5.1 m long. The sensors were acoustic emiss
ion transducers,
namely Physical Acoustics, PAC, R6 sensitive at frequencies below 100 kHz. The a
cquisition
system was a 16 channel PAC, Mistras operated on a sampling frequency
of 1 MHz. The
sampling time of 1 s, resulted in an error lower than 0.3%, since ev
en for the shortest
distances of 1.2 m, the transit time was approximately 300 s. The sensors were at
tached on
the surface using electron wax.
Pulse velocity was measured by the time delay of the first detectable
disturbance of the
waveform. As to the Rayleigh velocity, the reference point used for t
he measurement was
the first peak of the Rayleigh burst that stands much higher than th
e initial, weaker
longitudinal arrivals (Sansalone & Streett 1997, Qixian & Bungey 1996,
Aggelis & Shiotani
2007a, 2008b) (see Fig. 3(b)).
Fig. 2. Pattern of wave measurements and examined wave paths on concrete surface
.
Fig. 3. (a) Frequency spectrum of the excitation, (b) Typical waveforms for sens
ors at
different distances from the excitation.
1.2 m
1.5 m
18 paths
of 1.S m
16 paths
of S.2S m
12 paths
of
2.8S m
Sensoi
position
4. Results
4.1 Wave velocity
Figure 4 depicts the longitudinal velocity as a function of the propagation dist
ance. It is clear
that a strong dependence on the distance exists. As already mentioned,
nominally wave
velocity depends on the elastic constant and not on the travel path.
However, there are
several reasons that impose this phenomenon. One is the typical crack density co
mpared to
the length of the wave path. The possibilities that a short path is crack-free,
are higher than a
longer one. Therefore it is reasonable that the highest velocities are
exhibited at short
quarters. Additionally, attenuation effects are accumulated reducing the
signal amplitude
for longer propagation. This hinders the correct identification of the
waveforms leading
edge for signals collected at large distances and thus the velocity is underesti
mated. This can
be considered as an indirect effect of damage on the velocity, throug
h the difficulty on
interpretation that the attenuation imposes.
The average velocity measured for the short paths of 1.2 m before re
pair averages at 4500
m/s, while for the longest paths of 5.1 m it is below 4000 m/s. After repair and
allowing for
a period of two weeks for the injection cement to hydrate properly, the curve is
elevated by
about 250 m/s, as seen in Fig. 4. This corresponds to an increase of more than 5
%, from 4228
m/s to 4455 m/s.
Fig. 4. Pulse velocity vs. propagation distance for different stages of repair.
As to Rayleigh waves, the dependence on the distance is much weaker both before
and after
repair, as seen in Fig. 5. The reason is connected to the Rayleigh wave measurem
ent, since it
uses a strong reference peak (see Fig. 3(b)), much higher than the noise level.
The Rayleigh
velocity averages at 2235 m/s before and 2363 m/s after repair, being
increased by more
than 5%, similarly to the longitudinal velocity.
Befoie, y = 4S68.2x
u.u92
R = u.S981
Aftei, y = 4821.2x
u.u92
R = u.SuS2
SSuu
S7uu
S9uu
41uu
4Suu
4Suu
47uu
49uu
S1uu
SSuu
SSuu
u 1 2 S 4 S 6
P
u
l
s
e
v
e
l
o
c
i
t
y
(
m
/
s
)
Distance (m)
Befoie
Aftei
Fig. 5. Rayleigh velocity vs. propagation distance for different stages of repai
r.
It is mentioned that the average increase is the result of the veloc
ity change for the
individual travel paths which can differ considerably from point to po
int. Fig. 6 shows the
velocity change for all individual measurement positions. The large maj
ority of points
showed an increase, while a smaller population exhibited decrease. Ther
efore, while the
overall increase is considered satisfactory, examination of the change point by
point enables
further investigation of specific points that exhibited velocity decrease unexpe
ctedly.
Fig. 6. Difference of velocity between before and after repair for all measureme
nt points.
Befoie, y = 2244.6xu.uu9
R = u.uu2
Aftei, y = 2477.4xu.u6
R = u.1S44
17uu
19uu
21uu
2Suu
2Suu
27uu
29uu
u 1 2 S 4 S 6
R
a
y
l
e
i
g
h
v
e
l
o
c
i
t
y
(
m
s
)
Bistance (m)
Befoie
Aftei
8uu
6uu
4uu
2uu
u
2uu
4uu
6uu
8uu
1uuu
12uu
u 1 2 S 4 S 6
V
e
l
o
c
i
t
y
c
h
a
n
g
e
(
m
/
s
)
Distance (m)
v
Longituuinal
Aveiage
inciease
longituuinal
v Rayleigh
Aveiage
inciease
Rayleigh
The average increase of longitudinal and Rayleigh velocities can be tr
anslated to elastic
modulus improvement by Eqs. (1) and (2). Before repair the effective
elastic modulus is
calculated at the value of 34.5 GPa, while after repair it is calcul
ated at 37.9 GPa,
corresponding to an improvement of approximately 10%. This shows the efficiency
of the
repair action. It is possible that after full hydration of the inject
ed cement, this increase
would be even higher. However, one point needs attention. This is the
environmental
temperature that influences the hydration rate. In cold environments hydration i
s delayed
and thus the material is gaining rigidity with a slow rate. It has
been seen that elastic
wave measurements immediately after injection and before the injection material
hardens
can show the opposite of the desired result (Aggelis & Shiotani 2009, Shiotani e
t al. 2009).
This is due to scattering on the soft pockets of grout which lowers
the velocity for a
limited time period. Although this phenomenon is normal, and has been
recently
explained, it contradicts the common knowledge that velocity should alw
ays rise after
repair. Therefore, it could be misleading and therefore, it is suggested that me
asurements
for repair evaluation are conducted after a sufficient period of time
for the cementitious
repair agent to cure.
4.2 Pulse frequency
As mentioned in the experimental part, the impact excited a waveform
with major
frequency components below 20 kHz. As any pulse propagates through inh
omogeneous
and possibly damaged concrete, the higher of its frequency components are influe
nced more
severely. This change of spectrum can be well monitored by the central
frequency, C
(Aggelis & Philippidis 2004, Shiotani and Aggelis 2009). In the specif
ic case the central
frequency is calculated as the centroid of the FFT of the waveforms up to 40 kHz
:
( )
( )
40
0
40
0
=
}
}
fM f df
C
M f df
(2)
where f is the frequency, and M(f) the magnitude of the FFT.
In Fig. 7, the dependence of central frequency on the distance is de
picted. Despite the
experimental scatter of the points which is expected due to the inhom
ogeneity of the
material, a certain decreasing trend is exhibited both before and afte
r repair. This
demonstrates the cumulative effect of the inhomogeneity of the travel path on th
e frequency
content.The average frequency before repair is 12.5 kHz for short prop
agation distance
while it is reduced to 6.8 kHz for the longest paths of 5.1 m. The
trend can be fitted quite
well with a decaying exponential curve, and is attributed to the stro
nger attenuation of
higher frequencies (Tsinopoulos et al. 2000, Shiotani & Aggelis 2009). Measureme
nts on the
same points after repair showed a frequency increase of approximately
2 kHz for any
distance, as seen again in Fig. 7. This can be attributed to the fi
lling of the cracks which
reduces the material scattering attenuation. It should be mentioned that the fre
quency is not
directly correlated with concrete strength or structural integrity. Howe
ver, in any case,
strength can be estimated only through empirical relations even for pu
lse velocity. The
importance lies on the comparison between the two stages (before and after repai
r). Since a
frequency upgrade was evident this shows that there was a certain imp
rovement on the
structures surface layer due to void elimination.
Fig. 7. Central pulse frequency vs. propagation distance for different stages of
repair.
4.3 Reliability aspects
At this point it is worth to mention the importance of reliability o
f the signal acquisition.
Strong attenuation due to inhomogeneity and spreading due to long distances decr
ease the
signal level very effectively. This is crucial especially for the puls
e velocity measurement,
which is conducted by the first detectable disturbance of the received waveform.
In case the
signal is weak, the first cycle may be of the same level or even lower than the
noise level and
therefore, the velocity is underestimated. This is a phenomenon that i
s not widely
considered in practical applications. The strength of the signal compar
ed to the noise is
measured by the signal to noise ratio, S/N, which is simply the ratio
of the waveforms
peak amplitude divided by the average noise level of each waveform. T
he noise level is
calculated by the amplitude of the pre-triggerperiod, the period after e
ach sensor starts
acquisition and before the actual wave arrives at the sensor point. In order to
see the effect
of noise on the measurement of wave parameters, the dependence of pul
se velocity,
Rayleigh velocity and central frequency are plotted vs. S/N in Fig. 8
(a), (b) and (c)
respectively. Concerning Fig. 8(a) before repair, an increasing trend of velocit
y can be found
with increase of S/N. It is obvious that for S/N lower than 1000, velocities ave
rage around
4000 m/s, while for higher S/N the velocities reach 4500 m/s in aver
age. After repair, as
already discussed, the cloud of points is elevated to higher velocity (
due to elastic
modulus restoration) and higher S/N values due to the decrease of sca
ttering attenuation.
At the same time, the correlation between velocity and S/N is sufficiently weake
ned (from
Befoie, y = 1S.994x
u.466
Aftei, y = 1S.76Sx
u.42S
S
7
9
11
1S
1S
17
19
u 1 2 S 4 S 6
F
i
e
q
u
e
n
c
y
(
k
B
z
)
Bistance (m)
Befoie
Aftei
uioup A
uioup B
R
2
of 0.26 to 0.07). It can be concluded that when S/N is low (e.g.
less than 1000) the
velocities are certainly underestimated, while the measurement becomes a
lmost
independent for S/N ratios higher than 3000. This is an aspect that
should always be
considered in field measurements, especially for long distance measurements.
As aforementioned the Rayleigh velocity is calculated by possibly the strongest
peak of the
waveform (the first peak of the Rayleigh burst) and therefore, it doe
s not crucially depend
on the S/N ratio. This is shown in Fig. 8b where the correlation coefficient bet
ween velocity
and S/N is close to zero. Similar results with pulse velocity with e
ven higher correlation
coefficients can be seen in Fig. 8c for the central frequency of the pulse. Befo
re repair there is
a very clear increasing trend exhibiting also a quite high correlation
coefficient R
2
of 0.61.
This shows that when the signal is strongly attenuated and its streng
th decreases with
respect to noise, the higher frequencies are the first to be influenc
ed. The increasing trend
holds for the measurements after repair, though in this case the correlation is
much weaker.
The discussion of S/N is concluded with its dependence on distance. In any case
it is normal
that the S/N ratio decreases with distance in any material, since the effect of
attenuation is
accumulated for longer travel paths. However, it is remarkable to see
that the S/N
undergoes a change from approximately 6000 for short wave paths (1.2 m
) to less than 500
for longer (5.1 m) as depicted in Fig. 9. This change of more than 12 times indi
cates that the
data should be divided in smaller groups in order to exclude the effect of atten
uation, when
discussing elasticity modulus or dispersion effects, as will be analyze
d in the next section.
Additionally, after repair, it is evident that the S/N increased and the correla
tion to distance
was certainly reduced from 0.59 to 0.25. It is also characteristic that for the
shortest distance
(1.2 m) the S/N is quite similar for both conditions (before and after repair) w
ith values 5900
and 6600 respectively. Larger differences are observed for longer paths
(>4 m) where the
effect of attenuation is accumulated, i.e. less than 400 before and 1250 after r
epair.
4.4 Dispersion relation
Both longitudinal velocity and frequency decrease exponentially with dis
tance; thus they
are correlated with each other, as seen in Fig. 10. The clusters consist of 130
points which is
the total number of possible paths between the sensors. Due to inhomogeneity and
locality
effects there is a certain experimental scatter. However, it is clear that both
before and after
repair there is a positive correlation showing a certain dependence of velocity
on frequency.
Nevertheless the average velocity and frequency both increase after rep
air. Apart from the
average longitudinal velocity increase of more than 5%, central frequency increa
ses from 9.6
kHz to 11.1 (change of 15.6%). This kind of simultaneous examination
of different features
can enhance repair characterization since frequency increase seems to b
e more sensitive to
repair than velocity of longitudinal and Rayleigh waves.
It is understood that the correlations in Fig. 10 include the influen
ce of distance since the
measurements are taken for any different distance between 1.2 m and 5
.1 m. In order to
examine the frequency effect on the propagation velocity excluding the
influence of
attenuation, the information was processed in small groups of data col
lected over close
distances. Two cases will be indicatively discussed. One group (group A of Fig.
7) includes
the data collected at the shortest distances of 1.2 m and 1.5 m (to
tally 34 points), while the
other (group B of Fig. 7) includes the approximately double distances
of 2.83 m and 3 m
Fig. 8. Wave parameters vs. signal to noise ratio (a) pulse velocity, (b) Raylei
gh velocity, (c)
central frequency.
Befoie, y = SS9S.Sx
u.u292
R = u.2S97
Aftei, y = SS88.1x
u.u2S8
R = u.u677
Suuu
SSuu
4uuu
4Suu
Suuu
SSuu
6uuu
P
u
l
s
e
v
e
l
o
c
i
t
y
(
m
/
s
)
Befoie
Aftei
y = 21S7.Sx
u.uu4S
R = u.uuS1
y = 2uu7.2x
u.u192
R = u.u4S9
1Suu
17uu
19uu
21uu
2Suu
2Suu
27uu
29uu
R
a
y
l
e
i
g
h
v
e
l
o
c
i
t
y
(
m
/
s
)
Befoie
Aftei
y = 276ux
u.16S9
R = u.6149
Aftei, y = S1S6x
u.1498
R = u.2S6S
u
S
1u
1S
2u
u Suuu 1uuuu 1Suuu
C
e
n
t
r
a
l
F
r
e
q
u
e
n
c
y
(
k
H
z
)
Signal to Noise ratio (-)
Befoie
Aftei
Fig. 9. Signal to noise ratio vs. propagation distance for different repair stag
es.
Fig. 10. Pulse (a) and Rayleigh (b) velocity vs. frequency for the total populat
ion of
measurements.
Befoie, y = 9296.6x
1.9S4
R = u.S91S
Aftei, y = 8S99.Sx
u.848
R = u.2S42
u
2uuu
4uuu
6uuu
8uuu
1uuuu
12uuu
14uuu
u 1 2 S 4 S 6
S
i
g
n
a
l
t
o
n
o
i
s
e
r
a
t
i
o
(
-
)
Distance (m)
Befoie
Aftei
Befoie, y = 828.7x
u.178
R = u.4218
y = 112u.4x
u.148S
R = u.1884
Suuu
SSuu
4uuu
4Suu
Suuu
SSuu
6uuu
P
u
l
s
e
V
e
l
o
c
i
t
y
(
m
/
s
)
Befoie
Aftei
Befoie, y = 828.7x
u.178
R = u.4218
y = 8u9.98x
u.1148
R = u.u984
1Suu
17uu
19uu
21uu
2Suu
2Suu
27uu
29uu
S 1u 1S
R
a
y
l
e
i
g
h
v
e
l
o
c
i
t
y
(
m
/
s
)
Frequency (kHz)
Befoie
Aftei
Fig. 11. Pulse velocity vs. Frequency for different length of wave paths (a) bef
ore and (b)
after repair.
(totally 24 points). For each group, the velocity vs. frequency correlation is d
epicted in Figs.
11a and b. Since the data in each group are taken from similar dist
ances, the effect of
attenuation is eliminated allowing examination of pure dispersion effects within
each group.
The experimental scatter is high since each travel path in damaged concrete is u
nique due to
the randomness of inhomogeneity. The total population shows a trend th
at should not be
ignored; there is a positive correlation between velocity and frequency
before repair. This
implies that an increase in frequency content results in a slight inc
rease of pulse velocity.
This trend is in agreement with recent dispersion studies on concrete
(Chaix et al. 2006,
Philippidis & Aggelis 2005, Aggelis and Shiotani 2007, 2008, Aggelis 2009). They
reveal that
S7uu
4uuu
4Suu
46uu
49uu
S2uu
SSuu
6 8 1u 12 14 16
P
u
l
s
e
v
e
l
o
c
i
t
y
(
m
s
)
Fiequency (kBz)
uioup A, 1.2 to 1.S m
uioup B, 2.8 to S m
y = u.uuuSx + 4SS7.7
S7uu
4uuu
4Suu
46uu
49uu
S2uu
SSuu
6 8 1u 12 14 16
P
u
l
s
e
v
e
l
o
c
i
t
y
(
m
s
)
Fiequency (kBz)
uioup A, 1.2 to 1.S
m
uioup B, 2.8 to S m
below 100 kHz the increasing slope of the dispersion curve is steep,
while for frequencies
around or above 100 kHz the curve reaches a plateau. This dispersion
is stronger as the
inhomogeneity of the material increases and has been measured for wide
frequency bands
up to almost 1 MHz (Philippidis & Aggelis 2005, Chaix et al. 2006). In general i
n all typical
particulate composites or porous materials (Kinra & Rousseau 1987, Saye
rs & Dahlin 1993,
Aggelis et al. 2004), phase velocity is lower than the velocity of t
he matrix for low
frequencies, but rises with frequency until reaching approximately the
velocity of the
matrix. For higher frequencies the velocity does not exhibit serious alterations
according to
the above mentioned studies.
It is interesting to examine the same correlation after repair. Apart
from the general
translation of the points to higher velocity and frequency levels for
any propagation
distance, the slight positive correlation is eliminated for both groups. Cement
injection filled
cavities and cracks eliminating inhomogeneity to a large extend. Thus,
the structure
behaved in a less dispersive way. This repair-dependent dispersion should not be
ignored as
it shows the potential to enhance the rough characterization performed so far in
concrete. It
is mentioned that all individual distance groups of data exhibited the
velocity-frequency
dependence. Fig. 11 contains the groups with the largest population. S
ame behavior
concerning the frequency dependence of pulse velocity was observed for
through the
thickness measurements of elastic waves in concrete pier before and af
ter repair. A certain
dispersive trend was exhibited initially, while after the repair materi
al was properly
hydrated into the structure, this trend was weakened (Aggelis et al. 2011).
The dependence of Rayleigh velocity on distance and frequency is much weaker tha
n in the
case of longitudinal as was seen in Figs. 5 and 10. One possible ex
planation is the long
wavelength compared to the depth of surface-opening cracks. Some surfac
e cracks with
wide openings (i.e. 0.2 0.4 mm) were targeted using longitudinal and Rayleigh wa
ves for
depth measurement (Aggelis et al. 2009). The depth of cracks never exceeded 100
mm while
an average value for the depth of the cracks was 35 mm. Rayleigh wa
ves propagating
through a depth of more than 200 mm (similar to their wavelength) ca
n dive below the
shallow defects. Therefore, their dispersion in this case is weaker, since the h
ighest degree of
damage is concentrated near the surface, while the propagation of Rayleigh takes
place also
through deeper, more homogeneous zones. On the other hand, longitudinal waves tr
aveling
along the surface are influenced by each surface crack as it poses a
discontinuity on their
path. This makes their dependence on frequency stronger, as was seen
in Figs. 4 and 10.
Dispersion should be further studied for material characterization since
in a general case,
limited dispersion of Rayleigh should imply that defects are shallower
than the Rayleigh
wavelength offering a means of fast estimation for a large surface.
5. Rayleigh wave tomography
Analysis of elastic wave data certainly offers insight on the structur
al condition of the
material, as was shown above by the changes in wave properties and dispersion. H
owever,
one way to further enhance the understanding of the results is visual
ization. This is one
global goal concerning the inspection. Techniques like X-rays, radar, t
hermography and
elastic waves aim to produce tomograms of the structure (Aggelis et al. 2011, Diama
nti et
al. 2008, Ito et al. 2001, Ohtsu and Alver 2009) in order for the
results to be more easily
understood by a general audience (owner of structure or user). For the specific
case of near
surface damage, the establishment of Rayleigh wave-based tomography tech
nique is
beneficial since it materializes single-side assessment, which is particularly u
seful for certain
structures with open access only on one surface, such as tunnels and
bridge deck panels.
Also using the tomography technique, examination of in-situ structures
becomes more
realistic because it provides general indications on problematic spots
within a relatively
short time of execution. These spots could then be assessed by a pin
-point approach in a
detailed manner to characterize the defect.
5.1 Analytical study of Rayleigh waves propagating in concrete with horizontal c
rack
To clarify the behaviour of Rayleigh waves impinged by a horizontal crack lying
within the
concrete sub-surface, a series of wave motion simulations have been co
nducted. The
simulations were conducted using commercial software developed for solvi
ng two-
dimensional elastic wave propagation based on finite difference method.
As illustrated in
Fig. 12, the analytical model was generally composed of the concrete
medium with four
sensors located on the top side of the model at a uniform distance of 100 mm. A
point source
for generating elastic waves was located 50 mm away from the trigger sensor to t
he left. The
trigger sensor initiate simultaneous recording of waveforms once the incoming of
wave was
detected. Sensors R1, R2 and R3 were placed to record waveforms at d
ifferent locations as
the generated waves propagate further from the source. In the simulations, the c
oncrete was
modeled with the first and second lame constants of 10 GPa and 15 G
Pa, respectively, and
density of 2300 kg/ m
3
. Damping of elastic waves in concrete was appropriately defined to
simulate waveforms reasonably close to those observed in the experiment
al measurement.
The configuration resulted in a corresponding longitudinal wave velocity of appr
oximately
4200 m/s, a typical value for homogeneous concrete of normal strength. The left
, right and
bottom sides of the concrete model were configured with infinite bound
ary conditions in
order to avoid reflections and simulate a larger structure geometry. B
eside the model of
homogeneous concrete, models with an additional 150 x 2 mm void place
d parallel to the
concrete surface at varying depths from the top side of concrete were
also prepared to
constitute delamination in concrete.
(a) (b)
Fig. 12. Wave motion simulation models, (a) Homogeneous concrete and (b) Inhomog
eneity
in the form of a horizontal crack
A single-cycle excitation of elastic waves from the point source was
generated for all the
analytical models. To study the relation between wavelength of Rayleigh
waves and depth
of delamination that causes distortion, excitations at different frequen
cies were performed.
Hence the simulations produced waveform data for different combinations
of excitation
frequency and crack depth to serve further analysis purposes.
Fig. 13 presents snapshots of typical simulated cases of elastic waves
propagation for the
homogeneous model. Knowing that velocity of Rayleigh waves is approximately 57%
that of
the primary waves for a normal strength concrete (Sansalone and Streett 1997), i
n the case for
150 kHz excitation (Fig. 13(a)), in which the wavelength was relatively shorter,
the separation
of Rayleigh wave energy component (with the larger amplitudes) from th
e primary waves
was evident from sensor R1 onwards. The separation became greater as the waves p
ropagated
further. For low frequency excitations, namely the 10 kHz one, the se
paration could not be
clearly seen at propagation distance equivalent to that observed for t
he 150 kHz excitation,
because the wavelengths of both wave modes are quite long and therefore overlap
at the early
distances of propagation before the different propagation velocities separate th
em (Fig. 13(b)).
The separation of Rayleigh waves from longitudinal waves can be justified from t
he recorded
waveforms for the homogeneous concrete model, indicating clearer separation of o
ne full cycle
for the 150 kHz excitation when reaching sensor R1. For the waveforms obtained f
rom 10 kHz
excitation, even though no complete separation can be observed, Rayleig
h waves can be
extracted without much difficulty since their arrival was marked by significantl
y large energy
that initiated abrupt increase of amplitude (Qixian & Bungey 1996).
(a) (b)
Fig. 13. Snapshots of wave motion and waveform of homogeneous concrete for (a) 1
50 kHz
and (b) 10 kHz excitation frequencies
Rayleigh
waves
wave motion
wave motion
For the cracked case, when the elastic waves impinged on the crack tip of the de
lamination,
as indicated by snapshots given in Fig. 14, wave diffraction and scattering occu
r, with some
portion of wave energy being reflected upwards from the crack-concrete i
nterface to
eventually converge with the ones propagating on the surface. Distorted
waveforms
recorded by sensor R1 indicated a more apparent decrease compared to
those from the
receivers of the homogeneous concrete model.
(a) (b)
Fig. 14. Snapshots of wave motion and waveforms of concrete model with a horizon
tal crack
located at 50 mm from top for (a) 150 kHz and (b) 10 kHz excitation frequencies
5.2 Processing for phase velocity
Since the distortion of waves due to the defect could reduce the ene
rgy of Rayleigh waves
relatively to that of the body waves, which was bound to hamper accu
rate analysis, the
characteristic amplitudes of Rayleigh waves was extracted. The extractio
n process was
carried out simply by muting the other wave components so that only th
e one-cycle
Rayleigh waves would remain. The relevant wave components of could be recognized
as the
maximum amplitudes in both the positive and the negative phases. The
processed
waveform then yielded a more clear-cut frequency response by the fast
Fourier transform
(FFT), indicating a characteristic peak frequency belonging to the prop
agating Rayleigh
waves. Observation revealed that the processed waveform for the first sensor hav
e the peak
Rayleigh
waves
wave motion
wave motion
frequency almost equivalent to the frequency of excitation as configure
d in the simulation,
suggesting that this frequency value could be regarded as the characteristic fre
quency of the
generated Rayleigh waves.
Figure 15 plots the travel time of Rayleigh waves with excitation fre
quency of 20 kHz
against distance of propagation for the homogeneous concrete model. The
travel time was
obtained by taking the time of positive peak arrival as acquired by the receiver
s R1, R2 and
R3 with reference to the trigger. The gradient of linear regression could be inv
ersed to yield
a propagation velocity of 2398 m/s.
Fig. 15. Linear correlation between arrival time and propagation of Rayleigh wav
es with 20
kHz excitation frequency
In order to take into account the possible dependence between velocity and frequ
ency which
is expected in materials like concrete, especially when they contain i
nhomogeneity, the
phase velocity curve can be calculated. The whole procedure is explained
in (Sachse &
Pao, 1978 and Philippidis & Aggelis, 2005). It includes Fourier transf
ormation of the time
domain signals recorded at different positions on the surface of the
material. The phase of
the signals is unwrapped and phase difference () for each frequency is
calculated. Then
by the knowledge of the separation distance between the receivers, x, the velocit
y for each
phase (frequency component) of phase velocity V
ph
can be derived by the following
equation:
I
ph
= = _
2x
] (3)
where is the wavelength, f is the frequency for which the phase difference is ca
lculated. By
plotting the phase velocity versus the frequency within the bandwidth, i.e. disp
ersion curve,
the change of phase velocity of Rayleigh waves with regards to frequency can be
examined.
The results can also be used to calculate theoretical wavelength of R
ayleigh waves for the
respective excitations. Fig. 16 shows exemplary dispersion curves for a
ll the three trigger-
receiver combinations in the case for the homogeneous concrete and tha
t with a horizontal
crack at 25 mm depth, which were acquired from data with 20 kHz of excitation fr
equency.
The phase velocity for the homogeneous concrete corresponding to 20 kHz was aver
aged to
a value of 2317 m/s, which was very close to that obtained by taking into accoun
t the arrival
time of Rayleigh wave peak amplitude. As can be seen from the figure
s, the dispersion
curves were seemingly uniform for the case of homogeneous concrete model, indica
ting that
the phase velocity for each frequency component was quite similar within the ban
dwidth of
excitation. On the other hand, for the models with horizontal crack, the dispers
ion curves for
all the three trigger-receiver combinations exhibited considerable differ
ences and were all
translated to lower values. The average phase velocity at the excitation frequen
cy of 20 kHz
was calculated as 1723 m/s, which was lower than that obtained for t
he homogeneous
concrete model. Considering that the penetration depth for Rayleigh wav
es can be
equivalent up to its wavelength, it can be inferred that the generated Rayleigh
waves in this
particular case, which has a calculated wavelength of approximately 110
mm, was
effectively distorted by the horizontal crack at 25 mm depth to exhib
it translation of phase
velocity to lower values as a result of the change in propagation behavior.
(a)
(b)
Fig. 16. Dispersion curve computed for data by excitation frequency of 20 kHz, (
a)
homogeneous concrete and (b) horizontal crack at 50 mm from top
5.3 Velocity change
The average phase velocity values at respective excitation frequencies
were obtained from
the models with delamination and normalized with those of the homogene
ous concrete
model to obtain the velocity index for facilitating comparison. The velo
city indices were
arranged in accordance with the depth of delamination depth, as plotted in the f
orm of bar
charts given by Fig. 17. The bars in the charts were classified under two distin
ct groups: data
with the theoretical wavelengths equal or greater than the delamination
depth (d ); and
those of the opposite condition (d < ). The data with d gave lower
velocity index
compared to the data with d < in general. Among the simulated cases
, the velocity index
dropped most noticeably for d = 25 mm, where the soundness indices have decrease
d to as
low as 0.68. In cases where d = 150 mm and 200 mm, although velocity indices of
lower than
1.00 were also obtained for d < , the decrease was comparatively small. This coul
d bear the
inference that the generated waves have only been slightly disturbed by the hori
zontal crack
although the penetration depth was greater (assuming equivalent to one
wavelength) than
the crack depth. In other words, the penetration depth was essentially
less than the
wavelength.
Fig. 17. Velocity indices for different cases of crack depth
V
e
l
o
c
i
t
y
i
n
d
e
x
Frequency (kHz) Frequency (kHz)
V
e
l
o
c
i
t
y
i
n
d
e
x
V
e
l
o
c
i
t
y
i
n
d
e
x
V
e
l
o
c
i
t
y
i
n
d
e
x
V
e
l
o
c
i
t
y
i
n
d
e
x
V
e
l
o
c
i
t
y
i
n
d
e
x

Slight decrease in velocity indices in the cases of d was in gener
al observed in the
simulated results. The decrease could be attributed to slight change in the inhe
rent nature of
the simulated waves subject to the combinations of crack depths and excitation f
requencies.
As implied by the findings, inhomogeneity introduced by the presence of horizont
al crack in
concrete tended to produce variation of phase velocity, even though the decrease
would be
less significant in some cases. Further examination revealed that the
wavelength should be
at a minimum of 2.5 times greater than the depth of defect in order to constitut
e a decrease
of at least 10% (velocity index = 0.9) in phase velocity.
5.4 Development of Rayleigh wave tomography
5.4.1 Principle of travel time tomography
A typical travel time tomography involves a unique algorithm that perf
orms iterative
computations to reconstruct the velocity profile of a measured object from the p
icked travel
time of elastic waves, which can be obtained in the form of observed
data through
measurement at selected locations on the object. In ray theory, the travel time,
t is computed
from the integral of the slowness along a given ray path l
t = _ ut
I
= _
1
:
I
ul = _ sul
L
(4)
where v is the velocity, s is the slowness (reciprocal of velocity) and dl is th
e length element.
If a model is subdivided into M cells of constant slowness, s
i
for each corresponding segment
length l
i
, the integral product can be totalled as
t = l
M
=1
(5)
The above equation can be expressed in a matrix-vector form:
t = Ls (6)
where t is the travel time vector and s is the vector of s
i
. L is a matrix describing the ray path
segments. The matrix is generally sparse since each ray would cover only a limit
ed number
of cells. Equation (6) is solved to simulate a velocity distribution
that can explain the
measured travel times. Although the ray propagation itself is considere
d as a linear
operator, the inverse problem can become non-linear, which has to be solved usin
g iterative
process. With a fundamental model s
0
, new models are computed successively based on the
discrepancy of data and model response by solving
s
k+1
= s
k
+s
k
= s
k
+L
|
(s
k
)(t -L(s
k
)s
k
) (7)
with L recomputed and so on, whereby L
is the used inverse operator.

The computation process is commonly known as the ray tracing procedure
, for which the
simultaneous iterative reconstruction technique (SIRT) is adopted. The ray paths
are usually
restricted to the edges of the mesh and a weighted shortest path problem has to
be solved.
Since not every path can be used, or there is insufficient observed data, the co
mputed travel
times are always not accurately estimated in the first few iterations.
The accuracy is
improved following increasing refinement in the computation to reduce discrepanc
y.
5.4.2 Experimental measurement of concrete slab
To investigate the feasibility of Rayleigh wave properties as the obse
rved data for
tomography reconstruction to visualize defect in concrete, laboratory ex
periment was
carried out to measure and evaluate a concrete slab specimen. The spe
cimen was a 1500 x
1500 x 300 mm concrete slab of normal strength. Defects in the forms of horizont
al cracks in
concrete at different depths were modelled by including four 5 mm-thick circular
Styrofoam
plates of 300 mm in diameter. The Styrofoam plates were located at 30 mm, 60 mm,
100 mm
and 140 mm from the top of concrete surface, respectively, as depicted in Fig. 1
8. Shown in
the same figure also are the arrangement of 16 accelerometer sensors
for sensing and
recording of elastic waves. The arrangement formed a total of 240 dir
ect ray paths. During
the measurement, elastic waves were generated by manually hitting the
concrete surface
with a ball hammer that has a spherical head of 15 mm in diameter. A 16-channel
waveform
acquisition system was used to record waveforms at an interval of 5 s
for a total of 4096
samples. Each set of impact was performed by continuing hitting the specimen sur
face for 5
seconds, which resulted in approximately 20 impacts or waveforms to be recorded
by each
sensor. In the preliminary processing of signal, the multiple waveforms
recorded by the
(a) (b)
(c) (d)
Fig. 18. Tomography measurement on concrete slab: (a) Crack depth and arrangemen
t of
accelerometer sensors (() denotes depth from the opposite surface)), (b) Ray pat
h lay-out, (c)
Snapshot during measurement, and (d) Generation of elastic waves by hammer with
spherical head.
300
3
0
0
9
0
0
900
Embedded Styrofoam
Accelerometer sensor
60
(240)
30
(270)
140
(160)
100
(200)
Unit: mm
respective sensors would be stacked accordingly to eliminate random noi
se that could be
influential in hampering proper processing in the later stage. Each ti
me the impact was
made adjacent to one sensor which has been configured as the trigger,
with the others as
receivers to detect incoming waves. The acquisition was synchronized in such man
ner that
all 16 channels would start recording at the same time once the trigger channel
was excited
due to the arrival of elastic waves from the hitting point. The impact and recor
ding process
was repeated by subsequently setting the next sensor as the trigger i
n a pre-determined
order until all the sensors have been assigned.
In order to measure the propagation velocity of Rayleigh waves on sou
nd concrete, a
separate measurement utilizing nine accelerometer sensors at 150 mm distance was
carried
out on the sound concrete portion of the same specimen. From the recorded wavefo
rms, the
peak amplitudes could be identified because of the large increase in
relation to the
preceding ones that belonged to the body waves. Taking into account t
he time of Rayleigh
wave arrival, a typical travel time versus distance plot can be drawn as given i
n Fig. 19. The
velocity could be obtained as the inverse of gradient from linear reg
ression, giving an
approximate value of 2356 m/s for the sound concrete.
Fig. 19. Time vs. distance plot of Rayleigh waves to estimate propagation veloci
ty
In Fig. 20, an example of time series data and its corresponding FFT frequency s
pectrum is
shown. The frequency spectrum presented two frequency peaks, suggesting the exis
tence of
more than one dominant frequency in the waves. By using the similar
procedure adopted
for the processing of simulation data as discussed earlier, the Raylei
gh wave amplitudes
was extracted, based on the identification of time window containing t
he main Rayleigh
wave components as anticipated based on the typical propagation velocity. This r
esulted in
a frequency spectrum as given in Fig. 21, clearly indicating only one peak at ap
proximately
9 kHz. Using the measured velocity and peak frequency, the dominant w
avelengths was
calculated as 240 mm.
y = 0.4244x + 1.0437
1.00
1.10
1.20
1.30
1.40
1.50
1.60
0.00 0.25 0.50 0.75 1.00 1.25 1.50
T
r
a
v
e
l
T
I
m
e
(
m
s
)
Distance (m)
(a) (b)
Fig. 20. Typical measurement results: (a) Time series of original waveforms, (b)
Corresponding frequency spectrum
Fig. 21. Frequency spectrum of the proccessed waveform, giving characteristic pe
ak
frequncy of Rayleigh waves
5.5 Tomography reconstruction using phase velocity
All the recorded waveforms were processed for Rayleigh wave amplitudes,
followed by
transformation to frequency response to compute phase difference, which
were used to
calculate the phase velocity by utilizing Equation (1). The results we
re plotted against the
frequency, as exemplified in Fig. 22 for the case of Rayleigh waves
propagating directly
from one trigger to its surrounding eight receivers. In the figure, t
hree out of the eight
dispersion curves were obtained from propagation paths that contained artificial
horizontal
crack in between trigger and receiver. The crack was located 30 mm from the surf
ace. All the
three paths apparently yielded reduced phase velocity values for the e
ntire effective
frequency range of excitation, which was estimated at 5-15 kHz. The phase veloci
ty has been
decreased to less than 1900 m/s in general, registering an average di
fference of
approximately 500 m/s compared to the propagation paths containing no
crack. It was
interesting to note that the wavelength of Rayleigh waves as calculate
d based on the peak
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
0.0 1.0 2.0 3.0 4.0
A
m
p
l
i
t
u
d
e
(
v
o
l
t
)
Time (ms)
0 5 10 15 20
S
p
e
c
t
r
u
m
d
e
n
s
i
t
y
Frequency (kHz)
0 5 10 15 20
S
p
e
c
t
r
u
m
d
e
n
s
i
t
y
Frequency (kHz)
frequency and propagation velocity was greater than the crack depth. T
he finding was in
agreement with that of the simulation work in confirming that when th
e waves was
effectively impinged by a horizontal crack that lied within the region
of penetration,
distortion was bound to occur, resulting in the change of propagation
behaviour to gave
lower phase velocity and translation of dispersion curve in the effect
ive frequency range.
The decrease in phase velocity due to the horizontal crack was approx
imately 500 m/s
compared to the sound concrete and was found to be greater than that
observed in the
simulation work.
Fig. 22. Frequency spectrum of the proccessed waveform, giving characteristic pe
ak
frequncy of Rayleigh waves
For the purpose of tomography reconstruction, the measured area was di
vided into 6 x 6
square cells of 150 mm. The phase velocity data acquired from the peak frequency
of excitation
were used as the observed data and the computation process was carried out for 5
0 iterations
to achieve satisfactory convergence. Fig. 23 presents the results of tomography
reconstruction
expressed in the form of phase velocity distributions for measurements from the
surface with
relatively small crack depths (30 mm ~ 140 mm) and large crack depths (160 mm ~
270 mm).
For measurement from the surface with small crack depths, all the fou
r crack location were
successfully detected with good visualization effect; while from the su
rface where the crack
depths became greater, the visualization for cracks was less convincing
except for the crack
with 160 mm depth, which is the shallowest among the four. Based on the setting
for graphical
display, the cracks were indicative when the phase velocity was lower than 2000
m/s. It is also
to be noted that the circular shape of the horizontal cracks was not accurately
represented by
the low phase velocity region and in most cases the centre of defect
was indicated with the
lowest value, and the change of value from the centre to the edge of defects has
developed in a
non-uniform way. Cracks which were not successfully visualized were due to the f
act that the
generated Rayleigh waves did not penetrate sufficiently deep into the
concrete to be
effectively impinged by these cracks. From the results it was demonstrated that
the accuracy of
assessment by the Rayleigh wave-based tomography was essentially dependa
ble upon the
depth of crack as well as the penetration depth.
1500
2000
2500
3000
5.0 7.5 10.0 12.5 15.0
P
h
a
s
e
v
e
l
o
c
i
t
y
(
m
/
s
)
Frequency(kHz)
path on sound
concrete
path with crack
(a) (b)
Fig. 23. Rayleigh wave tomography resutls for (a) the side with relatively small
crack depth,
and (b) with greater crack depth from top surface of concrete
Based on literature study, however, it is found that the penetration
depth varied in
accordance to material as well as structural form. For example, the w
as a study reporting
satisfactory estimation for elastic properties of layered concrete slabs
by assuming the
effective Rayleigh wave penetration depth to be equivalent to half a
wavelength (Wardany
et. al. 1997). On the other hand, another study using concrete beams
concluded that at
vertical distances greater than half the beam depth, Rayleigh waves did not form
ed because
of the fundamental modes diverged at longer wavelengths and energy was
consumed in
flexural mode (Zewer et. al. 2005). Also, in determining the near-surf
ace profile of soil
structures (Long & Kacaoglu 2001), it was reported that the Rayleigh
wave velocity was
determined primarily by the shear wave (S-wave) velocity of material i
n a depth range of
1/4 wavelength. Based on the findings of the current experimental study, the wav
elength of
Rayleigh waves has to be greater than 1.5 times the depth to crack for successfu
l detection.
The estimated wavelength, was much less than that concluded from the
simulation work
discussed earlier herein (2.5 times the crack depth). Nevertheless, the feasibil
ity of utilizing
the phase velocity for tomography reconstruction to detect defect in c
oncrete has been
demonstrated. It is considered further work has to be directed to ref
ining quantitative
relation between the wavelength and the depth of crack so that the vertical loca
tion of defect
can be estimated in more accurately way. In any case however, it is worth mentio
ning that
even if a crack is not visualized at its actual dimensions using the
proposed method,
successful detection for the fingerprint of the defect would help provide useful
information
for minimizing the labour and increasing the effectiveness of other pi
npoint NDT
techniques to characterize in detail.
6. Conclusion
The present chapter discusses elastic wave propagation of low frequency
on large concrete
surfaces. The objective is characterization of damage and effectiveness
of repair by cement
injection. A pattern of piezoelectric transducers was mounted on a large part of
the surface
6u mm Su mm
14u mm
1uu mm
27u mm
24u mm
2uu mm
16u mm
24uu
22Su
21uu
19Su
18uu
m s
allowing a quick and reliable scanning of the velocity of the surface of the str
ucture. Repair
was connected to a velocity increase of the level of 5% to 6% for
both longitudinal and
Rayleigh waves. Furthermore, the frequency content of the pulses surviv
ing long
propagation distances increased by 15% exhibiting stronger sensitivity t
o the repair action.
Attenuation and dispersion effects impose a dependence of measured velo
city on the
distance, which is more clear for longitudinal waves that travel on a
shallow depth on the
surface. On the other hand Rayleigh waves have a propagation depth si
milar to their
wavelength which enables them to fly below the shallow cracks and limits their d
ispersion.
After repair the longitudinal wave dispersion was weakened offering an ad
ditional feature
sensitive to the repair effectiveness. Reliability considerations that a
re not normally taken
into account are discussed which seem to influence the measured wave parameters.
Additionally, the feasibility of Rayleigh waves for developing a single
-side access
tomography technique of concrete was investigated numerically and experi
mentally.
Tomography reconstructions results indicated the suitability of the meas
urement method
and data analysis procedure for visualizing subsurface defects. It was demonstra
ted that by
increasing the dominant wavelength of Rayleigh waves, wave penetration
could be
improved and deeper defect could be detected. Studies on the quantitative relati
on between
dominant wavelength and depth of defect, in addition to clarifying the
dispersion
characteristic of Rayleigh waves are imperative in future in order to
realize accurate
assessment of concrete structures using the proposed tomography techniqu
e. Numerical
simulations are expected to greatly enhance knowledge concerning the both the in
teraction
between dominant wavelength and depth of defect that can be identified
as well as the
dependence of velocity on frequency. Nowadays, since waveform acquisition is sta
ndard to
almost all ultrasonic equipment the rough characterization based on pul
se velocity can be
easily enhanced by features like frequency and dispersion of each wave mode. So
far these
have been applied successfully in laboratory conditions for material characteriz
ation but it is
expected that their application in real structures will definitely improve NDT c
apabilities.
7. References
Aggelis, D.G., Hadjiyiangou, S., Chai, H.K., Momoki, S., Shiotani, T.
(2011). Longitudinal
waves for evaluation of large concrete blocks after repair, NDT&E International
44
(2011) 6166
Aggelis D.G., Momoki S., Chai H. K. (2009). Surface wave dispersion i
n large concrete
structures, NDT&E International 42, 304307.
Aggelis D. G., Philippidis, T. P. (2004). Ultrasonic wave dispersion and attenua
tion in fresh
mortar. NDT & E International, 37(8), 617-631.
Aggelis D.G., Shiotani T. (2007a). Experimental study of surface wave
propagation in
strongly heterogeneous media. J Acoust Soc Am; 122(5):EL1517.
Aggelis, D.G., Shiotani, T. (2007b). Repair evaluation of concrete crac
ks using surface and
through-transmission wave measurements, Cement & Concrete Composites 29;
700711.
Aggelis, D. G., Shiotani, T. (2008a). Effect of inhomogeneity parameter
s on the wave
propagation in cementitious materials. American Concrete Institute Materi
als
Journal, 105(2), 187-193.
Aggelis, D. G., Shiotani, T. (2008b). Surface wave dispersion in cemen
t-based media:
inclusion size effect. NDT&E INT, 41, 319-325.
Aggelis, D.G., Shiotani, T., Polyzos, D. (2009). Characterization of su
rface crack depth and
repair evaluation using Rayleigh waves. Cement and Concrete Composites,
31(1),
77-83.
Aggelis, D.G., Tsimpris, N., Chai, H.K., Shiotani, T., Kobayashi, Y. (
2011). Numerical
simulation of elastic waves for visualization of defects, Construction
and Building
Materials 25 15031512.
Aggelis, D. G. and T. Shiotani. (2009). An experimental study of wave propagatio
n through
grouted concrete. American Concrete Institute Materials Journal, 106(1) 19-24
Anderson, D.A., Seals, R.K. (1981). Pulse velocity as a predictor of 28- and 90-
day strength,
ACI J. 78 79, 116 122.
Chai, H.K., Aggelis, D.G., Momoki, S., Kobayashi, Y., Shiotani, T. (20
10). Single-side access
tomography for evaluating interior defect of concrete, Construction and
Building
Materials 24 24112418.
Chaix, J. F., Garnier V., Corneloup G. (2006). Ultrasonic wave propagation in he
terogeneous
solid media: Theoretical analysis and experimental validation, Ultrasonics 44, 2
00
210.
Diamanti, N., Giannopoulos, A., Forde, M.C. (2008) Numerical modelling and exper
imental
verification of GPR to investigate ring separation in brick masonry ar
ch bridges.
NDT&E Int ;41:35463.
Doyle, P.A., Scala, C.M. (1978). Crack depth measurement by ultrasonics
: a review.
Ultrasonics 16(4):16470.
Graff, K.F. 1975. Wave motion in elastic solids. New York: Dover Publications.
van Hauwaert, A.; Thimus, J. F.; and Delannay, F. (1998). Use of Ultrasonics to
Follow Crack
Growth. Ultrasonics, 36 209-217.
Hevin, G., Abraham, O., Pedersen, H.A., Campillo, M. (1998). Characteri
sation of surface
cracks with Rayleigh waves: a numerical model. NDT&E Int ;31(4): 28997.
In, C-W., Kim, J.Y., Kurtis, K.E.,Jacobs, L.J. (2009). Characterization
of ultrasonic Rayleigh
surface waves in asphaltic concrete, NDT&E International 42 610617.
Issa, C.A., Debs, P. (2007). Experimental study of epoxy repairing of
cracks in concrete.
Constr Build Mater;21:15763.
Ito, F., Nakahara, F., Kawano, R., Kang, S-S., Obara, Y. (2001). Visu
alization of failure in a
pull-out test of cable bolts using X-ray CT. Construct Build Mater 15:26370.
Jacobs, L. J., Owino, J. O. (2000). Effect of aggregate size on attenuation of R
ayleigh surface
waves in cement-based materials, J. Eng. Mech.-ASCE. 126 (11) 1124-1130.
Jian, X., Dixon, S., Guo, N., Edwards, R.S., Potter, M. (2006). Pulsed Rayleigh
wave scattered
at a surface crack. Ultrasonics 44:11311134.
Jones R (1953) Testing of concrete by ultrasonic-pulse technique, Proce
edings of the thirty-
second annual meeting. Highway Res Board 32:258275
Kaplan, M.F. (1959). The effects of age and water/cement ratio upon t
he relation between
ultrasonic pulse velocity and compressive strength, Mag. Concr. Res. 11 (32) 85 9
2.
Kaplan, M. F. (1960). The relation between ultrasonic pulse velocity a
nd the compressive
strength of concretes having the same workability but different mix pr
oportions,
Magazine of Concrete Research, 12 (34) 3-8.
Kase, E.J., Ross, T.A. (2003). Quality assurance of deep foundation el
ements, Florida
Department of Transportation. In: Proceedings of the 3rd international
conference
on applied geophysics geophysics 2003, Orlando, Florida, December 812.
Keating, J.; Hannant, D. J.; and Hibbert, A. P. (1989). Correlation b
etween Cube Strength,
Ultrasonic Pulse Velocity and Volume Change for Oil Well Cement Slurries, Cement
and Concrete Research, 19(5); 715-726.
Kepler, W.F., Bond, L.J., Frangopol, D.M. (2000). Improved assessment o
f mass concrete
dams using acoustic travel time tomography, part IIa icatio. Construct Build
Mater 14:14756.
Kheder, G.F. (1999) A two stage procedure for assessment of in situ concrete str
ength using
combined non-destructive testing. Mater Struct 32:410417
Kinra V. K., Rousseau, C. (1987). Acoustical and optical branches of
wave propagation, J.
Wave Mater. Interaction 2, 141152 .
Kobayashi, Y.; Shiotani, T.; Aggelis, D. G.; and Shiojiri, H. (2007).
Three- Dimensional
Seismic Tomography for Existing Concrete Structures,Proceedings of the Se
cond
International Operational Modal AnalysisConference, IOMAC 2007, (April 30
-May 2,
Copenhagen), V. 2.pp. 595-600.
Long, L. T., Kacaoglu, A. (2001). Surface-wave group-velocity tomography
for shallow
structures. J Env Eng Geophys 6(2); 71-81.
Landis, E. N., Shah, S. P. (1995). Frequency-dependent stress wave att
enuation in cement-
based materials, J. Eng. Mech.-ASCE 121 (6) 737-743.
Liu, P.L., Lee, K.H., Wu, T.T., Kuo, M.K. (2001). Scan of surface-opening cracks
in reinforced
concrete using transient elastic waves. NDT&E Int 34:21926.
Mikulic, D., Pause, Z., Ukraincik, V. (1999). Determination of concrete quality
in a structure
by combination of destructive and non-destructive methods. Mater Struct 25:6569
Monteiro, P. J. M., Helene, P. R. L., Kang, S. H. (1993). Designing
concrete mixtures for
strength, elastic modulus and fracture energy, Materials and Structures, 26 443-
452
Naik TR, Malhotra VM, Popovics JS. (2004).The ultrasonic pulse velocity
method. In:
Malhotra VM, Carino NJ, editors. Handbook on nondestructive testing of concrete.
Boca Raton: CRC Press.
Ohtsu M, Tomoda Y. (2008). Phenomenological model of corrosion process
in reinforced
concrete identified by acoustic emission. ACI Mater J. 105(2):1949.
Ohtsu, M., Alver, N. (2009). Development of non-contact SIBIE procedure
for identifying
ungrouted tendon duct. NDT&E Int 42:1207.
Ono, K. (1988) Damaged Concrete Structures in Japan due to Alkali Sil
ica Reaction, The
International Journal of Cement Composites and Lightweight Concrete, 10;4 247-25
7.
Owino, J. O., Jacobs, L. J. (1999). Attenuation measurements in cement-based mat
erials using
laser ultrasonics, J. Eng. Mech.-ASCE. 125 (6) 637-647.
Pecorari, C. (2001). Scattering of a Rayleigh wave by a surface-breaki
ng crack with faces in
partial contact. Wave Motion ;33:25970.
Philippidis, T. P., Aggelis D. G. (2003) An acousto-ultrasonic approach for the
determination
of water-to-cement ratio in concrete. Cement and Concrete Research, 33(4), 525-5
38.
Philippidis, T. P., Aggelis, D. G. (2005). Experimental study of wave
dispersion and
attenuation in concrete. Ultrasonics, 43(7), 584-595.
Popovics, S. (2001). Analysis of the Concrete Strength versus Ultrasoni
c Pulse Velocity
Relationship, Materials Evaluation, 59(2) 123-130.
Punurai W, Jarzynski J, Qu J, Kurtis KE, Jacobs LJ. (2006). Characterization of
entrained air
voids in cement paste with scattered ultrasound. NDT&E Int 2006;39(6):51424.
Qasrawi HY (2000) Concrete strength by combined nondestructive methods
simply and
reliably predicted. Cem Concr Res 30:739746
Qixian L., Bungey J.H. (1996). Using compression wave ultrasonic transd
ucers to measure
the velocity of surface waves and hence determine dynamic modulusof el
asticity
for concrete. Construct Build Mater 4(10):23742.
Sachse, W., Pao, Y.-H. (1978). On the determination of phase and grou
p velocities of
dispersive waves in solids, J. Appl. Phys. 49 (8) 43204327.
Sansalone, M.J., Streett, W.B. (1997). Impact-echo nondestructive evaluation of
concrete and
masonry. Ithaca, NY: Bullbrier Press.
Sayers C. M., Dahlin, A. (1993). Propagation of ultrasound through hydrating cem
ent pastes
at early times, Advanced cement based materials, 1:12-21.
Sassa K. (1988). Suggested methods for seismic testing within and betwe
en boreholes. Int J
Rock Mech Min Sci Geomech Abstr;25(6):44972.
Shah, S.P., Popovics, J.S., Subramanian, K.V., Aldea, C.M. (2000). New direction
s in concrete
health monitoring technology, J. Eng. Mech.ASCE 126 (7) 754760.
Shiotani, T., Aggelis D. G. (2009). Wave propagation in concrete conta
ining artificial
distributed damage. Materials and Structures, 42(3), 377-384.
Shiotani, T., Momoki, S., Chai, H. K., Aggelis, D.G. (2009). Elastic
wave validation of large
concrete structures repaired by means of cement grouting, Construction
and
Building Materials 23 26472652.
Shiotani, T., Aggelis, D. G. (2007). Evaluation of repair effect for deteriorate
d concrete piers
of intake dam using AE activity. Journal of Acoustic Emission, 25, 69-79.
Song, W.J., Popovics, J.S., Aldrin, J.C., Shah, S.P. (2003). Measuremen
ts of surface wave
transmission coefficient across surface-breaking cracks and notches in c
oncrete. J
Acoust Soc Am 113(2):71725.
Thanoon, W.A., Jaafar, M.S., Razali, M., Kadir, A., Noorzaei, J. (2005). Repair
and structural
performance of initially cracked reinforced concreteslabs. Constr Build
Mater
19(8):595603.
Tsinopoulos, S. V., Verbis, J. T., Polyzos, D. (2000). An iterative e
ffective medium
approximation for wave dispersion and attenuation predictions in particu
late
composites, Adv. Composite Lett. 9, 193200.
Tsutsumi, T., Wu, J., Wu, J., Huang, X., Wu, Z. (2005). Introduction
to a new surface-wave
based NDT method for crack detection and its application in large dammonitoring.
In: Proceedings of international symposium on dam safety and detection of hidden
troubles of dams and dikes, 13 November, Xian, China (CD-ROM).
Yokota, O., Takeuchi, A. (2004). Injection of repairing materials to c
racks using ultrasonic
rectangular diffraction method. In: 16th World conference on non destru
ctive
testing 2004 (WCNDT), Montreal, Canada, August 30September 3.
Zerwer, A., Polak, M.A., Santamarina, J.C. (2005). Detection of surface
breaking cracks in
concrete members using Rayleigh waves. J Environ Eng Geophys;10(3): 295306.
9
Ultrasonic Testing of
HPC with Mineral Admixtures
R. Hamid, K. M. Yusof and M. F. M. Zain
Universiti Kebangsaan Malaysia
Malaysia
1. Introduction
High strength and high performance concrete (HSC and HPC) have been i
ntroduced
commercially for the columns of high-rise buildings in 1970s in the U
nited States of
America and now being made worldwide, one example being the tallest b
uilding in
Malaysia, the Petronas Twin Towers (Shah, 1997). Fly ash, silica fume,
or slag are often
mandatory in the production of high-strength and high performance concr
ete for the
reason that the strength gain obtained with these supplementary cementi
ng materials
cannot be attained by using additional cement alone. These mineral additives are
usually
added at dosage rates of 5% to 20% or higher by mass of cementing
material (Portland
Cement Association EB001 [PCA EB001], 2002). HSC and HPC with mineral
additives
have different microstructure than normal concrete. The matrix is dense
r and has lower
capillary porosity compared to ordinary concrete. In HSC at more than
105 MPa with
silica fume, the matrix is perfectly homogenous, apparently amorphous and the ca
pillary
porosity is diminished and discontinuous compared to that of other concretes w
here it is
interconnected (Malier, 1992).
There are a number of ultrasonic testing techniques applied to concret
e since 40 years ago
(International Atomic Energy Agency TCS 17 [IAEA TCS17], 2002). These
techniques are
relatively not as sophisticated as compared to other materials application such
as metals, but it
is found that these techniques are relevant in some cases. These techniques incl
ude ultrasonic
pulse velocity (UPV), ultrasound pulse echo, the amplitude or the pulse energy m
easurement,
impact-echo/resonance frequency/stress wave test (pulse-echo, impact-echo,
impulse
response and spectral analysis of surface waves), and relative amplitude method
(RA).
Ultrasonic pulse velocity measurement in concrete is applicable in asse
ssing the quality of
concrete. The UPV measurements are correlated with the concrete strength as a me
asure of
concrete quality. For normal concrete with ordinary Portland cement, wi
th strength range
between 5 MPa up to 60 MPa, the empirical correlations between streng
th and UPV are
recorded as in the form of exponential (Bungey, 1984; Facaoaru, 1969;
Popovics, 1986;
Teodaru, 1989; etc as cited in Ismail et al., 1996) with correlation
equation, R
2
higher than
90%. For strength between 5 MPa to 60 MPa, the UPV varies between a
s low as below 2.0
km/s and as high as more than 4.7 km/s. The exponential relationship
between the UPV
and strength is expressed as:
K = oc
b
(1)
where K is the compressive strength of concrete, v is the UPV and a and b are th
e empirical
constants.
Researches on improvement on the strength-UPV correlation are done vastly either
to have
more accurate relationship (there are many factors that affect the UPV measured
but not the
concrete strength) or to make it applicable in assessing concrete stru
ctures (Carino, 1994).
British Standard (BS 1881: Part 203. Testing Concrete Recommendation f
or measurement
of velocity of ultrasonic pulses in concrete) listed some factors that affect th
e ultrasonic pulse
measurement in concrete. These factors are path length; lateral dimensi
on of the specimen
tested; moisture content of the concrete; temperature of the concrete
and the presence of
reinforcing steel. The corrections for each factor are as suggested by BS 1881:
Part 203.
As the usage of mineral admixtures is inevitable in HPC, researches o
n influence of
additional cementitious materials on strength as well as UPV have been studied.
Demirbog
et al., 2004; Trkmen et al., 2010 reported the coefficient of correlation (R
2
) of more than 0.90
which indicate a very good exponential relationship between UPV and co
mpressive
strength of concrete containing Portland cement, fly ash (FA), natural zeolite (
NZ) and blast
furnace slag (BFS) and both NZ + BFS concrete (Figure 1). They confi
rmed the exponential
relationship of strengthUPV of concrete with these mineral admixtures (as express
ed in Eq.
1). The UPV measured varies between 3250 m/s at strength of 5 MPa u
p to 4400 m/s at
strength 55 MPa. As the concrete strength increase, as in HPC and HS
C, the exponential
relationship still applied (Khatib, 2008; Hamid et al., 2010), but for
strength range between
60 to 100 MPa, the variation of UPV is between 4.5 to 5.0 km/s, ma
king the UPV
measurement insensitive towards the variation of strength of HPC and HSC in whic
h their
strengths are in this range (Figure 2).
Fig. 1. Relationship between compressive strength and UPV for all results betwee
n 3 and 90
days of curing periods for Portland cement, natural zeolite (NZ) and blast furna
ce slag (BFS)
and both NZ + BFS concrete (Trkmen et al., 2010)
Relative amplitude (RA) method is an alternative method to assess the
concrete strength
(IAEA TCS17, 2002). The principle of this method will be discussed fu
rther in this chapter.
Ultrasonic Testing of HPC With Mineral Admixtures 223
Ismail et al., 1996 described the relationship between RA and strength
for normal concrete
up to 60 MPa. The RA-strength relationship of HPC with silica fume i
s also described
(Hamid et al., 2010). Meanwhile, researches on combined ultrasound method
were done to
have more accurate estimation of concrete strength (Galan, 1990). The
strength of normal
concrete has been successfully estimated with only 5% error using the
combination of
ultrasonic pulse velocity-strength and relative amplitude-strength correlation c
urves (Ismail
et al., 1996).
Legend: M1, M2, M3 - 0% FA; M4 20% FA; M5 40% FA; MA6 60%FA; MA7 80%FA
Fig. 2. Relationship between strength and UPV for concrete with different propor
tion of fly
ash (FA) (Khatib, 2008)
This chapter emphasizes on ultrasonic pulse velocity (UPV) method in a
ssessing the
strength of high performance concrete (HPC) with mineral admixtures par
ticularly fly ash
and silica fume. Adding mineral admixtures to replace certain percentag
es of cement
content increase the strength of HPC compare to the same concrete combination co
ntaining
only ordinary Portland cement (OPC). This chapter discusses further in
detail results of
experiments on the UPV and RA method in strength estimation of HPC with fly ash.
2. Experimental program, results and discussion
2.1 Experimental program
About 300 specimens, representing 32 trial mixes were prepared and tes
ted in this study.
Fabrication, curing and testing of specimens were done based on ASTM C 192, ASTM
C618,
ACI Committee 211, BS 1881: Part 116 and BS 1881: Part 203.
2.1.1 Materials
The cement used is Type I Portland cement produced locally and confor
ms with the
Malaysian Standard. Its specific gravity is 3.15 (SI unit) and the mo
isture content is 0.6%.
The finess modulus of the mine sand used is 2.48, with its maximum
size at 4.75 mm;
specific gravity 2.25 (SI unit) and total moisture content 2.1%. The maximum siz
e of granite
aggregate used is 20 mm, its specific gravity is 2.29 (SI unit) and the total mo
isture content is
at 0.7%. Superplasticizer (SP) used in this study is a naphthalene fo
rmaldehyde sulfate
based condensed superplasticizer with brand name Super-20 supplied by R
eady Mixed
Concrete Malaysia. Its specific gravity is 1.21 (SI unit) with 40% of
solid content and 0%
chloride content.
2.1.2 Mix design
To reduce the number of variants, the ratio of sand to aggregate is held constan
t at 0.4. Any
admixtures added in the form of liquid, is calculated as part of the
mixing water. Three
series of HPC were designed with 10, 20 and 30% replacement with cla
ss F Malaysian fly
ash plus one series of mixes with 0% fly ash content. Each series c
onsisted of 10 different
water/binder ratios (7 for mixes with 30% FA and 5 for mixes with 0% FA) to get
variation
of strength, with the percentage of SP added were changed for each w
/b ratio in
consideration for workability, but maintained within the series. The w/
b ratios are from
0.200 to 0.380 with 0.02 intervals or otherwise stated. Total cementit
ious materials are 600
kg/m
3
and the ratio of fine aggregate to coarse is 0.4. Table 1 shows t
he detailed mixes in
this study.
2.1.3 Fabrication, curing and testing
Mixing is done in a rotating mixer and enough time is allowed until the mix show
s consistency
visually. The actual time to achieve the mix consistency depends on the slump/wo
rkability of
the mixes. Slump test is done on each water/binder ratio mix in each
series. The test
specimens are 150 mm 150 mm 150 mm cubes. The specimens are demoul
ded after 24
hours. One third of the total samples are immersed in water for 28 days, the oth
er one third are
immersed in water for 7 days and the last one third is let dried i
n open air inside an
uncontrolled temperature and humidity building. The temperature inside t
he building is on
average 32C. Compression test are done with a compression machine with
the capacity of
3000 kN after the ultrasonic testing were done. All tests are done on day 91 aft
er casting with 3
samples tested for each test. The concrete specimens were tested in wet conditio
n (more than
48 hours immersed in water before testing) as required by ASTM C 39.
2.1.4 Method of ultrasonic testing
Ultrasonic pulse velocity (UPV) is measured in materials by the equation:
: =
L
1
m/s (2)
where v is the longitudinal (pressure) pulse velocity, L is the path
length and T is the time
taken by the pulse to traverse that length.
A set up of the ultrasonic pulse velocity measurement on laboratory s
pecimen is as shown
in Figure 3.
Mix Water/ binder
ratio
Binder (kg/m
3
) SP (Liter) Sand
(kg/m
3
)
Aggregate
(kg/m
3
)
Cement Fly ash
I 0.220
0.260
0.300
0.340
0.380
600
600
600
600
600
0
0
0
0
0
17.4
14.9
12.4
9.9
7.4
653
630
607
584
561
980
945
910
876
841
II 0.198
0.216
0.234
0.252
0.270
0.288
0.306
0.324
0.342
0.360
540
540
540
540
540
540
540
540
540
540
60
60
60
60
60
60
60
60
60
60
17.4
17.4
14.9
14.9
12.4
9.9
9.9
7.4
7.4
7.4
648
638
628
618
607
597
587
576
566
556
973
957
942
926
911
895
880
865
849
834
III 0.200
0.220
0.240
0.260
0.280
0.300
0.320
0.340
0.360
0.380
480
480
480
480
480
480
480
480
480
480
120
120
120
120
120
120
120
120
120
120
19.8
17.4
17.4
14.9
14.9
12.4
9.9
9.9
7.4
7.4
643
632
620
609
597
586
575
563
552
540
965
948
930
913
896
879
862
845
828
810
IV 0.260
0.280
0.300
0.320
0.340
0.360
0.380
420
420
420
420
420
420
420
180
180
180
180
180
180
180
14.9
14.9
12.4
9.9
9.9
7.4
7.4
602
590
579
567
556
545
533
903
885
868
851
834
817
800
Table 1. Details of Mix Proportions
Tx - Transmitter
Rx - Receiver
CRO - Cathode Ray Oscilloscope
Trig - Trigger
Fig. 3. Ultrasonic pulse velocity measurement set up
The piezoelectric transducers (lead zirconate-lead titanates ceramic type) with
frequencies of
54 kHz, 82 kHz and 180 kHz as the ultrasonic pulse transmitter and receiver are
employed.
The ultrasonic pulse is generated by alternate current using commercial
ultrasonic-scope
(brand name PUNDIT) with digital display of transit time in microsecond. Figure
1 shows a
digital storage oscilloscope connected to the pulse exciter to trace t
he transmitted wave
from where the pulse amplitude is obtained. The oscilloscope display shows that
the pulse
amplitude is represented as displacement in one axis (normally y-axis)
and the ultrasonic
pulse transit time is represented by the displacement at the other ax
is (x-axis). The arrival
time of the pressure wave (transit time) can be located as t
p
shown in Figure 3. In linear
amplification system, vertical displacement is proportional to the signal amplit
ude. By using
logarithmic scale amplification, y-axis is represented by logarithmic scale.
The relative amplitude method is basically an attenuation method, which measures
the ratio
of the wave amplitudes. Ultrasonic waves are attenuated as they pass through the
materials.
The attenuation of the sound pressure propagating in solid consists of
two components,
namely the geometric divergence and energy dissipation. For quality eva
luation of
materials, the dissipation factor is critical since it is related to the propert
ies of the material.
The dissipation effect consists of two processes, namely scattering and
absorption.
Scattering is due mainly to the non-homogenous composition of the medium and abs
orption
is the actual loss of sound energy. When the properties of material vary, the in
tensity of the
lost energy will also vary, thus the attenuation characteristic can be
used to determine the
properties of the material. For low amplitude sinusoidal wave, the attenuation o
f wave due
to both conversion processes is in an exponent form and the incident amplitude (
either the
particle displacement amplitude or the pressure amplitude) can be expre
ssed as
[Krautkramer & Krautkramer, 1983] :
P
u
= P
0
c
-ux
(3)
where P
0
and P
a
are the initial pressure wave amplitude and pressure wave amplitude at a
distance x in a cross section with length d and attenuation coefficient, (alpha)
. The natural
logarithm of Eq. (3) is:
ox = ln
P
c
P
0
(4)
articulated as the attenuation at a distance x, a dimensionless number
expressed in Nepers
(Np). Attenuation coefficient is in Np/cm or dB/m. Since log x = ln x/ln 10 = 0.
434 ln x and
1 Np = 8686.0 dB, Eq. (4) can be written as:
ox = 2ulog
P
c
P
0
dB (5)
The relative amplitude, , is expressed as:
[ = 2ulog
A
p
A
ps
dB (6)
where is the measurement of attenuation with A
p
taken as the first peak amplitude after the
arrival of pressure wave (at t
p
= transit time of UPV through the concrete specimen) and A
ps
as
the peak amplitude after the arrival of shear wave (at t
s
= two times the transit time of UPV).
The location where t
s
is located is based on the fact that shear wave is slower (about two times)
than the pressure wave in concrete. Figure 4 shows the typical oscill
oscope display defining
the transit time and wave amplitude that delineate the relative amplit
ude (). As is also a
ratio between two amplitudes, it is also known as relative amplitude ratio (RAR)
.
Fig. 4. Typical oscilloscope display defining the transit time and wave amplitud
e (IAEA
TCS17, 2002)
However, consistent and constant coupling pressure for the transducers
has to be
maintained and will affect the attenuation values. It is suggested tha
t a simple pressure
measurement mechanism based on load cell and dial is built to check this problem
, though,
results discussed here did not include this mechanism.
2.2 Results and discussion
2.2.1 Ultrasonic pulse velocity-strength relationship of HPC with mineral admixt
ures
At high strength, the well established exponential relationship between
strength and UPV
(Eq. 1) is not sensitive to estimate the strength accurately (UPV bet
ween 4.5 to 5 km/s for
strength range 50 to 100 MPa). The correlation between strength and U
PV is also poor.
Uysal & Yilmaz, 2011 reported the UPV values ranged from 4222 m/s to
4998 m/s for
strength value of 55 to 105 MPa for concrete containing limestone pow
der (LP), basalt
powder (BP) and marble powder (MP). They also found that the use of
limestone powder
(LP), basalt powder (BP) and marble powder (MP) as cement replacement
at various
contents had caused a reduction in UPV as well as the compressive st
rength compared to
the control mixture. Increasing amounts of these kind of mineral admix
tures (limestone
powder, basalt powder and marble powder) generally decrease the strength. The co
rrelation
between the compressive strength and UPV is poor (R
2
= 0.75) at high strength compare to
concrete with mineral admixtures (fly ash, natural zeolite and blast f
urnace slag) at low
strength (Demirbog et al., 2004; Trkmen et al., 2010).
Khatib, 2008 reported that concrete with 40% fly ash (FA) mix shows
the largest value of
UPV compared with the 20% FA mix. Generally there is a decrease in
strength with the
increase in FA content. The strength versus UPV values shows an expon
ential relationship
with an R
2
of 0.97 but the strength range covered is from 5 MPa up to 85 MPa. Khatib, 2008
result also shows that for strength range 50 to 85 MPa, the UPV only varies 4.5
to 4.7 km/s;
however the relationship seems to be independent of the FA content.
Silica fume is a mineral admixture that provides early strength gain and high st
rength gain
to HPC. Hamid et al., 2010 reported that at a given strength, specimens with hig
her content
of silica fume (30%) shows lower readings of UPV compared to specimen
s with lower
content of silica fume (10% and 20%) (Fig. 5). The argument was that the mixture
with lower
content of silica fume is denser than the mixture with higher content
of silica fume at the
same water/binder ratio, even though they bear the same strength. The
strength-UPV
relationship is also reported as exponential form which resulted in at
strength higher than
85 MPa, the variation of UPV reading reduced. The coefficient of corr
elation, R
2
, for
strength-UPV curve for HPC with 10%, 20% and 30% silica fume are goo
d which are 0.93,
0.92 and 0.96 respectively. Hamid et al, 2010 also reported that the
sensitivity of the UPV
measurement decreased with the increased on HPC with silica fume strength.
For HPC with FA at strength range 50 to 90 MPa, Hamid, 2004 found
that the correlation
between strength and UPV of FAHPC (strength range 50 90 MPa) is poo
r (Figure 6).
Results of experiments show that using different transducers frequency gave the s
ame best-
fitted line of strength-UPV curve, but the correlation of coefficient using tran
sducer 150 kHz
is higher than using transducer 54 kHz. Measuring the UPV on saturate
d samples increase
the value of UPV compared to measuring on dry samples. The correlations of the b
est-fitted
line strength-UPV curve for the saturated samples are higher than the correlatio
ns of those
on dry samples. The compressive strength of mixes with higher content of fly ash
decrease
compare to mixes with low content of FA. The increase packing density
influences the
measured UPV very significantly but in reverse to the effect towards
strength. Higher
content of FA produce higher measured UPV at the same strength level of HPC with
lower
fly ash content.
Fig. 5. Relationship between silica fume content with UPV (Hamid et al., 2010)
Fig. 6. Relationship between UPV and strength of HPC with fly ash (Hamid, 2004).
It can be concluded that, these phenomena are observed in the measure
ment of UPV and
their relationship with strength of HPC with mineral admixtures:
K
K
K
1. Ultrasonic pulse velocity (UPV) is not sensitive to estimate the strength ac
curately (UPV
between around 4.2 to 5.0 km/s for strength range 50 to 100 MPa).
2. The correlation of coefficient (R
2
) of the UPV-strength curve is poor (below 0.8) for all
mineral admixtures except silica fume.
3. The content of mineral admixtures in the HPC affects the measured
UPV such that at
different mineral admixtures content, but at the same strength level, the UPV me
asured
are different. Generally, the UPV is higher in HPC with higher conten
t of mineral
admixtures.
2.2.2 Relative amplitude-strength relationship in HPC with mineral admixtures
Relative amplitude (RA) method is described in IAEA TCS17, 2002 and general rela
tionship
between strength, K and relative amplitude, is an exponential form:
K = c
c-d[
(7)
where c and d are empirical constants.
The relative amplitude decreases as the strength increased. The factors which in
fluence the
relationship are the moisture content of concrete; concrete age; aggreg
ate size and type;
curing method; bar size and beam path length.
The RA, in contrast with the UPV, shows greater sensitivity towards t
he strength change,
that is, 6 20 dB for strength range between 5 MPa to 60 MPa of n
ormal concrete with
coefficients of correlation greater than 80% (Ismail, 1996).
Fig. 7 shows that for HPC with silica fume at strength higher than 85 MPa, the v
ariation of
RA value is 14% for variation of strength of 40 MPa to 100 MPa. Th
e coefficients of
correlation, R
2
, for strength-RA curve for all three series of mixes are 0.54, 0.80
and 0.48
respectively (Hamid et al., 2010).
Fig. 7. The relationship between the silica fume content with the relative ampli
tude (Hamid
et al., 2010)
Figure 8 shows that each series of HPC with different fly ash content exhibits a
n exponential
relationship with the relative amplitude ratio (RAR), which is in the form:
K = cc
d[
(8)
where K is the compressive strength, is the RAR and c and d are the empirical co
nstant.
The RAR, in contrast with the UPV, shows greater sensitivity towards the strengt
h change,
that is, 80% (2 10 dB) for the same strength rage. The coefficients of correlati
on are greater
than 80% as shown in Figure 8.
Fig. 8. The relationship between the strength of HPC with FA and the RAR (Hamid
, 2004)
2.3 Factors affecting the UPV-strength relationship of HPC with fly ash
Tests were done to determine the effect of the transducers frequency,
moisture content,
samples shape, transmission type and FA content on the UPV-strength and
RA-strength
relationships. The frequencies of the transducers selected for the test
are 54 kHz and 150
kHz. Table 2 shows the effect of different transducers frequency on th
e correlation
coefficient (R
2
) of UPV-strength relationship of HPC with FA. It can be seen from
Table 2
that higher transducer frequency gives better correlation coefficient (R
2
) of UPV-strength
relationship of HPC with FA.
50
55
60
65
70
75
80
85
90
95
1 3 5 7 9 11
Relative amplitude ratio (dB)
C
o
m
p
r
e
s
s
i
v
e
S
t
r
e
n
g
t
h
(
M
P
a
)
FA 10% FA 20% FA 30%

R= 99.067e
-0.0512
R
2
= 0.8902
R= 85.543e
-0.0486
R
2
= 0.8388
R= 89.271e
-0.0427
R
2
= 0.8676

Fly ash
content
R
2
(54 kHz
transducer)
R
2
(150 kHz
transducer)
Empirical
constant (Eq.1)
Empirical
constant (Eq.1)
a b a b
10% 0.8921 0.7075 2.8076 0.6873 0.8252 0.9507
20% 0.4466 0.5317 9.2439 0.4142 9.0802 0.4195
30% 0.0007 0.4205 108.42 -0.0945 0.0034 1.9929
Table 2. Effect of transducers frequency on the correlation coefficient (R
2
) of UPV-strength
relationship of HPC with FA
The samples were then submerged in water for more than 72 hours and
tested. The
increases in moisture content recorded are between 0.4% and 0.8%. Table 3 shows
the effect
of moisture content on the correlation coefficient (R
2
) of UPV-strength relationship of HPC
with FA. Saturated samples give better correlation coefficient (R
2
) of UPV-strength
relationship of HPC with FA.
Fly ash
content
R
2
(54 kHz,
saturated)
R
2
(150 kHz,
saturated )
R
2
(150 kHz,
air dry )
10% 0.6253 0.8023 0.7075
20% 0.5521 0.6808 0.5317
30% 0.4503 0.4387 0.4205
Table 3. Effect of moisture content on the correlation coefficient (R
2
) of UPV-strength
relationship of HPC with FA
Cylindrical samples of 200 mm in height and 100 mm in diameter with
the same mix
proportions were tested and the resulting UPV measured are compared wi
th those of 150
mm
3
cube samples. Results show that UPV measured in cubes are higher than in cylinde
rs at
the same strength level. Equation 9 shows the relationship between the
UPV measured in
cubes and cylinders:
:
cyIIndcr
= u.611S:
cubc
+ 1.79SS (9)
for v
cube
> 4.6 km/s.
As described in IAEA TCS17, 2002, the ultrasonic pulse transmission ty
pes affect the value
of the measured UPV. The transmission types are the direct, semi dire
ct and indirect
transmission. Equations 10 and 11 show the relationship between the UPV measured
using
indirect method, semi direct method and the direct method:
:
dIrcct
= 1.2u89:
scmI dIrcct
(10)
:
dIrcct
= 1.Suuu:
IndIrcct
(11)
It is observed than there is an increase in the indirect (surface) transmission
correction factor
(CF) for HPC with fly ash compared to 1.05 reported for normal concr
ete in BS 1881:
Part 203.
2.4 Correction factors concerning UPV-strength and RA-strength relationships of
HPC
with fly ash
Tests were also done to determine the effect of the transducers frequency, moistu
re content,
samples shape, transmission type and FA content on the RA-strength. Tab
le 4 shows the
correction factor (CF) suggested from the analysis the test results.
Fly ash
content
Correction Factor (CF)
UPV
v150 kHz = CFv54
kHz
RA
150 kHz = CF54 kHz
10% 1.0041 1.6397
20% 0.9969 1.5463
30% 1.0211 1.5417
Table 4. Correction factors applied with different transducers frequency for UPV
and RA-
strength relationship of HPC with FA
On the other hand, effect of moisture content on UP-strength and RA-s
trength curve are
found to be insignificant as shown in Equations 12 and 13:
:
saturatcd
= 1.uuS4:
dry
(12)
[
saturatcd
= 1.uu21[
dry
(13)
The effect of fly ash content on the UPV and RA-strength correlation
curves for HPC with
fly ash compared to the same mix design without fly ash content (FA
0%) is as shown in
Table 5.
Fly ash
content
Correction Factor (CF)
UPV RA
10% vFA10%=1.4942vFA0%-2.521 [
FA10%
= u.2711[
FA0%
1.8769
20% vFA20%=1.3250vFA0%-1.4169 [
FA20%
= u.uS94[
FA0%
2.4494
30% vFA30%=0.6587vFA0%-1.8073 [
FA30%
= u.1199[
FA0%
2.2340
Table 5. Correction Factor (CF) applied with different fly ash content for UPV a
nd RA-
strength relationship of HPC with FA
3. Conclusion
This chapter emphasizes on ultrasonic pulse velocity (UPV) and relative
amplitude (RA)
method in assessing the strength of high performance concrete (HPC) wi
th mineral
admixtures such as fly ash. Adding mineral admixtures to replace certa
in percentages of
cement content increase or decrease the strength of HPC compare to th
e same concrete
combination containing only ordinary Portland cement (OPC). By using ex
tremely fine
pozzolanic material, such as fly ash, the compactness of the concrete
improves because the
fly ash fills the micro voids in grain packing. This feature results
in higher measured UPV
values in higher fly ash content concrete compared to lower fly ash
content concrete at the
same measured compressive strength. The strength-UPV correlation in HPC with fly
ash is
found to be poor (correlation coefficient, R
2
, reduces with increase of fly ash content). HPC
with silica fume samples produces good strength-UPV correlation but, as with the
case of fly
ash concrete, the silica fume content affects the measured UPV significantly. Th
e correction
factor for fly ash content is proposed. It is recommended that when establishing
iso-strength
curve for concrete with mineral admixtures, correction factor for each mineral a
dmixture is
included.
4. References
Carino, N.J. (1994). Nondestructive Testing of Concrete: History and Ch
allenges, In:
Concrete Technology Past, Present and Future, Mehta, P.K., (Ed.), 623-
678, ACI SP-
144, American Concrete Institute, Detroit, MI.
Demirbog, R., Trkmen, . and Karakoc, M. B. (2004). Relationship between
ultrasonic
velocity and compressive strength for high-volume mineral-admixtured conc
rete.
Cement and Concrete Research, Vol. 34, pp.23292336.
Galan, A. (1990). Combined Ultrasound Methods of Concrete Testing. Else
vier, Amsterdam,
North Holland.
Hamid, R. (2004). The Characteristics of Ultrasonic Wave in Hardened H
igh Performance
Concrete with Fly Ash and Silica Fume (in Malay), PhD Thesis, Univers
iti
Kebangsaan Malaysia, Malaysia.
Hamid, R., Yusof, K.M., Zain and M.F.M. (2010). A combined ultrasound method app
lied to
high performance concrete with silica fume, Construction and Building M
aterials,
Vol. 24, No, 1, (January 2010), pp. 9498, ISSN 0950-0618.
International Atomic Energy Agency TCS 17, Guidebook on Non-Destructive Testing
of Concrete
Structures, Vienna, 2002.
Ismail, M.P., Yusof, K.M. and Ibrahim, A.N. (1996). A combined ultraso
nic method on the
estimation of compressive concrete strength. INSIGHT, Vol. 38, No. 11, (November
1996), pp. 781-785.
Khatib, J.M. 2008. Performance of self-compacting concrete containing fl
y ash. Construction
and Building Materials, Vol. 22, No. 9, pp.1963-1971. ISSN 0950-0618.
Krautkramer, J. and Krautkramer, H. (1983). Ultrasonic Testing of Mater
ials. (3rd ed.).
Springer-Verlag, Berlin.
Malier, Y. (1992). Introduction, In: High Performance Concrete: From Material to
Structure, Yves
Malier, pp. (xiii xxiv), E & F Spon, ISBN 0419176004, London, UK.
PCA EB001, Design and Control of Concrete Mixtures, 14th edition, Port
land Cement
Association. 2002.
Price, W.F. and Hynes, J.P. 1996. In-situ testing of high strength co
ncrete. Magazine of
Concrete Research. Vol. 48, No.176, pp. 189-197.
Shah, S.P. 1997. High performance concrete: controlled performance concr
ete. Magazine of
Concrete Research. Vol. 49, No: 178, pp. 1-3.
Trkmen, I., z, A. and Aydin, A. C. (2010). Characteristics of workabili
ty, strength, and
ultrasonic pulse velocity of SCC containing zeolite and slag. Scientifi
c Research and
Essays, Vol. 5, No.15, pp. 2055-2064.
Uysal, M. and Yilmaz, K. (2011). Effect of mineral admixtures on prop
erties of self-
compacting concrete. Cement & Concrete Composites, Vol. 33, pp. 771776, ISSN 0958
-
9465.
10
Imaging Methods of Concrete
Structure Based on Impact-Echo Test
Pei-Ling Liu and Po-Liang Yeh
National Taiwan University, Institute of Applied Mechanics
Taiwan
1. Introduction
The impact echo method is a non-destructive test of concrete structure
s. This method was
developed in 1980s (Sansalone & Carino, 1986) and nowadays is extensively used b
ecause of
its simplicity and robustness. In the impact echo test, an impact for
ce is applied on the
surface of the target structure. Then, a transducer is used to measur
e to response of the
structure. Transforming the time signal to the frequency domain, one c
an determine the
depth of the reflector beneath the test point.
The impact echo test is a pointwise detection method. The non-destructive examina
tion of
a structure usually requires a large amount of tests. How to integrate the tes
t results to get
an overall picture about the condition of the structure is a challeng
ing task. In order to
simplify the interpretation, many imaging methods were proposed, providi
ng direct
information on the concrete interior.
This chapter introduces several imaging methods based on the impact ec
ho test, including
spectral B- and C-scan, spectral tomography, surface rendering, and vol
ume rendering.
These imaging methods can be used to determine the size and location of internal
defects in
concrete structures. Several numerical and experimental examples are given to il
lustrate the
results of these imaging methods. The features of these methods are also compare
d.
2. The impact echo test and depth spectrum
2.1 Impact echo test
In the impact echo test, a steel ball or a hammer is used to produ
ce a wave source on the
surface of the structure. Consequently, stress waves are generated and
propagate in the
structure. A transducer is placed near the impact point to measure th
e response of the
structure. Then, the received signals are recorded for data analysis.
If there is an interface beneath the test point, the longitudinal wav
es generated by the
impact will bounce between the top surface and the interface. Such echo waves wi
ll form a
peak in the spectrum of the signal. The peak frequency f and the depth of the in
terface d is
related by the following formula (Sansalone & Carino, 1986):
d=C
p
/2f (1)
236
where C
p
is the velocity of the longitudinal wave. Equation 1 is valid only when the aco
ustic
impedance of the material on the other side of the interface is less
than that of concrete. If
the material is stiffer, e.g., steel, the factor 2 in Eq. 1 should be replaced b
y 4.
It is proposed in the literature (Gibson & Popovics, 2005; Lin & Sansalone, 1992
; Sansalone &
Streett, 1997; Schubert & Khler, 2008; Zhu & Popovics, 2007) that the depth obtai
ned by Eq.
1 be multiplied by a correction factor to eliminate the discrepancy b
etween the actual and
predicted depths. The correction factor is dependent on the geometry o
f the structure. Its
value has been determined for a few types of structural members. When imaging me
thod is
adopted to examine structures with defects, the geometry of the target structure
is complex.
It is almost impossible to choose an appropriate value for correction
factor. Hence, Eq. 1 is
used directly in the imaging process.
The Fourier analysis is the most widely used approach to construct the spectrum
of the test
response. However, the Fourier spectra usually contain ripples and mult
iple peaks
generated by the transform process. Such artificial interferences may j
eopardize the
interpretation of test results (Yeh & Liu, 2008). Several timefrequency
techniques have
been proposed in the literature to prevent the interferences, for exam
ple, the wavelet
transform and the HilbertHuang transform (Abraham et al., 2000; Algernon
&
Wiggenhauser, 2005; Lin et al., 2009; Shokouhi et al., 2006; Yeh & Liu, 2008). H
owever, there
is a trade-off because none of these methods would produce echo peaks
as sharp as the
Fourier transform does.
2.2 Depth spectrum
No matter which transform is adopted, the horizontal axis of the spectrum is the
frequency,
not depth. For the imaging purpose, it is preferable to have depth as the horizo
ntal axis. The
horizontal axis of the spectrum can be easily transformed to the depth axis by a
pplying Eq.
1, as proposed by Yeh and Liu (2009). The frequency-depth transformati
on procedure is as
follows:
Suppose ( ) a f is the original spectrum of a signal.
1. Select an appropriate depth interval z A .
2. Apply Eq. 1 to , 1, 2,... i z i A = to find the corresponding fre
quency
i
f .
3. Determine the maximum amplitude
i
a in each interval ( )
1
,
i i
f f

, 1, 2,... i = , as shown
in Fig. 1.
4. Plotting
i
a versus i z A yields the depth spectrum of the signal.
Notice that
i
a , instead of ( )
i
f a , is used to plot the depth spectrum. This is to insure that no
peak is left out in the depth spectrum since peaks are the most important inform
ation in the
spectrum. One should also know that the transformation cannot start fr
om z=0 because it
maps to = f .
The depth spectrum constructed as above has a constant depth interval, which is
convenient
for image processing. However, the data points in the Fourier spectrum
are not totally
retained. Therefore, some of the details in the Fourier spectrum are
not reproduced in the
depth spectrum. The peaks, which contain the most important information
, are preserved,
Imaging Methods of Concrete Structure Based on Impact-Echo Test
237
but they may not appear at the precise depth. The maximal depth devi
ation of a peak is
2 z A . Therefore, it is advisable to make the depth interval as short as possib
le.
Fig. 1. Frequency-depth transformation
The depth spectrum is useful for the imaging of impact echo data. It has another
advantage
if the depth D of the concrete structure is known. In the impact echo test, seve
ral vibration
modes of the structure will be induced. The impact echo is one of them. Unfortun
ately, the
peaks caused by the lowermodes are often higher than the echo peak in the spectru
m. This
complicates the interpretation of results. The frequencies of the lower
modes usually
correspond to depths greater than D. Hence, if the depth spectrum is
drawn only for the
depth range 0 z D, the lowermode peaks do not appear. That simplifi
es the image
interpretation.
3. Imaging methods
In general, the imaging of impact echo data contains three steps: dat
a acquisition, data
construction, and image rendering. The details of these steps will be
introduced in the
following sections.
3.1 Data acquisition and construction
To acquire data for image rendering, one has to conduct a series of impact echo
tests on the
target structure. Firstly, select a coordinate system such that the x-
y plane coincides with
surface of the concrete. Secondly, draw a mesh on the concrete surfac
e, and perform the
238
impact echo test at each grid of the mesh, as shown in Fig. 2(a). The location o
f the impact
and receiver can be chosen arbitrarily inside the grid. However, their
midpoint should be
located at the center of a grid.
Then, apply the Fourier transform or other transforms to the test sig
nals to obtain the
frequency spectra. In tomography and 3D imaging, the spectra are further transfo
rmed into
the depth spectra.
After obtaining the spectra, one can proceed with the construction of data. Assu
me that the
numbers of test points along the x and y directions are n
x
and n
y
, respectively, and each
spectrum contains n
z
data points. All the spectra are assembled into an
x y z
n n n matrix,
| | , , V i j k . The array
| | , , 1
z
V i j k n s s is simply the amplitude of spectrum under test
point (i, j).
After the volume data is constructed, one may apply any of the imaging methods t
o depict
the interior of concrete structure.
Notice that in tomography and 3D imaging, each vector
| | , , 1
z
V i j k n s s is a depth
spectrum. Suppose the grid size of the test mesh is x y A A and th
e depth interval of the
depth spectrum is z A , each element in | |
, , V i j k corresponds to a voxel (volume element) in
the space with voxel size x y z A A A . Hence, the matrix maps to a
rectangular solid with
side lengths L
x
, L
y
, and L
z
, where
x x
L n x = A ,
y y
L n y = A , and
z z
L n z = A . In this case, | |
, , V i j k
represents the reflection energy from voxel | |
, , i j k in the rectangular solid.
In spectral B- or C-scan, on the other hand, each vector
| | , , 1
z
V i j k n s s is a frequency
spectrum. Hence, the third axis is frequency, not depth. Therefore, the matrix d
oes not map
to a rectangular solid in the space.
(a)
(b)
Fig. 2. (a) Test mesh on the concrete surface and (b) side view of the example s
pecimen
In this chapter, both numerical simulations and model tests are presen
ted to illustrate the
imaging methods. A concrete specimen with an slant internal crack is
considered. The
dimension of the specimen is 80 cm (L) 80 cm (W) 20 cm (H). The depth of the int
ernal
crack ranges from 6 cm to 14 cm beneath the surface, as shown in Fig. 2(b). The
coordinates
of the four corners of the crack are [x, y, z] = [24, 24, 14], [24, 56, 14], [56
, 24, 6], and [56, 56, 6]
(in cm).
239
The test mesh on the specimen is a 76 cm 76 cm square with 19 19 grids, leaving
2 cm of
margins on the four sides. In each test, the impact source and the
receiver were located at
the upper right and lower left corners of a grid, respectively.
In the numerical simulation, the finite element code LSDyna970 (Hallquis
t, 2003) was
adopted to simulate the response of the concrete specimen due to the impact of a
steel ball.
Threedimensional solid elements with side length 1cm were used in the
numerical
simulation. The mass density, Youngs modulus, Poissons ratio, and the longitudinal
wave
speed of the concrete are 2300 kg/m
3
, 33.1 GPa, 0.2, and 4000 m/s, respectively.
A timevarying pressure was applied to the surface to simulate the impa
ct of a steel ball
with a diameter of 6 mm. According to Goldsmith (1960), the pressure was approxi
mated by
a halfsine function with a contact time t
c
= 25 s. The total time of simulation was 3 ms, and
the time increment was 3 msec/1024 = 2.93 s.
Nonreflecting boundary conditions were applied on the four sides of the blocks to
prevent
waves reflecting from the boundaries. As such, the examples could focu
s on the results of
imaging methods, excluding the effect of specimen geometry.
In the model test, the dimension of the concrete specimen and the cr
ack location are
identical to those of the numerical model. The longitudinal wave velocity of the
concrete is
3890 m/s. In the experiment, a steel ball with diameter 6 mm was dropped on the
concrete
surface to produce the impact source. A conical transducer, developed
at the Institute of
Applied Mechanics, National Taiwan University, was adopted to measure t
he vertical
displacement on the concrete surface. The impact and receiver locations were the
same as in
the numerical examples. The received voltage signals were enhanced by
an amplifier and
then recorded by a digital oscilloscope (LeCroy WavePro940). In the mo
del tests, the
sampling rate was 1 MHz, and the total sampling time was 5 msec.
Unlike the numerical tests, the energy of the impact source in the model tests m
ay vary from
test to test. Therefore, the experimental data were normalized based on the ampl
itude of the
surface wave.
3.2 Spectral B-scan and C-scan
The spectral B and Cscan methods were firstly proposed by Liu and Yiu (2002) to d
etect
the internal cracks in concrete. The spectral Bscan was adopted by Schubert et al
. (2004) to
measure the thickness of concrete specimens. Kohl et al. (2005) further extended
the idea so
that one could combine impact echo data with ultrasonic data to const
ruct B and Cscan
images.
The concepts of spectral B- and C-scans are similar to the ultrasonic
B- and C-scans. They
respectively generate the images of vertical and horizontal cross-sections of a
test specimen.
The data acquisition and construction for spectral B- and C scans are the same a
s described
in Section 3.1. However, in spectral B-scan one only has to perform impact echo
tests along a
test line. Therefore, n
y
=1 in B-scan.
For the purpose of imaging, the matrix | | , , V i j k needs to
be transformed into a matrix of
color scales | | , , c i j k :
240
| |
| |
| |
| |
| |
max max
min
max min max
max min
min
, ,
, ,
, , , ,
0 , ,
c V i j k V
V i j k V
c i j k c V V i j k V
V V
V i j k V
>
= < <
<

(2)
where
max
c is the upper bound of the color scale, and | |
min max
, V V defines the range in
which | | , , V i j k is mapped linearly to | | , , c i j k
. If | | , , V i j k exceeds
max
V , the color scale is
set to
max
c ; if | | , , V i j k is less than
min
V , the color scale is set to 0. Unless specified
otherwise,
max
V and
min
V are chosen to be the maximum and minimum of the volume data,
respectively. However, one can increase the value of
min
V to suppress the noise and
enhance the contrast of the image (Liu & Yeh, 2010). The contrast ratio
min max
V V that yields
a satisfactory result can be obtained by adjusting the ratio manually.
In spectral B-scan, the image of the vertical section under the test
line can be obtained
simply by using | | k j i c 1, , = to generate a 2D density plot.
Notice that | | k j i c , , reflects
spectral amplitude. If a peak appears in the ith spectrum at frequenc
y f kA , the
corresponding color scale is high and one can see a clear spot at location x i
x A = , f k f A = on
the B-scan image. The clear spots will connect into a clear stripe i
f a crack exists. As such,
one can detect the size and location of the crack in the concrete.
The spectral C-scan constructs an image for a horizontal section. Hence, one has
to select the
depth of the horizontal section to be examined. Suppose the depth cor
responds to
frequency f K f A = , according to Eq. 1. The spectral C-scan image
of the horizontal section
can be obtained by using | | K k j i c = , , to generate a 2D density plot
.
Theoretically, the color of the image is homogeneous if no defect exi
sts. If the section does
contain a defect, there will be peaks in the spectra. Hence, one can
determine if there is a
defect simply by examining the color variation of the image.
In the following, numerical examples are given to illustrate the spect
ral B- and C-scan
methods. Figure 3 shows the spectral B-scan images of the vertical se
ctions under test
line (a) x=16 cm and (b) x=40 cm. The crack occurs in the range 24 56 y s s c
m. Hence, the
first section contains no crack while the second section does. One ca
n see that there are
only horizontal stripes in Fig. 3(a). The stripe with the highest color scale (r
ed) appears at
f=10 kHz. Since the longitudinal wave velocity C
P
4000 m/s, this frequency
corresponds to the depth of the bottom, 20 cm, according to Eq. 1. The bright st
ripe near
20 kHz is caused by the multiple peaks of the Fourier spectra. There are
also red stripes
at the bottom of the image. They are induced by the lower-mode vibra
tions of the
specimen.
Figure 3(b) looks quite different from Fig 3(a). One can find an inclined red st
ripe occurring
in the crack range. Outside that range, the image resumes the no-crac
k pattern, that is, red
horizontal stripes appearing near f =10 kHz, denoting the bottom of the specimen
. Although
the B-scan image reveals the existence of the inclined crack, it does not exhibi
t the profile of
the specimen under the test line. This is certainly because the verti
cal axis is frequency,
not depth.
241
Figure 4 shows the spectral C-scan images of the horizontal sections at depth (a
) 4 cm, (b) 10
cm and (c) 20 cm. Figure 4(a) looks just like a blue rectangle, implying that th
e cross-section
contains no reflector. The image in Fig. 4(b) contains a red zone, depicting the
crack on the
cross-section. The yellow and cyan zones in the image both denote con
crete, but the cyan
zone is the concrete under the crack. Figure 4(c) shows the C-scan image at the
bottom of the
specimen. A cyan zone appears at the center of the image, surround b
y red zone. It can be
considered as the shadow cast by the crack because the waves are blocked by the
crack. The
cyan zone is not a square because stress waves may go around the crack edge and
reach the
bottom when the test point is near the top of the edge. This zone provides a sup
plementary
evidence for the existence of a defect above this region.
Y (cm)
F
r
e
q
u
e
n
c
y
(
k
H
z
)
0 20 40 60 80
70
60
50
40
30
20
10
0
Y (cm)
F
r
e
q
u
e
n
c
y
(
k
H
z
)
0 20 40 60 80
70
60
50
40
30
20
10
0
(a) (b)
Fig. 3. Spectral B-scan of cross-sections under test lines (a) x=16 cm and (b) x
=40 cm
X (cm)
Y
(
c
m
)
0 20 40 60 80
80
60
40
20
0
X (cm)
Y
(
c
m
)
0 20 40 60 80
80
60
40
20
0
X (cm)
Y
(
c
m
)
0 20 40 60 80
80
60
40
20
0
(a) (b) (c)
Fig. 4. Spectral C-scan of cross-sections at depth (a) 4 cm, (b) 10 cm, and (c)
20 cm
3.3 Spectral tomography
In the spectral Bscan method, the Fourier spectra of the test signals
are assembled to
construct an image of the test crosssection. Such image certainly provi
des useful
information about internal defects. However, the vertical axis of the
image is frequency.
Hence, the spectral B-scan does not provide a picture of the test section.
In order to provide a more intuitive image, the vertical spectral tomography was
proposed
by Liu and Yeh (2010) . The imaging process of the vertical spectral tomography
is the same
as in spectral B-scan except that the Fourier spectra are replaced by
depth spectra. Liu and
Yeh (2011) extended the notion further to tomography of arbitrary cros
ssections. As such,
the inspector could examine the interior of a structure from various
angles to get better
understanding of its condition. The spectral C-scan is only a special
case of spectral
tomography.
242
In spectral tomography, the volume data is constructed using the depth
spectra of the
impact echo test. Similar to the spectral B- and C-scan, the matrix
| | , , V i j k is transformed
into a matrix of color scales | | , , c i j k according to Eq. 2. Then,
the matrix | | , , c i j k is used to
produce tomograms for designated crosssections.
Consider a crosssection defined by 0 = + b
T
x n , where n is the outward normal of the
crosssection, and | | z y x , , = x is the position vector. The spec
tral tomogram can be
generated as follows:
1. Define a new coordinate system x such that the x y plane
coincides with the
crosssection, as shown in Fig. 5(a). The coordinates in the new and o
ld systems are
related by
( )
T
= x Q x t
(3)
where Q is the rotation matrix with component ( , )
i j
Q i j = e e , and t is the translation
vector.
The coordinate transformation defined in Eq. (3) is not unique. t can
be easily defined by
choosing a point on the cross-section as the origin of the new coord
inate system. Q can be
constructed as follows. Suppose | |
, ,
T
p q r = n is the outward normal of the crosssection,
where
2 2 2
1 p q r + + = . The new base vectors can be selected as
z
= e n ,
x z z
= e e e , and
y z x
= e e e . Hence,
2
1
1
0
x
q
p
r
(
(
=
(
(

e
2
2
1
1
1
y
pr
qr
r
r
(
(
=
(

e
z
p
q
r
(
(
=
(
(

e
(4)
and
2 2
2 2
2
1 1
1 1
0 1
q pr
p
r r
p qr
q
r r
r r
(
(

(
(
=
(
(
(
(
(

Q
(5)
2. Obtain the orthographic projection of the
x y z
L L L solid on the x y plane, and find
a coordinate rectangle to enclose the projection, as shown in Fig. 5(b).
The orthographic projection of the solid can be obtained by projecting
the vertices of the
solid on the x y plane. The location of a projected vertex is
simply the new coordinates
( ) , x y of that vertex obtained by Eq. (3). Suppose ( ) ,
x y satisfies
min max
x x x s s and
min max
y y y s s for every vertex. Then, the bounding rectangle is for
med by the coordinate
lines
min
x x = ,
max
x x = ,
min
y y = , and
max
y y = .
243
3. Draw a mesh of square grids on the rectangle. For each grid, using Eq. (3) t
o transform
the location of its center
| | , , 0
c c c
x y = x back to the original coordinate system, i.e.,
c c
= + x Qx t
(6)
4. Use
c
x to determine in which voxel the center is located. Then, determine
the color
scale of the grid
( )
c
c x from the volume data. If
c
x is outside the volume, no color scale
is assigned and the pixel is transparent.
5. Construct the tomogram of the crosssection by filling each grid with its colo
r
( )
c
c x .
Notice that in Step 3, each grid in the mesh corresponds to a pixel
in the tomogram. It is
advisable to adopt a fine mesh so that one can obtain a high-quality image. The
resolution of
the tomogram is certainly limited by the voxel size of the volume data. However,
a pixel in
the tomogram may pass through more than one voxel in the solid. A f
ine mesh helps to
display the boundary between adjacent voxels more precisely. Furthermore
, it helps to
better delineate the borderline of the cross-section in the tomogram.
Take the cross-section
in Fig. 5 for example. The borderline of the cross-section is a hexagon. Since o
nly pixels with
its center located inside the volume are colored, the tomogram of the cross-sect
ion appears
as the colored polygon in Fig. 5(b). It is seen that some segments
of the borderline become
zigzag. Apparently, the zigzag borderline approximates the true boundary
better if a finer
mesh is adopted.
The proposed spectral tomography does not provide the velocity profile of a test
section as
most conventional nondestructive techniques do. Instead, the spectral tomogram s
hould be
considered as a profile of reflection energy due to the impact.
In the following, a numerical example is presented to illustrate the
spectral tomography.
Figure 6 shows the horizontal tomograms constructed at depth (a) 20 cm, (b) 12 c
m, and (c) 9
cm, respectively. Similar to Fig. 4, the crack casts a shadow on the bottom tomo
gram in Fig.
6(a). Bright stripes appear in Figs. 6(b) to (c), depicting the crack on each cr
oss-section. Since
this is a slant crack, the bright stripes shift along the y direction a
s the cross-section moves
up. In this case, it is difficult to get an overall picture of the
crack using horizontal
tomograms alone.
Figure 7 shows the vertical tomograms along x= (a) 18 cm and (b) 40
cm, respectively. No
crack presents on the first cross-section. Hence, one only finds bright stripes
at the bottom of
Fig. 7(a). In contrast, the slant crack is clearly depicted on the tomogram in F
ig. 7(b). One can
use this tomogram to determine the length and inclination of the crack.
It should be mentioned that the thickness of the bright stripe in Fig. 7(b) does
not represent
the thickness of the crack. In fact, the impact echo test cannot provide informa
tion about the
thickness of the crack. The thickness of the bright stripe results fr
om the width of the echo
peak in the depth spectra. Hence, the true depth of the crack should be determin
ed based on
the location of the brightest pixels in the tomogram.
One may construct oblique tomograms to get a picture of the whole crack, as show
n in Fig.
8. The slopes of the cross-sections vary from 0 to 90. One can see that the area o
f the crack
image is maximal when the slope is around 14, coincident with the cra
ck orientation.
Hence, Fig. 8(c) provides the best picture of the crack among the to
mograms. The crack
244
image is narrower near the shallow edge of the crack. Nevertheless, on
e can use the bright
zone to estimate the size of the crack. This is quite difficult if
only vertical or horizontal
tomograms are available.
Fig. 5. (a) Coordinate transformation in tomography and (b) orthographic project
ion of
specimen on the x y plane
Fig. 6. Horizontal tomograms at depth (a) 20 cm, (b) 12 cm and (b) 9 cm
) , (
max
max
y x
) , (
min
max
y x
) , (
max
min
y x
) , (
min
min
y x
(b)
(c) (a) (b)
(a)
245
Fig. 7. Vertical tomograms along x = (a) 18 cm and (b) 40 cm
Fig. 8. Oblique tomograms with slopes (a) 0, (b) 8, (c) 14, (d) 24, (e) 60, and (f) 9
0
This numerical example demonstrates that the inspector can use spectral
tomography to
examine any cross-section of a specimen. To get an overall assessment
of the interior
condition, it is advisable to examine the specimen in a systematic wa
y rather than scan
randomly. An inspection procedure is proposed herein: Firstly, construct
a horizontal
tomogram at the bottom of the specimen to find the defect zone. Then, construct
a series of
horizontal or vertical tomograms to find the location, size, shape, an
d orientation of the
defect. Finally, based on the scanning results, perform oblique tomogra
phy to get a better
image of the defect if necessary.
Figure 9 shows the results of spectral tomography in the model test. From Fig. 9
(a), one can
see the dark zone also forms beneath the crack. Figure 9(b) is the
vertical tomogram along
x=44 cm. The result is similar to that of numerical test. Bright zon
es occur at the crack and
bottom. Despite noise, one can easily detect the inclination of the crack using
this tomogram.
Figure 9(c) shows the oblique tomogram of the crack plane. A bright
zone appears in the
crack area.
0
0
0
8
0
14
0
24
0
60
0
90
(a) (b)
(b)
(a) (c)
(d)
(e)
(f)
246
Fig. 9. Model test, (a) horizontal tomogram at the bottom, (b) vertical tomogra
m at x=44 cm,
and (c) oblique tomogram.
Although the test data have been normalized by the amplitude of the
surface wave, the
tomograms still look mottled. The deterioration of tomogram quality may
come from
variation of impact force, random noise, measurement error, and non-uni
form material of
the model test. Therefore, it is advisable to compare the tomograms o
f different angle and
sectioning before making judgment. Regardless of the noise, the bright
and dark zones in
these tomograms still reveal the location, orientation, and size of the crack.
3.4 Surface rendering
With the progress of computer graphics, 3D display becomes a trend in
the processing of
volume data. The methods that render 3D images are called volume visu
alization
techniques. There are two branches techniques, namely, surface rendering
and volume
rendering. The surface rendering technique was proposed by Yeh and Liu
(2009) to depict
the internal cracks in concrete structures.
The surface rendering method is equivalent to drawing contour lines in
a 2D density plot.
Consider the vertical tomogram in Fig. 10(a) for example. If one selects an iso-
value 104, one
gets contour line 1; if one chooses a different iso-value 208, contou
r line 2 is obtained, as
shown in Fig. 10(b). Apparently, if the iso-value is chosen properly,
the contour line will
depict the location of an interface. The same idea can be extended to the 3D cas
e.
Fig. 10. 2D analogy of surface rendering (a) vertical tomogram and (b) contour
plot
The main idea of surface rendering is to abstract the iso-surface from the volum
e data, i.e., to
find a surface with the same spectral amplitude. Let ( , , ) V x y z denote
the volume data. The
iso-surface corresponding to iso-value C can be represented as:
{( , , ) : ( , , ) } x y d V x y d C = (7)
X (cm)
D
e
p
t
h
(
c
m
)
20 40 60
0
4
8
12
16
20
0 80
1
2
1 1
(b)
X (cm)
D
e
p
t
h
(
c
m
)
0 20 40 60 80
0
4
8
12
16
20
(a)
(a) (c) (b)

247
Once an iso-value is assigned, one can use triangular patches to gene
rate an iso-surface.
Similar to the contour lines in Fig. 10, if the iso-value is chosen properly, th
e iso-surface will
depict the location of an interface in the specimen.
After the iso-surface is generated, it is projected to a 2D view plane. Notice t
hat when a 3D
object is projected to a 2D plane, a sphere will appear as a circle and a cube i
s turned into a
hexagon. In order to obtain a stereograph, shading and lighting are necessary.
The Phong reflection model is a popular and effective approach for th
is end (Angel, 2006).
This model assumes three types of light-material interactions, namely, ambient,
diffuse, and
specular reflection. The intensity of the reflected light is dependent
on four vectors: the
normal vector of the surface N, the viewer vector V, the light sourc
e vector L, and the
reflected ray vector R.
The ambient light has the same intensity in the space. When it encou
nters a surface, it is
absorbed and reflected. The intensity of the ambient reflection I
a
is as follows:
a a a
I K L = 0 1
a
K s s (8)
where K
a
is the ambient reflection coefficient and L
a
is the intensity of the ambient light.
The diffuse reflection is characterized by the roughness of the surfac
e. The intensity of the
reflected light depends on the material and incident direction of the
light. Since each point
on the surface has a different normal vector, it will reflect differe
nt amount of light. The
intensity of the diffuse reflection I
d
is as follows:
d d d
I K L = L N 0 1
d
K s s (9)
where K
d
is the diffuse reflection coefficient and L
d
is the intensity of the incident diffuse
light.
The specular reflection is used to highlight the shiny part of surface. Although
the ambient
and diffuse reflection make the image look three-dimensional, the lack of specul
ar reflection
would make the surface look dull. The intensity of the specular reflection is as
follows:
( )
s s s
I K L
|
= R V 0 1,
s
K s s 1 | > (10)
where K
s
is the diffuse reflection coefficient, | is the shininess coefficient
, and L
s
is the
intensity of incident specular light. Usually,
d s
L L = . As | increases, the reflected light tends
to concentrate on a smaller region.
The total intensity of the reflected light from an object is the sum
of I
a
, I
d
, and I
s
. The
coefficients K
a
, K
d
, K
s,
and | are taken as 0.6, 0.8, 0.5, and 150, respectively, in the following
examples.
Figure 11 shows the top, side, and oblique views of the surface rend
ering image of the
numerical model. An iso-surface denoting the crack is observed in all
three images at the
correct location. One can also find the bottom of the specimen in the image. How
ever, a hole
is formed in the bottom beneath the crack, as shown in Fig. 11(a). This is becau
se the waves
are blocked by the crack and cannot reach the bottom.
248
From the side view in Fig. 11(b), one can see that the crack is th
icker than the real crack.
Actually, the impact echo test cannot provide thickness of the crack. The thickn
ess of crack
in the image results from the width of echo peak in the spectra. Therefore, the
true depth of
the crack is around the center of the iso-surface.
(a) (b) (c)

Fig. 11. Surface rendering of numerical model, (a) top view, (b) side view, and
(c) oblique
view
In practical applications, it is critical to choose a proper iso-value
because different iso-
values result in different images. The iso-value of the images in Fig
. 11 is 15. Figure 12
shows a series of surface rendering images with iso-values 12, 18, an
d 22, respectively.
Clearly, the crack iso-surface shrinks as the iso-value increases. If
the iso-value is too low,
the iso-surface denoting the crack becomes a large layer. On the other hand, if
the iso-value
is too high, the crack iso-surface becomes excessively small. Since the hole can
be considered
as the shadow of the crack, the iso-value should be chosen such that the sizes of
the crack
and the hole match.
The surface rendering images of the experimental model are shown in F
ig. 13. Because the
experimental signals are contaminated by noise, the 3D images are not
as clear as in the
numerical example. The crack is not enveloped in a single iso-surface.
Several iso-surfaces
appear around the crack location instead. This is mainly because the
impact source is
unsteady. Normalization of the signals may reduce the influence of the
intensity of the
source function, but not the shape. Therefore, it is hard to find an
appropriate iso-value to
surround the crack by a single iso-surface.
Nevertheless, one can still manage to find the crack by viewing the
specimen at different
angles. For example, the side view in Fig. 13(b) provides a clear picture of the
inclined crack.
Furthermore, the hole beneath the crack is still visible in Figs. 13(a) and (c).
3.5 Volume rendering
The volume rendering technique was proposed by Yeh and Liu (2008) to
construct 3D
images based on impact echo data. The data acquisition and construction procedur
es are the
same as described in Section 3.1.
249
(a) (b) (c)
Fig. 12. Surface rendering of the numerical model with iso-values (a) 12, (b) 18
, and (c) 22
(a) (b) (c)
Fig. 13. Surface rendering of the experimental model, (a) top view, (b) side vie
w, and (c)
oblique view
In the volume rendering method, each voxel is assigned an opacity o based on t
he volume
data such that 0 1 o s s . The opacity represents the level of diffi
culty that light goes
through a voxel. If a voxel is complete opaque, 1 o = , and if it i
s complete transparent,
0 o = . The opacity of a voxel can be determined as follows:
| |
| |
max
min
min max
max min
min
1
, ,
, ,
0
V V
V i j k V
i j k V V V
V V
V V
o
>
= < <
<

(11)
As in spectral tomography, the default values of
max
V and
min
V are the maximum and
minimum of the volume data, respectively.
The process of volume rendering is analogous to X-ray examination. To
construct the
volume rendering image, let parallel rays emitted from a light source
behind the volume
transmit through the volume and reach a projection plane, as shown in
Fig. 14(a). When a
ray passes through the volume, it accumulates the opacity of the voxe
ls it encounters, as
shown in Fig. 14(b). The compositing operation is a recursion of opac
ity, as shown in the
following (Angel, 2006):
(1 )
out in
o o o o = + (12)
250
where o is the opacity of the current voxel,
in
o is the accumulated opacity entering the
voxel, and
out
o is the accumulated opacity leaving the voxel. If 1 o = , then 1
out
o = and the
light is totally obstructed. On the other hand, if 0 o = , then
out in
o o = and the light remains
unchanged. Therefore, o is an indicator of the degree that light penetrates th
e voxel.
Equation 12 is applied recursively to determine the accumulated opacity
of a ray until it
leaves the volume and reaches the projection plane. After the accumula
ted opacity is
obtained for each ray penetrating the volume, one can draw a density
plot of the
accumulated opacity on the projection plane. This method is called the
ray casting (Watt,
2000). Although the ray casting method describes the concept of imaging clearly,
it is time-
consuming. Hence, the texture mapping technique has been proposed to s
peed up the
imaging process (Engel et al., 2006).
Recall that in the impact echo test, an interface will induce a peak in the spec
trum. Hence, if
a defect appears in the volume, the spectral amplitude along the defe
ct is large, so is the
opacity. Apparently, the rays that pass through the defect will be dimmer than t
he rays that
do not. Hence, one will find a shadow in the image if there is a defect in the v
olume.
Fig. 14. The volume rendering method

The signals obtained in the impact echo tests inevitably contain noise. That may
downgrade
the quality of image and make the diagnosis difficult. This problem c
an be tackled by
adjusting the relation between the volume data and the opacity, as defined in Eq
. 11. It does
not help to alter the value of
max
V . Hence, one can simply use its default value. The value of
min
V can be increased to suppress noise and enhance the contrast of image.
Figure 15 shows the influence of
min max
V V on the volume rendering image. Generally
speaking, as
min max
V V increases, the crack image gets clearer and the hole at the
bottom
gets larger. With
min max
V V below 20%, the image looks blurry, as seen in Figs. 15(a) and
(b). When
min max
V V is increased to 30%, the crack and the hole at the bottom beco
me
visible. When
min max
V V = 40%, one gets a very good image of the specimen.
However, the contrast ratio should not be over raised. As
min max
V V reaches 50%, the crack
starts to shrink and the size of the hole exceeds that of the crack. The situati
on is even worse
as
min max
V V = 60%.
light
(a)
projection plane
light
(b)
251
Notice that the optimal value of
min max
V V may change from case to case. The inspector has
to try different value to see which value yields the best result.
Fig. 15. Volume rendering of numerical model using

min max
V V = (a) 10 %, (b) 20 %, (c) 30 %,
(d) 40 %, (e) 50 %, (f) 60 %
Figure 16 show the volume rendering images of the experimental model with variou
s view
angles. The contrast of image is
min max
V V =30%. The test data contain a lot of noise. Hence,
the crack appears as a cluster of dark patches in the image.
In Fig. 16(a), one can find a cluster of dark patches in the central area of the
image, denoting
the crack. As one rotates the model, the hole at the bottom becomes
visible, indicating the
existence of a defect. The side view in Fig. 16(d) provides a clear view of the
crack. The size
and location of the crack can be estimated based on this image.
Although these images are not as clear as in the numerical examples, one can sti
ll locate the
crack by viewing the specimen from different angles. This is best don
e with an interactive
imaging program that allows the inspector to interactively adjust the
view angle and the
contrast ratio. With the aid of such program, one can easily manipula
te the image.
Furthermore, as one adjusts the view angle gradually, the image becomes stereosc
opic. That
helps the inspector to interpret the image. Unfortunately, such effect
cannot be
demonstrated in this book.
(f) (d) (e)
(c)
(b) (a)
252
Fig. 16. Volume rendering of experimental model test with various view angles
4. Conclusion
This chapter introduces several methods to construct the image of conc
rete interior using
impact echo data, including the spectral B-scan, spectral C-scan, spect
ral tomography,
surface rendering, and volume rendering. With these imaging methods, th
e inspector may
examine the interior of a structure to get better understanding of its health co
ndition.
The imaging procedure contains three steps: data acquisition, data cons
truction, and image
rendering. Basically, the first two steps are the same for all the i
maging methods. Firstly, a
series of impact echo tests are performed at the grids of a mesh on the surface
of the concrete.
Then, the time signals are transformed into frequency spectra or depth spectra.
Assembling the
spectra into a 3D matrix yields the volume data, which could be used to construc
t images.
The spectral B- and C-scan are derived from ultrasonic scan. The spectral B-scan
constructs a
2D density plot of the spectral amplitude on the vertical section under a test l
ine; while the
spectral C-scan constructs the plot for a horizontal section. Because
the vertical axis of the
spectral B-scan is frequency, it does not provide the profile of a v
ertical section. One may
judge whether there is an internal defect by examining the discontinui
ty of horizontal
stripes in B-scan. However, the size and location of the defect canno
t be determined from
the image directly.
The spectral C-scan, on the other hand, provides the profile of a ho
rizontal section. Thus,
one can use the image to determine the size and location of an internal defect.
However, it is
sometimes difficult to get an overall picture of the concrete interior
by viewing horizontal
sections alone.
The spectral tomogram can be considered as an extension of the spectr
al C-scan. It can be
used to construct the profile image for arbitrary crosssections. The in
spector can observe
the interior of a structure from different angles and by different se
ctioning. As such, the
(a)
(a)
(c)
(b)
(d)
253
internal defects in concrete structures can be easily located. Through
the numerical and
experimental examples, it is seen that the spectral tomography can depict the in
ternal crack
of the concrete specimen successfully. When a crack exists in the con
crete, it appears as a
bright zone in the tomogram. Furthermore, no bright stripes appear beneath the c
rack at the
bottom of the tomogram because the waves are blocked by the crack. T
his provides
supplementary information about the size and location of the crack.
Surface rendering and volume rendering are 3D imaging techniques. The
idea of surface
rendering is to abstract the iso-surface from the volume data. In surface render
ing, a defect
is represented by one or several iso-surfaces, which can be used to
estimate its size and
location. Same as in spectral tomogram, a hole, approximate the size of defect,
appears at the
bottom beneath the defect.
Volume rendering is analogous to X-ray examination: Parallel rays are
generated, transmit
through a specimen, and reach a projection plane. If a defect exists
in the specimen, the
voxels covering the defect would have high opacity. Hence, a dark zon
e forms in the
volume rendering image. If the view angle is chosen properly, one could also fin
d a hole at
the bottom beneath the defect.
Comparing these methods, one can see that each method has its own st
rength and
weakness. Volume rendering is a robust technique; it is not sensitive to the int
erferences in
spectra. Surface rendering is sensitive to noise, but it can depict t
he details of a defect.
Spectral tomography is robust and insensitive to noise. However, it do
es not provide a 3D
image and one can only view the specimen by sectioning.
To maintain good balance between robustness and precision, the inspecto
r should take
advantage of the strength of each method. One may apply volume render
ing to get an
overall picture of the specimen and to find the approximate location
of the defect, if any.
Then, use surface rendering or spectral tomography to observe the details.
It is seen in the experimental examples that the presence of noise downgrades th
e quality of
images, no matter which method is adopted. Unfortunately, the test dat
a is always noisy
and the impact source is unsteady in real applications. The images obtained are
sometimes
difficult to interpret. In that situation, an interactive imaging program is ind
ispensable. With
the interactive graphic interface, one can adjust the imaging parameter
s or view angle
arbitrarily and get an updated image instantly. As such, one may attain a better
view of the
specimen easily. More importantly, the 3D image appears stereoscopic as
one changes the
view angle gradually. Therefore, one can tell which object is in the front and w
hich is in the
back. This is useful especially when the quality of the image is poor.
The imaging methods presented in this chapter may provide the most di
rect information
about the defects in concrete structures. However, its practical applications ar
e hindered by
two issues. Firstly, it is very time-consuming because a vast amount
of tests need to be
conducted. Secondly, the unsteadiness of the impact source deteriorates
the quality of the
image. It seems that an automatic test system is the solution to the
se problems. It is hope
that such system can be developed in the near future so that the imaging techniq
ues can be
widely applied in the inspection of concrete structures.
5. Acknowledgment
The works presented in this chapter was supported by the National Sci
ence Council of
Taiwan under grant NSC 98-2211-E-002-104-MY3.
254
6. References
Abraham, O., Leonard, C., Cote, P. & Piwakowski, B. (2000). Time-frequ
ency Analysis of
Impact-Echo Signals: Numerical Modeling and Experimental Validation. ACI
Materials Journal, Vol.97, No.6, pp. 645-657
Angel, E. (2006). Interactive Computer graphics: a top-down approach us
ing OpenGL 4th
Ed., Addison Wesley, ISBN 0-321-3125-2X, MA
Engel, K., Hadwiger, M., Kniss, J. M., Rezk-Salama, C. & Weiskopf, D.
(2006). Real-Time
Volume Graphics, A K Peter, Ltd., ISBN 1-56881-266-3, Wellesley, MA
Gibson, A. & Popovics, J. S. (2005). Lamb wave basis for impact-echo method anal
ysis. ASCE
Journal of Engineering Mechanics, Vol.131, No.4, pp. 438-443
Goldsmith, W. (1960). ImpactThe Theory and Physical Behavior of Collidin
g Solids,
Edward Arnold Ltd., London
Hallquist, J. O. (2003). LS-DYNA Keyword User s Manual, Livermore Software Techn
ology
Corporation, Livermore
Kohl, C., Krause, M., Maierhofer, C. & Wostmann, J. (2005). 2D- and 3D-visualisa
tion of NDT-
data using data fusion technique. Materials and Structures, Vol.38, No.9, pp. 81
7-826
Lin, C. C., Liu, P. L. & Yeh, P. L. (2009). Application of empirical mode decomp
osition in the
impact-echo test. NDT & E International, Vol.42, No.7, pp. 589~588
Lin, Y. & Sansalone, M. (1992). Transient Response of Thick Circular
and Square Bars
Subjected to Transverse Elastic Impact. J. Acoustical Society of Americ
a, Vol.91,
No.2, pp. 885-893
Liu, P. L. & Yeh, P. L. (2008). Imaging of internal cracks in concr
ete structures using the
volume rendering technique, 17th WCNDT, Shanghai, China, Oct. 25-28, 2008
Liu, P. L. & Yeh, P. L. (2010). Vertical Spectral Tomography of Concrete Structu
res Based on
Impact Echo Depth Spectra. NDT & E International, Vol.43, No.1, pp. 45-53
Liu, P. L. & Yiu, C. Y. (2002). Imaging Of Concrete Defects Using E
lastic Wave Tests,
Proceedings, the 2002 Far-East Conference on Nondestructive Testing, Tok
yo,
Japan, Oct. 21-24, 2002
Liu, P.-L. & Yeh, P.-L. (2011). Spectral Tomography of Concrete Structures Based
on Impact
Echo Depth Spectra. NDT & E International, Vol.44, No.8, pp. 692~702
Sansalone, M. & Carino, N. J. (1986). Impact-Echo: A Method for Flaw Detection i
n Concrete
Using Transient Stress Waves, Gaithersburg, MD: National Bureau of Standard
Sansalone, M. J. & Streett, W. B. (1997). Impact-echo: nondestructive
evaluation of concrete
and masonry, Bullbrier Press, Ithaca, N.Y.
Schubert, F. & Khler, B. (2008). Ten Lectures on Impact-Echo. Journal
of Nondestructive
Evaluation, Vol.27, pp. 5-21
Schubert, F., Wiggenhauser, H. & Lausch, R. (2004). On the Accuracy o
f Thickness
Measurements in Impact-echo Testing of Finite Concrete Specimens-numerical and
Experimental Results. Ultrasonics, Vol.42, pp. 897-901
Watt, A. (2000). 3D Computer Graphics, Addison Wesley, ISBN 0-201-39855
-9, Edinburgh
Gate, British
Yeh, P. L. & Liu, P. L. (2008). Application of the Wavelet Transform and the Enh
anced Fourier
Spectrum in the Impact Echo Test. NDT & E International, Vol.41, No.5, pp. 382-3
94
Yeh, P. L. & Liu, P. L. (2009). Imaging of internal cracks in concr
ete structures using the
surface rendering technique. NDT & E International, Vol.42, No.3, pp. 181-187
Zhu, J. Y. & Popovics, J. S. (2007). Imaging concrete structures usin
g air-coupled impact-
echo. ASCE Journal of Engineering Mechanics, Vol.133, No.6, pp. 628-640

NDT XT

Uploaded by

Copyright:

Available Formats

NDT XT

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NDT XT

Uploaded by

Copyright:

Available Formats

What are some of the nondestructive testing methods discussed in the document?

What are some of the nondestructive testing methods discussed in the document?

What are some of the applications of nondestructive testing mentioned?

What are some of the applications of nondestructive testing mentioned?

NONDESTRUCTIVE

TESTING METHODS AND

Nondestructive Testing Methods and New Applications

Copyright 2012 InTech

Nondestructive Testing Methods and New Applications, Edited by Mohammed Omar

Fig. 8. Illustration showing an ultrasound beam focusing on a defect of area A.

Nondestructive Inspection Reliability: State of the Art 19

which can be rewritten

Fig. 11. The SQUID based eddy current NDE system

SQUID Based Nondestructive Evaluation

Fig. 1. a) Representative examples of industrially-made dental devices: 1) ortho

Nondestructive Testing Methods and New Applications

Fig. 3. Attenuation of neutrons by specimen

Table 2. Neutron sources and their general characteristics

Nondestructive Testing Methods and New Applications

Converter Screen Emitted

Table 5. Common neutron converter screens and image recording media

Neutron(Gd/film) Neutron (imaging plate) X-Ray

[13] R. Funklin. Application of Neutron radiography using Neutron Imaging

(E) represents the attenuation of

is the total path length through material i. Fig.2 is an example

)millimetres. It means that the projected

millimetres vertically during

(e) (f) (g)

Flaw Simulation in Product Radiographs 115

(a) (b) (c)

parallel to the 0y direction, varying at the 18uEz frequency, with

is given under the form of a double Fourier expansion

Fig. 4. Rectangular aluminium plate with a calibrated crack having a width (w =

Nondestructive Testing Methods and New Applications 134

in the case of a defect

, a variation of the vertical component of the induced

. This allows to evaluate the z

. Then we modify either

to decrease e and repeat this process until e is as small as possible. This is

Nondestructive Testing Methods and New Applications 138

> p, (B) theoretical

Magnetic Adaptive Testing

Fig. 17. Peak values of signals of two most ductile (T

0.0 1.0 2.0 3.0

Nondestructive Testing Methods and New Applications 210

is the used inverse operator.

FA 10% FA 20% FA 30%

Nondestructive Testing Methods and New Applications 232

(a) (c) (b)

(a) (b) (c)

Fig. 14. The volume rendering method

Fig. 15. Volume rendering of numerical model using

You might also like