Article

Corrosion Assessment of Steel Bars Used in

Reinforced Concrete Structures by Means of Eddy
Current Testing
Naasson P. de Alcantara Jr. *, Felipe M. da Silva, Mateus T. Guimarães and Matheus D. Pereira
Received: 25 September 2015; Accepted: 19 December 2015; Published: 24 December 2015
Academic Editor: Piervincenzo Rizzo

Department of Electrical Engineering, São Paulo State University—Unesp, Bauru 17033-360, Brazil;
[email protected] (F.M.S.); [email protected] (M.T.G.);
[email protected] (M.D.P.)
* Correspondence: [email protected]; Tel.: +55-14-3013-6115; Fax:+55-14-3231-2820

Abstract: This paper presents a theoretical and experimental study on the use of Eddy Current
Testing (ECT) to evaluate corrosion processes in steel bars used in reinforced concrete structures.
The paper presents the mathematical basis of the ECT sensor built by the authors; followed by
a finite element analysis. The results obtained in the simulations are compared with those obtained
in experimental tests performed by the authors. Effective resistances and inductances; voltage
drops and phase angles of wound coil are calculated using both; simulated and experimental
data; and demonstrate a strong correlation. The production of samples of corroded steel bars;
by using an impressed current technique is also presented. The authors performed experimental
tests in the laboratory using handmade sensors; and the corroded samples. In the tests four gauges;
with five levels of loss-of-mass references for each one were used. The results are analyzed
in the light of the loss-of-mass and show a strong linear behavior for the analyzed parameters.
The conclusions emphasize the feasibility of the proposed technique and highlight opportunities
for future works.

Keywords: reinforced concrete structures; corrosion process; nondestructive testing; eddy current
testing; accelerated corrosion techniques

1. Introduction
Reinforced concrete structures are, nowadays, the main construction element in most countries.
However, despite flexibility and other construction advantages, reinforced concrete presents some
problems that need constant monitoring. One of the main problems that fall within this scope is
the process of corrosion of the reinforcements. In fact, the corrosion process dramatically affects
the long-term performance of reinforced concrete structures, because it affects the flexural strength,
deformation behavior, ductility, bond strength and mode of failure of the structures (El Maaddawy
and Soudky [1]).
Corrosion of steel in concrete structures is as an oxidation process, followed by the breakdown
of the passive film of the steel, due to the entry of chloride ions or carbon dioxide. In the initial
phase the corrosion crack doesn’t happen directly on the surface of the concrete structure, but only
shows up when the corrosion product reaches its threshold value. In addition, after the appearance of
corrosion cracks on the surface, the rate of corrosion increases significantly due to the increased inflow
of chloride ions or carbon dioxide through the cracks. In conclusion, this acceleration of the corrosion
process threatens the safety of reinforced concrete structure (Maruya et al. [2]).

Sensors 2016, 16, 15; doi:10.3390/s16010015 www.mdpi.com/journal/sensors

Sensors 2016, 16, 15 2 of 18

As a result of the corrosion process, the corrosion product volume is two to six times greater
than theSensors
original volume of the steel bar; so, this volume expansion causes cracking and spalling of
2015, 15, page–page
the concrete cover, reduction of the cross-sectional area of the reinforcing steel volume, beyond that
of theofalready mentioned
the already mentioned negative
negativeeffects.
effects. Consequently,
Consequently, the thereinforcement
reinforcement corrosion
corrosion reduces
reduces the the
load-carrying
load-carrying capacity
capacity of the
of the structure,and
structure, andbrittle
brittle failure
failuremay
mayoccur
occurwithout warning.
without According
warning. to
According to
Roqueta et al. [3], quoting Arndt and Jalinoos [4], there are six phases in the concrete corrosion
Roqueta et al. [3], quoting Arndt and Jalinoos [4], there are six phases in the concrete corrosion process
process for nondestructive monitoring of the service life of a concrete structure, as depicted in Figure 1.
for nondestructive monitoring of the service life of a concrete structure, as depicted in Figure 1.

CORROSION LEVEL DURING THE CONCRETE STRUCTURE

SERVICE LIFE

Free rust Stress Concrete

Chloride penetration and corrosion initiation
expansion initiation cracking

O2 Cl- HO2

Figure 1. Schematic
Figure illustration
1. Schematic ofofthe
illustration thevarious
various steps ofthe
steps of theconcrete
concrete deterioration
deterioration duedue tocorrosion
to the the corrosion
of theofreinforcement
the reinforcement (adapted
(adapted from
from [3]).
[3]).

The aim of this paper is to present an experimental study on the use of Eddy Current Testing
The aim of this paper is to present an experimental study on the use of Eddy Current Testing
(ECT) to evaluate corrosion processes in the steel bars used in reinforced concrete structures. The
(ECT) to evaluate corrosion processes in the steel bars used in reinforced concrete structures.
following sections will present a survey of the techniques used to evaluate the corrosion in
The following sections will
reinforced structures, the present a survey
mathematical basis,ofand
thedetails
techniques
of the used to evaluate
ECT sensors thethe
built for corrosion
tests, in
reinforced structures,
computational the mathematical
simulations basis,with
and comparisons and details of results,
experimental the ECTthesensors built
production for the tests,
of corroded
computational simulations and comparisons with experimental results,
samples of steel bars, results of experimental tests and their analysis. the production of corroded
samples of steel bars, results of experimental tests and their analysis.
2. A very Brief Survey of the Corrosion Assessment of Reinforced Concrete Structures
2. A very Brief Survey of the Corrosion Assessment of Reinforced Concrete Structures
2.1. Electrochemical Techniques
2.1. Electrochemical
Although, Techniques
in general, visual inspection is the most common practice for the evaluation of the
conservation status of reinforced concrete structures, it is an inappropriate choice for checking the
Although,
existence of in general,
corrosion visual Signs
processes. inspection is thesuch
of damage, most common
as cracks and practice for the
spalling, when evaluation
they appear, of
the conservation status of reinforced concrete structures, it is an inappropriate choice
are indicative of an extensive corrosion process, so it is desirable to monitor the corrosion process infor checking
the existence of corrosionstarung
the reinforcement, processes.
withSigns
the of damage,
structure such as cracks
construction andby
phase, spalling, when periodic
conducting they appear,
are indicative ofand
inspections, an extensive corrosion
keeping a record process, so it is desirable to monitor the corrosion process in
of data.
There are different
the reinforcement, starung methods
with thetostructure
assess theconstruction
reinforcementphase,
corrosion on existing structures
by conducting periodic such as:
inspections,
(1) open circuit potential
and keeping a record of data. measurements; (2) surface potential measurements; (3) linear polarization
resistance
There are measurements;
different methods(4) galvanostatic
to assess thepulse transient methods;
reinforcement corrosion(5) on
electrochemical impedance
existing structures such as:
spectroscopy; (6) harmonic analysis and (7) noise analysis. This is not an exhaustive list, but it
(1) open circuit potential measurements; (2) surface potential measurements; (3) linear polarization
reflects the most common methods studied and used in recent years. Their basic principles remain
resistance measurements; (4) galvanostatic pulse transient methods; (5) electrochemical impedance
unchanged, but some new technological contributions have been added over time. Below there is a
spectroscopy; (6) harmonic
brief description analysis
of each cited and (7) noise analysis. This is not an exhaustive list, but it reflects
method:
the most common methods studied and used in recent years. Their basic principles remain
(1) A metal body in contact with the surrounding media develops an electric potential. In
unchanged, but some new technological contributions have been added over time. Below there is
reinforced concrete structures, the concrete acts as an electrolyte, generating an electrostatic
a brief description of each cited method:
potential, which can vary from place to place, depending on the state of the concrete. The
(1) A metal bodyinvolved
principle in contact with
in the thecircuit
open surrounding
potentialmedia developsisan
measurements electric the
essentially potential. In reinforced
measurement of
the corrosion
concrete potential
structures, of the rebar
the concrete actsto as
a standard referencegenerating
an electrolyte, electrode. This is the most typical
an electrostatic potential,
procedure
which can vary for from
the routine
placeinspection
to place,ofdepending
reinforced concrete
on the structures
state of the(Erdogdu et al. [5]).
concrete. The principle
(2) During the corrosion process, an electrical current flows through the
involved in the open circuit potential measurements is essentially the measurement concrete between the of
anodic and cathodic regions. Measurements of the difference of potential at the concrete surface
the corrosion potential of the rebar to a standard reference electrode. This is the most typical
procedure for the routine inspection of reinforced concrete structures (Erdogdu et al. [5]).
2
Sensors 2016, 16, 15 3 of 18

(2) During the corrosion process, an electrical current flows through the concrete between the
anodic and cathodic regions. Measurements of the difference of potential at the concrete surface
detect this current flow. Surface potential measurements are a non-destructive test to identify
anodic and cathodic regions in concrete structures and, indirectly, to detect corrosion processes
of the reinforcement. Two reference electrodes are used for the measurements, and no electrical
connection to the rebar is required. An electrode is held fixed on the structure in a symmetrical
point. The other electrode, called moving electrode, is moved to the nodal points of a grid,
along the structure. The measurements are done using a high-impedance voltmeter. A positive
reading of the voltage represents an anodic area where corrosion is possible. The higher the
potential difference between the anodic and cathodic areas, the higher is the probability of
corrosion (Song and Saraswathy [6]).
(3) The unique electrochemical technique with quantitative ability regarding the corrosion rate is
the so-called polarization resistance, Rp . This technique is based on the application of a small
electrical perturbation to the rebar by using a counter electrode and a reference electrode. If the
electrical signal is uniformly distributed throughout the reinforcement, the ∆E/∆I ratio defines
Rp . The corrosion current, Icorr , is inversely proportional to Rp , or, Icorr = B/Rp , where B is
a constant. Rp can be measured employing direct current or alternating current techniques
(Andrade and Alonso [7]).
(4) The galvanostatic pulse method is a transient polarization technique working in the time
domain. A short time anodic current pulse is impressed galvanostatically on the reinforcement
from a counterelectrode placed on the concrete surface. The reinforcement is polarized in
the anodic direction compared to its free corrosion potential. A reference electrode records
the resulting change of the electrochemical potential of the reinforcement. Applying a constant
current to the system, an intermediate ohmic potential jumps, and a slight polarization of
the rebars occur. Under the assumption that a simple Randles circuit describes the transient
behavior of the rebars, the potential of the reinforcement, V(t), at a given time t, can be expressed
by an exponential expression, plus a constant resistance (Sathiyanarayanan et al. [8]).
(5) Measurement of the electrochemical impedance is done by imposing a sinusoidal voltage
(or current) signal of small amplitude, and by measuring the response signal of voltage and
current. The amplitudes and the phase difference between the two signals are then analyzed.
The frequencies vary between 10´5 and 105 Hz, and the amplitudes between 10 mV and 10 V
(MacDonald et al. [9]).
(6) The harmonic analysis method is an extension of the impedance method. Its execution is faster
and leads to results that are more straightforward than those of the electrochemical impedance
method. This technique is carried out by imposing an A.C. voltage perturbation at a single
frequency and measuring the A.C. current density, i1 . Two higher harmonics i2 and i3 are also
measured. The harmonic analysis uses the fact that the corroding interface acts as a rectifier,
in that the second harmonic current response is not linear about the free corrosion potential
(Vedalakshmi et al. [10]).
(7) In the electrochemical noise method, measurements of the spontaneous fluctuations of
the corrosion potentials and currents, which are observed as electrically coupled pairs, are taken.
This method is random in nature. The range of frequency is typically from 10´3 to 1.0 Hz.
Typical amplitudes are of the order of µV to mV, for voltage, and from nA to µA, for current.
Electrochemical noise is a low-cost nondestructive technique reasonably straightforward,
although attention must be paid to avoid problems, such as instrument noise, extraneous noise,
aliasing, and quantization (Sheng et al. [11]).

2.2. Electromagnetic Techniques

The previous section presented some techniques for the identification of corrosion process in
reinforced concrete structures, based on electrochemical phenomena. In this section, we present
Sensors 2016, 16, 15 4 of 18

the use of techniques for rebar inspection, as well as to detect corrosion processes, based on
electromagnetic phenomena. Of course, this is not a state-of-the-art review, but it will serve to
contextualize the present work in this scenario.
Electromagnetic fields are classified in stationary fields, low-frequency varying fields,
and high-frequency varying fields. These three types of electromagnetic fields are used to develop
non-destructive technology (NDT) techniques to assess the reinforcement of concrete structures.
Makar and Desnoyer [12], and Wolf and Vogel [13] presented examples of the use of magnetic
flux leakage (MFL) method, produced by magnetostatic fields, to detect failures in concrete rebars.
In both papers, the MFL method was used to detect breaks in steel tendons of prestressed structures.
Eddy current testing (ECT) is the best known technique in the NDT area based on low-frequency
electromagnetic fields. Shull [14] and Garcia-Martin et al. [15] have presented very well the principles
of ECT. In the reinforced concrete inspection context, Rubinacci et al. [16] presented an example of
the use of this technique. They used the ECT principles to develop a numerical model, based on the
finite element method, to locate and identify the size of steel bars under the concrete. Alcantara [17],
built differential electromagnetic sensors, produced dozens of reinforced concrete samples, and
performed laboratory tests, using the results to construct ANN training vectors, to locate and identify
steel bars under the concrete cover.
Ground penetrating radar (GPR) is the best known technique in the NDT area based
on high-frequency electromagnetic fields. Annan [18] and Blindow [19] are good references to
understand the principles of GPR. Farnoosh et al. [20] presented electromagnetic and computational
aspects of the technique, using a numerical analysis. According to these authors, one of the biggest
difficulties in the use of the GPR technique is that skilled personnel should analyze the results,
and it involves a considerable amount of post-processing work. Shaw et al. [21] used GPR results
to construct ANN training vectors to locate and identify steel bars in reinforced concrete structures.
Concerning corrosion assessment in the reinforcement of concrete structures, there are few
works using GPR techniques. In [22] the authors describe laboratory experiments on the influence
of moisture and chloride contents on the amplitude of radar signals. In reference [3], the authors
used low-profile ultra-wide-band antennas and twelve concrete samples with induced corrosion,
to correlate electromagnetic signatures with the corrosion level of the steel bars. The results were
compared with numerical simulations, to verify their consistency.
Radiography is one of the earliest NDT techniques used for imaging the steel reinforcements
immersed in the concrete. Due to their very small wavelengths, they propagate through the material
along straight paths without any significant diffraction. X- and gamma-ray methods are capable of
producing accurate two-dimensional images of the concrete interior. However, their use in concrete
testing is limited, due to their high initial costs, relatively low speed, heavy and expensive equipment,
need for extensive safety precautions and highly skilled operators, and perhaps most important of all,
the requirement of accessing both sides of the structure (Buyukozturk [23]).
Thermography is another NDT technique based on electromagnetic principles used in
the inspection of rebar-reinforced concrete structures. Baek et al. [24] proposed an integration of
electromagnetic heat induction and infrared thermography to detect steel corrosion in concrete.
They used an inductive heater to heat the steel rebar remotely from the concrete structure surface,
integrated with an IR camera to capture the heat signatures.

3. Description and Basics of the Developed ECT Sensors for Steel Bar Inspections

3.1. Mathematical Basics

The sensors developed in this work use the eddy current principles. Their electromagnetic parts
are, basically, RLC series circuits. An RLC series circuit is an association in series of a resistor,
an inductor, and a capacitor. In this case, L is the inductance of a sensor coil, Lc , R is the sum of
the coil resistance Rc and any other additional resistances Ra , and C is the capacitance of an external
Sensors 2016, 16, 15 5 of 18

capacitive array, Ca , placed in series with the coil. Figure 2 shows an electrical equivalent circuit for
the sensor.
The following simplifying assumptions limit the use of this equivalent circuit: (1) the capacitors
Sensors
do not 2015, 15,
present page–page
electrical resistances; in other words, only displacement currents are considered within
the capacitors; (2) if external resistors are added to the sensor, they do not present both inductive and
inductive and
capacitive capacitive
effects and onlyeffects and only
conduction conduction
current current
will be presentwill
in be present
these in these
resistors; (3) resistors; (3) the
the operational
operational frequency of the sensors will be in the range of 7.5–15 kHz, so no capacitive
frequency of the sensors will be in the range of 7.5–15 kHz, so no capacitive effects will be considered effects will
be considered in the sensor coil; (4) the model does not consider parasitic capacitances; (5)
in the sensor coil; (4) the model does not consider parasitic capacitances; (5) if contact resistances are if contact
resistances
known, theyarecanknown, theytocan
be added thebe added to the model.
model.

Rc + jωLc Ra 1/(jωCa)

isensor
Vcap
Vsource

~
Figure 2. The electrical equivalent circuit of the ECT sensor for reinforcement inspections.
Figure 2. The electrical equivalent circuit of the ECT sensor for reinforcement inspections.

In Figure 2,
In Figure 2, an input VVsource
an input , with
, with
source angular
angular frequency
frequency ω ω = 2πf,
= 2πf, is the
is the voltage
voltage applied
applied to theto
the terminals of the RLC circuit, V
terminals of the RLC circuit, Vcap is the is the voltage at the capacitive array terminals, and
cap voltage at the capacitive array terminals, and isensor is the i loop
sensor is
the loopincurrent
current in theInitially,
the sensor. sensor. Initially,
the analysisthe analysis of the electrical
of the electrical circuit ofcircuit
Figureof 2Figure
will be2 will
donebefordone
the
for the no-load condition (no ferromagnetic material placed under the sensor).
no-load condition (no ferromagnetic material placed under the sensor). The second law of KirchoffThe second law of
Kirchoff allows to express, for the voltage at the source
allows to express, for the voltage at the source terminals: terminals:

11 ) 𝑖
ˆ ˙
𝑉
Vsource = (𝑅 + 𝑅 )𝑖
𝑐 R a q𝑎i sensor + 𝑗 (𝜔𝐿
` j ωLc ´ − (1)
𝑠𝑜𝑢𝑟𝑐𝑒
“ pRc ` 𝑠𝑒𝑛𝑠𝑜𝑟 𝑐
𝜔𝐶a𝑎 isensor
ωC
𝑠𝑒𝑛𝑠𝑜𝑟 (1)

and for the current:

and for the current:
Vsource
𝑉𝑠𝑜𝑢𝑟𝑐𝑒
isensor “ =
𝑖𝑠𝑒𝑛𝑠𝑜𝑟 ˆ ˙ (2)
11 (2)
pR(𝑅
c `𝑐 + q 𝑎`) +
R a𝑅 j 𝑗ωL
(𝜔𝐿c´ −
𝑐 ωC )
𝜔𝐶
a𝑎

Before proceeding with with thethe analysis

analysisofofthe theequivalent
equivalentcircuit circuitit it
is is necessary
necessary to to
carry carry
outout
an
an analysis
analysis of the
of the behavior
behavior of the
of the electromagnetic
electromagnetic field
field in in
thethe region
region of of interest,
interest, with
with thethe presence
presence ofof
a
asteel
steelbar.
bar.AsAsthethesensor
sensorisisfed fedbybyaatime-varying
time-varyingvoltagevoltageat at its
its terminals,
terminals, thethe resultant time-varying
electromagnetic field will induce eddy current current loops
loops inin the
the conducting
conducting body. body. The magnitude and
behavior of the eddy current will depend on the magnetic flux density distribution, the metal
conductivity,
conductivity, metalmetalpermeability,
permeability,and andthethe
electrical frequency.
electrical frequency. The The
induced eddy eddy
induced currents will create
currents will
acreate
countera counter time-varying magnetic field that will disturb the original field. As an illustration,3
time-varying magnetic field that will disturb the original field. As an illustration, Figure
shows
Figure a3field
shows mapping
a fieldfor a 900 turn
mapping forcoil, fedturn
a 900 by a coil,
voltagefedsource of 5.0 Vrms
by a voltage , withofa 5.0
source frequency
Vrms, withequala
to 8.05 kHz.
frequency equalThetonon-commercial FEMM software
8.05 kHz. The non-commercial [25] was
FEMM softwareused [25]
to perform
was used the
to 2D simulations.
perform the 2D
Figure 3a shows
simulations. Figurethe3aflux distribution
shows in the regioninunder
the flux distribution the coil,
the region without
under thewithout
the coil, presencethe of presence
the steel
bar.
of theFigure 3b shows
steel bar. Figurethe3bflux
showsdistribution in the region
the flux distribution inunder
the regionthe coil
underwiththethecoil
presence
with the of presence
the steel
bar, and
of the Figure
steel 3c shows
bar, and Figuredetails
3c showsof the flux of
details within the within
the flux bar andthe inbar
the and
region surrounding
in the it.
region surrounding it.
Inspecting the the maps
maps of of Figure
Figure3,3,it itis is possible
possible to to
seesee some some expected
expected phenomena
phenomena fromfrom
the
the electromagnetic
electromagnetic theory:
theory: (1)high
(1) the the permeability
high permeability of the of thedistorts
steel steel distorts
the lines the
of lines of flux the
flux around around
bar;
the
as isbar;
known,as isinknown, in a magnetic/non-magnetic
a magnetic/non-magnetic interface, theinterface,
lines of flux theare
lines of flux are to
perpendicular perpendicular
the interface
to the non-magnetic
in the interface in the non-magnetic
medium; medium;
(2) the high (2) theofhigh
conductivity steelconductivity
does not permit of steel does not permit
the penetration of the
the penetration
magnetic of thethe
flux within magnetic
bar; the flux
flux within
densitythe (andbar; the flux density
consequently the eddy(andcurrents)
consequently the eddy
is confined to a
currents)
tiny region is around
confinedthe tointerface
a tiny region
betweenaround
the barthe and
interface between thenon-conductor
the surrounding bar and the surrounding
medium; in
fact, the skin depth for the steel bar (calculated using the formula 𝛿 = 1/√𝜇𝑓𝜎), considering a
relative permeability, 𝜇𝑟 , equal to 1000, electric conductivity, 𝜎, equal to 5.9 MS/m [16], and
frequency 𝑓 equal to 8.05 kHz is about 0.13 mm, is very consistent with the figures in Figure 3.
Finally, magnitude and phase of the flux density are disturbed point by point. Table 1 shows the
components of the magnetic induction vector at the center of the red line in Figure 3b.
Sensors 2016, 16, 15 6 of 18

non-conductor medium; in fact, the skin depth for the steel bar (calculated using the formula
a
δ “ 1{ µ f σ), considering a relative permeability, µr , equal to 1000, electric conductivity, σ, equal to
5.9 MS/m [16], and frequency f equal to 8.05 kHz is about 0.13 mm, is very consistent with the figures
in Figure 3. Finally, magnitude and phase of the flux density are disturbed point by point. Table 1
shows the2015,
Sensors components of the magnetic induction vector at the center of the red line in Figure 3b.
15, page–page

(a) (b)

(c)

Figure 3.
Figure 3. Simulation
Simulationof of
thethe
fluxflux
distribution under
distribution an ECT
under an sensor. (a) Without
ECT sensor. the steelthe
(a) Without bar;steel
(b) With
bar;
the steel bar; (c) Details of the flux in the steel bar and region around it.
(b) With the steel bar; (c) Details of the flux in the steel bar and region around it.

Table
Table 1.
1. 2D
2D components
components of
of the
the magnetic
magnetic induction
induction vector, with and
vector, with and without
without the
the steel
steel bar.
bar.

Bx (Wb/m2) By (Wb/m2)
Bx (Wb/m2 ) By (Wb/m 2)
Without the bar −3.665 × 10−6 − j6.346 × 10−7 2.526 × 10 − j1.224 × 10−6
−4

Without
With thethe
barbar ´3.665
−4.240 ˆ 10´6
× 10 −8 −´ j6.346׈10
j5.172
´7
10−11 2.526 ˆ 10
2.522
´4
× 10−4´+j1.224
j1.815ˆ×10
´6
10−7
With the bar ´4.240 ˆ 10´8 ´ j5.172 ˆ 10´11 2.522 ˆ 10´4 + j1.815 ˆ 10´7
As a result of the above discussion, the parameters of the electrical equivalent circuit will be
affected
As ainresult
the following way: discussion,
of the above (1) ohmic losses will occur in
the parameters of the steel bar; this
electrical fact cancircuit
equivalent be taken
willinto
be
account by
affected adding
in the a resistance
following ∆Rohmic
way: (1) e in thelosses
equivalent circuit;
will occur (2) The
in the steelcoil
bar;inductance is no
this fact can be longer the
taken into
original value,
account 𝐿𝑐 . A
by adding a resistance ∆Reinductance
new effective 𝐿𝑒𝑓 will
in the equivalent be defined
circuit; (2) Theas:coil inductance is no longer the
original value, Lc . A new effective inductance Le f will be defined as:
𝐿𝑒𝑓 = 𝐿𝑐 + ∆𝐿𝑒 (3)
where ∆𝐿𝑒 is a little change of the coil inductance, ∆Le by the changes in the original magnetic
Le f “ Lc `caused (3)
field, by the presence of eddy currents in the steel bar.
where ∆Lvoltage
The e is a little change
at the sensorofterminals
the coil inductance,
will be now caused by theas:
be expressed changes in the original magnetic
field, by the presence of eddy currents in the steel bar.
1
𝑉𝑠𝑜𝑢𝑟𝑐𝑒 = (𝑅𝑐 + 𝑅𝑎 + ∆𝑅𝑒 )𝑖𝑠𝑒𝑛𝑠𝑜𝑟 + 𝑗 (𝜔(𝐿𝑐 + ∆𝐿𝑒 ) + )𝑖 (4)
𝑗𝜔𝐶𝑎 𝑠𝑒𝑛𝑠𝑜𝑟
or:
1 (5)
𝑉𝑠𝑜𝑢𝑟𝑐𝑒 = (𝑅𝑐 + 𝑅𝑎 + ∆𝑅𝑒 )𝑖𝑠𝑒𝑛𝑠𝑜𝑟 + 𝑗 (𝜔𝐿𝑐 −
)𝑖 + 𝑗𝜔∆𝐿𝑒 𝑖𝑠𝑒𝑛𝑠𝑜𝑟
𝜔𝐶𝑎 𝑠𝑒𝑛𝑠𝑜𝑟
The sensor is designed to operate at its resonant frequency at no load. In other words, the
inductive and capacitive reactances within the parentheses in Equation (5) will have the same value.
Sensors 2016, 16, 15 7 of 18

The voltage at the sensor terminals will be now be expressed as:

ˆ ˙
1
Vsource “ pRc ` R a ` ∆Re q isensor ` j ω pLc ` ∆Le q ` isensor (4)
jωCa
or: ˆ ˙
1
Vsource “ pRc ` R a ` ∆Re q isensor ` j ωLc ´ isensor ` jω∆Le isensor (5)
ωCa
The sensor is designed to operate at its resonant frequency at no load. In other words, the
inductive and capacitive reactances within the parentheses in Equation (5) will have the same
value. or:
1
ωLc “ (6)
ωCa
and:
ω 1
f “ “ ? (7)
2π 2π Lc Ca
At the resonant frequency, the current in the sensor is:

Vsource
isensor “ (8)
pRc ` R a ` ∆Re q ` jω∆Le

which after some algebraic operations are expressed as the sum of a real and an imaginary parts:

Vsource
isensor “ pRc ` R a ` ∆Re ´ jω∆Le q (9)
pRc ` R a ` ∆Re q2 ` pω∆Le q2

The voltage at the capacitor (after some algebraic operations) is:

Vsource
Vcap “ ´ ” ı rω∆Le ` j pRc ` R a ` ∆Re qs (10)
ωCa pRc ` R a ` ∆Re q2 ` pω∆Le q2

The phase angle for the sensor current, φc , is tg´1 p´ω∆Le { pRc ` R a ` ∆Re qq, and phase angle
for the voltage at the capacitor, φv , is tg´1 ppRc ` R a ` ∆Re q {ω∆Le q. The phase angle for the relation
Vcap {isensor will be always φ “ ´π{2.
At no-load condition, ∆Le “ ∆Re “ 0, and Equations (9) and (10) become:

Vsource
isensor “ (11)
Rc ` R a

and:
Vsource
Vcap “ ´j (12)
ωCa pRc ` R a q
Equations (11) and (12) express the maximum values of the current at the sensor, and of the
voltage at the probe capacitor, respectively. Connecting a potentiometer in series with the sensor coil,
the no-load condition (values of isensor and Vcap without the presence of a steel bar under the sensor)
can be periodically calibrated.

3.2. The Frequency-Adjustment Method to Calculate the Effective Resistance and Inductance
The mathematical development presented in the previous subsection shows that the current in
the sensor and the voltage drop at the terminals of the capacitive array depend on variations of both
resistance and inductance. Therefore, for a better use of the results obtained, it is important to have
a way to calculate the resistances and inductances explicitly, after the measurements.
From finite element simulations, the authors observed that variations on the effective
inductances between the load and no-load condition are less than 0.5%. Including this variation in
Sensors 2016, 16, 15 8 of 18

the calculation of a new resonant frequency, the frequency variation is less than 0.25%. Based on this
fact, the authors propose a simple method to extract the resistance and inductance values from the
measurements. After reaching the resonant frequency of the no-load condition, the sensor is placed
on the steel bar, and the frequency is adjusted until the new resonant condition is attained:

1
Lc ` ∆Le “ (13)
p2π f n q2 Ca
Sensors 2015, 15, page–page
and:
Vsource
pRc ` R a ` ∆Re q “
where fn is the new resonant frequency. (14)
2π f n Ca Vcap
where fn isElement
3.3. Finite the newSimulations
resonant frequency.
and Experimental Comparisons for an ECT Sensor
3.3. Finite
FiniteElement
elementSimulations
analysis isand Experimental
a very Comparisons
interesting for an ECTthe
way to investigate Sensor
behavior of electromagnetic
devices. Through it, a good understanding of the phenomena involved can be obtained, in addition
Finite element analysis is a very interesting way to investigate the behavior of electromagnetic
to the mathematical modeling of the problem. Moreover, prototypes are built with more confidence,
devices. Through it, a good understanding of the phenomena involved can be obtained, in addition
if the expected results for their operation can be accurately predicted.
to the mathematical modeling of the problem. Moreover, prototypes are built with more confidence,
This subsection will present the construction details of an ECT sensor built from the
if the expected results for their operation can be accurately predicted.
mathematical model presented in the previous section. Figure 4 shows the electromagnetic
This subsection will present the construction details of an ECT sensor built from
component of this sensor. It is composed of a multi-turn coil with 900 turns of 24 AWG wire,
the mathematical model presented in the previous section. Figure 4 shows the electromagnetic
connected in series with a capacitive array with capacitance equal to 5 nF and an additional
component of this sensor. It is composed of a multi-turn coil with 900 turns of 24 AWG wire,
resistance of 50 Ω (not shown in the figure for clarity). The dimensions of the coil are also provided
connected in series with a capacitive array with capacitance equal to 5 nF and an additional resistance
in this figure.
of 50 Ω (not shown in the figure for clarity). The dimensions of the coil are also provided in this figure.
The authors performed 3D finite element frequency-domain simulations for this sensor using
The authors performed 3D finite element frequency-domain simulations for this sensor using
the commercial software COMSOL Multiphysics [26]. Figure 5 shows the magnetic flux density at
the commercial software COMSOL Multiphysics [26]. Figure 5 shows the magnetic flux density at
the coil surface and the steel bar surface. The gauge of the bar is 20 mm, and the distance between the
the coil surface and the steel bar surface. The gauge of the bar is 20 mm, and the distance between
top of the bar and the sensor is 25 mm. Figure 6 shows the mapping of the eddy current induced in
the top of the bar and the sensor is 25 mm. Figure 6 shows the mapping of the eddy current induced
the steel bar. As can be seen from these pictures, the magnetic induction is very low elsewhere
in the steel bar. As can be seen from these pictures, the magnetic induction is very low elsewhere
(magnetic saturation is not present), and the eddy currents are concentrated in the region of the steel
(magnetic saturation is not present), and the eddy currents are concentrated in the region of the steel
bar right below of the sensor.
bar right below of the sensor.

70 mm

40 mm

110 mm
140 mm

Winding height: 20 mm

(a) (b)

Figure 4.
Figure 4. The
The electromagnetic
electromagnetic components
components ofof an
an ECT
ECT sensor
sensor to
to inspect
inspect the
the reinforcement
reinforcement of
of concrete
concrete
structures. (a) Perspective view; (b) Coil dimensions.
structures. (a) Perspective view; (b) Coil dimensions.
 Winding height: 20 mm

(a) (b)c

SensorsFigure 4. 15
2016, 16, The electromagnetic components of an ECT sensor to inspect the reinforcement of concrete
9 of 18
structures. (a) Perspective view; (b) Coil dimensions.

Figure 5.
Figure 5. Magnetic
Magnetic flux
flux density
density at
at the
the surface
surface of
of the
the coil
coil and
and the
the steel
steel bar.
bar.
Sensors 2015, 15, page–page
8

(a)

(b) (c)
Figure 6. Induced current in the steel bar represented by the red arrows. (a) Perspective view; (b) Top
Figure 6. Induced current in the steel bar represented by the red arrows. (a) Perspective view;
view;
(b) Top(c)view;
side (c)
view.
side view.

These figures are very interesting to understand the behavior of the field quantities involved.
These figures are very interesting to understand the behavior of the field quantities involved.
However, to evaluate if this sensor will produce the expected results other variables should be
However, to evaluate if this sensor will produce the expected results other variables should be
analyzed, such as the effective resistance and inductance of the winding, changes in the phase angle
analyzed, such as the effective resistance and inductance of the winding, changes in the phase angle
of the current in the sensor, and the voltage drop at the capacitive array. COMSOL Multiphysics was
of the current
prepared in the sensor,perform
to automatically and the simulations
voltage dropfor at two
the capacitive
bar gauges,array. COMSOL
as well Multiphysics
as for different was
positions
prepared to automatically perform
of the steel bar in relation to the sensor. simulations for two bar gauges, as well as for different positions
of theSimulations
steel bar in were
relation
doneto the
for asensor.
steel bar with gauge of 20.0 mm, placed at 25.0 and 45.0 mm under
Simulations
the sensor. Figurewere done
7 show theforresults
a steelforbar
thewith gauge
effective of resistance,
coil 20.0 mm, effective
placed atcoil
25.0inductance,
and 45.0 mm
the
under the sensor. Figure 7 show the results for the effective coil resistance, effective
voltage at the capacitive array and phase angle of the current in the sensor. The graphics also coil inductance,
present
the voltage at thevalues
the experimental capacitive arrayforand
obtained thisphase angle
sensor, but of
thethe current in the
methodology usedsensor.
for the The graphics
experimental
also
tests present
will be the experimental
present values obtained
in the subsequent sections.forAsthis
cansensor, butthe
be seen, thesimulated
methodology used for the
and experimental
results agree very well each other.

Resistance (Simulated and Measured) Inductance (Simulated and Measured)

86 78.9
85
84
of the current in the sensor, and the voltage drop at the capacitive array. COMSOL Multiphysics was
prepared to automatically perform simulations for two bar gauges, as well as for different positions
of the steel bar in relation to the sensor.
Simulations were done for a steel bar with gauge of 20.0 mm, placed at 25.0 and 45.0 mm under
the sensor. Figure 7 show the results for the effective coil resistance, effective coil inductance, the
Sensors 2016, 16, 15 10 of 18
voltage at the capacitive array and phase angle of the current in the sensor. The graphics also present
the experimental values obtained for this sensor, but the methodology used for the experimental
tests will be present
experimental in be
tests will thepresent
subsequent
in thesections. As can
subsequent be seen,
sections. Asthe
cansimulated andsimulated
be seen, the experimental
and
results agree very
experimental well
results eachvery
agree other.
well each other.

Resistance (Simulated and Measured) Inductance (Simulated and Measured)

86 78.9
85
84 78.8
Resistance (Ohm)

83

Inductance (mH)
82
78.7
81
80
78.6
79
78
77 78.5
76
75 78.4
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Sensor Displacement (mm) Sensor Displacement (mm)
Sensors 2015, 15, page–page
(a) (b)
Voltage at the Capacitive Array Phase Angle for the Current at the Sensor
53 Figure 7. Cont. 83.2

52
9 Phase Angle (deg) 83
51
Voltage (V)

50 82.8

49 82.6
48
82.4
47

46 82.2
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Sensor Displacement (mm) Sensor Displacement (mm)

(c) (d)

Figure
Figure 7.
7. Simulated
Simulated(color
(colormarks)
marks)and
andexperimental
experimentalresults
results(hollow
(hollowblack
blackmarks)
marks)for
for the
the effective
effective
resistance
resistance (a),
(a); effective
effective inductance
inductance (b),
(b); voltage
voltage (c)
(c);and
andphase
phaseangle
angle (d)
(d) for
for aa 20
20 mm
mm steel
steel bar.
bar. Red
Red
marks: Steel bar placed 25 mm under the sensor. Blue marks: Steel bar placed 45 mm under
marks: Steel bar placed 25 mm under the sensor. Blue marks: Steel bar placed 45 mm under the sensor.the sensor.

4. The Production of Corroded Samples of Steel Bars

4. The Production of Corroded Samples of Steel Bars
The
The corrosion
corrosion process
process of
of the
the reinforcement
reinforcement ofof concrete
concrete structures
structures is, is, in
in general,
general, quite
quite slow.
slow.
Corrosion acceleration techniques are an important part of the studies on this subject. The
Corrosion acceleration techniques are an important part of the studies on this subject. The impressed
impressed
current technique is the most suitable one for this purpose. References [1–4,27,28] describe some
current technique is the most suitable one for this purpose. References [1–4,27,28] describe some
research on this technique.
research on this technique.
In this work, the impressed current technique was used to produce corroded samples of steel
In this work, the impressed current technique was used to produce corroded samples of steel
bars. Figure 8 shows the schematic arrangement of the experiment and Figure 9 shows four concrete
bars. Figure 8 shows the schematic arrangement of the experiment and Figure 9 shows four concrete
samples in the laboratory during the corrosion process.
samples in the laboratory during the corrosion process.
The purpose of the experiment was to obtain corroded samples of steel bars, with different levels
of corrosion. In the context of this paper, corrosion level is correlated with the loss-of-mass of the steel
sample. The concrete samples were immersed 12Vin a solution composed of 5 g of NaCl for each liter
of water. A 12V DC battery was used to provide DC the electricalanode
current. Care was taken to not allow
the current in each tank to exceed 1.0 A, renewing the saline solution periodically. The concrete
flex cables between
samples remained within the solution for periods A between one and two months, to achieve different
the steel bars
levels of corrosion. After the period of corrosion, the bars were removed from the concrete and the
cathod
rust carefully cleaned. Finally, the bars were weighed, and their weight compared with the weight of
samples of the same gauge and length,e but not corroded.
(steel

bar)

Figure 8. Schematic arrangement for the corrosion process.

 Corrosionacceleration
Corrosion accelerationtechniques
techniquesare
arean
animportant
importantpart
partofofthe
thestudies
studieson
onthis
thissubject.
subject.The
Theimpressed
impressed
currenttechnique
current techniqueisisthe themost
mostsuitable
suitableone
onefor
forthis
thispurpose.
purpose.References
References[1–4,27,28]
[1–4,27,28]describe
describesome
some
research on this technique.
research on this technique.
InInthis
thiswork,
work,thetheimpressed
impressedcurrent
currenttechnique
techniquewas
wasusedusedtotoproduce
producecorroded
corrodedsamples
samplesofofsteel
steel
bars.Figure
bars. Figure 8 shows the schematic arrangement of the experiment and Figure 9 shows four concrete
Sensors 2016, 16,815shows the schematic arrangement of the experiment and Figure 9 shows four concrete 11 of 18
samples ininthe
samples thelaboratory
laboratoryduring
duringthe
thecorrosion
corrosionprocess.
process.

12V
12V
DC
DC anode
anode
flex cables between
flex cables between
AA
the steel bars
the steel bars
cathod
cathod
e
e
(steel
(steel
bar)
bar)

Figure8.
Figure
Figure 8.8.Schematic
Schematicarrangement
Schematic arrangementfor
arrangement forthe
for thecorrosion
the corrosionprocess.
corrosion process.
process.

Sensors 2015, 15, page–page

The purpose of the experiment was to obtain corroded samples of steel bars, with different
levels of corrosion. In the context of this paper, corrosion level is correlated with the loss-of-mass of
the steel sample. The concrete
Figure
Figure
samples
9. Fourconcrete
concrete were in
samples immersed
the processinofacorrosion,
solution at
corrosion, composed of 5 g of NaCl for
thelaboratory.
laboratory.
Figure 9.9. Four
Four concrete samples
samples in
in the
the process
process of
of corrosion, at
at the
the laboratory.
each liter of water. A 12V DC battery was used to provide the electrical current. Care was taken to
not allow the current in each tank to exceed 1.0 A, renewing the saline solution periodically. The
10
From
concrete the produced
samples remainedmaterial,
within thesixteen
solution for 10
samples of corroded
periods betweensteel barstwo
one and andmonths,
four samples of
to achieve
non-corroded
different levelssteel bars wereAfter
of corrosion. chosen
thefor this paper.
period The gauges
of corrosion, the barsused were:
were 10.0, 12.5,
removed from16.0
the and 20.0
concrete
mm. For each bar gauge, bars were chosen with corrosion levels close to 10%, 15%,
and the rust carefully cleaned. Finally, the bars were weighed, and their weight compared with the 20% and 25%.
Figure 10
weight of shows
samples theof16.0
the mm
samegauge
gaugesteel
andbar samples.
length, but not corroded.

(a) (b)
Figure
Figure 10.
10. (a)
(a) Concrete
Concrete debris;
debris; (b)
(b) Samples
Samples of
of corroded steel bars.
corroded steel bars.

From the produced material, sixteen samples of corroded steel bars and four samples of
It should be pointed out that in the context of this work corrosion level is correlated with
non-corroded steel bars were chosen for this paper. The gauges used were: 10.0, 12.5, 16.0 and 20.0
the loss-of-mass of the steel samples, but according to the simulations presented in the previous
mm. For each bar gauge, bars were chosen with corrosion levels close to 10%, 15%, 20% and 25%.
section, a more tighter correlation would be with the loss-of-cross-section-area of the bar,
Figure 10 shows the 16.0 mm gauge steel bar samples.
since the electro-magnetic field does not penetrate into the body of the samples.
It should be pointed out that in the context of this work corrosion level is correlated with the
loss-of-mass of the steel samples, but according to the simulations presented in the previous section,
a more tighter correlation would be with the loss-of-cross-section-area of the bar, since the
electro-magnetic field does not penetrate into the body of the samples.

5. Results
Sensors 2016, 16, 15 12 of 18

5. Results

5.1. Experimental Setup

For this paper, an ECT-RLC sensor, similar to those presented in Section 4, was built.
The capacitive array is a 3 ˆ 3 array of 5.0 nF capacitors connected series-parallel, resulting in
an equivalent capacitance of 5.0 nF. The winding and the capacitive array were connected in series,
inserted in a plastic box especially built for this, and after the internal connections, its interior was
filled with a plastic resin. A U1733C handheld LCR meter (Agilent, São Paulo, Brazil) was used to
measure the resistance, inductance and capacitance of the sensor at 10 kHz, and the measured values
were: 73 Ω, 78.31 mH and 5.0 nF, respectively. Figure 11a shows a picture of the prototype used in
the measurements.
Figure 11b shows the experimental set-up used for this paper. It is composed of a signal
generator to excite the probe at desired values of voltage and frequency, a digital desk-multimeter
to measure the voltage at the capacitive array, an oscilloscope to inspect the quality of the signals and
a laptop
Sensorsequipped with a LabView application to control all the equipment.
2015, 15, page–page

(a)

(b)

Figure
Figure 11.
11. (a)
(a) An
An ECT
ECT sensor
sensor for
for corrosion
corrosion inspection; (b) Experimental
inspection; (b) Experimental setup
setup for
for the
the measurements.
measurements.

5.2. The
5.2. The Movement
Movement of of the Sensor
the Sensor
The measurements
The measurementswere weretaken for two
taken for distances between
two distances the sensor
between theand the bars
sensor and(e the
= 25 bars
and
45 mm). For the measurements already shown in Figure 7, the sensor was placed
(e = 25 and 45 mm). For the measurements already shown in Figure 7, the sensor was placed atat nine positions
(d = 0, 10, 20, 30, 40, 50, 60, 70, and 80 mm), in relation to the bar axis, as illustrated in Figure 12. The
nine positions (d = 0, 10, 20, 30, 40, 50, 60, 70, and 80 mm), in relation to the bar axis, as illustrated
base frequency used in the test was 8043 Hz, the measured resonant frequency of the sensor.
in Figure 12. The base frequency used in the test was 8043 Hz, the measured resonant frequency of
the sensor.

sensor movement

e
d

sensor movement
 5.2. The Movement of the Sensor
The measurements were taken for two distances between the sensor and the bars (e = 25 and
45 mm). For the measurements already shown in Figure 7, the sensor was placed at nine positions
(d = 0, 10, 20, 30, 40, 50, 60, 70, and 80 mm), in relation to the bar axis, as illustrated in Figure 12. The
Sensors 2016, 16, 15 13 of 18
base frequency used in the test was 8043 Hz, the measured resonant frequency of the sensor.

sensor movement

e
d

sensor movement

(a) (b)
Figure 12. Schematic representation of the movement of the sensor: (a) top view; (b) side view.
Figure 12. Schematic representation of the movement of the sensor: (a) top view; (b) side view.
5.3. The Procedure for the Measurements
5.3. The Procedure for the Measurements
The procedure for the measurements was as follows:
The procedure for the measurements was as follows:
- At no-load condition (no steel bar under the sensor):
- At no-load
(1) Thecondition (nois steel
equipment baron,
turned under the sensor):
the resonant frequency is set in the signal generator and
Sensors 2015, 15, page–page
slightly changed until to reach the real resonant frequency for the sensor, 8043 Hz in this
(1)The equipment is turned on, the resonant frequency is set in the signal generator and
slightly
case. This changed
frequencyuntilwas to reach
used the realfrequency
as starting resonant for
frequency for the sensor,
all measurements 8043 Hz in
with corroded or
this case. This 12
non-corroded steel frequency
bars, from was used on.
this point as starting frequency for all measurements with
corroded or non-corroded steel bars, from this point on.
(2) The coil inductance is calculated, using Equation (13).
(2) The coil inductance is calculated, using Equation (13).
(3) The coil resistance is calculated, using Equation (14).
(3) The coil resistance is calculated, using Equation (14).
- At load condition:
- At load condition:
(4) A steel bar is placed under the sensor aligned with its main axis. The voltage at the
(4) capacitive
A steel bar is is
array placed underand
measured therecorded.
sensor aligned withthe
After this, its frequency
main axis.slightly
The voltage at the
changed, up
capacitive array is measured and recorded. After this, the frequency slightly changed,
to the new resonance condition, and steps (2) and (3) are repeated, to calculate the effective
up to the new resonance condition, and steps (2) and (3) are repeated, to calculate
resistance and effective inductance.
the effective resistance and effective inductance.
(5)
(5) The
Thephase
phaseangle
angleofofthe
thecurrent
currentininthe
thesensor
sensorisis calculated
calculated using
using the
the extracted
extracted values of
values of
resistance and
resistance andinductance.
inductance.
Figure 13
Figure 13 shows
shows the
the voltage
voltage at
at the
the capacitive
capacitive arrays
arrays when
when the
the bars
bars are
are placed
placed at
at the
the reference
reference
levels of
levels of 5,
5, 25
25 and
and 45
45 mm
mm to
to the
the sensor.
sensor. Circle
Circle marks
marks are
are the
the results
results obtained
obtained before
before the
the adjustment
adjustment of
of
the frequency, and square marks are the results recorded after this adjustment.
the frequency, and square marks are the results recorded after this adjustment.

Distance to Sensor: 25 mm
52
Voltage at the Capacitor (V)

51

50

49

48

47

46
0 10 15 20 25
Loss of Mass (%)

Figure 13. Measured voltage

13. Measured voltage at
at the
the capacitive
capacitive array
array for
for corroded
corroded (loss of mass equal zero) and
non-corroded steel bars, as
non-corroded steel bars, as a function of the distance of
of the bar to the sensor.
the sensor. Red curves—20.0 mm
bar gauge;
gauge. Blue curves—16.0 mm mm bar bar gauge,
gauge, magenta
magenta curves—12.5
curves—12.5 mmmm barbar gauge;
gauge and black
curves—10.0 mm bar
bar gauge.
gauge.

Distance to Sensor: 25 mm
6

5
rence (V)

4
 Distance to Sensor: 25 mm
52

Voltage at the Capacitor (V)

51

50
Sensors 2016, 16, 15 14 of 18
49

48

As can be seen, the differences 47 between the values measured before and after the frequency

adjustment become smaller, as the 46distance from the sensor increases. For a better understanding of
0 10
the behavior of the sensor as a function of the barof15
Loss gauge, 20 25
Mass (%) corrosion level and distance from the bar to

the sensor, Figure 14 shows the difference between the voltage at no load condition and the voltage
with Figure
a steel 13.
barMeasured
under thevoltage
sensor.at As
the can
capacitive
be seen,array
theseforvoltage
corroded (loss of mass
differences equal well
are very zero)stratified,
and
non-corroded steel bars, as a function of the distance of the bar to the sensor. Red curves—20.0
both in relation to the distance from the bar to the sensor, as in relation to the bar gauge. In a real mm
field bar gauge. Blue
inspection, if thecurves—16.0
gauge of themm bar bar gauge, the
is known, magenta
corrosioncurves—12.5
level andmm the bar gauge of
thickness and
theblack
concrete
curves—10.0 mm bar gauge.
cover can be identified with enough confidence.

Distance to Sensor: 25 mm
6

5
Voltage Difference (V)

0
0 10 15 20 25
Loss of Mass (%)

Figure 14. Voltagedifference,

14. Voltage difference,asasa afunction
function
of of
thethe
barbar gauge,
gauge, lossloss of mass
of mass and and reference
reference distance.
distance. Red
Red curves—20
curves—20 mmgauge.
mm bar bar gauge; Blue curves—16.0
Blue curves—16.0 mm barmmgauge,
bar gauge;
magenta magenta curves—12.5
curves—12.5 mm bar mm bar
gauge
gauge; and black curves—10.0 mm
and black curves—10.0 mm bar gauge. bar gauge.

As can
To be seen,
complete thistheanalysis,
differences between
Figures theshow
15–17 valuesthe
measured
effectivebefore and after
resistances andthe frequency
inductances,
adjustment become smaller, as the distance from the sensor increases. For a better understanding
calculated according to the procedure shown in the beginning of this section when the bars are placed of
thethe
at behavior of distances
reference the sensorofas5,a25
function of theThe
and 45 mm. barcode
gauge, corrosion
color for theselevel and is:
graphic distance from the bar
red curves—20.0 mmto
the sensor,
bar gauge; Figure
blue 14 shows the difference between the voltage at no load condition
curves—16.0 mm bar gauge; magenta—12.5 mm bar gauge; black curves—10.0 mm and the voltage
Sensors 2015, 15, page–page
withgauge.
bar a steel bar under the sensor. As can be seen, these voltage differences are very well stratified,
Concerning
both in relation to thethedistance
resistance, the variations
from the bar toarethesignificant
sensor, asdepending
in relationon:to(1)the
thebar
corrosion
gauge.level;
In a real
(2) the gauge of the steel bar and; (3) the distance of13
the bar to the sensor. Concerning the inductance,
field inspection, if the gauge of the bar is known, the corrosion level and the thickness of the concrete
the variations are insignificant, depending on: (1) the corrosion level; (2) the gauge of the steel bar
cover can be identified with enough confidence.
and (3) the distance of the bar to the sensor. The most significant changes occur for the distance of
To complete this analysis, Figures 15–17 show the effective resistances and inductances,
5 mm. However, this would not be a usual distance between the bar and the sensor in a field test.
calculated according
The thickness toconcrete
of the the procedure
cover mustshown
be at in theequal
least beginning of this
to the gauge of section
the bar, when the bars
which does not are
placed at the
happen reference
in this distances
case. With regard of 5, 25
to the and 45
distance mm.
of 25 mm,Thetherecode color changes,
are slight for thesedepending
graphic on
is: red
curves—20.0
the gaugemm barbar,
of the gauge; blueremains
but that curves—16.0 mm
constant, bar gauge;
regardless magenta—12.5
of the corrosion level. mmWith
bar regard
gauge;toblack
curves—10.0 mm
the distance of bar gauge.
45 mm, apparently there is no significant change in the inductance values.

(a) (b)
Figure 15. Resistance
Figure (a)(a)
15. Resistance and inductance
and inductance(b)
(b) for
for the steelbars
the steel barsatatthe
the reference
reference distance
distance of 5 of
mm.5 mm.

Distance to Sensor: 25 mm
uctance (mH)

78.65
 (a) (b)
Figure 15. Resistance (a) and (a)
inductance (b) for the steel bars at(b)
the reference distance of 5 mm.
Sensors 2016, 16, 15 15 of 18
Figure 15. Resistance (a) and inductance (b) for the steel bars at the reference distance of 5 mm.
Distance to Sensor: 25 mm

Distance to Sensor: 25 mm

(mH)(mH)
78.65

Inductance
78.65

Inductance
78.6

Effective
78.6

Effective
78.55
78.55
0 10 15 20 25
0 10 Loss15 20
of Mass (%) 25
Loss of Mass (%)

(a)(a) (b)
(b)
Figure 16.16.
Figure
Figure
Resistance (a)(a)
16.Resistance
Resistance and
(a) inductance
and
and
(b)
inductance(b)
inductance
for
(b)for thesteel
forthe steel
steel bars
bars
bars
atthe
atatthe
the reference
reference
reference
distance
distance
distance
ofmm.
25 mm.
ofof2525mm.

Distance
Distanceto Sensor: 45 mm
to Sensor: 45 mm
79.5
79.5

(mH)
79

(mH)
79
Inductance
Inductance
78.5
78.5

78
Effective

78
Effective

77.5
77.5
77
0 10 15 20 25
77 Loss of Mass (%)
0 10 15 20 25
Loss of Mass (%)
(a) (b)
(a)
Figure 17. Resistance (a) and inductance (b) for the steel bars at the reference distance of 45 mm.
(b)
Figure 17.17.
Figure Resistance
Resistance(a)
(a)and
andinductance (b)for
inductance (b) forthe
thesteel
steel bars
bars at at
thethe reference
reference distance
distance of 45 of
mm.45 mm.
Concerning the resistance, the variations are significant depending on: (1) the corrosion level;
(2) the gauge of the steel bar and; (3) the distance of the bar to the sensor. Concerning the inductance,
Concerning the resistance, the variations are significant depending on: effective
(1) the corrosion level;
the As a conclusion
variations of this analysis
are insignificant, it is apparently
depending on: (1) thesufficient
corrosiontolevel;
calculate the
(2) the gauge of theresistance,
steel bar
(2) the
fromgaugedataof the steel bar and; (3) the distance of the bar to the sensor.
for the Concerning of the inductance,
and the obtained
(3) the distance of in
thethe
barmeasurements, to develop
to the sensor. The a method
most significant changes inspection
occur for the reinforced
distance of
the concrete
variations are insignificant,
structures,this
5 mm. However, both depending
as regards
would not be the
on: (1) the
identification
a usual
corrosion
and location
distance between
level;
the barof the(2) the gauge
andreinforcement,
of the
for steel
the sensor in aasfield the
test.
bar
anddetection
(3) the distance of the
of corrosion bar of
process to the
thesteel
sensor.
bars.The most significant changes occur for the distance of
5 mm. However, this would not be a usual distance
14 between the bar and the sensor in a field test.
5.4. The Lift-Off Effect
In eddy current NDT tests, “lift-off effects” are 14the effects caused by undesired variations of
the distance between the sensor and the specimen, and can easily mask the test results. In this article,
a fast investigation was conducted to evaluate the lift-off effect on the resistance and inductance
values. The experiment to investigate the lift-off effects was done with non-corroded bars with
a gauge of 20 and 16 mm. The bars were placed at the reference distance of 25 and 45 mm, and for
each one, four lift-off values were used: 2, 4, 6 and 8 mm. The results were compared with those
obtained for the original distance.
Figure 18 shows the effect of the lift-off on the resistance values, and Figure 19 shows the effect
on the inductance values. Again, red curves stand for 20 mm bar gauge and blue curves stand for
16 mm bar gauge. Circle marks stand for 25 mm, and square marks stand for 45 mm between the
sensor and the bar. As can be seen, the effects are greater for the resistance. However, in a real field
inspection, if the measures are taken with the sensor performing small offsets in the axial direction,
probably the lift-off should not affect the results substantially. However, this is an issue that must be
carefully considered in the development of a real system for the inspection of concrete structures.
distance between the sensor and the specimen, and can easily mask the test results. In this article, a
distance between the sensor and the specimen, and can easily mask the test results. In this article, a
fast investigation was conducted to evaluate the lift-off effect on the resistance and inductance
fast investigation was conducted to evaluate the lift-off effect on the resistance and inductance
values. The experiment to investigate the lift-off effects was done with non-corroded bars with a
values. The experiment to investigate the lift-off effects was done with non-corroded bars with a
gauge of 20 and 16 mm. The bars were placed at the reference distance of 25 and 45 mm, and for each
gauge of 20 and 16 mm. The bars were placed at the reference distance of 25 and 45 mm, and for each
one, four lift-off values were used: 2, 4, 6 and 8 mm. The results were compared with those obtained
one,2016,
Sensors four16,
lift-off
15 values were used: 2, 4, 6 and 8 mm. The results were compared with those obtained
16 of 18
for the original distance.
for the original distance.

(a) (b)
(a) (b)
Figure
Figure 18. Lift-off
Lift-off
18.18. effects
effects on the
onon resistance
thethe
resistance values. (a)
values. absolute values;
(a)(a)
absolute values; (b)
(b) Percent
Percent variation.
variation.
Figure Lift-off effects resistance values. absolute values; (b) Percent variation.

Inductance in Function of the Lift-off Variation of the Inductance in Function of the Lift-off
79 Inductance in Function of the Lift-off 0.25Variation of the Inductance in Function of the Lift-off
79 0.25

78.9 0.2
0.2
Percent Variation (%)
78.9
Percent Variation (%)
Inductance (mH)
Inductance (mH)

78.8 0.15
78.8 0.15

78.7 0.1
78.7 0.1

78.6 0.05
78.6 0.05

78.5 0
0.0
78.5 2.0 4.0 6.0 8.0 2 mm/0
0 4mm/0 6 mm/0 8 mm/0
0.0 2.0 Lift-off (mm)
4.0 6.0 8.0 2 mm/0 4mm/0
Lift-off relation 6 mm/0 8 mm/0
Lift-off (mm) Lift-off relation
(a) (b)
(a) (b)
Figure 19. Lift-off effects on the inductance values. (a) absolute values; (b) Percent variation.
Figure
Figure 19.19. Lift-off
Lift-off effects
effects onon
thethe inductance
inductance values.
values. (a)(a) absolute
absolute values;
values; (b)(b) Percent
Percent variation.
variation.
Figure 18 shows the effect of the lift-off on the resistance values, and Figure 19 shows the effect
Figure 18 shows the effect of the lift-off on the resistance values, and Figure 19 shows the effect
5.5. Discussion
on the inductance values. Again, red curves stand for 20 mm bar gauge and blue curves stand for 16
on the inductance values. Again, red curves stand for 20 mm bar gauge and blue curves stand for 16
mm bar gauge. Circle marks standtheforconclusion
25 mm, andissquare marks
use stand for current
45 mm betweento theidentify
sensor
mmBybar analyzing these
gauge. Circle results,
marks stand for 25 mm, andthat the
square marks ofstand
eddyfor 45 mmtesting
between the sensor
and
the the bar.
process As can
of corrosionbe seen,
in the the effects
reinforcement are greater for
of concrete the resistance.
structures However,
can lead However, in a
to a reasonable real field
level of
and the bar. As can be seen, the effects are greater for the resistance. in a real field
inspection,
success. Theif the measures
experimental are taken
results with
presented the sensor
here are performing
the first small
approach offsets
of the in the
authors,axial
and direction,
of course,
inspection, if the measures are taken with the sensor performing small offsets in the axial direction,
they can be improved. The sensor was taken to operate in its frequency of resonance and minor
15
variations around it. The frequency-adjustment method 15 proposed in the mathematical development
was successful. In a general way, the results showed a consistent behavior, with the highest values
being obtained for the highest levels of corrosion. Comparing the results on the gauge of the bars, was
possible to perceive a logical sequence, with the lowest values for the gauge of 10 mm, and the highest
values for the gauge of 20 mm.
A practical methodology for the use of ECT in the identification of corrosion processes in the
reinforcement of concrete structures can be outlined as follows: first, measurements can be made
over non-corroded parts of the reinforcement. In this way, the gauge and concrete cover (if not yet
known) can be determined. After, measurements can be made successively along the reinforcement,
comparing the results between then and with those for the non-corroded parts.
The way forward is now to conduct extensive laboratory and field measurements, to establish
large datasets that can be used to feed artificial intelligent tools, like artificial neural networks
or fuzzy logic, to develop expert systems for the detection of corrosion in the reinforcement of
concrete structures.
Sensors 2016, 16, 15 17 of 18

6. Conclusions
A theoretical and experimental study was carried out for determining the corrosion level
of steel bars used in reinforced concrete structures, using eddy current testing. The following
steps were followed: theoretical review, with an overview of the main types of methods in the
analysis of the corrosion of reinforcement of concrete structures; mathematical development of the
circuit theory, to obtain expressions for the parameters and electrical variables of interest for the
problem.; finite element simulations to understanding the electromagnetic phenomena involved in
the analysis, and to predict the behavior of the parameters and electrical variables of the proposed
sensors; experimental tests for the acceleration of the corrosion of steel bars in reinforced concrete
structures; experimental measurements with the sensor using corroded and non-corroded samples;
comparisons between simulated and experimental results; analysis of the results obtained for the
corroded samples. All these steps were successful.
The methodology presented here is not a substitute for the well-established electrochemical
methods already in use. It can be used as a preliminary assessment of the reinforcement of concrete
structures in search of corrosion processes, and further work can be done, using other methods,
to obtain a more comprehensive diagnosis of the problem. ECT sensors based on the principles
presented here can be easily built, and operated by personnel with basic professional training,
without the need for further knowledge in electrochemistry, for example. Finally, there is a wide field
to be explored on this subject, as new levels of frequencies, multi-frequency sensors, improvement in
the dimensions of the sensors, sensors with ferrite cores, etc., are all topics worth pursuing.

Acknowledgments: The authors express their gratitude to the FAPESP – São Paulo Research Foundation, for the
financial support of this research, under the grants number 2014/08797-8, 2014/09622-7, 2014/09624-0 and
2014/09609-0.
Author Contributions: Naasson P. de Alcantara Jr.: Supervision of the research. Definition of the system for the
process of acceleration of corrosion of steel bars, idealization of the ECT sensors, Conduction of the experimental
tests. Analysis of the experimental results. Wrote the article. Felipe M. da Silva: Built the electromagnetic
sensors and electronic circuits, performed experimental measurements. Mateus T. Guimarães: Performed the
experiments for the acceleration of the corrosion of the steel bars in the concrete samples and experimental
measurements. Matheus D. Pereira: Performed computational simulations and experimental measurements.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. El Maaddawy, T.A.; Soudky, K.A. Effectiveness of impressed current techniques to simulate corrosion of
steel reinforcement in concrete. J. Mater. Civil Eng. 2003, 15, 41–47. [CrossRef]
2. Maruya, T.; Takeda, H.; Horiguchi, K.; Koyama, S.; Hsu, K.L. Simulation of Steel Corrosion in Concrete
Based on the Model of Macro-Cell Corrosion Circuit. J. Adv. Concrete Technol. 2007, 6, 343–362. [CrossRef]
3. Roqueta, G.; Jofre, L.; Feng, M.Q. Analysis of the Electromagnetic Signature of Reinforced Concrete
Structures for Nondestructive Evaluation of Corrosion Damage. IEEE Trans. Instrum. Meas. 2012, 61,
1090–1098. [CrossRef]
4. Arndt, R.; Jalinoos, F. NDE for corrosion dectection in reinforced concrete structures—A benchmark
approach. In Proceedings of Non-Destructive Testing in Civil Engineering (NDTCE’09), Nantes, France,
30 June–3 July 2009; pp. 1–6.
5. Erdogdu, S.; Kondratova, I.L.; Bremner, T.W. Determination of Chloride Diffusion Coefficient of Concrete
Using Open-Circuit Potential Measurements. Cem. Concr. Res. 2004, 34, 603–609. [CrossRef]
6. Song, H.W.; Saraswathy, V. Corrosion Monitoring of Reinforced Concrete Structures—A Review. Int. J.
Electrochem. Sci. 2007, 2, 1–28.
7. Andrade, C.; Alonso, C. Test Methods for on Site Corrosion Rate Measurement of Steel Reinforcement in
Concrete by Means of the Polarization Resistance Method. Mater. Struct. 2004, 37, 623–643. [CrossRef]
8. Sathiyanarayanan, S.; Natarajan, P.; Saravanan, K.; Srinivasan, S.; Venkatachari, G. Corrosion Monitoring
of Steel in Concrete by Galvanostatic Pulse Technique. Cem. Conc. Compos. 2006, 28, 630–637. [CrossRef]
Sensors 2016, 16, 15 18 of 18

9. MacDonald, D.D.; El-Tantawy, Y.A.; Rocha-Filho, R.C. Evaluation of Electrochemical Impedance

Techniques for Detecting Corrosion on Rebar in Reinforced Concrete. Available online:
http://onlinepubs.trb.org/onlinepubs/shrp/SHRP-91-524.pdf (accessed on 20 Setember 2015).
10. Vedalakshmi, R.; Manoharan, S.P.; Song, H.W.; Palaniswamy, N. Application of harmonic analysis in
measuring the corrosion rate of rebar in concrete. Corros. Sci. 2009, 51, 2777–2789. [CrossRef]
11. Sheng, Y.F.; Luo, J.J.; Wilmott, M. Spectral analysis of electrochemical noise with different transient shapes.
Electrochimica Acta 2000, 45, 1763–1771.
12. Makar, K.; Desnoyers, R. Magnetic Field Techniques for the Inspection of Steel Under the Concrete Cover.
NDT&E Int. 2001, 34, 445–456.
13. Wolf, T.; Vogel, T. Detection of Reinforcement Breaks: Laboratory Experiments and an Application of the
Magnetic Flux Leakage Method. In Proceedings of the 14th International Conference and Exhibition on
Structural Faults and Repair, Edinburgh, UK, 3–5 July 2012; pp. 1–9.
14. Shull, P.J. Nondestructive Evaluation: Theory, Techniques and Applications; CRC Press: New York, NY,
USA, 2002.
15. García-Martin, J.; Gomez-Gil, J.; Vazquez-Sanchez, E. Non-Destructive Techniques Based on Eddy Current
Testing. Sensors 2011, 11, 2525–2565. [CrossRef] [PubMed]
16. Rubinacci, G.; Tamburrino, A.; Ventre, S. Concrete Rebar Inspection by Eddy Current Testing. Int. J. Appl.
Electromagn. Mech. 2007, 25, 333–339.
17. Alcantara, N.P., Jr. Identification of Steel Bars Immersed in Reinforced Concrete Based on Experimental
Results of Eddy Current Testing and Artificial Neural Network Analysis. Nondest. Test. Eval. 2013, 28,
58–71. [CrossRef]
18. Annan, A.P. Electromagnetic Principles of Ground Penetrating Radar. In Ground Penetrating Radar: Theory
and Applications; Elsevier: Amsterdam, The Netherlands, 2009; pp. 3–40.
19. Blindow, N.; Eisenburger, D.; Illich, B.; Petzold, H.; Richter, T. Ground Penetrating Radar. In Environmental
Geology; Springer: Berlin, Germany, 2007; pp. 283–335.
20. Farnoosh, N.; Shoory, A.; Moini, R.; Sadeghi, S.H.H. Analysis of Thin-Wire Ground Penetrating Radar
Systems for Buried Target Detection, Using a Hybrid MoM-FDTD, Technique. NDT&E Int. 2008, 41,
266–272.
21. Shaw, M.R.; Millard, S.G.; Molyneaux, T.C.K.; Taylor, M.J.; Bungey, J.H. Location of Steel Reinforcement in
Concrete Using Ground Penetrating Radar and Neural Networks. NDT&E Int. 2005, 38, 203–212.
22. Hugenschmidt, J.; Loser, R. Detection of chlorides and moisture in concrete structures with ground
penetrating radar. Mater. Struct. 2007, 41, 785–792. [CrossRef]
23. Buyukozturk, O. Imaging of Concrete Structures. NDT&E Int. 1998, 31, 233–243.
24. Baek, S.; Xue, W.; Feng, M.Q.; Kwon, S. Nondestructive corrosion detection in RC through integrated heat
Induction and IR thermography. J. Nondest. Eval. 2012, 31, 181–190. [CrossRef]
25. FEMM—Finite Element Method Magnetics. Available online: http://www.femm.info/ (accessed on
20 November 2015).
26. COMSOL Multiphysics. Available online: http://comsol.com (accessed on 20 September 2015).
27. Ahmad, S. Techniques for Inducing Accelerated Corrosion of Steel in Concrete. Arab. J. Sci. Eng. 2009, 34,
95–104.
28. Shetty, A.; Venkataramana, K.; Gogoi, I. Performance Evaluation of Rebar in Accelerated Corrosion by
Gravimetric Loss Method. Int. J. Earth Sci. Eng. 2012, 5, 154–159.

© 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open
access article distributed under the terms and conditions of the Creative Commons by
Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).