A Theoretical Model for Corrosion Assessment
in Overhead Line Conductors
Celio F. Barbosa
Gerncia de Solues Tecnolgicas para o Setor Eltrico
Centro de Pesquisa e Desenv. em Telecomunicaes - CPqD
Campinas, SP, Brasil
grcelio@cpqd.com.br
AbstractThis paper presents an electromagnetic model for
assessing the zinc layer thickness of galvanized steel strands that
form cables used in overhead power lines, i.e., aluminum
conductor steel-reinforced (ACSR) and overhead ground-wire
(OHGW). The status of the steel galvanization provides an
important indication of remaining cable life. The model is based
on the evaluation of eddy currents induced in the zinc layer and
its influence on a test coil impedance that has the cable as part of
its core. The theoretical results of the model agreed well with
results of experiments, both for the a cable containing only steel
strands (e.g., OHGW) and a complete ACSR cable. This model
was applied in the analysis of data collected by an inspection
robot to be used in overhead power lines.
Index TermsACSR cable, corrosion detection, galvanization,
overhead line, eddy currents.
I. INTRODUCTION
T HE overhead power lines are usually made of aluminum
conductors steel reinforced (ACSR), where the inner steel
strands provide the desired mechanical strength and the outer
aluminum strands conduct the line current. One important
factor that limits the service life of these conductors is internal
corrosion, which is initiated by the loss of the zinc layer that
covers the internal steel strands. When this layer is lost, the
corrosion process between aluminum and zinc strands
progresses fast, compromising the integrity of the conductor.
As this corrosion takes place inside the conductor, it is not
detectable by visual or infra-red inspections, except when the
corrosion products are large enough to bulge the conductor
but, at this stage, the conductor's failure may be imminent.
In order to detect the corrosion of ACSR conductors in its
early stages, an inspection system based on the eddy currents
induced in the zinc layer was developed in the 80's [1, 2]. This
system is based on a remote-controlled robot that sweeps the
line and detects the reactance of a coil that has the conductor
as part of its core. Similar systems using the same principle are
described in the literature [3, 4, 5], as well as estimations of
the conductor remaining life [6, 7].
Other techniques to detect corrosion in ACSR conductors
This work was financed, in part, by Brazilian National Council for
Scientific and Technological Development (CNPq).
Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014
can also be found in the literature, and some of them were
described by the Electric Power Research Institute (EPRI) [8,
9, 10]. One of these techniques uses permanent magnets to
magnetize the steel wires and magnetic field sensors to detect
non-uniformities in the field caused by corrosion of the steel
strands [11, 12, 13]. This technique is based on empirical
references and detects corrosion on an advanced stage, where a
substantial steel cross-section area is lost.
Although the eddy current technique has been widely used
in the field [14, 15, 16], to the best of this author's knowledge,
there is no published theoretical approach in order to correlate
the sensor readings to the zinc layer thickness. Indeed, the
inspection results seem to be interpreted in an empirical way,
i.e., the sensor output from an inspected conductor is
compared with outputs from reference samples of the same
conductor, but with known flaws. Therefore, the aim of this
paper is to provide a theoretical model for the corrosion
assessment of overhead line conductors based on the eddy
current method. This model is intended to provide the zinc
layer thickness from the signal picked-up by a sensing coil
surrounding the cable. The theoretical results are compared
with experimental data obtained from samples produced in
laboratory, which support the proposed model.
II. THE GALVANIZED STEEL STRAND
A. Theoretical Model
This section analyzes the response of a steel strand when it
is subjected to a time-varying longitudinal magnetic field. Let
us consider a strand of length s, as shown in Fig. 1. The steel
core is a cylinder of radius r, covered by a zinc layer of
thickness d. It is considered that the zinc has conductivity
and permeability 0, while the steel permeability is
0 r' j r" , (1)
where 0 is the free-space permeability, 'r and "r are the real
and imaginary parts of the relative permeability, respectively,
and the dash above the symbol denotes a complex variable.
The steel permeability takes into account the eddy currents
within the steel, so that it may be regarded as an effective
permeability. As the intensity and frequency of the applied
magnetic field are kept constant in an inspection system, the
effective steel permeability is assumed to remain constant.
ISSN 2177-6164
 Substituting = 2 f in (11), where f is the frequency, and
dividing by0 leads to the normalized total field:
H N 1 j f r d .
1
(12)
Equation (12) gives the total normalized magnetic field in a
steel strand. Fig. 2 shows the dependency ofN modulus with
the zinc layer thickness, for representative values of the
Fig. 1. Section of a galvanized steel strand and its dimensions. relevant parameters. As the output of a sensing device is a
function ofN , it is clear from Fig. 2 that zinc layer thickness
For a current i flowing in the zinc layer around the steel core in the range of interest (e.g., from 0 to 80 m) can be assessed
of length s, the corresponding magnetic field in the steel is: from the total magnetic field.
i
H . (2) 1,2
s
1

Normalized magnetic field

Considering that the current has a harmonic time-dependence,
the magnetic flux in the steel is: 0,8

Hr . 2
(3) 0,6

Substituting (2) into (3), dividing by the current, and 0,4

multiplying by the angular frequency gives the frequency-
0,2
dependent impedance of the zinc layer:
0
j r2
X
0 20 40 60 80 100
. (4)
Zinc layer thickness (um)
s
The resistance of the zinc layer is given by: Fig. 2. Total normalized magnetic field in a steel strand according to (12) as
2 r a function of zinc layer thickness. r = 1 mm, = 1.69 107 S/m,
R . (5) f = 100 kHz, 0 = 1.257 10-6 H/m, 'r = 3.1, and "r = 2.5.
sd
B. Steel Strand Magnetic Permeability
The total zinc layer impedance is obtained from (4) and (5):
A key parameter in the use of (12) is the magnetic
r 2 permeability of a steel strand. This permeability is defined in
Z j r . (6)
s d (1) as a complex variable, where the real part (') is related to
the energy stored in the magnetic field and the imaginary part
If the steel core is excited with a harmonic longitudinal (") is related to the losses.
magnetic field0, the corresponding flux in the steel core is: In order to obtain experimentally the steel strand
0 H 0 r 2 , (7) permeability, some samples of strands were taken from
different ACSR conductors and also from galvanized steel
and the induced voltage in the zinc layer is: conductors used as OHGW. The test procedure consisted in
0 selecting a strand sample of about 0.5 m in length and
V j H0 r 2 , (8) removing the galvanization by means of acid attack. A test coil
t
was applied in the middle of the sample, wrapping tightly a
This voltage will drive a current in the zinc layer that, by its thin insulated copper wire (0.16 mm diameter) around the
turn, will give rise to a magnetic field given by: strand, in order to obtain a coil with 120 turns. The real and
V . imaginary parts of this coil was measured with a precision
HR (9)
LRC meter (HP 4284A). In order to have reference values, a
sZ
similar coil was built by wrapping the wire around a plastic
Substituting (6) and (8) into (9) gives: core having the same diameter as the steel strand. The steel
j H0 r d . permeability components were obtained as follows:
HR (10)
2 j r d r'
Xs , (13)
Xp
The total magnetic fieldT in the steel core is given by the
superposition of the incident field0 and the reaction R Rp , (14)
r"
s

fieldR : Xp
2 H0
HT H 0 H R . (11) where Xs and Xp are the reactances of the steel-core and
2 j r d plastic-core coils, and Rs and Rp are the resistances of the steel-
core and plastic-core coils, respectively.

Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014 ISSN 2177-6164
 Fig. 3 shows the values obtained for the steel strands from a P H N ' r ' H N " r "1 , (16)
3/8" diameter OHGW. It can be seen that the real component
Q H N " r ' H N ' r " , (17)
('r) falls fast as the frequency increases, and then goes
A
asymptotically to 1. The imaginary component ("r) reaches a k s , (18)
maximum value around 2 kHz and then also falls close to the Ac
real component. These measurements were carried out at HN' and HN" are the real and imaginary parts of the normalized
ambient temperature (23 C) and with an applied magnetic magnetic field, respectively, and As and Ac are the cross-
field equal to 45 A/m. section areas of the steel strand and the test coil, respectively.
In order to assess the dependency of the permeability with For the test coil considered, Ac = 46 mm2.
the intensity of the exciting magnetic field, some results are AsN is function of the zinc layer thickness, the
shown in Fig. 4, for the frequency of 40 kHz. It can be seen measurement ofc for steel strands with zinc layers of
that the permeability is relatively stable in a wide range of different thicknesses can provide a verification of (12).
magnetic field intensity. It is important to highlight that in a Therefore, a set of steel strands was prepared taking a 0.15 m
detection device the frequency and the applied magnetic field sample from the core of different ACSR cables. Each strand
intensity are kept constant. was weighted in a precision scale. The galvanization was
completely removed from one strand of each core, and the
30 respective zinc layer thickness was calculated from the strand
25 diameter and its mass reduction. The other strands were
subjected to the same chemical treatment to remove the zinc,
Relative permeability

20
but the process was interrupted at different stages. The
15 remaining zinc layer thickness was calculated from the mass
reduction and the assumption that all strands of the same cable
10
core had the same original zinc layer thickness.
5 Fig. 5 shows the modulus of the coil impedance as a
0
function of zinc layer thickness for steel strands taken from the
0,1 1 10 100 1000 core of an ACSR 556,5 MCM cable (Dove), as calculated
Frequency (kHz) from (15) and (12). The steel strand radius is 1.45 mm. Fig. 5
also shows the experimental values obtained from the strand
Fig. 3. Relative permeability of a steel strand as a function of frequency. samples. It can be seen that the correlation between theoretical
Bold line: real part ('r); Thin line: imaginary part ("r).
and experimental data is very good, which supports the
electromagnetic model contained in (12).
6

5 6
Relative permeability

4 5
Coil impedance ()

3 4

2 3

1
2
0
1
0 20 40 60 80 100
Magnetic field intensity (A/m)
0
0 10 20 30 40 50 60
Fig. 4. Relative permeability of a steel strand as a function of magnetic field Zinc layer thickness (m)
intensity at 40 kHz. Bold line: real part ('r); Thin line: imaginary part ("r).

C. Experimental Verification
At this point, it is interesting to perform an experimental
verification of (12), as it is a key equation for the model
presented in this paper. In order to do so, a small coil was
built, which has at 40 kHz the resistance R0 = 0.28 and
reactance X0 = 2.93 . It is easy to show that, when a steel
strand is introduced into this coil, its impedance changes to:
Z c R0 X 0 k Q j X 0 1 k P ,
where:
Fig. 5. Modulus of the test coil impedance for steel strands taken from the
ACSR 556,5 MCM cable. Continuous line: calculated from (15) and (12);
Dots: experimental data obtained from treated samples.
The data in Fig. 5 are for the modulus of the coil
impedance. Fig. 6 shows the comparison of the real and
imaginary parts of the coil impedance. Again, there is a good
correlation between theoretical end experimental results.
(15)

Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014 ISSN 2177-6164
 5 For a longitudinal inducing magnetic field, it is possible to
4
Coil impedance ()
3
2 effect of each strand. Assuming that the strands are identical,
1
0 Ac
0 10 20 30 40 50
Zinc layer thickness (m)
Fig. 6. Real and imaginary parts of the test coil impedance for steel strands
taken from the ACSR 556,5 MCM cable. Thin line: calculated real part
(resistive); Bold line: calculated imaginary part (inductive); Crosses:
experimental real data; Triangles: experimental imaginary data.
In the calculation shown in Figs. 5 and 6, the zinc
conductivity considered is = 11,7 MS/m. This value was
obtained experimentally and it is somewhat lower than the
conductivity of pure zinc ( = 16,9 MS/m). This difference is
likely to be due to impurities in the zinc used in the
galvanization process. The zinc conductivity was obtained by
weighting a 1 m sample of galvanized steel strand and also
measuring its DC resistance with a milliohmmeter
(HP 4328A). The sample was subjected to a chemical
treatment to remove all the galvanization and its weight and
resistance were measured again. Based on the strand
dimensions (length and diameter), mass loss and increase of
resistance, it was possible to obtain the zinc conductivity.
This experimental investigation was carried out for steel
wires from the following cables: ACSR 1113,0 MCM (Blue
Jay), ACSR 795,0 MCM (Drake), ACSR 556,5 MCM (Dove),
and galvanized steel cable (3/8" OHGW). The results were
similar to the ones shown in Figs. 5 and 6, which validate (12).
III. THE ACSR CABLE
The behavior of a galvanized steel strand was investigated
in the previous section. This section analyzes the behavior of a
group of steel strands joined together to form a cable core, as
well as the effect of the aluminum strands of an ACSR cable.
A. Joining Steel Strands
Excepting the smaller ACSR cables that have only a single
steel strand, most of ACSR cables have several steel strands in
the core. A common formation is seven strands, with one in the
center and six strands twisted around the central one, as shown
in Fig. 7.
Fig. 7. Seven-strand steel core and the induced currents.
Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014
identify two modes of induced currents: the intra-strand and
the inter-strand modes, as shown in Fig. 7. If the inter-strand
currents are neglected, the effect of having several strands
inside a sensing coil may be modeled the superposition of the
(18) can be adapted in order to consider n strands in the coil:
As .
k n (19)
60
The validity of (19) was verified experimentally using a test
coil of 180 turns, 67 mm length, and 13 mm diameter. The
coil impedance with air core at 40 kHz is R0 = 1.74 and
X0 = 19.76 . A seven-strand steel core sample was introduced
into the coil and the coil impedance was measured. Then one
strand was removed from the steel core and the six-strand core
was measured. This process was repeated, until only the
central strand remained. The obtained data shows the coil
impedance as a function of the number of steel strands. Fig. 8
shows the experimental data and the theoretical values
calculated from (15) and (19). It can be seen that an excellent
agreement was obtained, which supports the approximations
used in the development of (19).
70
60
Coil impedance ()
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8
Number of strands
Fig. 8. Coil impedance modulus as a function of the number of strands.
Continuous line: calculated values from (16); Dots: experimental data.
B. The Aluminum Strands
Equation (15) gives the response of a galvanized steel cable
that is commonly used for OHGW. However, an ACSR cable
has also one or more layers of aluminum strands around the
steel core. Therefore, it is necessary to model the effect of the
aluminum strands on the response of a sensing coil.
The inter-strand induced current on the aluminum layers
could provide some shielding effect to the steel core. In order
to assess this effect, some experiments were carried out with
samples of typical ACSR cables. Each sample had its strands
held tightly by using plastic straps and the inner steel strands
were removed. The hollowed sample was placed inside a coil
excited with alternating current. A small probe coil was
inserted inside the sample, and the voltage induced in this
probe was measured. The experiment was repeated without the
cable sample, for the same current in the inducing coil.
ISSN 2177-6164
 The results obtained for different types of cables show that
the hollowed cable does not change significantly the magnetic
field intensity inside the coil. The presence of the aluminum
strands produced a reduction on the magnetic field of less than
2% of the field measured without the hollowed cable. This
result shows that the shielding effect provided by the inter-
strand currents can be neglected. One possible explanation is
that the contact resistance between the strands is large enough
to prevent the flow of significant currents among the strands.
However, the currents induced inside the aluminum strands
need to be taken into account (see the intra-strand current if
Fig. 7). This phenomenon can be modeled by the same
approach used in Section II, assuming that the current in the
aluminum strand is confined in a layer defined by the skin
depth . For a non-magnetic conductor, the skin depth is [17]:
f 0 1 / 2 .
For instance, considering the frequency f = 60 kHz and the
aluminum conductivity ( = 35.5 MS/m), the skin depth is
345 m. As the skin depth is not much shorter than the strand
radius, an approximate assumption is necessary in order to
obtain an effective strand radius for the application of (12). It
is assumed that 50% of the induced eddy currents flow
between the effective radius re and the real strand radius r and,
of course, the remaining 50% flows within re. It is easy to
show that this condition leads to:
re r 0.69 .
The validity of this approximation can be assessed by
measuring the response of a test coil with aluminum strands. In
order to do so, the steel core was removed from samples of
Grosbeak (ACSR 636.0 MCM) and Partridge (ACSR 266,8
MCM) cables. The resulting hollowed cables were introduced,
one at a time, in a 32 mm diameter test coil. The coil
impedance was measured for different frequencies with and
without the hollowed sample.
Fig. 9 shows the impedances measured, as well as those
calculated by (15) and (21). It can be seen a good agreement
between the theoretical and experimental results, which
supports the proposed model for the aluminum strands.
Fig. 9 also gives the coil impedance without the cable,
showing that the aluminum strands actually reduce the coil
impedance. A closer look into the real and imaginary data
shows that the resistance increases due to the losses in the
aluminum and the inductance decreases due to the
neutralization of the magnetic field within the strands.
Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014
20
15
Coil impedance ()
10
5
0
0 20 40 60 80 100 120
Frequency (kHz)
Fig. 9. Impedance of a coil containing a hollowed ACSR cable. Bold line:
coil impedance without cable; thin line: theoretical values for 266.8 MCM
cable; crosses: experimental data for 266.8 MCM cable; dashed line:
theoretical values for 636,0 MCM cable; triangles: experimental data for
(20)
636,0 MCM cable.
C. The Complete ACSR Cable
Once the effects of the steel and aluminum strands have
been modeled, (15) may be adapted in order to provide the
sensing coil impedance for a complete ACSR cable:
Z c R0 X 0 k s Qs k a Qa j X 0 1 k s Ps k a Pa , (22)
where P, Q, and k are given by (16), (17), and (18),
respectively, using the data for the relevant strand, i.e., the
(21) subscripts s and a apply to the steel and aluminum strands,
respectively. It is worth to remember that, for aluminum
strands, r' = 1 and r" = 0.
In order to verify (22) experimentally, two ACSR cables
were used. For each cable, two samples were prepared: one
remained intact and the other had its galvanization totally
removed. The zinc layer thickness was calculated based on the
mass variation, as explained before. The samples were
introduced into sensing coils and the resulting impedances
were measured. Fig. 10.a shows the experimental values
obtained for the cables considered, while Fig. 10.b shows the
corresponding theoretical values. It can be seen a good
agreement between the experimental and theoretical values:
the difference remained below 1% for the steel strands without
galvanization and below 2% for the galvanized strands. The
higher difference for the galvanized strands is likely due to the
uncertainties related to the assessment of the zinc layer
thickness by the loss-of-mass process. Fig. 10 also shows the
results for a 3/8" (9.2 mm) galvanized-steel overhead ground
wire (OHGW), for comparison.
ISSN 2177-6164
 12 method for detecting and locating flaws in overhead transmission lines,"
IEEE Trans. on Power Delivery, vol. 5, pp. 1484-1490, Jul. 1990.
[4] Y. Kojima, J. Fukuda, T. Kumeda, J. Iinuma, and M. Endo, "Corrosion
10
detector robot for overhead transmission line," Fujikura Technical
Test coil impedance ()

Review, n. 21, pp. 74-83, 1992.

8 [5] S. D. Kim and M. M. Morcos, "An application of solenoid sensor for
inspecting deterioration of ACSR conductors due to forest fires," IEEE
6 Power Engineering Review, pp. 50-53, Oct. 2001.
[6] D. G. Havard, G. Bellamy, P. G. Buchan, H. A. Ewing, D. J. Horrocks,
4 S. G. Krisnasamy, J. Motlis, and K. S. Yoshiki-Gravelsins, "Aged
ACSR conductors, Part I: Test procedures for conductors and line
2 items," IEEE Trans. on Power Delivery, vol. 7, n. 2, pp. 581-587, April
1992.
[7] D. G. Havard, G. Bellamy, M. K. Bissada, C. G. Fajardo,
0
D. J. Horrocks, J. R. Meale, J. Motlis, M. Tabatabai, and K. S. Yoshiki-
(a) Dove Partridge OHGW
Gravelsins, "Aged ACSR conductors, Part II: Prediction of remaining
life," IEEE Trans. on Power Delivery, vol. 7, n. 2, pp. 588-595 April
1992.
12
[8] "Inspection & assessment of overhead line conductors: A state-of-the-
science report," EPRI, Palo Alto - CA, 2000, 1000258.
10 [9] "Guide to help reduce deterioration & extend life of field aged ACSR
Test coil impedance ()

conductors: Application of available technologies to assess condition of

8 conductors," EPRI, Palo Alto - CA, 2002, 1001741.
[10] "New tools for conductor evaluation: State of the art review and
6 promising technologies," EPRI, Palo Alto - CA, 2003, 1002002.
[11] L. B. Geller, D. Pophenroth, J. E. Udd, and D. Hutchinson, "Evaluation
of electromagnetic rope testers: Joint Canadian-US work," Materials
4
Evaluation, vol.. 50, no. 1, pp. 56-63, 1992.
[12] A. V. Pinto Jr., M. Z. Sebrao, C. R. S. C. Loureno, I. S. Almeida,
2 J. Saad, and P. M. Loureno, "Remote detection of internal corrosion in
conductor cables of power transmission lines," 1st Int. Conf. on Applied
0 Robotics for the Power Industry, Montreal, Canada, Out. 2010.
(b) Dove Partridge OHGW [13] X. L. Jiang, Y. F. Xia, J. L. Hu, F. H. Yin, C. X. Sun, and Z. Xiang,
"Optimal design of MFL sensor for detecting broken steel strands in
Fig. 10. Impedance of a coil containing a complete ACSR or OHGW cable. overhead power line," Progress in Elect. Research, vol. 121, pp. 301-
Bars with dashes: new galvanization; Bars with bricks: no galvanization; 315, 2011.
(a) Experimental data; (b) Theoretical data from (22). [14] "Diagnosis of internal corrosion technology for ACSR conductor of
overhead power line", Fujikura News, n. 372, pp. 2, July 2012.
[15] "Overhead line corrosion detector", Cormon data sheet n. PTP002.
IV. CONCLUSION Available: www.cormon.com/products/datasheets\ptp002.pdf (acces.
08 Jan. 2013).
The electromagnetic model described in this paper allows [16] "Overhead transmission line corrosion surveying", Attar datasheet.
the assessment of the zinc layer thickness of galvanized steel Available: www.attar.com.au/overhead-line-corrosion-detection (acces.
08 Jan. 2013).
strands that form ACSR or OHGW cables, in order to provide [17] E. Halln, "Electromagnetic Theory," John Wiley & Sons Inc., 1962.
an indication of the remaining conductor life. The theoretical
results agreed well with results of experiments, which can be
easily reproduced.
A technique employing this model has been successfully
applied to a remote-controlled robot to inspect overhead
conductors. The field trials have been carried out in de-
energized and energized overhead lines, with consistent
results. In order to apply the eddy current inspection in
energized lines, the acquired data had to be processed in order
to remove the power current influence on the steel strand
permeability. This data processing technique will be subject of
a future publication.

REFERENCES
[1] J. Sutton and K. G. Lewis, "The detection of internal corrosion in steel-
reinforced aluminum overhead transmission lines," Proc. of UK
Corrosion'86, vol.1, pp. 345-359, Birmingham, UK, 1986.
[2] P. H. Schwabe and D. Pike, "The measurement of corrosion in overhead
power line," Anti-corrosion, July, 1988.
[3] K. Komoda, T. Kawashima, M. Mineruma, A. Mineyama, M. Aihara,
Y. Ebinuma, T. Kanno, and M. Kiuchi, "Electromagnetic induction

Anais do V Simpsio Brasileiro de Sistemas Eltricos, Foz do Iguau PR, Brasil. 22-25/04/2014 ISSN 2177-6164