UNIVERSITY OF GAZIANTEP

FACULTY OF ENGINEERING
CIVIL DEPARTMENT

CE-547
Corrosion of Plain &Reinforced concrete

Report #5 About :
(An article research about galvanostatic

pulse method to determine the

corrosion status of reinforcement in
concrete )
Submitted to:

Do.Dr. Mehmet GESOLU

Prepared by:

Chalak Ahmed Mohammed

[email protected]

2014 45056

Date : 13.04. 2015

1

Corrosion Science 42 (2000) 675686

An algorithm of galvanostatic pulse method to

determine the corrosion status of
reinforcement in concrete
Cui Lu, Yan Peiyu*
Depavtment of Ci1il Fngineeving, Zzinghua Uni1evzit5, Beijing 100084, Peoplez Republic of China
Received 22 December 1998; accepted 12 August 1999

Abstrast
An algorithm was developed to Fourier-transform the measuring data of galvanostatic
pulse method (GPM) from the time domain into the frequency domain. The polarization
resistance and ohmic resistance are determined in the time domain; the impedance spectrum
and disperse parameter are obtained in the frequency domain to characterize the corrosion
status of reinforcement in concrete. Based on the investigation upon the diRerent corrosion
status of rebars in solution and in concrete, the criterions of depassivation are
discussed. g 2000 Elsevier Science Ltd. All rights reserved.
Ke5wovdz Rebar; Concrete; Galvanostatic; Time domain; Frequency domain

1. Introdustion
The corrosion status of steel rebar in concrete is commonly examined by several
electrochemical techniques [1], such as linear polarization response (LPR) and
electrochemical impedance spectrometry (EIS). Linear polarization response is a
relatively mature method which can be achieved galvanostatically [2]. LPR has
been used extensively on site and its data are processed normally in time domain.
The LPR results can reveal only limited information about steel corrosion in
* Corresponding author. Tel. +86-1062785836; fax +86-1062771132.
F-mail addvezz [email protected] (Y. Peiyu).
0010-938X]00]$ - see front matter g 2000 Elsevier Science Ltd. All rights reserved.
PII S 0 01 0 - 9 38 X ( 9 9 ) 00 1 02 - X

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

676

concrete and can not distinguish the contributions of various processes in the
whole sample system, e.g. charge-transfer resistance, concentration polarization
and ohmic resistance. EIS is a traditional measuring method in frequency domain
and a powerful tool for identification of corrosion mechanism [3]. Repeated
perturbations upon the system and long measuring time in low frequency range
may result in change of the corrosion status and error to the outcome. EIS
measurement needs more sophisticated instruments. Therefore, EIS is mainly used
for the researching work in laboratory. Owing to the distinct advantages of the
two methods, both of them have been developed, respectively. However, the
diRerence between the time domain and frequency domain makes great diAculties
to compare and inoculate the experience results of EIS and LPR.
Besides the steady-state measuring methods, some non-steady-state measuring
methods are also developed to determine the corrosion status of rebar in concrete,
such as galvanostatic pulse method (GPM) [48] and coulostatic transient method
[9]. Since the extensive review of the electrochemical technique by Gonzalez in
1985 [10,11], transient methods have made a big progress. Among them, GPM
gains special attention. Bohni used GPM successfully on site [5]. He simplified the
reinforced concrete system to a modified Randles equivalent circuit consisting of
an ohmic resistance in series with a parallel combination of the double-layer
capacitance Cdl and the linear polarization resistance Rp. Newton [6] suggests a
complex model of multi-series RC elements to depict the system. He fitted the
potential response curve by a multi-exponential function
.
AV = Iapp R0

n
..

Ri Ri e

t/Ri Ci

.
(l)

i=1

However, in some cases, especially when the steel is badly corroded, too many RC
elements can be found to give them appropriate physical meanings. A complex
analysis tends to be subjective and sometimes does not result in a unique
interpretation of measured data. Millard [7] used the similar program of Newtons
to characterize the corrosion rate of rebar in concrete.
The data of transient measure are analyzed mainly in time domain at present,
which limit the practical information obtained in the experiments. As EIS shows
that the resistances and capacities of the steelconcrete interface equivalent circuit
are frequency dependent, precise analysis about the corrosion mechanism in time
domain will become diAcult. It is well known that the impedance spectrum may
be obtained by Laplace or Fourier transformation of transient data recorded in
the time domain. Glass [9] made a good example based on the coulostatically
induced transient by Laplace transformation. But the great impedance of concrete
cover may induce large error to the measuring result of coulostatic transient
method. In this paper, a new algorithm is presented. It can Fourier-transform the
data of GPM measurement from time domain into frequency domain and
determine the corrosion status of the rebar in concrete.

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

677

2. Wheory
A modified Randles equivalent circuit (Fig. 1) is adopted to describe the
electrochemical system in reinforced concrete. Rp(v) and Cdl(v) are frequencyindependent [12]. The input signal of GPM, a galvanostatic pulse (Fig. 2), is the
same as that of galvanostatic linear polarization method. The main diRerence
between the two methods is the technique used to record the potential transient.
GPM records the entire potential charging procedure, while LPR records mainly
the steady potential. The sampling rate of GPM is much higher than that of LPR.
On the basis of experience of LPR and EIS, the galvanostatically-induced
potential transient (Fig. 3) can be divided into three parts [12,13]
1. The first part of the transient is related to the concrete electrolyte or the
corrosion product. It can be considered as an ohmic resistance with a time
constant 0 in the range of s, or much smaller. It is far less than the sampling
period used in GPM measurement. Thus, in the galvanostatically induced
response curve, a sharp potential step, AV, is gained at the beginning of
transient, from which the concrete resistance can be got
R =

AV

(2)

Iapp

where AV is the potential step in the potential, Iapp the applied polarization
current.
2. The second part of the transient corresponds to the charging process of double
layer. The curve will eventually attain a limitation. This part contains the
information of the corrosion reaction.
3. The third part of the transient, which continuously increases without
attainment of steady state, corresponds to diRusion eRect. Ualter [14] found
that the transient corresponding to the diRusion eRect developed in a
characteristic way
AVdiff =

. .1/2
8
Iapp o t1/2
n

Fig. 1. Simplified Randles circuit [12].

(3)

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

678

Fig. 2. The galvanostatic pulse input into the sample system.

where AVdiff is the diRusion over-potential, Iapp, is the applied polarization

current, o is Uarburg coeAcient, and t is the time.
The diRusion transient AVdiff becomes linear when it is plotted against t 1/2. Thus,
this part can be easily removed by subtracting the t1/2 element from the charging
curve.
Then the rest of potential transient can be attributed to the linear polarization
resistance and double layer capacity. From the steady state value of the curve, we
can get the polarization resistance
Rp =

AVp
Iapp

(4)

where AVp is the polarization potential of the interface of corrosion product and
concrete.
The GPM measuring data in time domain can be Fourier-transformed into an
impedance spectrum in frequency domain

Fig. 3. The potential response of the sample system.

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

V(v)
X(v) =

U( v)

= R

Rp
1 (jv )1a

679

(3)

where U(v) and V(v) are the frequency function of input and output signal,
respectively, R is the concrete resistance, Rp is the polarization resistance, v is
the angular frequency, = RC is the time constant of circuit, and a is the
dispersion parameter ranged from 0 to 1.
Similar to EIS, the calculated impedance, X(v), can be expressed in terms of its
real, X', and imaginary, X", components. Thus, the data collected by GPM can be
translated into the same form as EIS. Uith the aid of the algorithm, a few
minutes detection by GPM can give similar results as those from several hours
EIS experiments. It has been proved that the error produced by data
transformation from the time domain into the frequency domain is less than 4%
[15].
Hachani [12] found by EIS that the dispersion parameter a is related with the
homogeneity of the product formed directly on the steel surface in concrete.
Therefore, the dispersion parameter a was recommended to characterize the
corrosion of the specimens. The depassivation of rebar begins when a > 0.1. The
larger the a value, the heavier the corrosion, especially the more dangerous the
pitting corrosion. McCaRertys method [16] is used to extract the dispersion
parameter a to characterize the corrosion status of rebar in concrete. Defining the
arc chord vectors U and V (Fig. 4)
V = X R

(6)

.
.
U = R Rp X

(7)

Combining Eqs. (5), (6) and (7), it can be obtained

Fig. 4. Definition of the arc chord vectors U and V and the colecole figure with diRerent a values.

680

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

.
. .=
.U
.
vC R (a1)
. V . . dl p

(8)

.
.
. = (a 1)ln v (a 1)ln CdlRp
ln.... V ..
.
.U .

(9)

Thus, a plot of log(v) vs. log(U/V ) gives a straight line (Fig. 5). a can be
obtained from the slope, while Cdl from the intercept as Rp has been determined
in time domain.

3. Experimental method
The test specimens are $8 mm ~ 80 mm mild steel bars. The bars were
subsequently polished to 1000 grit with SiC paper and then degreased with
acetone. Both ends of the bar were masked by adhesive tape to expose 1000 mm2
area. The testing systems were bars immersed in solution and reinforced concrete.
$25 mm ~ 100 mm concrete cylinders with bars embedded in center were prepared.
The composition of used concrete was shown in Table 1. The testing samples and
conditions were shown in Table 2. The used cement was PS-425 blended Portland
cement. The sand was natural river sand with a fineness modulus of 2.43. The size
range of coarse aggregate was 510 mm. After curing the concrete cylinders with
bar at 20 2C and 90% humidity for 28 days, the specimens were immersed in
solution. A 100 A]cm2 anode current was applied to sample No. 4 to accelerate
the corrosion. After 72 h, cracks produced by the expansion of corrosion product
could be seen on the surface of the concrete cylinder.
The testing bars in solution or concrete acted as working electrode (UE) during
the electrochemical measurement. An external saturated calomel electrode was
used as reference electrode (RE). The counter electrode (CE) was an aluminum
cylindrical shell ($80 mm ~ 60 mm) to obtain a better current distribution.
A galvanostatic pulse supplied by a 18 V DC battery was controlled by a series-

Fig. 5. Determination of disperse parameter a.

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

681

Table 1
The composition of concrete
E]C

Uater
(kg]m3)

Cement
(kg]m3)

Aggregate
(kg]m3)

Sand
(kg]m3)

Superplasticizer
(wt% of cement)

CaCl2
(wt% of cement)

0.44
0.44

268
268

609
609

456
456

1065
1065

0.2
0.2

0
3

wound resistor and a relay. The amplitude of the input signal was adjusted to
make the output signal not higher than 25 mV. The amplified potential response
was converted to digital signal by a 12-bit A]D converter and then recorded into a
personal computer. The polarization resistance and the ohmic resistance were
determined in time domain; other parameters were obtained in frequency domain.
The accuracy of the measurement was tested by analyzing a dummy Randles type
combination of electrical resistors and capacitors.

4. Results and dissussion

The potential response curves of sample Nos. 1 and 2 are shown in Figs. 6 and
7. Their calculation results are shown in Table 3. Sample No. 2 that is immersed
in Cl contaminated solution has smaller time constant of electrochemical
reaction and larger diRusion eRect than No. 1, that is immersed in saturated
Ca(OH)2 solution. The impedance spectrum of No. 1 (Fig. 8) is close to an ideal
semicircle. Therefore, there is basically one RC element corresponding to the
electrochemical reaction on the interface of steel and solution. The rebar in No. 1
must be passive now. The impedance spectrum of No. 2 (Fig. 9) contains two arcs
with diRerent curvatures. There are two RC elements in the system. One
represents the electrochemical reaction and another, the eRect of corrosion
product layer that induces the diRusion eRect. The rebar in No. 2 must be active
now.
The corrosion current of No. 1 is 0.36 A]cm2, calculated from the Stern

Fig. 6. The potential response of sample No. 1.

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

682

Table 2
Samples and the experiment conditions
No. Sample
1
2
3
4

Electrolyte solution

Measuring time

Rebar in Ca(OH)2 saturated

Ca(OH)2 saturated solution At 4 h after rebar was immersed in
solution
solution
Rebar in Ca(OH)2 saturated
Ca(OH)2 saturated solution At 4 h after rebar was immersed in
solution
solution containing 1%CaCl2 containing 1% CaCl2
Rebar in concrete without Cl Ca(OH)2 saturated solution At 4 h after rebar was immersed in
solution
Rebar in concrete with Cl
Ca(OH)2 saturated solution After the current is input for 72 h
containing 1% CaCl2

Geary equation
Icorr =

(lO)

Rp

where Tafel constant B is defined as 52 mV for the passive state.

Since corrosion current of No. 1 is larger than 0.1 A]cm2, the rebar comes into
activation. It is contrary to the fact and the criterion of the disperse parameter
that is 0.077 and remains in the range of passivation. The corrosion current of
No. 2 is 5.75 A]cm2 (Tafel constant B is defined as 26 mV for the active state).
Its disperse parameter is 0.28. Both of them show that the rebar begins to corrode
and the surface non-uniformity increases because Cl ion accelerates the corrosion
rate and increases the possibility of pitting corrosion.
The potential response curve of the rebar in plain concrete (No. 3) (Fig. 10) has
a large time constant and an obvious diRusion eRect because oxygen diRuses with
diAculty in uncracked and water-saturated concrete. Its impedance spectrum
contains two arcs corresponding to the electrochemical reaction and the interface
between rebar and concrete, respectively (Fig. 11). The disperse parameter is very
small (Table 3). It means that the rebar in plain concrete is passive and its surface

Fig. 7. The potential response of sample No. 2.

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

Fig. 8. The impedance spectrum of sample No. 1.

Fig. 9. The impedence spectrum of sample No. 2.

Fig. 10. The potential response of sample No. 3.

683

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

684

Table 3
The GPM results of corrosion status of rebar in solution or concrete
No.

Ohmic
resistance
(kM cm2)

Polarization
resistance
(kM cm2)

Double-layer
capacitance
(F]cm2)

Corrosion
current
(A]cm2)

Disperse
parameter

1
2
3
4

=0
=0
0.15
8.48

144.69
4.52
267.97
2.71

48.6
247.2
28.8
83.8

0.36
5.75
0.19
9.60

0.077
0.28
0.0075
0.422

is smooth and uniform. But there is once again a contradiction between the
disperse parameter and the corrosion current that is 0.19 A]cm2 larger than the
limit of passivation.
After acceleration of corrosion by 72 h input of anodic current, the rebar in the
concrete cylinder containing Cl (No. 4) corrodes heavily and the concrete cover
has cracks. At this time, both the diRusion resistance of oxygen in concrete and
the polarization resistance of the electrochemical reaction decrease obviously. The
potential response curve has small time constant and low diRusion eRect (Fig. 12).
The ohmic resistance increases because there is a thick isolate layer of corrosion
product on the surface of rebar. Therefore, there is a significant jump in the
beginning of response. The impedance spectrum contains three arcs with diRerent
curvatures (Fig. 13). They represent three RC elements of electrochemical
reaction, the interface of rebar and corrosion product and the interface of
corrosion product and concrete, respectively. Both the disperse parameter of 0.422
and the corrosion current about 10 A]cm2 show that there is badly non-uniform
corrosion on the surface of rebar.

Fig. 11. The impedance spectrum of sample No. 3.

11

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

685

Fig. 12. The potential response of sample No. 4.

5. Conslusion
An algorithm is developed to Fourier-transform the measuring data of
galvanostatic pulse method from the time domain into the frequency domain. The
polarization resistance and ohmic resistance are determined in the time domain;
the impedance spectrum and disperse parameter are obtained in the frequency
domain to assess the corrosion status of the rebar in concrete.
The criterion of corrosion current may over-estimate the beginning time of
depassivation when the status of rebar is just below the critical depassive level. In
this case, the criterion of disperse parameter should be prior used. Both criterions
of corrosion current and disperse parameter can judge accurately the corrosion
status of rebar in the case of heavy corrosion.

Fig. 13. The impedance spectrum of sample No. 4.

686

C. Su, Y. Pei5u ] Covvozion Science 42 (2000) 675686

Asknowledgements
This work was funded by Chinese National Hey Projects on Basic Research and
Applied Research Applied Research on Safety and Durability of Major
Construction Projects.

Referenses
[1] A. Bentur, S. Diamond, A.S. Beerke, in Steel corrosion in concrete, 1st ed, E&FN SPON,
London, 1997, p. 68.
[2] H.R. Gowers, S.G. Millard, J.S. Gill, R.P. Gill, Programmable linear polarization meter for determination of corrosion rate of reinforcement in concrete structure, British Corrosion Journal 29
(1994) 2532.
[3] C. Andrade, L. Soler, X.R. Novoa, Advances in electrochemical impedance measurements in reinforcement in reinforced concrete, Material Science Forum 192194 (1995) 843.
[4] G.H. Glass, C.L. Page, N.R. Short, S.U. Yu, An investigation of galvanostatic transient method
used to monitor the corrosion rate of steel in concrete, Corrosion Science 35 (1993) 1585.
[5] B. Elsener, H. Uojtas, H. Bohni, Galvanostatic pulse measurements rapid on-site corrosion
monitoring, Proc. Inter. Conf. Corrosion and Corrosion Protecting of Steel in Concrete 1 (1994)
236.
[6] J.G. Newton, J.M. Sykes, A galvanostatic pulse technique for investigation of steel corrosion in
concrete, Corrosion Science 28 (1989) 1051.
[7] S.G. Millard, H.R. Gowers, J.H. Bungey, Galvanostatic pulse techniques a rapid method of
assessing corrosion rate of steel in concrete structures, Corrosion 95, Paper no. 525.
[8] A.A. Sagues, S.C. Hrang, E.I. Moreno, Evaluation of electrochemical impedance with constant
phase angle component from the galvanostatic step response of steel in concrete, Electrochimica
Acta 41 (1996) 1239.
[9] G.H. Glass, A.M. Hassanein, N.R. Buenfeld, Low frequency impedance data by Laplace transformation of coulostatically induced transients, Electrochimica Acta 43 (1998) 1863.
[10] J.A. Gonzalez, A. Molina, M.L. Escudero, C. Andrade, Errors in the electrochemical evaluation
of very small corrosion rate, part I, Corrosion Science 25 (1985) 917.
[11] J.A. Gonzalez, A. Molina, M.L. Escudero, C. Andrade, Errors in the electrochemical evaluation
of very small corrosion rate, part II, Corrosion Science 25 (1985) 519.
[12] L. Hachani, E. Triki, J. Grandet, A. Raharinaivo, Comparing the steelconcrete interface state
and its electrochemical impedance, Cement and Concrete Research 26 (1996) 253.
[13] L. Hachani, C. Fiaud, E. Triki, A. Raharinaivo, Characterization of steel]concrete interface by
electrochemical impedance spectroscopy, British Corrosion Journal 19 (1994) 122.
[14] G.U. Ualter, Corrosion rates of zinc, zinc coatings and steel in aerated slightly acidic chloride
solutions calculated from low polarization data, Corrosion Science 16 (1976) 573.
[15] P.Y. Yan, L. Cui, Analysis of galvanostatic pulse method in frequency domain used in corrosion
studies, Journal of Tsinghua University (Science and Technology) 39 (6) (1999) 124.
[16] E. McCaRerty, On the determination of distributed double-layer capacitance from colecole plots,
Corrosion Science 39 (1997) 243.

13