Вы находитесь на странице: 1из 7

Materials Transactions, Vol. 47, No. 4 (2006) pp.

1240 to 1246
#2006 The Japan Institute of Metals

EXPRESS REGULAR ARTICLE

Development of AE Monitoring Method for Corrosion Damage of the Bottom


Plate in Oil Storage Tank on the Neutral Sand under Loading
Sosoon Park1; * , Shigeo Kitsukawa2 , Kenji Katoh3 , Sigenori Yuyama4 ,
Hiroaki Maruyama5 and Kazuyoshi Sekine1
1
Division of Materials Science and Engineering, Faculty of Engineering, Yokohama National University,
Yokohama 240-8501, Japan
2
High Pressure Institute of Japan, Tokyo 101-0025, Japan
3
Nippon Steel Corporation, 20-1, Shintomi, Futtsu 293-8511, Japan
4
Nippon Physical Acoustics Corporation Ltd., Tokyo 150-0011, Japan
5
Japan Oil, Gas and Metals National Corporation, 212-8554, Japan

In Europe and USA, in-service diagnosis technique by AE method has already been used for evaluating the corrosion damage of bottom
plate of oil storage tank. However, the diagnosis technique is dependent on the empirical method without physical bases such as the mechanism
of AE generation due to the corrosion of bottom plates of oil storage tank. And also the relation between the corrosion rate and AE activity has
not been claried yet. This study was designed to clarify AE source and the relation between corrosion rate and AE activity during corrosion of
bottom plate in oil storage tank, in particular, on the neutral sand bed. For modeling the corrosion condition which is similar to that of an inservice tank bottom on the neutral sand, the static stress of 382 MPa had been applied on the specimen plates. We investigated experimentally the
relation between the corrosion rate and AE activity during the corrosion. Furthermore, the AE source and mechanism were examined based on
the SEM observation of corroded surface. From these experimental results, the validity and physical basis for quantitative evaluation of
corrosion rate by means of AE was presented.
(Received October 19, 2005; Accepted February 24, 2006; Published April 15, 2006)
Keywords: oil storage tank, acoustic emission source, acoustic emission activity, corrosion, cracking

1.

Introduction

In-service diagnosis technique for corrosion damage


evaluation of oil storage tanks is increasingly needed to
reduce the maintenance costs. AE method has been considered as a cost-eective one to meet the requirements of inservice inspection of structures. Therefore, in Europe and
USA, in-service diagnosis technique by AE method has
already been adopted for evaluating the corrosion damage of
bottom plate of oil storage tanks (hereafter, tank bottom).
Concerning AE corrosion monitoring, there have been basic
research works on the following subjects: 1) AE due to the
fracture of rust in the air.1) 2) AE due to the corrosion in an
acid liquid2,3) or in NaCl solution.4) 3) Application of AE
inspection to evaluate the corrosion damage of tank bottoms.511) 4) Source location and the propagation of AE
waves.1214) However, diagnosis of corrosion damage by AE
method for real tank bottoms has been conducted by
empirical procedures based on the eld data and is still
conned to a qualitative evaluation. Besides, AE sources due
to the corrosion of tank bottom and the relation between
corrosion damage and AE activity are not yet clear. Therefore, the mechanism of AE generation caused the corrosion
of them was discussed, and the relation between these two
factors were also examined in this study.
Regarding the in-service tank bottoms, two kinds of
corrosion processes would be considered. One is corrosion in
the acidic environment.1520) The other is corrosion in the
neutral environment. As possible mechanisms of AE generation due to the corrosion of in-service tank bottom, the
followings can be considered: (a) Hydrogen gas generation or
*Graduate

Student, Yokohama National University

bursting of hydrogen gas bubbles (hereafter, bursting) under


the strong acid condition, (b) Cracking of corrosion products
according to the corrosion progress under neutral conditions,
(c) Stress corrosion cracking under the external load. The
mechanism of (a) and the related AE activity behaviors in
acidic environment were claried in the previous paper by
the authors.14,15) This paper is concerned with the mechanism
of (b).
Corrosion Reaction of tank bottoms under the neutral
environment
2Fe ! 2Fe2 4e (anodic reaction)


O2 2H2 O 4e ! 4OH (cathodic reaction)


Fe(OH)2 $ Fe

2OH

Fe

2
3

2Fe O2 2H2 O ! 2Fe(OH)2


2

2H2 O ! Fe(OH)2 2H

6Fe(OH)2 O2 $ 2Fe3 O4 (ferrous oxide) 6H2 O 6


4Fe(OH)2 O2 ! 2Fe2 O3 (ferric oxide) 4H2 O

The electrochemical reactions that occur during the


corrosion of iron under neutral conditions can be described
with eq. (1)(7). Equations (1)(3) explain the rst stage of
the corrosion reaction of the tank bottom which was built of
SS400 steel plate. Equation (4) exhibits the result of
reactions (1)(3). As the reaction proceeds, the second stage
by hydrolysis reaction denoted by eqs. (3) and (5) would
occur. Under neutral conditions, the corrosion progress
moves to the third stage which is presented in eqs. (6) and
(7). As a result of these reactions, the corrosion products,
ferrous oxide and ferric oxide are generated at this stage.
According to corrosion progress, the corrosion products layer
grows up and their volume expansion give rise to cracking

Acoustic Emissions during Corrosion of the Bottom Plates of Oil Tank on the Neutral Sand

under a certain environmental condition. Cracking of the


corrosion products layer is considered to be more activated
by the weight of oils contained in an in-service tank. The
cracking can be a signicant as AE source due to the
corrosion of tank bottom on the neutral sand. If there are
chloride ions in the sand bed, the surface of tank bottom in
contact with the sand will become gradually more acidic at
the second stage. Furthermore, the bursting, which is
considered as another AE source due to the corrosion of
the tank bottom on strongly acidied may occur.14,15)
In this experiment, corrosion condition was controlled to
be similar to an in-service tank bottom. Specimen was set on
neutral sand and 380 MPa was loaded on it. During the
corrosion, corrosion rate and AE activity were measured
simultaneously. From the experimental results, the relation
between them was examined. Detected AE signals were
analyzed and classied into some groups using AE parameters such as peak frequency and RMS (AE Root Mean
Square value of the eective voltage [V]) of AE waveforms.
The characteristics of AE waves originated from dierent AE
source were analyzed by the group classication of AE waves
and the AE activity of each peak frequency. To verify the AE
corrosion monitoring for real above ground tanks, AE source
from corroded tank bottom was discussed with the obtained
results.
2.

Experimental Procedures

Our experiment is composed of two parts of measurement.


One is AE behavior during the corrosion of the tank bottom
specimen. The other is the corrosion behavior of specimens.
These two behaviors were estimated using the same specimens under the similar corrosion condition to in-service tank
bottom. Figure 1 illustrates the experimental set up for
measuring the behavior of corrosion and AE.
Table 1 shows the corrosion conditions of specimen and
the reference names of each experiment. Eective corrosion
area, that is, the area of the plate in contact with the sand was
10  10 cm2 . Side surface and top surface of the specimen
were coated with resin to protect them from corrosion.
Specimens were set on neutral sand and loaded by a static
stress of 382 Pa, which was estimated from the stress by the
load of oil contained in an in-service oil tank of a diameter of
10 m. During the volume expansion of the corrosion product
growth, this stress works as a compression stress to the
corrosion product. It is considered to make the layer ner to
Steel weight
2cm

Initial moisture contents of sand : 12 mass%

Vinyl box for sealing up

20cm
20cm

380MPa

AE measurement

Potentiostat

Pre-amplifer(40dB)
Load

FRA

AE sensor
R.
E.

Specimen

DISP TM
Digital AE system
with 4 channels
PC for
signal analysis

PC
C. E. (SUS304)
sand

Corrosion rate
measurement

Rubber for absorbing room noise

Fig. 1 Experimental set up (for estimation of corrosion rate by AE


method).

Table 1
ment.

1241

Corrosion conditions and the reference names for each experiLoad

380 MPa (30 kN/1000 kl Tank)

Moisture contents

12 mass% of Sands

Experiment name

The sound SS4001

The corroded SS4002

1

It is for a specimen without corrosion productions, non-corroded


SS400.
2
It is for a specimen with corrosion products similar to corroded tank
bottom, corroded SS400.
2
This specimen was prepared from seriously corroded tank bottoms in
the eld.

vertical direction and also considered to make the cracking of


the layer more active during the growth of corrosion product.
In this experiment, moisture content was controlled to 12%
of sands mass (hereafter, 12 mass%) because it was known
that the corrosion rate was very high and the corrosion
products had been produced in the largest amount under this
condition.1620) The moisture content of sand was dened as
the mass ratio of distilled water to the dried sand. The
chamber for the corrosion environment was sealed up to
retain the moisture. Detailed congurations of the specimen
and apparatus for loading are shown in the center of Fig. 1.
The specimen was prepared from a seriously corroded tank
bottom which has been used for decades (hereafter the
corroded SS400). From this, it was conrmed that the
physical and chemical characteristics of corrosion products
on the corroded SS4400 are the same to those of in-service
tank bottoms. For the purpose of specifying the AE behavior
during corrosion, a non-corroded SS400 plate (hereafter, the
sound SS400) was also measured in the same corrosion
environment as the corroded one having initially corrosion
products. From the comparison of the results from the
corroded SS400 and the sound SS400, the AE behavior
responding to the corrosion behavior was examined.
There are many aecting factors like metallic condition,
acidity, moisture content, temperature, inhibitor and activator, etc. However, it is dicult to consider the inuence of
these factors at the same time. So, these environmental
factors were kept the same condition during this experiment.
However, a deeper consideration can be conducted by
comparing the results that were obtained during the changing
of these environmental factors, with the results in this
research.
2.1 Measurement of polarization resistance
Electrochemical impedance spectroscopy (hereafter, EIS)
was analyzed using Solartron Instruments to estimate the
polarization resistance during corrosion. Solartron Instruments are composed of SI 1287 for electrochemical interface
and SI 1260 for impedance/gain-phase analyzer with Z-Win
and Z-Plot software. Frequency response to an alternating
current (hereafter, AC) was measured by a frequency
response analyzer (FRA) and plotted as EIS. Applied
potential was modulated as sinusoidal signal with a frequency of 0.01 Hz30,000 Hz within the voltage  10 mV. (Refer
to Fig. 1).
Polarization resistance (Rp) was calculated from the EIS
(Nyquist plot). Reciprocal of polarization resistance (Yp) and
corrosion current density (icorr ) is given by the equation

S. Park et al.

Yp 1=Rp, icorr K  Yp respectively. K is the proportional


constant determined by the corrosion system.21) Therefore,
Yp means the representative value used for corrosion rate as
well as icorr . The corrosion potential (Ecorr ) was measured
with a saturated Ag/AgCl electrode as a reference electrode
(R.E.). SUS304 was used as a counter electrode (C.E.). The
corrosion state could be estimated from the values of Yp and
Ecorr . Experimental conguration for estimating the corrosion rate is described in detail in the left part of Fig. 1.
2.2 Measurement of AE
A digital AE system, mDiSP with a 4-channel board
(Physical Acoustics Corporation Ltd.) and 150 kHz AE
sensor of a narrow band was used for acquisition of AE
signals. Since this paper was focused on only the corrosion of
the tank bottom, the propagation of AE waves was not
considered. An AE sensor was set on the top surface of the
specimen. Therefore, the AE propagation distance is very
short and so the attenuation of AE waves during propagation
can be ignored. Furthermore, the 150 kHz sensor used in this
experiment has high receiving sensitivity in the range of 30
200 kHz. Hence there is no problem for the measurement of
the AE waves having the frequency range from 30 kHz to
200 kHz. AE signal acquisition was performed with a
threshold of 37 dB and a sampling rate of 2 MHz.
The AE experimental set-up is described in the right side
of Fig. 1. From the AE waves detected, AE activity was
estimated in terms of mean hit rate (hereafter, MHR) during
the AE acquisition period. And also MHR is determined by
(accumulated hit count)/(acquisition time). An AE signal of
which the maximum amplitude was over 37 dB was counted
as 1 hit. MHRt (MHR during entire AE acquisition time) and
MHR1hr (MHR at every 1hour step) are presented as AE
activity values. MHRt is considered to be suitable for
estimating the AE activity of which variations are not so
drastic as corrosion behavior in neutral sand. MHR1hr is
considered to be suitable for estimating the AE activity when
variation is drastic. The peak frequency [kHz] of the fast
Fourier transformation method and the RMS of AE waves
were analyzed. All signicant AE waves were classied in
terms of the value of the peak frequency and the RMS.
3.

Results and Discussion

3.1 Estimation of corrosion rate on the neutral sand


A typical result of EIS by Nyquist plots of the corroded
SS400 is shown in Fig. 2. Rs could be calculated from
impedance at higher frequency range of current. (Rp Rs)
could be estimated from impedance at lower frequency range
of current. Therefore, Rp could be calculated from the
diameter of the elliptical curve denoted by the dotted line.
From Fig. 2, the equivalent circuit of this corrosion system
could be estimated as shown in Fig. 3, where. Rp is the
polarization resistance to the ionic current ow in the SS400
plate during electrochemical reaction, and Rs corresponds to
the ohmic resistance to current ow in sand. An electrical
double layer which works as condenser was constructed on
the contacting surface between the specimen and the sand.
And its capacitance (Cdl in Fig. 3) causes the impedance
variation which responds to the frequency of the AC. From

-150

the corroded SS400 at first day

-130
Imaginary part /

1242

30 kHz

-110

Rp = 2(510-375) (/mm)
Rs = 375

-90

0.1Hz
0.05 Hz

-70

19 kHz

0.01 Hz

0.6 Hz

-50
-30

510

-10
378 Hz

300

Fig. 2

400
Rs = 375

500

600

700

Real part /

EIS of the corroded SS400 8 hours later (Nyquist representation).

Rp
SS400
Rs

SUS304

Cdl

Fig. 3

Equivalent circuit of corrosion system.

the phase analysis, the left part of Nyquist plot presented in


Fig. 2, which ranged from 30 kHz to 378 Hz, was considered
to be noise due to the deviation of phase.15) Therefore, it was
not considered as important. Phase analysis was conducted
from the phase shift which responds to the angular frequency
change of AC.
The Yp was calculated from Rp and Ecorr were also
measured directly. The changes of them with the corrosion
time are plotted in Fig. 4. Figure 4(a) shows the result of the
corroded SS400. Yp decreased from beginning time to the
100th hour but Ecorr did not exhibit any observable change in
this time range. From the 200th hour to the 272nd hour, Yp
suddenly increased while Ecorr decreased sharply. After the
280th hour, Yp showed a sudden turn for the decrease while
Ecorr kept decreasing tendency until the 500th hour. After the
500th hour, both Yp and Ecorr did not give a remarkable
change until the 1039th hour.
For the prediction of the corrosion state, the behavior of Yp
and Ecorr according to the time can be used as illustrated in
Fig. 4(c). Each corrosion state in Fig. 4(c) can be dened as
follows; 1) State A is corrosion acceleration by activation of
cathodic reaction. 2) State B corresponds to corrosion
acceleration by activation of anodic reaction. 3) State C
means corrosion reduction by reduction of cathodic reaction.
4) State D is corrosion reduction by reduction of anodic
reaction. These corrosion state representations are dened to
specify the corrosion behavior and also the reason of the
behavior change.
In Fig. 4(c), icorr increased from beginning to the 200th
hour under the State A. This is considered to originate in the
supplying of H2 O, that is, the electron ion on the surface of

15
-200

10

-300

-400
0

272hr
200 400

0
800 1000 1200 1400

600

the sound SS400


0
-100

-300

-400
200

400

600

800

0
1000

-2

Time, t / h

(b) The sound SS400 (without corrosion products)

30
the corroded SS400

I
II
III
IV
V

272 h

25
-2

-200

Time, t / h

205 h

20
III

15

II

340 h

Yp /10 Scm

Ecorr
Yp

-5

10

100

(a) The corroded SS400 (with corrosion products)


Reciprocal of polarization Resistance,

1243
Polarization Conductance,Yp / 10 Scm

-100

20

-2

Ecorr
Yp

-5

25

Corrosion Potential Voltage versus


SHE, Ecorr / mV

30
the corroded SS400

Yp /10 Scm

100

Reciprocal of polarization Resistance,

Corrosion Potential Voltage versus


SHE, Ecorr / mV

Acoustic Emissions during Corrosion of the Bottom Plates of Oil Tank on the Neutral Sand

10 991 h
5

0
-200

: State A
: State B
: State C
: State B
: State D

Yp

225 h

(+)

Yp

IV
317 h

1040 h

-150

-100

-50

185 h

465 h

100 (-)

50

Corrosion Potential Voltage versus SHE, Ecorr / mV

Ecorr(-)

Ecorr(+)

(c) Prediction for corrosion state by the corrosion potential voltage and reciprocal of polarization resistance

Behavior of corrosion by the corrosion potential voltage and reciprocal of polarization resistance by time.

MHR1hr
MHRt

60
294h

40

80
60
40
20

20
0
0

200

400

600

800

1000

0
1200

Time, t / h

-1

the sound SS400


MHR1hr
MHRt

300

100

200
50
100

0
0

100

200

300

400

Time, t / h

(a) The corroded SS400 (with corrosion products)


Fig. 5

MHR1hr, hit rate / hith

80

150

400
-1

-1

MHR1hr, hit rate / hith

the corroded SS400

MHRt, hit rate / hith-1

100

100

MHRt, hit rate / hith

Fig. 4

(b) The sound SS400 (without corrosion products)

Behavior of AE activity by MHR1hr and MHRt by time.

the corroded SS400 reacted to H2 O (Refer to eq. (2) of


corrosion reaction). From the 200th hour to the 272nd hour,
icorr increased sharply under State B. This corrosion acceleration is considered to be originated from the lack of
electron ion which is associated with the sucient supply of
H2 O. Therefore, anodic reaction is more activated and this
tendency might keep while there is sucient O2 . After the
272nd hour until the 465th hour, icorr exhibited a sharp
decrease according to the movement of corrosion state into
State C. This is considered to originate in the lack of O2 .
After the 465th hour, Yp and Ecorr increased gradually till the
1000th hour. After the 1000th hour, Yp began to decrease.
Afterwards, Yp was so low that it was impossible to measure
it with the measurement system mentioned in 2.1. Ecorr did
not show observable change during the corresponding period.
These results for the corroded SS440 mean that the corrosion
system was in the State A. Fig. 4(b) shows Yp and Ecorr of the
sound SS400 according to time. On the rst day, Yp of the
sound SS400 was only 40% of the corroded SS400s one.
Furthermore, Yp and Ecorr of the sound SS400 continued to
decrease gradually according to time. This means that the

corrosion state was in the State C and the corrosion rate was
very low. This can be explained by the existence of oxide
layer on the surface of the sound SS400, which kept the
surface of it in a passive state. Therefore, it is considered that
the corrosion scarcely progressed. The result of Fig. 4(b) also
indicates that it is hardly possible that corrosion product was
produced enough to emit AE from the undersurface of the
sound SS400 in 900 hours.
3.2 Analysis of AE activity
Figure 5 presents AE activity measured during the
corrosion experiment. Here, Fig. 5(a) shows MHR1hr and
MHRt of the corroded SS400 by corrosion time for 43 days.
The AE activity from the beginning time to the 72th hour (the
3rd day) exhibited a rapid decrease. From the 96th hour (4th
day) to the 168th hour, AE activity kept extremely low values.
However, AE activity tended to increase after the 200th hour
(8th day) and had peak value at the 294th hour (12th day).
After the 314th hour, the activity of AE started to decrease
and this tendency kept until the 500th hour. From the 500th
hour to the 1000th hour, the AE activity was very low.

th

-2
5

Yp /10 Scm

20
10
0
0

10

20

30

40

-1

MHRt, hit rate / hith

(a) Yp and MHRt at same time

-2

st

(from 1 day to 43 day)

30

30

y = 1.96x + 6.1
R2 = 0.73

20

at same time

Yp /10 Scm

40

Reciprocal of polarization Resistance,

S. Park et al.

Reciprocal of polarization Resistance,

1244

10
after 22hr
(from 9 th day to 43 th day)

0
0

10

20

30
-1

MHRt, hit rate / hith

(b) Yp and MHRt measured 22 hour later

Fig. 6 Relationship between Yp and MHRt.

Fig. 5(b) shows MHR1hr and MHRt of the sound SS400 on


the 8th day. The AE activity measured on the rst day was
very high, however it showed sharp decrease until the 82nd
hour (the 3rd day). In the time range from the 96th hour (the
4th day) to the 202nd hour (the 8th day), AE activity was very
low. Afterward, to the 1039th hour (the 43rd day), no AE
wave was detected. In this case, there was no corrosion
product layer on the surface of the sound SS400. From
Fig. 4(b), the corrosion rate of the sound SS400 for 41 days
was suggested to be very low. Therefore, for rst 8 days, the
possibility that the corrosion progressed, or that the cracking
of the corrosion products layer occurred is extremely low.
Therefore, AE waves which were detected from the sound
SS400 for rst 8 days are considered to originate in the
movement of sand which was pressed down by the load.
Considering the result of AE activity from the sound SS400,
we can guess that the movement of sand was almost settled
after 4 days, and was completely stabilized after 8 days.
3.3 Relationship between AE activity and corrosion rate
Figure 6(a) shows the relation between MHR1hr and Yp of
the corroded SS400 measured at the same time. MHR1hr and
MHRt were used for estimating the AE activity and the Yp
was used for estimating the corrosion rate. From Fig. 6(a), no
relation was found between these two factors. However, from
Figs. 4(a) and 5(a), it can be noticed that there was a time lag
of 22 hours between the times of their maximum peaks.
Hence, Yp was compared with MHRt measured 22 hours
later. Here, it can be know that AE activities in the period
from the rst day to the 3rd day presented in Fig. 5(a) (from
the rst day to the 8th day in Fig. 5(b)) include the false AE
signals emitted from the movement of sand. Therefore,
MHRt values from the rst day to the 8th day were excluded
to conrm that false signals are not included in the values for
estimating the relation between Yp and AE activity.
Figure 6(b) shows the relation between Yp and MHRt
measured 22 hour later. A proportional relation was found
between these two factors, taking into consideration the time
lag of AE activity such as MHR1hr and MHRt. MHR1hr
measured 22 hours later and Yp tended to the same result.
From these results, it is clear that the corrosion rate of the

tank bottom in the eld can be estimated. From this result, the
physical basis for quantitative evaluation of corrosion
damage is suggested. As for the further usage of this, the
possibility of a quantitative global diagnosis of the corrosion
damage in the tank bottoms by AE method was presented.
3.4

Pattern and waveform analysis of AE signals and


AE sources.
Various AE parameters were analyzed for pattern classication. However, most of AE parameters except frequency
and RMS had continuous values. Therefore, peak frequency
and RMS were used to classify the AE signals into patterns.
Figure 7 shows that AE waves were classied into some
patterns using the values of peak frequency and RMS.
AE waves of Group C1 for the corroded SS400 were
supposed to be originated from the corrosion. AE waves
having high RMS values of Group S2 of the corroded SS400
were considered to be caused by the movement of sand. The
reasons of this conclusion are as follows. 1) For both
specimens, the decrease of AE waves having a high peak
frequency and high RMS is obvious according to the increase
of time. 2) After the 8th day, no AE wave was detected from
the sound SS400. 3) The corrosion rate has a maximum value
around the 12th day. 4) The bursting, which was caused by the
hydrogen gas generated during the cathodic reaction,
happens under low pH corrosion condition and a high
corrosion rate in the State A.15,16) However, the initial pH of
the sand in this experiment was neutral, and the pH of the
sand under the specimen was 78 when the 1000 hours had
passed. Therefore, considering the values of Ecorr during this
experiment from the Pourbaix diagram, the possibilities that
hydrogen gas was generated and the bursting happened were
very low.21) 5) At the 12th day, there was no AE waves due to
the movement of sand (refer to 3.2).
Figures 7(a1) and 7(a2) are the AE patterns of the sound
SS400, Group S1 and Group S2. These patterns of signals
shown in Fig. 7(a1) and (a2) were false signals emitted from
the movement of the sand as mentioned in 3.3. Figure 8
shows MHRt of each peak frequency in each experiment.
From the Fig. 7(a1), (a2) and Fig. 8 it can be known that the
AE waves having high peak frequencies and high RMS

Acoustic Emissions during Corrosion of the Bottom Plates of Oil Tank on the Neutral Sand
0.002

the sound SS400

0.0015

Group S1
Group S2

0.001

Group S1
RMS, voltage / V

RMS, voltage / V

0.002

0.0005

1245

the sound SS400

0.0015

Group S2
75kHz

0.001

125kHz
167kHz

0.0005
0

50

100

150

200

50

100

150

200

Peak Frequency, frequency / kHz

Peak Frequency, frequency / kHz

(a2) at the 8th day

(a1) at the first day

(from the sound SS400, without corrosion products)


0.002

0.002

0.0015

Group S2

0.001

the corroded SS400


RMS, voltage / V

RMS, voltage / V

the corroded SS400

0.0005

0.0015
0.001

Group C1
0.0005

Group C1
0

0
0

50

100

150

200

Peak Frequency, frequency / kHz

50

100

150

200

Peak Frequency, frequency / kHz

(b2) at the 12th day

(b1) at the first day

(from the corroded SS400, with corrosion products)


Fig. 7

Pattern classication of AE waves by RMS and frequency which is due to the corrosion and the sand.

MHRt, hit rate / hith -1

25

corroded SS400(1st day)


corroded SS400(12th day)
sound SS400(1st day)
sound SS400(3rd day)
sound SS400(8th day)

20
15
10
5
0
0

50

100

150

200

250

Peak Frequency, frequency / kHz

Fig. 8

Acc hit for each peak frequency of each experiment.

values decreased at the 8th day. Our both seeing of the


Fig. 7(b1) and Fig. 8 leads to that the AE waves from the
corroded SS400 on the rst day also are included in Group
S2. This means that Group S2 of the corroded SS400 on the
rst day had the false signals emitted from movement of sand
as like those of the sound SS400. However, it is possible to
nd from the Fig. 7(b2) and Fig. 8 that AE waves from the
corroded SS400 on the 12th day are contained in just Group
C1, which had 0.0005 RMS and 2050 kHz peak frequency.
These waves were originated in the corrosion as mentioned in
3.3. The reason why the peak frequencies of AE waves from
the sound SS400 on the st day are higher than those of the
corroded SS400, is that the corrosion products layer on the
corroded SS400 absorbed the high frequency elements of the

AE wave. False signals in Group S1 of the sound SS400 had


two kinds of waveform, continuous type and burst type. RMS
is eective to detect continuous type one. From this fact,
false signals of continuous type can be eliminated with
classication of RMS from the AE signals due to the
corrosion. AE waves due to the corrosion had low peak
frequencies, from 25 to 50 kHz. This suggests that the false
signals having high frequency can be eliminated from AE
signals due to corrosion by means of classication of peak
frequency. Therefore, AE waves due to the corrosion are
considered to be characterized by RMS and peak frequency.
Through this pattern classication, the improvement of S/N
rate can be performed.
As an additional experiment, the corroded surface was
observed by SEM. Figure 9 shows a SEM micrograph of the
corroded surface. Cracking and aking of the corrosion
products layer were observed. According to the growth of
corrosion products, two neighboring corrosion product grains
would pushed each other. Finally, cracking or aking
occurred when the compressive force became higher than
the strength of the corrosion products or the adhesive force of
it. This means that AE due to the cracking or aking on the
corroded SS400 might have higher amplitude and can be
easily detected. AE activity of the corroded SS400 was
actually higher than that of the auxiliary experiment, of
which stress was not applied at the same corrosion environment although the detailed result of auxiliary experiment is
not presented in this paper. To examine the eect of stress
exactly, further work on auxiliary experiment should be

1246

S. Park et al.

Cracking

100 m

Flaking off

According to the growth of corrosion


products, the boundary of them approach
to and push each other. Finally cracking
or flaking of corrosion products occur.
Fig. 9 Observation of corroded surface by SEM.

cracking or aking of the corrosion products layer. The


movement of sand due to the growth of corrosion
products can also be an AE source due to the corrosion.
Characteristics of the AE waves due to the corrosion
are: a) Peak frequencies of AE waves were 2250 kHz.
b) Values of RMS were 0.00020.0016. c) Its waveform
was burst type.
(2) There is a positive proportional relationship between
the corrosion rate and AE activity due to the corrosion
of the corroded SS400 with initial corrosion products on
the neutral sand.
The results obtained in this study conrm that the active
corrosion of SS400 plates in neutral corrosion conditions
under load can be detected by the AE method. Furthermore,
the corrosion rate of the tank bottoms in the eld can also be
estimated using this method. AE sources due to the corrosion
of tank bottoms were specied and the physical basis for
quantitative evaluation of corrosion damage was suggested.
As for further usage of these results, the possibility of a
global diagnosis technique by AE method for evaluating the
corrosion damage in tank bottoms was presented.

REFERENCES

(+)

Fig. 10 Typical AE waveform and frequency contour by STFT method for


Group C1.

required.
As the discussion on Fig. 9, AE waves with high RMS
values of Group S2 from the corroded SS400 at the 12th day
are caused by the movement of sand and ones in Group C1 of
the corroded SS400 on the 12th day are supposed to be due to
cracking or aking of the corrosion products. Figure 10
shows an example of typical AE waveform in Group C1,
which has a low RMS and its waveform was the burst type
and these characteristic of Group C1 can conrm the facts.
4.

Conclusions

The sources of AE due to the corrosion of tank bottoms in


the eld were examined in this study. During the corrosion
experiment, the corrosion rate and AE activity were
measured and the relationship between these two factors
has been estimated. The results are as follows:
(1) AE waves detected from the corroded SS400 after the
8th day, at which the corrosion activation state started,
were due to the corrosion. A possible AE source is

1) M. Takemoto: Zairyo-to-Kankyo 51 (2002) 256261.


2) S. Yuyama: Boshoku Gijutsu 35 (1986) 163170.
3) S. Yuyama, T. Kishi and Y. Hisamatsu: The Iron and Steel Institute of
Japan 68 (1982) 155164.
4) Y. P. Kim, M. Fregonese, H. Mazille, D. Feron and G. Santarini:
NDT&E International 36 (2003) pp. 553562.
5) A. Kamiya, K. Morofuji, K. Enuma, M. Yamada and S. Yuyama:
J. High Pressure Institute of Japan 40 (2002) 228236.
6) M. Yamada, S. Kitsukawa, S. Yuyama and K. Sekine: Japan Society for
Non-Destructive Inspection 51 (2002) 1216.
7) G. Dai, W. Li, Y. Zhang and F. Long: Materials Evaluation 60 (2002)
976978.
8) I. Morita, T. Arakawa, H. Hatanaka and M. Hagiwara: J. High Pressure
Institute of Japan 40 (2002) 203212.
9) M. Yamada, S. Kitsukawa, S. Yuyama, A. Kamiya, K. Sekine and
H. Maruyama: J. High Pressure Institute of Japan 40 (2002) 222227.
10) S. Yuyama, M. Yamada and K. Sekine: J. High Pressure Institute of
Japan 40 (2002) 176184.
11) A. Maie, M. Yamade, S. Zama, K. Hatayama and H. Maruyama:
J. High Pressure Institute of Japan 40 (2002) 185193.
12) M. Takemoto: J. High Pressure Institute of Japan 40 (2002) 195202.
13) A. Kaneko and M. Takemoto: J. Japanese Society for Non-Destructive
Inspection 51 (2002) 718725.
14) S. S. Park, S. Kitsukawa, K. Katoh, S. Yuyama, H. Maruyama and K.
Sekine: The 17th International Acoustic Emission Symposium, (2004)
pp. 179185.
15) S. S. Park, S. Kitsukawa, K. Katoh, S. Yuyama, H. Maruyama and
K. Sekine: Mater. Trans. 46 (2005) 24902496.
16) K. Katoh, S. Itoh, H. Ishimoto and T. Yashiki: Zairyo-to-Kankyo 51
(2002) 99104.
17) K. Katoh, K. Suzumura, H. Ishimoto and T. Yashiki: J. High Pressure
Institute of Japan 41 (2003) 2835.
18) K. Katoh, H. Ishimoto and T. Yashiki: J. High Pressure Institute of
Japan 41 (2003) 5563.
19) K. Katoh, C. Katoh, H. Ishimoto and T. Yashiki: J. High Pressure
Institute of Japan 42 (2004) 8187.
20) The Japan Society of Corrosion Engineering: Corrosion Handbook,
(Maruzen, Tokyo, 2000) pp. III-4-1III-4-5.
21) M. Pourbaix: Atlas of Electrochemical Equlilbria in Aqueous Solutions, (National Association of Corrosion Engineers, Huston, 1974)
pp. 307321.