Professional Documents
Culture Documents
Corrosion under insulation (CUI) is one of the degradation phenomena that have become a serious problem in
recent years especially in chemical plants that have been operating for a long time. Development of a CUI inspection
technique which doesn’t require the removal of insulation and which is applicable to explosion-proof petrochemical
plants is strongly needed. So we focused attention on optical fiber Doppler sensors which already have the explo-
sion-proof characteristics, and we tried to develop a new CUI inspection technique using them. The development
of this new inspection technique is explained.
This paper is translated from R&D Report, “SUMITOMO KAGAKU”, vol. 2010-I.
Development of CUI Inspection Techniques moved as described previously, but even if complete tear-
Using Fiber Optic AE2) down inspections are carried out at great expense, corro-
sion is only found in 2 and 3 systems out of every 1000
CUI in carbon steel equipment and pipes has become systems. The problem is that the efficiency is very poor.
a deterioration phenomenon that has intensified in re-
cent years particularly in chemical plants that have 2. Principles of AE monitoring using fiber optical
been operating for many years. In particular, exterior vi- Doppler3)
sual inspection of pipes installed outside in pipe racks We focused on the relationship between corrosion
that are in high places is difficult, and also, since the and AE to establish an inspection method that was ef-
total distance is large, no CUI inspection methods have fective for pipe CUI. Fig. 2 is a conceptual diagram of
been established that are more effective than visual in- the AE generating mechanism. First of all, peeling and
spection with the insulation removed. In addition, the cracking of local corrosion products (corrosion
fact that 70 to 80% of the costs for these inspections is swelling) occur because of the progress in active corro-
scaffolding and tearing down the insulation is one major sion. At this time, the strain energy that has accumu-
problem. Therefore, there was a strong requirement for lated inside is released as minute elastic waves. Since
development of CUI inspection techniques for pipes these elastic waves are comparatively low frequency
that did not require the work of removing the insulation elastic waves from the audible to 500kHz, it is known
and worked for plant facilities requiring protection from for propagating in a wide range. Therefore, the pres-
explosions. We focused on fiber optical Doppler (FOD ence of corrosion can be sensed by detecting the AE
in the figures) sensors, which have explosion preven- caused by the peeling and cracking of the corrosion
tion properties from the beginning and attempted to de- using an AE sensor to detect the elastic waves as AE. In
velop new CUI inspection techniques. other words, we can consider improving the CUI in-
spection efficiency by tearing down the insulation and
1. Comparison and problems in CUI inspection carrying out visual inspections only on pipes where AE
methods for conventional pipes has been observed.
Table 1 gives the characteristics of various inspection
methods that have been applied to pipes up to now.
Corrosion Tubercles
AE sensor
Methods that have good inspection accuracy can only of Iron Rust
Crack…
carry out inspections at short distances, and methods ca-
pable of long-distance inspections (approximately 5m)
have poor inspection accuracy. Therefore, most opera-
tions carry out inspections where the insulation is re-
AE (audible sound~about 500kHz)
have explosion prevention properties. (5) The range of acteristics of being able to obtain an output proportional
temperature applications is limited. To solve these to the rate of expansion and contraction of the fiber,
problems, attention has been given to the fiber optic AE having a broad receiving band of 1 Hz to 1 MHz and
technology that has been developed in recent years. having a broad temperature range of –200°C to 250°C.
The image of telecommunications systems is strong In addition, advantages of fiber optical Doppler sen-
for fiber optics, but optical fibers can be used as sensors sors include (1) having high insulation properties, (2)
by using their Doppler effect. Now, when light waves having high resistance to electromagnetic noise, (3)
from light source with acoustic velocity C and fre- having explosion prevention properties and not generat-
quency ƒ0 are incident to an optical fiber, we let the opti- ing electrical sparks, (4) being capable of long distance
cal fiber be extended length L at velocity v (see Fig. 3). measurements and (5) having a broad range of applica-
Letting frequency ƒ0 be changed to ƒ1 because of the ble environments. Therefore, disadvantages (2)
Doppler effect at this time, frequency ƒ1 following the through (5) for the conventional piezoelectric AE sen-
change can be expressed as in Eq. 1 by the formula for sors described previously can in principle be resolved
the Doppler effect. by using fiber optical Doppler sensors.
out
Fig. 3 Model of Doppler effect of optical fiber
in f 0 – fd
C–v v f0
f1 = f0 = f0 – · f0 (Eq. 1)
C C
Fig. 4 Principle of FOD sensor
Therefore, if we consider modulation of frequency ƒ1
following the change from frequency ƒ0 prior to inci-
dence by ƒd, we get Eq. 2. Fig. 5 is a conceptual diagram of a fiber optic AE
monitoring system for pipe CUI. First, frequency ƒ0
v
fd = f0 · (Eq. 2) lightwaves from a laser Doppler vibrometer light
C
source in a measurement circuit are introduced into the
ƒd can be expressed as in Eq. 4 using the formula for to the optical fiber. The frequency of the lightwaves that
the wave given in Eq. 3. are incident to the fiber optical Doppler sensor receive
the AE arising because of the corrosion flaking and
C = f0 · λ (Eq. 3) cracking and are modulated to ƒ0 through ƒd. On the
other hand, the amount of frequency modulation is de-
v f0 1 dL
fd = f0 · = ·v= · (Eq. 4) tected using the heterodyne interferometric method.
C C λ dt
Specifically, a reference light with frequency ƒM
Eq. 4 shows that the rate of expansion and contrac- (80MHz) is added using a frequency modulator
tion of the optical fiber can be detected as and fre- (AOM), creating ƒ0 + ƒM modulation. Furthermore, the
quency modulation of the lightwave. difference in frequency ƒM + ƒd of the laser light from
In other words, we can detect the strain (elastic the sensor and the laser light from the measurement
waves, changes in stress, etc.) applied to the optical circuit is introduced and ƒd is detected. It is converted
fiber by reading the frequency modulation ƒd of the opti- to voltage V by a frequency-voltage converter and out-
cal fiber. Fiber optical Doppler sensors have been de- put. A frequency analysis (FFT) is used on the original
veloped for sensors that make use of the Doppler effect waveform data obtained at this time, and it is converted
when these optical fibers expand and contract (see Fig. into extracted data such that the horizontal axis is fre-
4). To increase the sensitivity of these sensors and quency and the vertical axis is spectral power. This
make reception from all directions possible, the optical waveform analysis is important as a technique for dis-
fiber is wound into a coil; therefore, the it has the char- criminating noise and the AE caused by corrosion.
6000
3.00E+08
4000
Power spectral (a.u.)
2.50E+08
Voltage (mV)
The noise and AE are sorted. 2000
2.00E+08
0
1.50E+08
1.00E+08 –2000
Fast Fourier
5.00E+07 –4000
transform (FFT)
0.00E+00 –6000
0 50000 100000 150000 200000 0 500 1000 1500 2000
Sampled data Frequency (Hz) Original data Time (µsec)
3. Mock-up piping for investigating CUI inspec- wetting and drying to efficiently generate CUI and fur-
tions ther by heating to 60 to 70°C using silicon oil. A fiber
Up to now there have been actual records of fiber optical Doppler sensor is positioned at any location on
optic AE evaluations of areas with loosening base rock this mock-up piping and data collected regarding AE
that accompanies underground storage tank excava- from this corrosion site.
tion, but there are no records of use in CUI evaluations. In this investigation, we used a commercial layered
Therefore, we fabricated mock-up piping like that fiber optical Doppler sensor that was a 65m optical fiber
shown in Fig. 6 to investigate the possibilities of devel- wound into a layered coil shape (see Fig. 7). This sen-
oping CUI inspection technology using fiber optic AE. sor is characterized by having an increased sensitivity
Insulation was applied to a 5m carbon steel pipe, and because the optical fiber is installed in a layered form.
heated silicon oil was circulated inside the pipe by a Even though it has a broad band, it has the same or
heating device. In addition, corrosion was accelerated greater receiving sensitivity than a narrow band con-
artificially by pure water and salt where the amount ventional piezoelectric AE sensor. When the layered
dripped was finely adjusted to an extent giving rise to fiber optical Doppler sensor is installed on the piping, it
70°C
FOD sensor
Piping Flange
FOD sensor
80 400
is secured using a U-bolt, and when it is installed on a AE hits/1 hour
70 350
Total AE hits
flange, it is secured with a clamp. Either is extremely
AE hits/1 hour (hits) 60 300
easy to install and remove, and like piezoelectric AE
Total AE hits
50 250
measurements, sending and receiving is possible 40 200
increase in
through a commercial contact medium. 30 temperature 150
decrease in
20 Pure temperature 100
Water
4. Results of measurements with mock-up piping4) 10
Salt 50
Voltage (mV)
2000
2.00E+08
0
1.50E+08
–2000 1.00E+08
–4000 5.00E+07
–6000 0.00E+00
0 500 1000 1500 2000 0 50000 100000 150000 200000
Time (µsec) Frequency (Hz)
2000
2.00E+08
0
1.50E+08
–2000 1.00E+08
–4000 5.00E+07
–6000 0.00E+00
0 500 1000 1500 2000 0 50000 100000 150000 200000
Time (µsec) Frequency (Hz)
400
200 5000000
0 4000000
–200 3000000
–400 2000000
–600 1000000
–800 0
0 500 1000 1500 2000 0 50000 100000 150000 200000
Time (µsec) Frequency (Hz)
pattern at as low a frequency as possible that does not 3,900mm from the site of the corrosion. Fig. 12 shows
overlap with the noise. the AE detection results at the 3,900mm position. In
this manner we confirmed that it was possible to detect
(2) Results of investigations into distances where AE is AE at a sufficient sensitivity even at maximum distance
detectable of 3,900mm. In addition, we found that the 50k to
In January 2008 we conducted verifications of the dis- 100kHz frequencies were detected the most out of the
tances at which AE could be detected. The state of cor- three AE patterns. If no noise is present in this range of
rosion on the piping at this time is shown in Fig. 11. frequencies in measurements on actual equipment,
Corrosion swelling has arisen, and in the progress of which are a problem we will take up later, we can effi-
the corrosion can be seen. We installed the fiber optical ciently detect AE caused by corrosion.
Doppler sensor at positions 2,000mm, 3,000mm and
Distance 3,900mm
60
100k-150kHz
AE hits/30 minutes (hits)
50 50k-100kHz
10k-50kHz
40
30
20
10
0
0-30 30-60 60-90 90-120 120-150 150-180 180-210 210-240
(3) Comparison of AE detection results on the piping confirmed that there was excellent AE detection on the
and flange parts flange part. If it is possible to detect AE on the flanges,
In March 2008 we compared AE detection on the pip- we can expect to only remove the insulation from the
ing and flange parts. The state of corrosion on the pip- flange parts to make AE measurements.
ing at this time is shown in Fig. 13. The corrosion
swelling has grown further, and cracking heads ap- (4) Investigation of corrosion progress and number of
peared in part of the corrosion swelling. Fiber optical AE hits
Doppler sensors were installed on the pipe at 3,900mm To make a comparison with the number of AE hits
and on a flange part at 3,950mm from the site of the cor- from the corrosion site in January 2008 and the number
rosion is shown in Fig. 14. of AE hits from the corrosion site in March 2008 after
the corrosion had progressed further, we installed a
fiber optical Doppler sensor at the same 3,900mm posi-
tion and compared the results of measurements. The
results of that comparison are shown in Fig. 16. How-
ever, the AE measurements in January 2008 were only
made up to 240 minutes. As is clear from this figure, we
can see that the number of AE hits in March has clearly
increased with the progress in the corrosion compared
with those in January (an approximately tenfold AE hit
number at 240 minutes). The AE method cannot quan-
Fig. 13 Corrosion of piping (3rd monitor: 03/2008) tify the area of the corrosion or depth of material reduc-
tion, but if the volume of the corrosion swelling
increases, the probability of AE hits increases. There-
FOD sensor 3,950mm fore, we can consider the possibility of evaluations
3,900mm using the degree of correlation with the progress of the
corrosion by counting the number of AE hits.
2500
Fig. 15 shows a comparison of the results for AE de- 400
350 Total AE hits
2000
tection on the piping and flange parts. While these re- 300
sults show poorer sensitivity than on the pipe itself, we 250 1500
200
1000
150
2500
400
350 Fig. 16 Comparison of the number of AE hits
Total AE hits
2000
300
01/2008 and 03/2008
250 1500
200
1000
150
100
500 Fiber Optic AE Measurements in Actual Equip-
50
0 0 ment5)
0-30 60-90 120-150 180-210 240-270 300-330
30-60 90-120 150-180 210-240 270-300 330-360
Elapsed time (minute)
1. Selection of stationary equipment and environ-
Fig. 15 Comparison of the situation of AE hits mental noise separation
piping and flange As was discussed earlier, it is possible to inspect
2. Attachment of fiber optical Doppler sensors to As is shown in Fig. 19, the fiber optical Doppler sen-
vertical reactor sors were protected by a waterproof case. After the sur-
The positions for attaching the fiber optical Doppler face coating on the outside of the reactor was removed
sensors to the vertical reactor are shown in Fig. 17 and with sandpaper, they were glued using a heat resistant
Fig. 18. Four (four channels) fiber optical Doppler sen- epoxy resin adhesive and secured from the top using
sors were attached at intervals of approximately aluminum tape. The signal cables extending from the
3,000mm with a pitch of 90° around the circumference fiber optical Doppler sensors were each collected in a
positioned vertically at 9,500mm based on the girth line jack box as ch1, ch2, ch3 and ch4 (see Fig. 20), and a
for the lower mirror and body of the main body of the cable was extended from there to the ground. When AE
reactor. measurements are made, a terminal box installed on
the aboveground part and a fiber optical Doppler inter-
ferometer installed in a vehicle are connected by a cable
(see Fig. 21). One of the major merits of this method is
Pass partition that, once the fiber optical Doppler sensors are con-
A A' nected to the actual equipment, inspections can easily
Heat exchanger
17500
13500
Heat
exchanger
9500
Air nozzle
5500
1440
0
0–30 30–60 60–90 90–120 150–180
10:05–10:35 10:35–11:05 11:05–11:35 11:35–12:05 12:35–13:05
0
3. AE measurement results before and after repair 0–30 30–60 60–90 90–120 120–150 150–180
of corroded parts 11:00–11:30 11:30–12:00 12:00–12:30 12:30–13:00 13:00–13:30 13:30–14:00
The results of AE measurements before and after the Fig. 23 AE hits/30 minutes after removing rust
removal of the corrosion due to CUI that had arisen on
the equipment by surface preparation are shown in Fig.
22 and Fig. 23, respectively. As a result, a large amount 4. Investigation of environmental noise separation
of AE was detected in the state where the corrosion due The typical AE waveforms detected before and after
to CUI was present, and after the corrosion had been re- the removal of the corrosion and the AE waveforms
moved, the number of AE hits was reduced to approxi- from corrosion that were obtained in the investigations
mately 1/10. Therefore, we assumed that we were also with the CUI mock-up piping are shown in Fig. 24
able to detect AE that can be thought of as being caused through Fig. 26, respectively. Since the rise of the
by corrosion on actual equipment. In addition, there waveforms, the duration, etc. have a very similar shape,
was also a large difference in the number of AE detec- it can be assumed that separating the environmental
tions by each of the sensors before removal of the cor- noise out using the shape of the waveforms will be diffi-
rosion. In visual inspections after the insulation was cult.
removed, multiple corrosion sites were found to be Next, the frequency distributions for the peak fre-
present in the vicinity of the sensors where there were quency (frequency with the largest amplitude in the fre-
many AE detections. In this manner, we were able to quency spectrum) and the barycentric frequency
confirm that there was a good correlation between the (frequency that is the center of gravity for the area of
state of corrosion and the state of AE detection in tests the frequency spectrum) for a measurement time of
carried out using actual equipment. three hours of AE detected before and after the removal
100
0 1500
–100
–200 1000
–300
500
–400
–500
0 500 1000 1500 2000 0
~50 50–100 100–150 150–200 200–250 250–300 kHz
Time (µsec)
Fig. 27 Frequency distribution of AE obtained
Fig. 24 Representative Waveform of AE obtained
before removing rust
before removing rust
Frequency distribution
1000
800 120
Peak Frequency
600
200 80
0
60
–200
–400 40
–600
20
–800
–1000 0
0 500 1000 1500 2000 ~50 50–100 100–150 150–200 200–250 250–300 kHz
Time (µsec)
Fig. 28 Frequency distribution of AE obtained
Fig. 25 Representative Waveform of AE obtained after removing rust
after removing rust
600
ing and that the frequency distribution of the AE after
500 the removal of the corrosion differed. These results
AE hits/3 hours (hits)
200
tal noise by finding the m value.
However, to use this separation technique at actual
100
sites, we will require the accumulation of more funda-
0
mental data based on laboratory results and the accu-
49 0
00
55 0
58 0
00
13 0
16 0
19 0
22 0
25 0
28 0
00
34 0
37 0
40 0
43 0
46 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
31
52
10
40
70
10
Maximum amplitude (mV) mulation of separation data for AE due to corrosion and
environmental noise through the accumulation of tests
Fig. 29 Distribution of AE hits and maximum
amplitude obtained before removing rust with actual equipment on site.
Summary
40
35
The AE arising from corrosion swelling was captured
AE hits/3 hours (hits)
30
roughly in the frequency range of 10kHz to 150kHz. In
25
20 addition, it was shown that AE could be captured even
15 at a maximum length of 4,000mm. This shows that it is
10 possible to inspect a range of 8,000mm left him right
5
with a single sensor, and since there was plenty of mar-
0
gin in the sensitivity, we can assume that even longer
55 0
58 0
00
13 0
16 0
19 0
22 0
25 0
28 0
00
34 0
37 0
40 0
43 0
46 0
49 0
00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
31
52
10
40
70
10
Maximum amplitude (mV) distances can be inspected. In addition, even if the sen-
sor was installed on a flange part, it was possible to cap-
Fig. 30 Distribution of AE hits and maximum
amplitude obtained after removing rust ture the AE from the pipe part in the same manner.
Furthermore, AE that was thought to be caused by cor-
rosion could also be detected in actual equipment, and
Mock-up piping it was suggested that separation of corrosion AE from
4.0
Before remove rust
y = –2.6719x + 10.183 After remove rust environmental noise would be possible by carrying out
3.5
y = –2.2302x + 8.5901 evaluations of the m values (shape parameter). Since
3.0
fiber optic AE sensors have explosion prevention from
log AE hits
2.5
the beginning, normal installation of the sensor parts
2.0 y = –1.7142x + 6.0529
may also be performed in plants that have where explo-
1.5
sions must be prevented such as petrochemical plants.
1.0
0.5
We hope to accumulate more data in the future and
“m” means the slope of approximate straight-line
0.0
move toward making this practical rapidly.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
PROFILE
Hidehiko SUETSUGU
Sumitomo Chemical Co., Ltd.
Process & Production Technology Center
Senior Research Associate