Zou, L.; Ferrier, G. A.; Afshar Vahid, Shahraam; Yu, Q.; Chen, L.; Bao, Xiaoyi.
Distributed brillouin scattering sensor for discrimination of wall-thinning defects in steel pipe under
internal pressure, Applied Optics, 2004; 43 (7):1583-1588.
Copyright © 2004 Optical Society of America
This paper was published in Applied Optics and is made available as an electronic reprint with
the permission of OSA. The paper can be found at the following URL on the OSA website
http://www.opticsinfobase.org/abstract.cfm?URI=ao-43-7-1583. Systematic or multiple
reproduction or distribution to multiple locations via electronic or other means is prohibited and
is subject to penalties under law.
OSA grants to the Author(s) (or their employers, in the case of works made for hire) the
(b)The right to post and update his or her Work on any internet site (other than the Author(s’)
personal web home page) provided that the following conditions are met: (i) access to the
server does not depend on payment for access, subscription or membership fees; and (ii) any
such posting made or updated after acceptance of the Work for publication includes and
prominently displays the correct bibliographic data and an OSA copyright notice (e.g. "© 2009
The Optical Society").
17th December 2010
Distributed Brillouin scattering sensor for
discrimination of wall-thinning defects in steel
pipe under internal pressure
Lufan Zou, Graham A. Ferrier, Shahraam Afshar V., Qinrong Yu, Liang Chen, and Xiaoyi Bao
A distributed Brillouin scattering sensor has been employed to identify several inner wall cutouts in an
end-capped steel pipe by measuring the axial and hoop strain distributions along the outer surface of the
pipe. The locations of structural indentations that constitute 50–60% of the inner pipe wall are found
and distinguished by use of their corresponding strain–pressure data.
terms of the fiber orientation, defect size and depth, and behavior relative to those of unperturbed pipe
sections. © 2004 Optical Society of America
060.2370, 290.5830, 290.0290, 060.2310.
These results are quantified in
Pipeline integrity and disturbance are generally not
monitored because of a lack of reliable and durable
Pipeline disturbances such as crack-
ing, corrosion, and tampering usually noticed only
when the output flow is affected or when they have a
severe effect on the surrounding environment.
fortunately, the lack of information available on the
types and locations of pipeline faults has created in-
efficient and potentially costly situations.
ample, as a result of corrosion fatigue cracking, a
28-in. ?71.12-cm-? diameter pipeline ruptured and re-
leased approximately 564,000 gallons of gasoline on 9
March 2000 in Greenville, Texas, which resulted in a
damage-cleanup cost of $18,000,000.2
corrosion was responsible for 63% of all pipeline fail-
ures in Canada from 1980 to 1997.3
Canadian pipeline industry spends more than
$50,000,000 a year for inspection and maintenance,
these pipeline failures occur with corresponding pro-
duction losses and environmental cleanup and litiga-
tion costs. Hence there is an urgent need for a real-
time structural-health monitoring system that can
intelligently monitor the integrity of underground
One of the major difficulties in monitoring pipe-
lines stems from the fact that tens of meters to hun-
dreds of kilometers of the pipeline length can be
buried underground.Conventional conductive sen-
sors have difficulties surviving their surrounding en-
vironments and have electrical noise problems.
Furthermore, numerous such point-sensing devices
are required for adequate monitoring of the health of
long pipelines, and at a substantial cost.
nately, with the advent of optical fiber technology
that uses low-cost optical fiber communication cables,
distributed structural-health monitoring of pipelines
can be achieved.
Fiber optic sensor technology has progressed at a
rapid pace over the past decade.
niques have been developed to monitor specific pa-
scattering–based sensor systems provide an excel-
lent opportunity for monitoring the structural health
of civil structures4by allowing measurements to be
taken along the entire length of the fiber, rather than
only at discrete points, by use of the fiber itself as the
sensing medium.One class of Brillouin-based sen-
sor is based on the Brillouin loss technique5whereby
two counterpropagating laser beams, a pulse and a
cw, exchange energy through an induced acoustic
field. When the beat frequency of the laser beams
equals the acoustic ?Brillouin? frequency vB, the
pulsed beam experiences maximum amplification
from the cw beam. By measuring the depleted cw
beam and scanning the beat frequency of the two
Many sensing tech-
The authors are with the Fiber Optics Group, Department of
Physics, University of Ottawa, 150 Louis Pasteur, Ottawa, Ontario
K1N 6N5, Canada.L. Zou’s e-mail address is lufan.zou@
Received 31 July 2003; revised manuscript received 10 Novem-
ber 2003; accepted 5 December 2003.
© 2004 Optical Society of America
1 March 2004 ? Vol. 43, No. 7 ? APPLIED OPTICS1583
lasers, one obtains a Brillouin loss spectrum centered
about the Brillouin frequency.
ity of Brillouin scattering arises from the dependence
of Brillouin frequency vBon the local acoustic velocity
and refractive index in glass, which has a linear tem-
perature and strain dependence6,7:
The sensing capabil-
vB?T0, ε? ? Cε?ε ? ε0? ? vB0?T0, ε0?, (1)
vB?T, ε0? ? CT?T ? T0? ? vB0?T0, ε0?, (2)
where Cεand CTare the strain and temperature
coefficients and ε0and T0are the strain and temper-
ature that correspond to a reference Brillouin fre-
quency vB0. One obtains spatial information along
the fiber length through optical time-domain analysis
by measuring the propagation times for light pulses
traveling in the fiber, thus allowing continuous tem-
perature and strain distributions to be obtained.4
This typeof sensinghas
structural-health monitoring because one can adjust
the spatial resolution for different applications sim-
ply by altering the pulse duration, even after the fiber
In this paper we report the results of our recent
experiments in using a distributed Brillouin scatter-
ing sensor to detect pre-embedded inner-wall cutouts
?defects? in an internally pressurized 1.83-m-long
end-capped steel pipe. The pre-embedded defects,
which constitute 50–60% of inner wall thickness, are
discriminated by use of the corresponding strain
measurements in the axial and hoop directions along
2. Experimental Procedures
To simulate the structural degradation of pipes we
entrenched rectangular indentations within the in-
ner wall of a steel pipe constructed at the Canmet
Materials Technology Laboratory in Ottawa, On-
tario, Canada, as shown in Fig. 1.
was 1.83 m long, with two supporting points and a
42-kg end cap located at each of the two ends.
material specifications of the pipe are as follows:
Poisson’s ratio, 0.3; Young’s modulus, 200 GPa ?equal
to 30 ? 106psi?; length, 183 cm; outer diameter, 11.4
cm; wall thickness, 1.1 cm.
with distilled water and pressurized with argon gas.
Preparation of Steel Pipe with Pre-Embedded
The steel pipe
The pipe was 90% filled
At predetermined locations along the pipe’s inner
wall, cutouts ?defects? that constituted 50–60% of the
wall thickness ?as shown in Fig. 2?a?? were purposely
implanted to simulate the condition in which a cor-
roded pipeline has approached its threshold for re-
placement ?compromise of 50% of the wall thickness?.
For comparison, we analyzed the behavior of a pipe
with cutouts that constituted 60% of the wall thick-
ness to determine whether the Brillouin sensor sys-
tem could differentiate the strain responses of pipe
regions that had different wall thicknesses.
pipeline defect occurs, a certain percentage of the
inner wall’s thickness is lost, and larger strains ap-
pear in the defective region at a constant pressure
within the pipe. The strain measurements were
conducted at regular pressure intervals from 0 to 500
A 30-m acrylate buffered SMF-28 optical fiber was
used for monitoring both the environmental temper-
ature conditions and the strain distributions along
the pipe.The distribution of defects and sensing-
fiber installations is shown in Fig. 2; the location and
dimensions of cutouts A, C, and D are tabulated in
Table 1; and region B constitutes the rest of the un-
perturbed pipe.To prepare the pipe for fiber instal-
lation we used sandpaper to remove irregularities
caused by excess welding and provide a smooth, uni-
form surface about its circumference.
Installation of Sensing Fibers
We then pre-
distributed Brillouin sensing experiment.
was contained by two bolted 42 kg end caps, and a frame near each
end supported the pipe.
Schematic diagram of the 1.83-m steel pipe used for the
The internal pressure
hoop installation of sensing fiber ?c?.
long cutout with 60% wall thinning, which starts at 23 and ends at
84 cm.Region B consists of the rest of the unperturbed pipe.
Regions C and D are 1.3 cm ? 10 cm cutouts with 50% and 60%
wall thinning, respectively.
Schematic diagram of the distribution of inner wall-
?a? axial installation of sensing fiber ?b? and
Region A is a 5.3 cm ? 61 cm
1584APPLIED OPTICS ? Vol. 43, No. 7 ? 1 March 2004
strained and mounted the optical fibers externally on
the pipe to monitor the strain changes on the outer
surface and secured them by using a special low-
viscosity glue made by MasterCraft, which required
approximately 3-h drying time.
cross talk and additional noise that might result from
fiber overlap we conducted the axial ?Fig. 2?b?? and
hoop ?Fig. 2?c?? strain measurements separately.
The temperature of the pipe was monitored by use of
three thermocouples at various locations ?Figs. 2?b?
and 2?c?? and was found to have a negligible effect
???5 ?ε? on strain measurements.
To prevent signal
The experimental setup for measurements of the
Brillouin loss spectrum was reported previously.8
The system is based on the interaction of a pulsed
laser and a counterpropagating cw laser operating at
a 1319-nm wavelength, as mentioned above.
pulsed beam is subjected to Brillouin amplification at
the expense of the cw beam.
drops in the cw beam are measured while the fre-
quency difference between two lasers is scanned, giv-
ing the Brillouin loss spectrum of the sensing fiber.
The Brillouin shift of the fiber is determined from the
spectrum and is used to evaluate the strain of the
sensing fiber. We acquired Brillouin spectra every 5
cm ?using a high-speed 2-GHz digitizer? along a 30-m
fiber every half-hour by using 8000 waveform aver-
ages.Hence we used a pseudo-Voigt function9for
peak fitting of each Brillouin spectrum to acquire
frequency and strain information.
tion is determined by the pulse duration, which has a
1-to-1 correspondence with fiber position through t ?
2nd?c, where d is the fiber position.
iment, a 1.3-ns pulse duration provided a 13-cm spa-
Distributed Brillouin Scattering Sensor System
The resultant power
For this exper-
3.Results and Discussion
The axial strain was measured by use of the config-
uration shown in Fig. 2?b?.
ranges from 23 to 84 cm, and the remaining pipe is
unperturbed along the 6-o’clock direction.
strain distribution along the longitudinal direction of
the pipe under 200-psi internal pressure is presented
in Fig. 3. The maximum ?46-?ε? and minimum ?14-
?ε? strains occur in the middle of the defective and
the unperturbed regions, respectively.
difference is outside the experimental error range
??15 ?ε based on a standard deviation? and therefore
is caused directly by wall thinning, the capability of
Measurement of Axial Strain
Large defective region A
As this 32-?ε
our system to distinguish defective and unperturbed
regions is successfully demonstrated.
strains depend sensitively on proximity, end-cap size
and type, and overall pipe design.10
steel pipe is asymmetric in terms of the distributions
end caps, and is supported near both pipe ends ?one is
closed to the defective region; the other is at the un-
perturbed region?, these complex boundaries produce
complicated and different strain distributions near
the boundaries at both sides of the pipe, i.e., strain
decrease at the beginning of the big defective region
A ??23 cm? and increase at the end of the unper-
compression that occurs at the beginning of defect A
arises from the combination of the support point, the
end cap, and 60% wall-thinning defect A.
creased tension that appears in the unperturbed re-
gion after 140 cm is related to the combination of the
support point and the end cap that causes local stress
The strain–pressure slopes of the defective region
exceed those within the unperturbed region, as
shown in Fig. 4.The strain–pressure slope obtains
a maximum of 0.48 ?ε?psi in the middle, which de-
creases toward the edges of the defect.
small local maximum arises near the edge between
the defective and unperturbed regions located at 84
cm that is due to a noncontinuous boundary condition
that results in a local stress concentration at the
edge.In addition, reduced strain measurements
near the defect boundary generally occur as a result
of the overlapping 13-cm pulse, which acquires the
average strain behavior from both regions and there-
fore leads to a strain–pressure ripple from 70 to 100
cm, as shown in Fig. 4.The strain–pressure slope
remains constant at 0.16 ?ε?psi within the experi-
mental error near the middle of the unperturbed re-
gion, as expected, because the influence of boundaries
is small there.
Table 1.Parameters of Cutouts ?Defects?
of the pipe through defect A and unperturbed region B.
strain difference between the maximum from the defective region
and the minimum from the unperturbed region is 32 ?ε at a
200-psi internal pressure.
Axial strain distribution along the longitudinal direction
1 March 2004 ? Vol. 43, No. 7 ? APPLIED OPTICS1585
When a uniform pipe is operated within its elastic
regime, its strain–pressure relationship remains lin-
ear. However, erosion or corrosion of the inner pipe
wall causes irregular defective regions with reduced
thickness. This increases their strain–pressure
slopes relative to that of neighboring unperturbed
regions, depending on defect size and thickness.
Thus a bigger strain–pressure slope is observed for
the middle of defect A ?60% of inner-wall thickness?
than for the middle of unperturbed region B and de-
fect C ?50% of inner-wall thickness?, as shown in Fig.
5.The fact that the strain–pressure slope of region
B is less than those of regions A and C demonstrates
that pre-embedded defects that consist of 50% and
60% of the reduced inner-wall thickness can be dis-
criminated from the corresponding axial strain mea-
with the fiber configuration shown in Fig. 2?c?.
widths of defects A, C, and D are 5.3, 1.3, and 1.3 cm,
respectively. As the spatial resolution of our mea-
surement is 13 cm, there is a signal overlap between
the unperturbed and the defective regions even in the
middle of the defects along the circumferential direc-
tion. However, we can still differentiate the hoop
strain measurements of the locations close to and far
from the middle of the defects, even for small defects
C and D.
Typical strain distributions about one pipe circum-
ference that encompasses defective region A are dis-
played in Fig. 6. Two maxima, which correspond to
one complete loop, have been observed in the hoop
strain and strain–pressure slope distribution of the
defective region under 300- and 500-psi internal pres-
sures. The differences between the maximum,
which corresponds to the defective region, and the
minimum, which corresponds to the unperturbed re-
gion, are ?20 and ?40 ?ε under internal pressures of
300 and 500 psi, respectively, which are outside the
experimental error range ??15 ?ε of the standard
deviation? and therefore are definitely caused by wall
thinning. The strain maxima and highest strain–
pressure slopes are observed approximately once ev-
We chose two points, one in the defective region
?6-o’clock position? and the other in the unperturbed
region ?12-o’clock position?; Fig. 7 shows a linear
strain–pressure relationship for these two points
with slopes of 0.27 and 0.21 ?ε?psi, respectively.
the minimum and maximum slopes of 0.21 and 0.27
?ε?psi are spaced approximately 180° apart ?as
shown in Fig. 6?, the defective and unperturbed re-
gions are clearly differentiated.
To compare defective regions that have the same
arc length but different penetration depths we mea-
sured their strain–pressure slopes directly, to see the
effects of wall penetration.
strain–pressure slopes of defective regions C and D
Measurement of Hoop Strain
As shown in Fig. 8, the
the pipe through defect A and unperturbed region B.
pressure slope with 0.48 ?ε?psi is highest in the defective region,
decreases at the edges of the defect, and remains constant at 0.16
?ε?psi near the middle of the unperturbed region.
Strain–pressure slope along the longitudinal direction of
bigger strain–pressure slope is shown for the middle of defect A
?60%-depleted wall? compared with the middle of unperturbed re-
gion B and defect C ?50%-depleted wall?.
Axial strain–pressure slopes of defects A, C, and D.A
encompassing defective region A.
one complete loop, can be observed.
Hoop strain distributions about one pipe circumference
Two maxima, corresponding to
1586 APPLIED OPTICS ? Vol. 43, No. 7 ? 1 March 2004
are 0.18 and 0.21 ?ε?psi, which correspond to 50%-
and 60%-depleted walls, respectively.
a larger strain–pressure slope indicates a thinner
pipe wall. The unperturbed wall in both fiber loops
experiences a constant strain–pressure slope of 0.15
The hoop strain results can also be used to dis-
tinguish among defective regions at different loca-
tions along the pipe.One can achieve this by
determining the average of all strain–pressure
slopes about a single fiber loop and comparing it
with other averages along the pipe.
all strain–pressure slopes obtained about one fiber
loop is a good indication of the arc length of the
defective region. Hence the average strain slopes
of 0.24 and 0.18 ?ε?psi for fiber loops wrapping
defective regions A and D indicate the dominance of
a larger defective region A ?5.3 cm wide? over the
smaller region D ?1.3 cm wide? for equal depth ?60%
The average of
of inner-wall thickness? on the net strain distribu-
tion about one fiber loop.
Certainly, the location near the middle of the de-
fects experiences the maximum slope, and the mini-
mum slope happens at locations farthest from the
middle of the defects, i.e., the unperturbed region
?12-o’clock position? because the deformations of de-
fective regions are larger than those of unperturbed
regions.The strain–pressure slope increases with
defect width and depth, which indicates that pre-
embedded large or small defects that constitute 50–
60% of the inner wall thickness can be discriminated
by hoop strain measurements.
and D are very small ?1.3 cm wide?, though they have
different wall thicknesses they do not affect the
Thus the corresponding unperturbed regions ?12-
o’clock positions? maintain a 0.15-?ε?psi strain–
pressure slope. The higher strain–pressure slope
from the unperturbed ?12-o’clock? region opposite the
large defect A ?0.21 ?ε?psi? is likely due to the bigger
deformation of the large defect.
Because defects C
We have demonstrated the capability of a distributed
Brillouin scattering sensor system to identify several
inner-wall cutouts in an end-capped steel pipe by
measuring the axial and hoop strain distributions
along the outer surface of the pipe.
strain measurement, maximum
strains that correspond to the middle of defective and
unperturbed regions have been measured; thus the
capability of our system to differentiate between
these two regions has been demonstrated.
mum and minimum strains, which correspond to de-
fective and unperturbed regions, respectively, have
also been observed in the hoop strain distribution of
one complete loop.This demonstrates the ability of
our system to differentiate between defective and un-
perturbed regions about the circumference of the
pipe.By measuring hoop strain–pressure slopes at
different points along the pipe we have also been able
to discriminate among defective regions of different
embedded defects that compose 50–60% of the inner-
wall thickness can be discriminated by both axial and
hoop measurements.Our results show that fiber
optic sensor technology based on distributed Bril-
louin scattering offers great potential as a figurative
nervous system for infrastructure elements that al-
low high-performance, cost-effective health and dam-
age assessment systems to be achieved.
For the axial
This research was supported in part by Intelligent
Sensing of Innovative Structures, Canada.
thors thank S. Papavinasam and A. Doiron of Can-
met Materials Technology Laboratory in Ottawa,
Ontario, Canada for providing the steel pipe and as-
nology Laboratory for useful discussions.
compassing defective region A obtained by hoop strain measure-
ments. The minimum and maximum slopes of 0.21 and 0.27 ?ε?
psi, spaced approximately180°
unperturbed and defective regions, respectively.
Strain–pressure slopes about one pipe circumference en-
apart, correspondto the
ments of defects C and D.
regions C and D are 0.18 and 0.21 ?ε?psi, corresponding to 50- and
60%-depleted walls, respectively.
pressure slope indicates a thinner pipe wall.
Strain–pressure slopes obtained by hoop strain measure-
The strain–pressure slopes of defective
As expected, a larger strain–
1 March 2004 ? Vol. 43, No. 7 ? APPLIED OPTICS1587
References Download full-text
1. E. Tapanes, “Fibre optic sensing solutions for real-time pipe-
line integrity monitoring,” presented at the Australian Pipe-
line Industry Association National Convention, 27–30 October
2. National Transportation Safety Board, “Pipeline accident
brief” ?National Transportation Safety Board, Washington,
D.C., 2001?; http:??www.ntsb.gov?publictn?2001?PAB0103.
3. Energy and Utilities Board, “Pipeline performance in Alberta
1980–1997” ?Energy and Utilities Board, Calgary, Alberta,
Canada, 1998?; http:??www.eub.gov.ab.ca?bbs?documents?re-
4. X. Bao, M. DeMerchant, A. Brown, and T. Bremmer, “Tensile
and compressive strain measurement in the lab and field with
the distributed Brillouin scattering sensor,” J. Lightwave
Technol. 19, 1698–1704 ?2001?.
5. X. Bao, D. J. Webb, and D. A. Jackson, “32-km distributed
temperature sensor based on Brillouin loss in an optical fiber,”
Opt. Lett. 18, 1561–1563 ?1993?.
6. T. Horiguchi, T. Kurashima, and M. Tateda, “Tensile strain
dependence of Brillouin frequency shift in silica optical fibers,”
IEEE Photon. Technol. Lett. 1, 107–108 ?1989?.
7. T. Kurashima, T. Horiguchi, and M. Tateda, “Thermal effects
on Brillouin frequency shift in jacketed optical silica fibers,”
Appl. Opt. 29, 2219–2222 ?1990?.
8. X. Zeng, X. Bao, C. Y. Chhoa, T. W. Bremner, A. W. Brown,
M. D. DeMerchant, G. Ferrier, A. L. Kalamkarov, and A. V.
Georgiades, “Strain measurement in a concrete beam by use of
the Brillouin-scattering-based distributed fiber sensor with
single-mode fibers embedded in glass fiber reinforced polymer
rods and bonded to steel reinforcing bars,” Appl. Opt. 41, 5105–
9. A. W. Brown, M. D. DeMerchant, X. Bao, and T. W. Bremner,
“Spatial resolution enhancement of a Brillouin distributed sen-
sor using a novel signal processing method,” J. Lightwave
Technol. 17, 1179–1183 ?1999?.
10. D. Heckman, “Finite element analysis of pressure vessels”
?Monterey Bay Aquarium Research Institute, Moss Landing,
1588 APPLIED OPTICS ? Vol. 43, No. 7 ? 1 March 2004