http://shm.sagepub.com/Structural Health Monitoring

http://shm.sagepub.com/content/11/6/684The online version of this article can be found at:

 DOI: 10.1177/1475921712455682

2012 11: 684 originally published online 15 August 2012Structural Health MonitoringHyeonseok Lee, Jinyeol Yang and Hoon Sohn

guided laser ultrasonics−Baseline-free pipeline monitoring using optical fiber  

Published by:

http://www.sagepublications.com

can be found at:Structural Health MonitoringAdditional services and information for    

  http://shm.sagepub.com/cgi/alertsEmail Alerts:

 

http://shm.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://shm.sagepub.com/content/11/6/684.refs.htmlCitations:  

What is This? 

- Aug 15, 2012OnlineFirst Version of Record  

- Nov 5, 2012Version of Record >>

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

Article

Structural Health Monitoring

11(6) 684–695

� The Author(s) 2012

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1475921712455682

shm.sagepub.com

Baseline-free pipeline monitoring usingoptical fiber–guided laser ultrasonics

Hyeonseok Lee, Jinyeol Yang and Hoon Sohn

AbstractThis study develops an optical fiber–guided laser ultrasonic system and baseline-free damage detection methodthat enables structural health monitoring of pipelines employed under a high-temperature environment of nuclear powerplants. First, this study designs embeddable optical fibers and fixing devices, so that the laser beams used for ultrasonicexcitation and sensing could be transmitted between laser sources and target pipe surfaces through the optical fibers.All devices are specially designed for a high-temperature environment up to 300�C. Then, a damage detection method isdeveloped to identify typical pipe defects without any previous baseline data. This technique is based on the premise thatmultiple identical wave paths exist in an intact axisymmetric pipe, but the similarity among the ultrasonic signals obtainedfrom these identical wave paths breaks down when there is a nonaxisymmetric defect. The feasibility of the proposedtechnique has been experimentally verified using a stainless steel pipe specimen under a high-temperature environment.

KeywordsPipeline, structural health monitoring, nuclear power plants, laser ultrasonics, optical fiber, high temperature, baseline-freedamage detection, outlier analysis

Introduction

It has been reported that structural integrity of nuclearpower plants (NPPs) continues to deteriorate. About78% of all NPPs worldwide have been in operationover 20 years and 30% over 30 years.1 Furthermore,many countries are facing growing public opposition tothe construction of new NPPs and under pressure toshut down existing NPP facilities. Therefore, nuclearregulatory authorities in many countries have recentlytightened their safety measures and implemented strictinspection criteria for operational NPPs.

Conventional NPP pipeline inspections have beenbased on periodic nondestructive testing (NDT).2,3

Despite many merits of NDT techniques, they have theirown limitations. First, NDT normally requires a periodicoverhaul of the NPP facilities, reducing its power pro-duction efficiency and causing the secondary damage.Also, NDT inspections of NPPs can be performed onlyby certified personnel, so the inspection process inevita-bly becomes labor-intensive and expensive. More impor-tantly, there are several critical points, which cannot beeasily accessed by conventional NDT techniques.4

In order to complement existing NDT techniques,the industry is developing structural health monitoring(SHM) systems for NPP management. Compared to

NDT, SHM techniques can provide the followingpotential benefits: (a) automated and continuous moni-toring of NPP facilities, (b) reduction in overhaul fre-quency, cost, and labor necessary for NPP inspections,and (c) monitoring of hidden critical spots, which havebeen difficult to inspect using conventional NDTtechniques.5,6

Current monitoring efforts for pipe structures can becategorized into two: global and local monitoring tech-niques. Global monitoring techniques identify struc-tural damage by measuring dynamic characteristics of astructural system such as accelerations7–9 and strain/stress.10–12 Global monitoring is performed based onthe premise that if a structural defect, such as wall thin-ning or cracking, were large enough to affect thedynamic characteristics of the structure, these would be

Department of Civil and Environmental Engineering, Korea Advanced

Institute of Science and Technology, Daejeon, Korea

Corresponding author:

Hoon Sohn, Department of Civil and Environmental Engineering,

Korea Advanced Institute of Science and Technology, 291 Daehak-ro,

Yuseong-gu, Daejeon 305-701, Korea.

Email: hoonsohn@kaist.ac.kr

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

reflected on the measurements. On the other hand, localmonitoring aims to identify, localize, and quantifystructural defects in an earlier stage than global tech-niques. For example, ultrasonic waves generated andmeasured by magnetostrictive sensors (MSSs)13–15 orelectromagnetic acoustic transducers (EMATs)16–18 areused to detect even a small defect based on the interactof the ultrasonic waves with the defect.

However, several challenges have hampered theapplication of SHM to pipe structure used in NPPs.First, it requires permanent installation of distributedsensors that can survive harsh operational conditionsof NPPs such as temperature and radiation.19 Second,SHM techniques are normally less accurate than NDTtechniques in quantifying the damage location andlevel, and they are vulnerable to false alarms inducedby the operational variations.

This study develops an optical fiber–guided laserultrasonic system and a damage detection method spe-cifically designed for NPP pipeline structures.Ultrasonic waves are generated using a Nd:YAG laserand sensed by another laser interferometer. The laserbeams from the laser sources to the target pipe surfacesare guided and transmitted using embeddable opticalfibers. Then, an autonomous damage detection methodtailored to the laser ultrasonic system is developedbased on the axisymmetric nature of pipe structures.The proposed techniques have the following advan-tages over the existing techniques: (a) the proposedoptical fiber–guided laser ultrasonic system can beapplied to NPP pipe network operating under high-temperature environment up to 300�C; (b) the fiber-guided system can be readily embedded to monitorpipelines covered with insulators; and (c) because theproposed damage detection algorithm does not requireany previous baseline data, environmental and opera-tional variations such as temperature change and vibra-tion due to water flow inside pipes barely affect thedamage detection performance.

Optical fiber–guided laser ultrasonicsystem

Figure 1 shows a schematic diagram of the opticalfiber–guided laser ultrasonic system. A pulse laser istransmitted through an optical fiber to generate ultra-sonic waves on a pipe surface, and an interferometerconnected to a separate optical fiber is used to measurethe corresponding response at another point on thepipe surface. Then, there are additional componentssuch as focusing modules for laser beam focusing andstainless steel strips used for fixing the focusing mod-ules to the pipe surface. More details of the measure-ment system and its components are described below.

Optical fiber–guided laser ultrasonic generation

Figure 2 shows the detailed configuration of the opticalfiber, the laser source, and the connecting componentsrequired for optical fiber–guided laser ultrasonic gener-ation. When a solid surface is illuminated by ahigh-power pulse laser, a localized heating and the cor-responding elastic expansion is produced at an infinite-simal area of the surface, acting as an ultrasonic wavesource.20 Here, the laser duration, power level, andbeam size need to be carefully tuned and maintained togenerate elastic waves without ablating the surface.Ablation typically occurs for metals with a power den-sity above 107 W/cm2,21 but this value varies dependingon their material properties as well as surface condi-tions. In this study, a 1064-nm Nd:YAG pulse laser(Brilliant Ultra; Quantel) is used. The excitation energyper unit pulse, the pulse duration, and the repetitionrate of the laser beam are set to 10 mJ, 8 ns, and 20 Hz,respectively, to avoid ablation.

A metal-coated multimode fiber with 1000 mm corediameter (HPSUV 1000CB; Oxford Electronics) is usedto carry the laser beam from the laser source to the tar-get point on the structure. Here, a multimode fiber isused for high-power laser transmission. This multi-mode fiber is made of step-index silica where the coreand the cladding are made of pure silica and dopedsilica, respectively. According to the manufacturer spe-cification, the fiber is also radiation resistant, and theCuBall metal coating allows the operation of the opti-cal fiber up to 500�C.

The high-power laser beam needs to be properlyfocused to the target surface using focusing lenses foran effective ultrasonic excitation. As shown in Figure 2,the Nd:YAG source laser beam is focused using aLak10 focusing lens with a focal length of 15 mm beforeentering the SubMiniature version A (SMA) connector.Then, the laser beam exiting from the SMA at the otherend of the fiber travels through a metal tube, and thelight is focused again using the identical Lak10 lens. Atube made of stainless steel and aluminum is insertedbetween the SMA and the focusing lens to minimize theheat transfer from the specimen to the SMA connector.As a final step, a stainless steel strip is used to fix thefocusing module to the specimen.

Optical fiber–guided laser ultrasonic sensing

Figure 3 shows the detailed configuration of the opticalfiber–guided laser ultrasonic sensing component. Thisstudy utilizes a phase-modulation fiber-coupled laserDoppler interferometer (OFV-551; Polytec) to measurethe out-of-plane ultrasonic response of the pipe surfacesince it is more sensitive and stable than conventionalintensity-modulation interferometers.22 The fiber

Lee et al. 685

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

interferometer normally uses a continuous wave (CW)laser as a power source, and its power level is muchlower than an excitation laser. The interferometer inthis study uses a He–Ne continual laser source with 633nm wavelength, 15 W maximum power, and 16 mm

beam size. When a continuous laser beam is pointed ata vibrating object and scattered back from it, the velo-city of the vibrating object produces a phase shift ofthe laser beam due to the Doppler effect. The frequencymodulation of the Doppler signal recovers the velocity

Figure 2. Configuration of the optical fiber–guided laser ultrasonic generation (all dimensions in millimeter).SMA: SubMiniature version A.

Figure 1. Schematic overview of the proposed optical fiber–guided laser ultrasonic system: (a) slide view and (b) cross section.

686 Structural Health Monitoring 11(6)

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

information, and the phase modulation reconstructsthe displacement.23,24

Since the laser beam emitted from the fiber interfe-rometer is less powerful than the pulsed excitation laserbeam, a panda-type single mode fiber is used for thetransmission of the sensing laser beam. The sensingfiber is made of polarization maintaining (PM) fiber,which can reduce the power loss induced by the polari-zation direction. This polarization-induced power lossis more critical for the sensing laser beam than the exci-tation beam, since the power loss directly governs thesignal-to-noise ratio of response signals and it cannotsimply be enhanced by raising the laser power level.Since the response signals contain background noiseand multimodal complexity from the pulse input, themeasured responses are amplified with a gain of 10and band-pass filtered in a range of 10–300 kHz usinga low-noise amplifier (SR560; Stanford ResearchSystems).

In order to enhance the transmission of the incidentand reflected laser beams, a sensor head is inserted atthe end of the optical fiber. A Lak10 focusing lens witha focal length of 15 mm is inserted close to the targetsurface similar to the previous case in section ‘‘Optical

fiber–guided laser ultrasonic generation.’’ One differ-ence from the previous wave generation component isthat an additional KG1 filter is placed close to the sen-sor head to prevent the infrared heat radiation from theheated pipe to the sensor head. Furthermore, the fix-ture strip and metal tube identical to the previous wavegeneration component are used.

Baseline-free damage detection technique

In this section, a damage detection algorithm specifi-cally tailored to the proposed laser ultrasonic system isdeveloped for pipe monitoring. There is a rich volumeof ultrasonic techniques developed for pipeline moni-toring. Some researchers have excited axisymmetricwave modes, L(0,1) or T(0,1), and then extracted non-axisymmetric and mode-converted flexural modes,F(m,n), induced by structural defects.25–28 Otherresearchers have developed beam-forming techniquesto steer ultrasonic waves for selective inspection of adesired point in a pipe.29,30 Based on a time reversalconcept, images of detects in pipes are constructed byexciting L(0,1) and sensing F(m,n) modes.31

Figure 3. Configuration of the optical fiber–guided laser ultrasonic sensing module (all dimensions in millimeter).

Lee et al. 687

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

There are challenges in using these conventionaldamage detection techniques in conjunction with theproposed laser ultrasonic system. The proposed lasersystem applies a broadband pulse input only to limiteddiscrete excitation points and simultaneously generatesmultiple nonaxisymmetric modes. Therefore, it cannotselectively excite axisymmetric wave modes to extractmode-converted flexural modes induced by structuraldefects. Also, the waveform that the proposed systemcan generate is limited only to a pulse input, and time-delayed multichannel excitation is difficult using theproposed system. That is, the aforementioned beam-forming and time reversal techniques cannot beused with the proposed laser ultrasonic system.Therefore, a new damage detection algorithm devel-oped and tailored specifically for the proposed opticalfiber–guided laser ultrasonic system is presentedbelow.

Figure 4 shows the configuration of ultrasonic wavegeneration and sensing used for the proposed damagediagnosis. For ultrasonic wave generation, multiplelaser excitation points (group A: A1, A2, ., An),equally spaced in a circumferential direction, are placedat one end of the pipe. Similarly, multiple laser sensingpoints (group B: B1, B2, ., Bn) are positioned at theother end of the pipe. Here, n is the number of excita-tion points (or the sensing points), and this number canbe increased to improve the spatial resolution of dam-age detection. In a round-robin manner, ultrasonicwaves are generated and measured from different pairsof excitation and sensing points.

Figure 5 shows several groups of identical signalsbased on the axisymmetric nature of the intact pipe.The pipe in Figure 5 is shown unwrapped for an illus-tration purpose. Due to the axisymmetric nature of thepipe, ultrasonic wave signals with the same pitch–catchdistance should be, in theory, identical for the intact

condition. The followings are examples of signal groupswith the equality.

(a) Signal A1B1 = Signal A2B2 = � � � = Signal An21

Bn21 = Signal AnBn;(b) Signal A1B2 = Signal A2B1 = � � �= Signal An21Bn

= Signal AnBn21;(c) Signal A1B3 = Signal A3B1 = � � �= Signal An22Bn

= Signal AnBn22

The basic premise of the proposed damage detectionalgorithm here is that this equality breaks down when anonaxisymmetric defect is introduced. Note that thereare other pairs of identical signals with the same pitch–catch distance depending on the size of n.

Based on this observation, the following damageindex (DI) can be computed in a round-robin fashion

DI i, jð Þ= 1� max corr fi, fj� �

ð1Þ

max corr fi, fj

� �= max

~t

1

N � 1

Xfi, fj

fi t +~tð Þ � �f 1

� �fj tð Þ � fj

� �sfi sfj

8<:

9=;ð2Þ

In equations (1) and (2), fi and fj are the timeresponse signals obtained from two wave paths withthe identical travel distance; fi is the mean of fi; sfi isthe standard deviation of fi; N is the total number ofdata points in fi; and ~t is the time shift. Therefore, allthe DI values will be close to zero for an intact case,but DI values for certain pairs of fi and fj will increasewhen there is a nonaxisymmetric defect in the pipe.

After computing all DI values, a sequential outlieranalysis with an adaptive threshold value32 is per-formed to identify the DI values that are obtained fromthe wave paths passing through a structural damage

Figure 4. Ultrasonic wave generation and sensing configuration for the proposed baseline-free damage detection algorithm.

688 Structural Health Monitoring 11(6)

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

and much larger than the ones obtained from the intactpaths. The procedure of the proposed sequential outlieranalysis shown in Figure 6 can be described as follows:

1. Arrange all DI values in an ascending order (1 < m< N).

2. Fit a parametric distribution function up to them 2 1 smallest DI values and compute a thresholdcorresponding to a user-specified one-sided confi-dence level. Initial m 2 1 is typically selected to be

the half of all available combinations defined inequation (2).

3. If the mth smallest DI value is larger than thethreshold value, the wave paths corresponding toall the mth smallest DI and the larger DI valuesare determined to pass through a defect location.If not, repeat steps 2 and 3 for the next smallestDI value m + 1, until it reaches N. If all the DIvalues are below the threshold value, the pipe isintact.

Figure 5. Group of equal signals based on the axisymmetric nature of the intact pipe: (a) equality among Signals AiBi, (b) equalityamong Signals AiBi + 1 and Signals AiBi21, and (c) equality among Signals AiBi + 2 and Signals AiBi22. Note that the pipe is shownunwrapped for a demonstration purpose here.

Lee et al. 689

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

Because the DI values are bounded between 0 and 1,the following b distribution is used to describe the sta-tistical distribution of the DI values33

f x; a, bð Þ=1

B a, bð Þ xa�1 1� xð Þb�1 ð3Þ

where B is the b function; a and b are the shape para-meters. A goodness-of-fit test using the Kolmogorov–Smirnov (K-S) test34 confirms that the b distributionproperly describes the population of the DI valuesobtained from the intact conditions. The threshold valueis set to the one-sided 99.7% (3s) confidence level.

Experimental results

Experimental description

Figure 7 shows the overall configuration of the experi-mental setup. The test is designed to simulate structuraland operational conditions of a typical secondary cool-ant system in NPPs. The target specimen (Figure 8) is astainless steel seamless-typed pipe (KS D 3562) com-monly used for high-pressure piping systems in NPPs.The outer diameter of the pipe is 114.3 mm and the wallthickness is 6 mm. The axial length of the target speci-men is 1000 mm, and the distance between ultrasonicwave generation and sensing points is set to 300 mm.

Figure 6. Process of the sequential outlier analysis: (1) plot DI values in an ascending order; (2) fit an extreme value distribution toDI values from the 1st to the m 2 1th smallest value; (3) set a threshold based on a user-specified confidence level; (4) check whetherthe mth smallest DI value is larger than the threshold value; (5) if not, repeat steps (2) and (3) using the m smallest DI values; (6)again, repeat step (4) for the next m + 1th smallest DI value. If all DI values are below the threshold value, the pipe is intact.DI: damage index.

690 Structural Health Monitoring 11(6)

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

As for structural damage, a longitudinal notch isengraved 575 mm away from the left end of the pipe tosimulate a longitudinal crack. The notch is 20 mm long,2 mm wide, and 2 mm deep.

In this experiment, there are four excitation points350 mm away from the left end and four sensing points350 mm away from the right end. Because there arefour equally spaced excitation points in the circumfer-ential direction (as well as sensing points), possiblewave path groups become as follows: (a) 6 pairs forSignals AiBi (4C2); (b) 28 pairs for Signals AiBi+ 1 andAiBi21 (8C2); and (c) 6 pairs for Signals AiBi+ 2 andAiBi22 (4C2). Therefore, there are a total of 40 wavepath pairs and 40 DI values (N = 40).

Baseline-free damage diagnosis

Figure 9 shows how several representative signalschange from the intact (dotted line) condition to thedamage (solid line) one. Each response signal is nor-malized by subtracting the mean and dividing by thestandard deviation for better comparison. Then, themaximum correlation (max corr) between the intactand damage cases is computed using equation (2) for

each signal as follows: 0.8849 for Signal A2B2; 0.6527for Signal A2B3; 0.2668 for Signal A3B2; and 0.6080 forSignal A4B2. Because Signal A3B2 goes directly throughthe introduced longitudinal notch, the maximum corre-lation for this path changed most drastically. However,it should be noted that the actual damage diagnosisproposed in this study is performed without this typeof pattern comparison with the previously obtainedintact signals.

Figure 10 shows the result of the sequential outlieranalysis for the damage case. Here, the DI values in they-axis are sorted in an acceding order and the indicesfor the corresponding pairs of wave propagation sig-nals are shown in the x-axis. For the outlier analysis ofthe mth DI value, a b distribution is fitted to the m 2 1smallest DI values and a threshold value (the dottedline) corresponding to a one-sided 99.7% (3s) confi-dence level of the b distribution is instantaneously com-puted. This process is repeated from m = 20 to 40.Note that m started from 20 (= N/2) with the assump-tion that the target defect is localized and at least thehalf of the DI values are not affected by the defect.

Among all DI values obtained from the damagecase, seven DI values associated with Signal A3B2

became noticeably larger than the other DI values.This observation indicates that Signal A3B2 is drasti-cally different from the other response signals due tothe existence of a structural defect along this path.Even though some other response signals are more orless influenced by the notch, Signal A3B2 is far moreaffected by the defect than the other signals becausethis signal directly passes through the defect.

Figure 11 shows the result of the sequential outlieranalysis performed for the intact case to check whetherthe proposed damage detection technique is vulnerableto a false alarm. As shown in Figure 11, no false alarmis triggered in this particular example.

Effect of insulator

The majority of pipes in NPPs are covered with insu-lators to prevent too much heat flow to the air. Anactual insulator commonly used in NPPs is employedhere, and the particular insulator used in this study iscomposed of two layers: a 50-mm inner layer made ofglass wool and a 20-mm outer layer made of glass felt.Also, it is designed to bring down temperature from300�C at the pipe surface to 40�C at the outer insula-tor surface. The effect of the insulator on responsesignals is examined in Figure 12. The dotted line indi-cates the response signal with the insulator and thesolid line without the insulator. The comparison ofthese two signals reveals a marginal effect of theinsulator.

Figure 8. Dimensions of the target pipe specimen and thenotch defect.

Figure 7. Configuration of the overall experimental setup.

Lee et al. 691

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

Figure 9. Comparison of the response signals obtained from the same pitch-and-catch distance and direction: (a) Signal A2B2, (b)Signal A2B3, (c) Signal A3B2, and (d) Signal A4B2. Dotted and solid lines are obtained from the intact and damage cases, respectively.Here, the ‘‘max corr’’ is computed between the intact and damage case signals.

Figure 10. Sequential outlier analysis for the damage case.DI: damage index.

692 Structural Health Monitoring 11(6)

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

Effect of high temperature

Next, the effect of temperature on the proposed opticalfiber–guided laser ultrasonic system is investigated. Fortypical NPPs, the nominal operational temperature of

the secondary coolant system is approximately 300�Cnear the reactor, and the temperature is maintainedaround 100�C–200�C in the secondary coolant system.To control the temperature of the test specimen, a tem-perature controller specially designed by SKICorporation is used (Figure 13). The controller has abuilt-in pipe and a thermal conductor inside the pipe toheat up the pipe. The axial length of the pipe is 600mm, the outer diameter is 114.3 mm, and the thicknessis 3 mm. The wave propagation distance between thewave generation and sensing points is set to 300 mm,which is same as the previous experiment. When thepipe is heated and the pipe surface temperature is stabi-lized at 300�C, the temperatures of the first and secondlayers of the insulator are 48.7�C and 41.2�C, respec-tively, demonstrating the effectiveness of the insulationlayer.

Figure 14 shows the response signals obtained fromthe pipe specimen at varying temperatures. The tem-perature was increased from the room temperature to300�C, and the response signals are measured from100�C to 300�C at an interval of 50�C. As the tempera-ture increases, the peak amplitude of the signaldecreases and the response is delayed in the timedomain. Note that the amplitude and the arrival timeof the response signal are 3.395 mV and 0.1214 ms,respectively, at 100�C. However, these values change to2.981 mV and 0.123 ms at 200�C, and 2.249 mV and0.1252 ms at 300�C, respectively. The attenuation andtime delay are mainly caused by the reduction ofYoung’s modulus due to the increase in temperature.35

Because significant response signal changes can beobserved due to temperature even without any struc-tural defects, a proper measure needs to be taken forreliable damage diagnosis. To compensate the signalchanges induced by temperature variations, a baselinesubtraction method has been proposed for damagedetection.36 This method identifies damage by subtract-ing the test signal from the baseline signal after

Figure 11. Sequential outlier analysis for the intact case.DI: damage index.

Figure 13. Experimental setup for high-temperatureexperiments.

Figure 12. Comparison of the response signals obtained fromthe pipe specimen with (dotted line) and without (solid line) theinsulator.

Lee et al. 693

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

correcting effect of temperature on group velocities ofspecific wave modes. However, this technique cannotbe applied to the proposed optical fiber–guided laserultrasonic system because a broadband pulse input usedin the proposed system excites multiple wave modessimultaneously and temperature compensation of allthese modes is practically impossible. On the otherhand, the proposed damage detection algorithm can beeffectively used even under varying temperature condi-tions since the axisymmetric nature, which should holdfor the intact case, is irrelevant to temperature changesand wave frequency ranges. The axisymmetric nature isbroken down only when a nonaxisymmetric defect isintroduced. However, large nonaxisymmetric tempera-ture gradient could also cause false alarms.

Conclusions

This study develops an optical fiber–guided laser ultra-sonic system and a baseline-free damage detection tech-nique specially designed for SHM of insulated pipesunder high-temperature environment in NPPs. An opti-cal fiber with a specially designed focusing lens and afixture are used to deliver a Nd:YAG pulse laser to anembedded target excitation point for ultrasonic wavegeneration. Another optical fiber is connected betweena fiber vibrometer and a sensing measurement point forultrasonic wave sensing with a similar focusing lens anda fixture. The proposed system allows generating andmeasuring ultrasonic waves at any desired locationseven when they have limited accessibility due to theinsulators or high-temperature environment. The per-formance of the proposed system has been successfullydemonstrated at varying temperatures up to 300�C. Abaseline-free damage detection algorithm is developedalong with the laser system based on the axisymmetricnature of multiple wave propagation paths in pipes.The axisymmetric nature is valid for the intact pipe, but

it breaks down when a nonaxisymmetric defect is intro-duced. The robustness of the proposed baseline-freedamage detection technique has been experimentallyverified both for the intact and damage cases of a pipespecimen commonly used in NPPs.

Several follow-up studies are required to improvethe performance of the proposed system. First, a radia-tion hardening test on the proposed measurement sys-tem is necessary for the system installation in realNPPs. In fact, we designed the proposed optical fiber–guided laser ultrasonic system by carefully choosingradiation-proof components, and a radiation hardeningtest to prove its robustness against radiation exposureis underway. Second, an optical switching deviceshould be incorporated to the proposed system forselective ultrasonic generation and sensing at multi-channels. Third, the proposed damage detection algo-rithm is applicable only to a simple straight pipe.Further research is warranted to explore damage detec-tion in more complex pipe geometries. Finally, the pro-posed approach is only intended for damageidentification. Therefore, once damage is detected,additional measures should be taken for detailed dam-age quantification.

Funding

This research was supported by Mid-career ResearcherProgram and National Nuclear R&D Program through theNational Research Foundation of Korea (NRF) funded bythe Ministry of Education, Science and Technology (No. 2011-0016470 and No. 2011-0018430) and by the Innovations inNuclear Power Technology (Development of Nuclear EnergyTechnology) of the Korea Institute of Energy TechnologyEvaluation and Planning (KETEP) grant funded by theKnowledge Economy (No. 2010T100101057) in Korea.

References

1. IAEA. Assessment and management of ageing of major

nuclear power plant components important to safety: PWR

vessel internals—2007 update (IAEA-TECDOC-1557).

Vienna: International Atomic Energy Agency (IAEA),

2007, pp. 10–32.2. Louis C. Nondestructive testing: radiography, ultrasonics,

liquid penetrant, magnetic particle, eddy current. Materials

Park, OH: ASM International, 1995, pp. 15–100.3. Jack B and Geoff S. Ultrasonic methods of non-destructive

testing. Cambridge: Chapman & Hall, 1996, pp. 101–123.4. Adams D. Health monitoring of structural materials and

components. Chichester: John Wiley & Sons Ltd, 2007,

pp. 78–83.5. Sohn H, Farrar C, Hemez F, et al. A review of structural

health monitoring literature: 1996-2001. Los Alamos,

NM: Los Alamos National Laboratory, 2003, pp. 19–32.6. Daniel J. Damage prognosis for aerospace, civil and

mechanical systems. Chichester: Wiley, 2005, pp. 123–289.

Figure 14. Comparison of the response signals obtained fromvarying pipe temperatures.

694 Structural Health Monitoring 11(6)

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from

7. Lee N, Bahn C, Lee S, et al. Development of an on-lineultrasonic system to monitor flow-accelerated corrosionof piping in nuclear power plants. Key Eng Mat 2004;270–273: 2232–2238.

8. Hashemian H. Aging management of instrumentation &control sensors in nuclear power plants. Nucl Eng Des

2010; 240: 3781–3790.9. Choi Y, Park J and Choi K. An impact localization tech-

nique for a nuclear power plant by using sensors of differ-ent types. ISA T 2011; 50: 111–118.

10. Auge L, Capra B, Lasne M, et al. Risk management andmaintenance optimization of nuclear reactor cooling pip-ing system. J Phys 2006; 136: 263–271.

11. Niewczas P and McDonald J. Advanced optical sensorsfor power and energy systems applications. IEEE Instru

Meas Mag 2007; 10: 18–28.12. Matsuba K, Ito C, Kawahara H, et al. Development of

fast reactor structural integrity monitoring technologyusing optical fiber sensors. J Power Energy Syst 2008;2(2): 545–556.

13. Luo W, Rose JL and Kwun H. Circumferential shearhorizontal wave axial-crack sizing in pipes. Res Nondes-

truct Eval 2004; 15: 149–171.14. Zhao X, Varma V, Mei G, et al. In-line nondestructive

inspection of mechanical dents on pipelines with guidedshear horizontal wave electromagnetic acoustic transdu-cers. J Press Vess-T ASME 2005; 127(3): 304–309.

15. Cheong Y, Jung H and Kim Y. Comparison of an arrayof EMATs technique and a magnetostrictive sensor tech-nique for a guided wave inspection of a pipe. Key Eng

Mat 2006; 321–323: 780–783.16. Hirao M and Ogi H. An SH-wave EMAT technique for

gas pipeline inspection. NDT&E Int 1999; 32: 127–132.17. Sun Z, Zhang L and Rose JL. Flexural torsional guided

wave mechanics and focusing in pipe. J Press Vess-T

ASME 2005; 127(4): 471–478.18. Tappert S, Allen D, Mann A, et al. Third-generation

EMAT tool enhanced for finding SCC and disbondedcoating in dry gas pipeline. Pipeline Gas J 2009; 236:26–34.

19. Upadhyaya BR and Lu B. Fault monitoring of nuclearpower plant sensors and field devices. Prog Nucl Energ

2003; 43(1–4): 337–342.20. Scruby CB and Drain LE. Laser ultrasonics: techniques

and applications. New York: Taylor & Francis, 1990, pp.76–85.

21. Pierce SG, Culshaw B and Shan Q. Laser generation ofultrasound using a modulated continuous wave laserdiode. Appl Phys Lett 1998; 72(9): 1030–1032.

22. Kundu T. Ultrasonic nondestructive evaluation: engineer-

ing and biological material characterization. London:CRC Press, 2004, pp. 121–138.

23. Staszewski WJ, Lee BC, Mallet L, et al. Structural healthmonitoring using scanning laser vibrometry: I. Lambwave sensing. Smart Mater Struct 2004; 13: 251–260.

24. Castellini P, Martarelli M and Toamasini EP. Laser Dop-pler vibrometry: development of advanced solutionsanswering to technology’s needs. Mech Syst Signal Pr

2006; 20(6): 1265–1285.25. Lowe MJS, Alleyne DN and Cawley P. The mode conver-

sion of a guided wave by a part-circumferential notch in apipe. J Appl Mech 1998; 65: 649–656.

26. Alleyne DN, Pavlakovic B, Lowe MJS, et al. Rapid, longrange inspection of chemical plant pipework using guidedwaves. Rev Prog Q 2001; 20: 180–187.

27. Hayashi T, Kwsashima K, Sun Z, et al. Guided wave

focusing mechanics in pipe. J Pressure Vessel Technol

2005; 127: 317–321.28. Davies J and Cawley P. The application of synthetic

focusing for imaging crack-like defects in pipelines usingguided waves. IEEE T Ultrason Ferr 2009; 56(4):759–771.

29. Barshinger J, Rose JL and Avioli MJ. Guided wave reso-nance tuning for pipe inspection. J Pressure Vessel Tech-

nol 2002; 124: 303–310.30. Li J and Rose JL. Natural beam focusing of non-

axisymmetric guided waves in large-diameter pipes.Ultrasonics 2006; 44: 35–45.

31. Deng F, Wu B and He C. A time-reversal defect-identify-ing method for guided wave inspection in pipes. J Pres-

sure Vessel Technol 2008; 130: 021503-1–021503-8.32. Sohn H, Park HW, Law KH, et al. Combination of a

time reversal process and a consecutive outlier analysisfor baseline-free damage diagnosis. J Intel Mat Syst Str

2007; 18(4): 335–346.33. Hayter A. Probability and statistics for engineers and

scientists. 3rd ed. Belmont, CA: Thomson Brooks/Cole,2007, pp. 184–212.

34. Ross SM. Introduction to probability and statistics for

engineers and scientists. 2nd ed. San Diego, CA: Aca-demic Press, 2004, pp. 504–508.

35. Raghavan A and Cesnik CES. Effects of elevated tem-perature on guided-wave structural health monitoring. JIntel Mat Syst Str 2008; 19: 1383–1398.

36. Lu Y and Michaels JE. A methodology for structuralhealth monitoring with diffuse ultrasonic waves in thepresence of temperature variations. Ultrasonics 2005; 43:717–731.

Lee et al. 695

at Korea Advanced Institute of Science and Technology (KAIST) on December 16, 2012shm.sagepub.comDownloaded from