quasi-distributed fiber-optic strain sensor: principle and experiment

Quasi-distributed fiber-optic strain sensor:principle and experiment

Yang Zhao and Farhad Ansari

Sensors capable of making distributed measurements allow for monitoring of the entire structure.Optical fiber sensors are especially attractive for this purpose, since they are geometrically versatile andcan be readily integrated within various types of structure and material. Development and character-istics of a quasi-distributed intrinsic fiber-optic strain sensor based on white-light interferometry aredescribed. The research presented here describes the development of a new optical fiber sensor systemfor measurement of structural strains based on double white-light interferometry. Individual segmentsof single-mode optical fibers forming a common-path interferometer are linked in series, and a scanningwhite-light interferometer provides for distributed sensing of strain signals from various locations in thestructure. The system is configured for automatic compensation of drift due to environmental effects,i.e., temperature and vibration. Strain gauges were employed for comparison and verification of strainsignals as measured by the new system. The experimental results demonstrate the linearity of thesystem and the capability for distributed sensing of strains. © 2001 Optical Society of America

OCIS codes: 060.2370, 120.3180.

te

i

1. Introduction

Fiber-optic sensors are emerging as sensors of choicein many applications. In sensing and measurementof strain, optical fiber sensors provide advantagesover the existing technologies. Because of their geo-metric adaptability and small size, optical fibers canbe easily embedded within concrete structures.Most importantly, they serve the dual purpose of be-ing the sensor as well as the pathway for the sensorsignal out of the structure. Localized fiber-optic sen-sors determine the strain over a specific segment ofthe optical fiber and are thus similar to conventionalstrain or temperature gauges. Distributed sensorsmake full use of optical fibers in which each elementof the optical fiber is used for both sensing and datatransmission. These sensors are most appropriatefor structural applications, because of their multi-point measurement capabilities.

Most of the fiber-optic sensor research activitieshave involved development of localized sensors.However, a survey of recent literature reveals a num-

The authors are with the Department of Civil and MaterialsEngineering, University of Illinois at Chicago, Chicago, Illinois60607-7023. F. Ansari’s e-mail address is [email protected].

Received 17 Noveber 2000; revised manuscript received 23March 2001.

0003-6935y01y193176-06$15.00y0© 2001 Optical Society of America

3176 APPLIED OPTICS y Vol. 40, No. 19 y 1 July 2001

ber of activities pertaining to the development of dis-tributed andyor multiplexed sensors.1 Brillouinscattering, optical switching, wavelength-divisionmultiplexing, and optical time-domain reflectometry~OTDR! have been used in the development of dis-ributed and multiplexed sensors. For instance, Gut al.2 developed a distributed sensor by connecting a

number of individual optical fibers of desired gaugelengths in series. The strain transduction mecha-nism was intensity based, and a high-resolutionOTDR was used for measurement of intensity lossesat the spliced joints. However, Michie andCulshaw3 employed the OTDR for the development ofa unique method for distributed detection of moistureingress in post tensioned grouted concrete beams. Aspecial hydrogel was used in conjunction with theoptical fiber sensors, which swelled and created lo-calized microbend losses within the optical fiber.

Other distributed and multiplexing sensing meth-ods involved techniques based on Brillouin scatter-ing, interferometry, and Bragg gratings. Lecoeucheet al.4 developed long-gauge-length distributed sen-sors, using Brillouin scattering, based on the Dopplershift in frequency. Kersey and Morey5 developed anumber of techniques for expanding the use of Bragggratings in structural measurements. One suchmethod involved wavelength-division multiplexing,through which a broadband source was employed forscanning a number of Bragg gratings in series andyorn parallel. The Bragg gratings in this application

sua

m

ddsb

Fifiom

tpmtswasobafwtu

strTs

tn

a

n

were designed with reflectivities at slightly differentwavelengths. Chen et al.6 employed an opticalswitch to provide multiplexing capabilities for awhite-light interferometric sensor. Liu et al.7 devel-oped a scanning interferometer based on a white-light Michelson interferometer. Rao and Jackson8

developed a system with multiplexing capability foras many as 32 extrinsic Fabry–Perot sensors.Brooks et al.9 reported on a multiplexed fiber-opticensor using the interferometer as both the sensingnit and the interrogating interferometer. Santosnd Jackson10 developed a multiplexed interferomet-

ric sensor in the time domain. Inaudi11 developed adouble-interferometer white-light quasi-distributedsensor for structural measurements. In the systemdeveloped by Inaudi,11 the sensing and reference sec-tions of the interferometer were separated by way ofa second coupler. Inaudi11performed separate ex-periments for measurement of deformations and tem-peratures in a three-sensor system and achievedmeasurement precision of better than 10 mm for the

easurements.The research presented here also describes the

evelopment of a double-interferometer quasi-istributed white-light sensor. In the currenttudy, however, in contrast to the system developedy Inaudi,11 the reference arm of the sensing mod-

ule is eliminated. Instead, a single optical fiberwas used for both sensing and referencing. Be-sides simplicity, this approach eliminated the errorassociated with the lead-in fiber phase changes thatwould inadvertently affect the signal. The config-uration of the double-arm interferometer module isdescribed in the following sections.

2. Distributed Sensing Methodology

Schematic representation of the quasi-distributedsensor consisting of m segments in series is shown in

ig. 1. The system consists of two parts: the sens-ng interferometer module and the receiving inter-erometer. The sensor is composed of a number ofndividual single-mode fibers coated on both ends andf desired gauge lengths. The individual fibers areechanically connected through ferrules, and a por-

Fig. 1. Schematic representation of the quasi-distributed sensorand interferometer assembly.

ion of the beam is reflected when the light waveasses through them. As noted above, the sensingodule performs both sensing and referencing func-

ions. Individual sensor segments, starting fromensor 1 immediately following the coupler, all theay to the final sensor in the series act as referencerms for the adjacent sensors. Since every sensorhares a common pathway through the coupler andther fiber segments, possible errors due to distur-ances caused by the lead fiber phase variations areutomatically compensated for. The receiving inter-erometer, as shown in Fig. 1, consists of a Michelsonhite-light interferometer with a scanning transla-

ion stage, signal processing, and the system controlnit.White light from a wideband LED is fed into the

ensor by way of a 2 3 1 coupler. The power isransmitted through the entire sensor system andeturned to the interferometer by way of a coupler.he returned optical field, E, from the individual sen-or segments is expressed in the following form:

E 5 (i51

m

Ei expH 2 jF2pvt 2 2k# SL0 1 (h51

h5i

nLhDGJ ,

(1)where Ei is the amplitude of the reflected beam; Ei ishe mean frequency of the signal; #k is the mean waveumber in vacuum; L0 is the initial optical path

length; Lj is the gauge length of each strain sensor; nis the effective index of refraction of the sensing fiber;m is the number of the sensors; i 5 1, 2, 3, . . . , m;nd h 5 1, 2, 3, . . . , i. The individual reflected sig-

nals contain the information pertaining to the mea-surand.

The set of reflected signals is then interrogated bythe white-light-receiving interferometer. The inter-ferometer consists of the reference arm and the sens-ing arm, which is composed of a gradient-index lensand a scanning mirror mounted on a motorized trans-lation stage. The reflected optical fields from thereference ~Er! and the sensing arms of the interferom-eter ~Em! can be expressed as

Er 5 (i51

m

Ei9 expH 2 jF2pv# t 2 2k#

3 SL0 1 (j51

j5i

nLj 1 LrDGJ , (2)

Em 5 (i51

n

Ei9 expH 2 jF2pv# t 2 2k#

3 SL0 1 (j51

j5i

nLj 1 LmDGJ , (3)

where Er9 and Em9 refer to the amplitude of the sig-als; Lr and Lm are optical path lengths for reference

and sensing arms in the interferometer; and Lm 5 Lr1 x, where x is the scanning displacement of thetranslation stage in the interferometer. We can ob-tain the output intensity Ic of the interference by

1 July 2001 y Vol. 40, No. 19 y APPLIED OPTICS 3177

sl

w

t

m

3

taking the time average of the production of the over-all output electric field from the interferometer andits complex conjugate,

Ic 5 ~^Er 1 Em&^Er 1 em&!*

5 (i51

n

Ei92 1 2(

i51

n

Ei9Ei119 cos 2k~nLi

2 x!n~nLi 2 x! 1 . . . , (4)

where n~nLi 2 x! is visibility of the fringe. Since thepectral distribution of the LED whose coherenceength is Lc used in these experiments is Gaussian,

then the visibility is given by12

n(nLi 2 x! 5 expH2F1.89~nLi 2 x!

LcG2J . (5)

White-light fringes develop every time when the op-tical path difference in the interferometer matchesLj~j 5 1,2,. . . ,m!, at which n~Li 2 x! becomes nonzero.The change of scan length, Dxi, for individual seg-ments of the optical fiber along the length of thesensor represents the gauge length change that isdue to measurand, i.e., strain.

The generalized strain–optic relationship betweenthe optical length change ~OLC! and the axial strain,exx, induced over the gauge length of a single segmentof the optical fiber is given below,13

OLC 5Dxi

25 nL Hn 2

12

n2@P12 2 nf~P11 1 P12!#J exx,

(6)

here P11 and P12 are Pockels constants, vf is thePoisson’s ratio, and L is the length ~gauge length! ofhe optical fiber.7 For single-mode silica optical fibers

at a wavelength of 1300 nm, the refractive index isn 5 1.46. Furthermore, P11 ' 0.12, P12 ' 0.27, vf 50.25, and therefore Eq. ~6! can be rewritten as

exx 5 Dxiy2.38L. (7)

3. Construction of the Sensors

The distributed strain optical fiber sensor is based onthe intrinsic property of a signal-mode optical fiber.The sensing section in the system consists of severalpieces of optical fiber representing various gaugelengths. Each piece of the optical fiber is polished ata right angle and coated with a thin layer of silver fora reflectivity of ;4%. The fiber ends are insertedinto a glass microcapillary to which they are gluedwith UV adhesive once the tips are brought into con-tact. Figure 2 details the construction of one seg-ment ~gauge length! of the sensor. Assuming that Ris the reflectance of the reflection coating, then theintensity of the beam form hth sensor at the inputterminal of the sensing section is

Ih 5 I0RT2h 5 I0R~1 2 R 2 A!2h, (8)

where A is the absorbancy of the coating and I0 is theintensity of the incident beam. For dielectric coat-


ings the absorbancy is zero ~A ' 0!, whereas A Þ 0 foretallic coatings.

4. Scanning Range and Scanning Speed

In white-light interferometry, fringes appear whenthe scanning distance of the sensing mirror matchesthe optical length of the sensor. Subsequent fringesoccur, each time, when the scanning distance equalsthe length difference between the adjacent sensors.Therefore, for a set of sensors with N individual sens-ing sections, the minimum scanning distance of theactuator Xmin is given by

Xmin 5 ~Ln 2 L0 1 D!n, (9)

where D is the maximum measurement range of sen-sors. The scanning range is determined by the dis-placement expected within the gauge-length andnumber of sensors within the linear segment of theoptical fiber. Scanning speed of the actuator willdetermine the response time, Ts, of each sensor.

5. Error Compensation

The quasi-distributed system is a double interferome-ter consisting of a sensing arm ~sensing interferome-ter! and the reference interferometer. All theindividual sensors in the sensing-arm–interferometersegment of the double interferometer act as sensors aswell as the reference arm for the adjacent sensors.They also share a common lead fiber to the source andthe second interferometer ~Fig. 1!. This configurationeliminates the ambient noise that otherwise wouldhave been generated in the sensing interferometer.However, unlike the sensing interferometer, the sens-ing and the reference arms of the receiving interferom-eter do not share the same optical path. If noteliminated, the phase shift that is due to the erroneouseffects of thermal fluctuations andyor ambient noisewill affect the strain measurements. Accordingly, Eq.~4! should include the unwanted phase shift, namely,

Fig. 2. Construction details of individual sensors and schematicrepresentation of reflected signals.

mbtTaapcmfrs

toctsspA

utwdbF

the system-error term, Dint, and in this case Eq. ~4! isrewritten as

Ic 5 (i51

n

Ei92 1 2 (

i51

n

Ei9Ei119 cos 2k~nLi 1 Dint 2 x!

3 n(nLi 1 Dint 2 x) 1 . . . . (10)

In the current configuration the system error is elim-inated by use of one of the sensors in the sensinginterferometer as a dummy sensor in a way similar tothat of the conventional strain gauges for structuraltesting. The dummy sensor is used solely for theacquisition of noise and not for measurement ofstrain. In the double interferometer we automati-cally eliminate the signal from this sensor by using itas a reference for all the other sensors.

6. Experimental Program

The experimental program encompassed two series oftests: ~1! direct tensile loading of the sensor and ~2!

easurement of strains in a steel beam subjected toending. The optoelectronics components of the in-erferometer consisted of a LED and a photodiode.he LED had a central wavelength of 1300 nm withspectral width of 60 nm. As shown in Fig. 3, data

cquisition and processing was controlled by a com-uter through a general-purpose interface bus. Theomputer was employed in driving the mirror-ounted translation stage in the white-light inter-

erometer system. Command programs includedoutines for detection of white-light fringes and mea-urement of fringe shifts.The experimental setup shown in Fig. 4 pertains to

he first series of tests, and it was designed to dem-nstrate the system’s linearity and self-compensationapabilities. As shown in Fig. 3, the four-sensor sys-em was tested under direct tension. Both ends ofensors 1 and 3 were fastened to two translationtages. One of the stages was fixed in place, and arecision micrometer was used to translate the other.

linear variable differential transformer ~LVDT!was used for measuring the tensile displacement.Sensor 2 was not used for measurement of strain.However, the signal received from this sensor wasemployed to examine the system’s self-compensation

Fig. 3. Optoelectrical signal processes within the system. GPIB,general-purpose interface bus.

capability. The fourth sensor was used as a dummyto monitor the ambient noise, temperature, and otherenvironmental effects.

Figure 5 pertains to the output of sensor 2 prior toand after compensation. Compensated output ofsensor 2 should indicate a flat response, since it wasnot employed for the measurement of strain. Figure5 demonstrates that, for all practical purposes, a flatresponse was observed following the application ofsignal compensation by way of sensor 4. The linearresponse of the sensor under tensile strain is shownin Fig. 6, where the strains measured by sensors 1and 3 are plotted as a function of the displacementmeasured by the LVDT. As shown in this figure, alinear response was observed over a range of 2000mm. The linear correlation coefficient is 0.998, andthe standard deviation is 1.1 mm. As demonstratedin Fig. 6, a one-to-one correspondence can be foundbetween the fiber-optic measured values and theLVDT measurements.

We performed the second experiment by loading asteel beam of dimensions 1828 mm 3 46 mm 3 7 mm

nder four-point bending. The four-point-bendingest arrangement is schematically shown in Fig. 7,here two concentrated loads are applied at the mid-le third points of the beam and it is supported atoth ends creating two additional reactive forces.igure 8 is a photo of the experimental setup. In

Fig. 4. Tensile test setup.

Fig. 5. Output of the sensor 2 before and after compensation.


3

this arrangement the middle third section of thebeam undergoes pure bending, which results in uni-form distribution of strain in the middle third sectionof the beam. Three sensor segments of the quasi-distributed fiber-optic sensor were fastened to thepure bending section of the beam. The fourth onewas used as a dummy gauge and was left unattachedto the beam. The fiber sensor segments had gaugelengths of 152.4, 154.94, 157.48, and 160.02 mm, re-spectively. A resistant strain gauge was used in the

Fig. 6. Calibration results for sensors 1 and 3.

Fig. 7. Instrumentation of the beam with various sensors

Fig. 8. Instrumented beam under four-point-bending test setup


same location for comparison of data with fiber-opticresults. By the same token, a LVDT was attached inthe pure bending zone of the beam measuring thedeformation within a gauge distance of 150-mm.The signals from strain gauge and LVDT were ac-quired through their individual conditioners by thecontrolling computer. A multichannel data-acquisition board was employed for recording ofstrain data. The measured strain by the distributedoptical fiber sensor and the strain gauge system isshown Fig. 9. The coefficient of correlation is 0.995,and the maximum standard deviation is 15 me. Fig-ure 10 demonstrates correlation between the fiber-optic sensor and the LVDT measured values ofdeformation. The coefficient of correlation in thiscase is 0.999, and the maximum standard deviation is0.6 me.

7. Conclusions

Development of a quasi-distributed optical fiber sen-sor for measurement of structural strains based onwhite-light interferometry is described. The config-uration of the sensor assembly is composed of a dou-ble interferometer for sensing and interrogation ofthe signal from all the sensors. The self-referencingcapabilities of the sensing interferometer togetherwith the allocation of one of the system sensors as adummy gauge provides self-compensation for the er-

Fig. 9. Comparison of the measured strains by fiber-optic sensorsand strain gauge system.

Fig. 10. Displacement of the beam as measured by the fiber-opticsensor ~FOS! and the LVDT

ic

ure along a 25-km optical fiber with 2 meter spatial resolu-
rors associated with the ambient noise and thermalfluctuations. The experimental results demonstratethe attributes of the system in terms of linearity andmultipoint sensing capability.
The research presented here has pertained to ex-perimentation with four sensors. However, addi-tional sensors and various gauge lengths can beintegrated into one line. The number of sensors islimited by the quality of the coating in terms of ab-sorbency and reflectivity. According to Eq. ~8!, thentensity from the fiber end of the tenth section inomparison with the first one will be 48% if A 5 0 and

22% if A 5 6%. Furthermore, the speed of strainmeasurements from all the sensors is dependent onthe number of sensors in the system ~scanning range!and on the speed of the scanning interferometer.

References1. F. Ansari, “Theory and applications of integrated fiber optic

sensors in structures,” in Intelligent Civil Engineering Mate-rials and Structures, F. Ansari, ed. ~ASCE Press, Reston, Va.,1997!, pp. 2–28.

2. X. Gu, Z. Chen, and F. Ansari, “Method and theory for multi-gauge distributed fiber optic crack sensor,” J. Intell. Mater.Syst. Struct. 10, 266–273 ~1999!.

3. W. C. Michie and B. Culshaw, “Detection of moisture in con-crete by optical fibers,” Cement Concrete Composites 19,35–44 ~1997!.

4. V. Lecoeuche, D. J. Webb, C. N. Pannel, and D. A. Jackson,“Distributed sensor for detection of impending structural fail-

tion,” J. Struct. Control 7, 23–34 ~2000!.5. A. D. Kersey and W. W. Morey, “Multiplexed Bragg grating

fiber-laser strain sensor system with mode-locked interroga-tion,” Electron. Lett. 29, 112–118 ~1993!.

6. Z. Chen, Q. Li, and F. Ansari, “Serial multiplexing of opticalfibers for sensing of structural strains,” J. Struct. Control 7,103–117 ~2000!.

7. T. Y. Liu, J. Cory, and D. A. Jackson, “Partially multiplexingsensor network exploiting low coherence interferometry,”Appl. Opt. 32, 1100–1103 ~1993!.

8. Y. J. Rao and D. A. Jackson, “A prototype multiplexing systemfor use with a large number of fiber-optic based extrinsicFabry–Perot sensor exploiting low coherence interrogation,” inDistributed and Multiplexed Fiber Optic Sensors V, J. P. Dakinand A. D. Kersey, eds., Proc. SPIE 2507, 90–98 ~1995!.

9. J. L. Brooks, R. H. Wentworth, R. C. Youngquist, M.Tur, B. Y.Kim, and H. L. Shaw, “Coherence multiplexing of fiber-opticinterferometric sensors, ” J. Lightwave Technol. LT-3, 1062–1072 ~1985!.

10. J. L. Santos and D. A. Jackson, “Coherence sensing of time-addressed optical-fiber sensors illuminated by a multimodelaser diode,” Appl. Opt. 30, 5068–5076 ~1991!.

11. D. Inaudi, “Coherence multiplexing of in-line displacementand temperature sensors,” Opt. Eng. 34, 1912–1915 ~1995!.

12. M. Born and E. Wolf, Principle of Optics, 7th ed. ~CambridgeUniversity, Cambridge, UK, 1999!, pp. 352–359.

13. J. S. Sirkis and H. W. Haslach, Jr, “Interferometric strainmeasurement by arbitrarily configured, surface-mounted, op-tical fibers,” IEEE J. Lightwave Technol. LT-8, 1497–1503~1990!.


quasi-distributed fiber-optic strain sensor: principle and experiment

Documents