当前位置: 首页 > >

Distributed sensor based on dark-pulse Brillouin scattering

发布时间:

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 17, NO. 7, JULY 2005

1501

Distributed Sensor Based on Dark-Pulse Brillouin Scattering
Anthony W. Brown, Bruce G. Colpitts, Member, IEEE, and Kellie Brown, Student Member, IEEE
Abstract—A novel dark-pulse-based technique has been used for the ?rst time in a Brillouin scattering-based distributed ?ber sensor. Experimentally obtained Brillouin spectra demonstrate that the dark-pulse con?guration is as capable of strain and temperature measurement as conventional pulse-based systems but at much higher spatial resolution. A spatial resolution of 50 mm is on a 100-m reported with a strain measurement accuracy of 6 sensing ?ber. Index Terms—Brillouin scattering, distributed sensor, ?beroptic sensor, spatial resolution, strain.

I. INTRODUCTION

T

HE SPATIAL resolution of a Brillouin optical timedomain analysis (BOTDA) distributed sensor system is generally limited by a combination of linewidth broadening and reduced signal strength associated with the use of short optical pulses [1], [2]. Maximum spatial resolutions of 1 m are typical of such systems, although some progress has been made toward further enhancements [2]–[4]; a resolution in the 500- to 100-mm range has been reported. For this reason, other techniques such as frequency-domain re?ectometry [5] and correlation-based methods [6] have been proposed to obtain higher spatial resolution. In both cases, however, improved resolution has come at the cost of acquisition speed and/or overall sensing length. In this letter, we demonstrate a novel method to obtain centimeter resolution from a time-domain pump and probe con?guration that does not compromise acquisition speed, and imposes only moderate restrictions on sensing length. II. THEORETICAL BACKGROUND

is that the strength of the Brillouin interaction is drastically reduced when the pulsewidth is shorter than the acoustic wave ns . The second is that the optical speclifetime trum of such a short pulse will be considerably broader than the 30–50-MHz typical width of the Brillouin gain pro?le. As a result, the Brillouin interaction spectrum will be signi?cantly broadened. This will reduce the measurement accuracy of the Brillouin frequency and, hence, the strain or temperature measurement accuracy. The behavior of the optical and acoustic signals in a BOTDA system is governed by the following coupled wave equations , and , for the pump, Stokes and acoustic ?elds respectively, (1) (2) (3) where is the frequency detuning parameter [7]. It was noted by Lecoeuche et al. [7] that because of the term in (1), the scattering interaction ceases immediately upon ) even given the ?nite the end of the pulse (i.e., when . It was suggested that the presence of a small CW component in the pulse signal could prepump the phonon ?eld before the arrival of a pulse, resulting in increased scattering for the duration of the pulse, an effect that had been observed previously [2]. This would make practical the use of shorter optical pulses resulting in higher resolution, at the cost of some distortion of the optical signal. If instead of a pulse of Stokes light (what we will call a bright pulse), one uses a CW Stokes wave, power will be continuously transferred from the pump to the Stokes wave at any localies within the Brillouin tion where the frequency difference spectrum of the ?ber. This will result in continual depletion of the pump. The received pump signal will still be CW, but at a lower overall power. Further, if an interacting CW Stokes wave is suddenly (and brie?y) switched OFF, all interaction will also momentarily cease. The depletion of the pump wave will stop, resulting in an increase in the pump signal above its mean level. The increase in pump signal will occur for the duration of the extinction of the Stokes light (what we will call a dark pulse). When the Stokes wave is restored, interaction will also resume. Moreover, if the period of extinction is much shorter than the phonon lifetime, it may be supposed that the acoustic ?eld will not be much changed. Therefore, the pump strength will

A BOTDA loss-based system operates by measuring the interaction intensity of two counterpropagating laser signals within a test ?ber, the continuous-wave (CW) pump at a fre. quency , and the Stokes pulse at a lower frequency Power will be transferred from the CW pump to the Stokes pulse through an acoustic wave at any location within the test lies within the local Brillouin gain pro?le. ?ber where The position of such a location may be determined through conventional optical time-domain re?ectometry techniques. Spatial resolution in a BOTDA system is determined by the duration of the Stokes pulse. For this reason, it is desirable to use short pulses to enhance spatial resolution. Unfortunately, two deleterious effects result from the use of short pulses. The ?rst
Manuscript received December 28, 2004; revised March 11, 2005. The authors are with the Department of Electrical and Computer Engineering, University of New Brunswick, Fredericton, NB E3B 5A3, Canada (e-mail: abrown@unb.ca; colpitts@unb.ca; k.brown@unb.ca). Digital Object Identi?er 10.1109/LPT.2005.848400

1041-1135/$20.00 ? 2005 IEEE

Authorized licensed use limited to: Beijing Institute of Petrochemical Technology. Downloaded on February 19, 2009 at 01:37 from IEEE Xplore. Restrictions apply.

1502

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 17, NO. 7, JULY 2005

Fig. 1. Brillouin loss waveforms of test ?ber for bright (solid line) and dark (dotted line) pulses at a beat frequency of 13 100 MHz, the resonant frequency for the strained section. Test conditions were identical in each case.

Fig. 2. Brillouin spectra taken with 2-ns bright and dark pulses. Test conditions were identical for both data sets, the dark-pulse spectrum has been inverted for clarity.

return to its original (partially depleted) level after the passage of the dark pulse. Because scattering only ceases at the location where the Stokes signal is not present (i.e., within the dark pulse), one may obtain spatially resolved Brillouin spectra using dark pulses. Since the optical spectrum of the quasi-CW Stokes wave is dominated by the narrow linewidth spectrum of the laser, the dark-pulse spectra will not suffer from spectral broadening. III. EXPERIMENTAL DETAILS A standard BOTDA system [4] was used for the following experiments. The optical sources were two diode pumped Nd : YAG lasers operating at a nominal 1319-nm wavelength. Pulses were made from the CW Stokes laser output by an electrooptic modulator (EOM). With the EOM set to extinction, an electrical pulse from a pulse generator produced bright optical pulses of approximately 80-mW peak power. Conversely, adjusting the dc bias of the EOM for maximum transmission resulted in a dark optical pulse being generated on an 80-mW background. The extinction ratio of the modulator was approximately 25 dB in either case. A pump laser power of approximately 3 mW was used. The power of the pump signal was monitored in the time domain by an ac-coupled photodetector. This effectively removed the background CW signal, showing only the change in pump level due to the Brillouin interaction. A 100-m-long test ?ber was used with a 2.54-m (100 in) section anchored between a translation stage and a ?xed block to provide a variable strain section. Brillouin spectra were taken using both bright and dark pulses of 2-ns duration. Averaging of 400 waveforms was used at each frequency. Time-domain waveforms for both cases are shown in Fig. 1 at a frequency difference of 13 100 MHz at which the strained section of ?ber is clearly visible. The Brillouin spectra of the strained section taken using the two methods are shown in Fig. 2. All test conditions other than the modulator bias point were identical in the two experiments and the spectra are shown to the same scale

(with the dark-pulse spectrum in Fig. 2 inverted for clarity). Both spectra were ?t to a Lorentzian function to obtain the center frequency. It is evident from the time-domain waveforms that the essential difference between the dark and bright pulse techniques is a simple inversion of the signal. The spectra of Fig. 2 demonstrate the advantage of the dark-pulse technique. The spectrum taken with the bright pulse shows poor signal-to-noise ratio (SNR) and a very broad linewidth (in the range of 200–300 MHz, albeit with a narrow central peak, as suggested previously [8]). In contrast, the dark-pulse spectrum has a much higher SNR and has a linewidth of about 40 MHz, the natural Brillouin linewidth of the ?ber. The result of the narrower linewidth and stronger signal is improved accuracy of the Brillouin frequency. Accuracy may inferred from the standard error of the center frequency parameters found in ?tting the spectra, 0.2 MHz for the dark pulse but only 0.7 MHz for the bright pulse. This corresponds to strain accuracies of 4 and 12 , respectively. It is dif?cult to obtain bright pulse spectra using pulses much shorter than 2 ns due to sensitivity limitations of the test equipment. It was, however, possible to take spectra using dark pulses as short as 500 ps, as shown in Fig. 3. Some degradation in the signal strength is noted, however, much of this may be attributed to bandwidth limitations of the sensor system components rather than reduced interaction. With a 500-ps pulsewidth, the strain measurements represented by these data have a spatial resolution of 50 mm which, to the best of our knowledge, is the highest resolution ever reported with a BOTDA system. The strain error for this measurement was only 6 . The measured linewidth of 40 MHz was also maintained. The 500-ps limit is imposed by the particular pulse generator used, however, there is no reason to assume that measurements could not be taken with shorter dark pulses if suitable equipment were available. In the form presented here, this technique is promising for extremely high resolution, as we have demonstrated. However, one

Authorized licensed use limited to: Beijing Institute of Petrochemical Technology. Downloaded on February 19, 2009 at 01:37 from IEEE Xplore. Restrictions apply.

BROWN et al.: DISTRIBUTED SENSOR BASED ON DARK-PULSE BRILLOUIN SCATTERING

1503

posed of a concatenation of 160 m of SMF-28, 110 m of OFS TrueWave XL, and 110 m of OFS TrueWave SRS. The total length exceeds the SBS threshold length for SMF-28 (about 200 m). Data was taken with both bright and dark pulses. The inset shows a horizontal magni?cation of 20 m of the TrueWave XL section, clearly indicating that the measured Brillouin frequency is substantially the same with either technique. The slight variation in the two measurements observed in the TrueWave SRS section is due to the narrower linewidth obtained with the dark pulses more clearly resolving the main peak of the multipeaked TrueWave SRS Brillouin spectrum. The ?uctuations in Brillouin frequency observed on the TrueWave ?bers is due to nonuniform tension applied as these ?bers were wound onto a spool.
Fig. 3. Brillouin spectrum taken with a 500-ps dark pulse. Experimental conditions otherwise identical to those used for Fig. 2.

IV. CONCLUSION It has been shown that dark pulses can be used to overcome the acoustic lifetime barrier which tends to limit the spatial resolution of bright pulse-based BOTDA distributed sensor systems. Brillouin linewidth is pulsewidth independent for dark pulses, making high accuracy measurements possible at arbitrary spatial resolutions. A 50-mm spatial resolution, limited only by the electronic bandwidth of available equipment, has been demonstrated with a strain measurement accuracy of 6 . A method has been demonstrated to allow dark pulses to be used on long sensing ?bers successfully overcoming the limitations imposed by SBS. REFERENCES
[1] A. Fellay, L. Thévenaz, M. Facchini, M. Niklès, and P. Robert, “Distributed sensing using stimulated Brillouin scattering: Toward ultimate resolution,” in Proc. 12th Int. Conf. Optical Fiber Sensors, OSA Tech. Dig. Series, vol. 16, 1997, pp. 324–327. [2] X. Bao, A. Brown, M. DeMerchant, and J. Smith, “Characterization of the Brillouin-loss spectrum of single-mode ?bers by use of very short ( 10-ns) pulses,” Opt. Lett., vol. 24, no. 8, pp. 510–512, 1999. [3] A. W. Brown, M. D. DeMerchant, X. Bao, and T. W. Bremner, “Spatial resolution enhancement of a Brillouin-distributed sensor using a novel signal processing method,” J. Lightw. Technol., vol. 17, no. 7, pp. 1179–1183, Jul. 1999. [4] A. W. Brown, J. P. Smith, X. Bao, M. D. DeMerchant, and T. W. Bremner, “Brillouin scattering based distributed sensors for structural applications,” J. Intell. Mater. Syst. Structures, vol. 10, pp. 340–349, 1999. [5] D. Garus, T. Gogolla, K. Krebber, and F. Schliep, “Brillouin optical?ber frequency-domain analysis for distributed temperature and strain measurements,” J. Lightw. Technol., vol. 15, no. 4, pp. 654–662, Apr. 1997. [6] K. Hotate and M. Tanaka, “Distributed ?ber Brillouin strain sensing with 1-cm spatial resolution by correlation-based continuous-wave technique,” IEEE Photon. Technol. Lett., vol. 14, no. 2, pp. 179–181, Feb. 2002. [7] V. Lecoeuche, D. J. Webb, C. N. Pannell, and D. A. Jackson, “Transient response in high-resolution Brillouin-based distributed sensing using probe pulses shorter than the acoustic relaxation time,” Opt. Lett., vol. 25, no. 3, pp. 156–158, 2000. [8] S. Afshar, V. G. Ferrier, X. Bao, and L. Chen, “Effect of the ?nite extinction ratio of an electro-optic modulator on the performance of distributed probe-pump Brillouin sensor systems,” Opt. Lett., vol. 28, no. 16, pp. 1418–1420, 2003. [9] K. Shiraki, M. Ohashi, and M. Tateda, “SBS threshold of a ?ber with a Brillouin frequency shift distribution,” J. Lightw. Technol., vol. 14, no. 1, pp. 50–57, Jan. 1996.

<

Fig. 4. Dark and light pulse frequency versus position data for SMF-28TrueWave XL-TrueWave SRS concatenation. Inset horizontally magni?es region from 160 to 180 m, showing agreement between both data sets.

drawback will be readily apparent: The large CW power levels involved may result in stimulated Brillouin scattering (SBS) if long ?bers are used. A quasi-CW power of 80 mW will exceed the SBS threshold for standard ?bers that are over a few hundred meters in length. This is still acceptable for many sensing needs, especially for small structures. If longer lengths are required, any of the standard methods for reducing SBS may be employed; the CW power may be reduced or a large effective area ?ber may be used. Alternatively, an SBS suppressing ?ber may be used [9]. SBS suppression through variation of the Brillouin frequency is easily implemented in a strain sensor through variation in the initial strain placed on the sensing ?ber and/or by concatenating sections of ?ber having different Brillouin frequencies. A demonstration of this is shown in Fig. 4 which depicts the Brillouin frequency distribution of a sensing ?ber com-

Authorized licensed use limited to: Beijing Institute of Petrochemical Technology. Downloaded on February 19, 2009 at 01:37 from IEEE Xplore. Restrictions apply.




友情链接: