1. Introduction

During the last few decades, fiber hydrophones have been intensively researched due to their exceptional advantages in comparison to the traditional piezoelectric type sensors, such as high sensitivity, compact size, immunity to electromagnetic interference and capable multiplexing [1,2]. Generally, fiber hydrophones are widely used in the underwater environment, such as oil exploration, natural disasters warning and military technology. The operation principle of the current fiber optic hydrophones can be classified into several techniques: optical reflection at the fiber end, fiber Bragg gratings, distributed feedback fiber laser, Michelson interferometers, Mach-Zehnder interferometers, and Fabry-Perot interferometers [3–7]. Among these techniques, fiber Michelson hydrophone which can provide an optimum solution to a combination of high sensitivity and sufficient multiplexing ability is much more likely to satisfy the basis requirements of the next generation of the sonar systems [8].

The fiber Michelson hydrophone (FMH) is based on the acoustic pressure sensitivity of the optical fiber, in which the phase of the injected light can be changed via the strain which is induced by the stretching and the compressing the optical fiber caused by acoustic pressure changes [9–11]. It is found that the phase sensitivity of a fiber optic hydrophone is strongly dependent on the strain configuration [12]. In order to improve the strain configuration, many design schemes of the sensor head have been proposed, such as multilayer fiber coil hydrophone, air-backed mandrel hydrophone and flatted mandrel hydrophone [13]. Currently, in most of design schemes of the FMH sensor head, it is still a scientific challenge to characterize the strain property that induced in the design processes. To the best of our knowledge, up to now some reported methods only can give the length difference of the two arms of the FMH [14,15]. In this study, we propose an online distributed strain measurement of the sensor arm of the FMH based on stimulated Brillouin scattering (SBS). We believe that the proposed scheme can offer an attractive solution to design optimization and sensitivity enhancement for the FMH sensor.

Fortunately, the distributed strain measurement along the fiber has been fully explored via the Brillouin scattering whose central frequency of the reflection spectrum named the Brillouin frequency shift (BFS) is dependent on the local strain. Several kinds of Brillouin-based optical fiber sensors have been studied,such as Brillouin optical time domain reflectometry (BOTDR) [16] and Brillouin optical time domain analysis (BOTDA) [17–19]. In a BOTDR system, only an optical pulse is employed as the injected wave, but its spatial resolution of ~1 m cannot satisfy the requirement of the centimeter spatial resolution which is determined by the centimeter diameter of the mandrel [20]. While a 2-cm spatial resolution can be realized through using differential pulse-width pair Brillouin optical time-domain analysis (DPP-BOTDA). However, a continuous wave (CW) probe is required in the conventional DPP-BOTDA system, which would be reflected by the sensor arm and the reference arm of the FMH and the reflected lights would generate interferences at the output of the FMH. As a result, the Brillouin signal at the detector in a conventional DPP-BOTDA system will be severe disturbed, because the injected light will be phase-modulated by the environmental variations e.g. the mechanical stress, the pressure, and the temperature change when it propagates along the sensor arm and reference arm in the FMH [11], resulting in a significant intensity fading.

In this research, to overcome the severe interference problem for the Brillouin signal, two optical pulses (i.e. a probe pulse and a pump pulse) are employed in the BOTDA system to measure the distributed strain along the sensor arm of the FMH. The width of the probe pulse should be adjusted to match the fiber length difference between the sensor arm and the reference arm of the FMH. The time delay between the pump pulse and the probe pulse is precisely controlled to achieve that they can interact with each other over the sensor arm. Furthermore, an 8/8.5 ns pump pulse pair is used providing a 5-cm spatial resolution for the distributed strain measurement. In experiment, the temperature and hydrostatic pressure property of the sensor arm in FMH are investigated by the pulsed-probe DPP-BOTDA. The spatial period of the mandrill and the fiber length of each layer can be clearly distinguished; the map of compression or stretch along the sensor arm can also be revealed.

2. Operation principle and experimental setup

2.1. Operation principle

The schematic diagram of the FMH is shown in Fig. 1. The light output from the laser is split into two branches (i.e. the sensor arm and the reference arm) via an optical coupler (OC). Subsequently, the optical waves in two branches are reflected back by two Faraday reflect mirrors (i.e. FRM1 and FRM2), and the two reflected optical waves are recombined by the OC. Then the output optical signal is directly received by a photoelectric detector.

 

Fig. 1 The schematic diagram of the FMH. OC: optical coupler; FRM: Faraday reflect mirror.

Download Full Size | PDF

The two reflected optical waves are interfered with each other, and the total optical intensity before the photoelectric detector is abided the following equation [2]:

Based on the Eqs. (1) and (2), a large intensity fluctuation for the output optical signal is generated since the phase delay $\phi$can be easily modulated by a small variation of the refractive index n and the length difference$\Delta L$, which are induced by the variation of the environmental pressure and temperature [11]. In order to measure the distributed strain along the sensor arm and to avoid the signal fading, the traditional CW-probe wave is intensity-modulated into the optical pulsed-probe wave in our DPP-BOTDA system. The width $\Delta t$of pulsed-probe wave is determined by the length difference $\Delta L$as Eq. (3):

Based on the Eq. (3), the optical wave reflected from the sensor arm can avoid being interfered with that reflected from the reference arm through precisely control the delay between the pump and probe pulses.

For the traditional BOTDA system, the Brillouin gain spectra (BGS) over the sensing fiber can be obtained through scanning the frequency offset (~11 GHz) between the pump pulse and the counter-propagating probe wave. Then, the BFS can be calculated by Lorentzian- or Gaussian-curve fitting of the BGS, which is linearly proportional to the strain and temperature in the sensing fiber.

As shown in Fig. 2(a), both the pump pulse and the probe pulse are launched into the sensor arm in the same direction. Subsequently, the pump pulse and the reflected probe pulse can be counter-propagated by precisely controlling their delay. In addition, point A is marked in the sensor arm as the mirror symmetric point of the end of the reference arm. As depicted in Fig. 2(a), the rising edge of the reflected pulsed-probe will meet the failing edge of the pulsed-probe at the point A in the sensor arm when the width $\Delta t$of the pulsed-probe is selected based on the Eq. (3), avoiding the overlap between the two probe pulses reflected from the sensor arm and the reference arm, respectively. By precisely controlling the time delay between the pump pulse and the probe pulse at an appropriate value, the counter-propagating pump pulse will be interacted with the rising edge of the reflected probe pulse at the point A (in the Fig. 2(a)), and then it will be interacted with the failing edge of the reflected probe pulse at the end of the sensor arm (in the Fig. 2(b)) after transmitting the whole$\Delta L$section. Thus, the optical interference at the output of the FMH, as well as the fading of Brillouin signal, can be avoided. Then the distributed BGS between the point A and the end of the sensor arm can be achieved through sweeping the frequency offset between the probe pulse and the pump pulse. Noted that the reflected pump from the reference arm will overlap with the probe signal from the sensor arm, which will seriously reduce the SNR of the measured Brillouin gain spectrum. As a result, the BFS at the first 85-cm fiber of the beginning section from point A cannot be measured accurately.

 

Fig. 2 The configuration of the pump pulse and the probe pulse. (a) the rising edge of the reflected pulsed-probe will meet the failing edge of the pulsed-probe at the point A, and the pump pulse is injected into the sensor arm with its rising edge aligned to the falling edge of probe pulse, (b) going through the whole $\Delta L$ section, the falling edge of the reflected probe pulse counter the pump pulse at the end of the sensor arm.

Download Full Size | PDF

2.2. Experimental setup

The experimental setup was assembled as shown in Fig. 3. The output of a narrow linewidth fiber laser at 1550 nm was spilt into two branches by a 70/30 optical coupler (OC1). The lower-branch with 30% power component was intensity-modulated into an optical pulse via an electric-optic modulator (EOM1), which was driven by the electrical pulse output from the channel 1 of an arbitrary-waveform generator (AWG). Then the optical pulse was amplified to 3 W by an Erbium-doped fiber amplifier (EDFA). A polarization scrambler (PS) was used to eliminate the polarization fading of the Brillouin signal since the SBS is a polarization-dependent effect. Consequently, the amplified and polarization-scrambled optical pulse used as the pump pulse was launched into the FMH through an optical circulator (C1).

 

Fig. 3 Experimental setup. Left part is the DPP-BOTDA system with the pulsed-probe and the right part (dashed box) is the fiber Michelson hydrophone configuration. OC: optical coupler; EOM: electro-optic modulator; EDFA: erbium doped fiber amplifier; PS: polarization scrambler; C: circulator; MG: microwave generator; AWG: arbitrary-waveform generator; ISO: isolator; PD: photo-detector; DAQ: data acquisition; FBG: fiber Bragg grating; FRM: faraday reflect mirror.

Download Full Size | PDF

To observe the cm-order strain variation in the FMH, an 8/8.5 ns pulse pair was pre-edited and output from channel 2 of the AWG, which has a pulse width difference of 0.5 ns corresponding to a 5-cm spatial resolution. The upper-branch with 70% power component was firstly modulated through the EOM2 which was driven by a sinusoidal microwave signal output from a microwave generator (MG). Two first-order sidebands were generated by EOM2 through adjusting its operation point with the carrier being suppressed. The lower frequency sideband was selected by a fiber Bragg grating (FBG1) and subsequently was intensity-modulated into an optical pulse via the EOM3 with a peak power of 2 mW. An isolator (ISO) was used to prevent the apparatus from being damaged by the reflection light of the Faraday reflect mirrors.

The fiber hydrophone configuration based on Michelson interferometer (inside the dashed frame in Fig. 3), which consisted of the OC2, a sensor arm and a reference arm, was tested by injecting the probe pulse and the pump pulse through the OC2. Then, both the pump pulse and the probe pulse were reflected by the FRM1 and the FRM2. After the pump pulse was filtered out by a fiber Bragg grating (FBG2) with a reflectivity of 97%, the transmitted light containing the Brillouin signal (the amplified probe pulse) and the residual pump pulse was detected by a photo-detector (PD).

Figure 4(a) shows a typical signal received by PD in Fig. 3, where the time delay 624.5 ns between the two probe pulses (two pump pulses) are equal to the round-trip time of the length difference $\Delta L=$62.45 m. Note that pulses A and C are the probe pulses reflected by FRM2 and FRM1, respectively, while pulses B and D are the pump pulses reflected by FRM2 and FRM1, respectively. Note that the probe pulse was launched into the FMH earlier than the pump pulse, and the time delay between A and C is 624.5 ns. As a result, the pump pulse and the reflected probe pulse was counter-propagated along the $\Delta L$segment and interacted with each other through the SBS effect. It is noted that the width of probe pulse and pump pulse are 95 ns and 8 ns in Fig. 4(a), respectively; and the time delay between the probe pulse and the pump pulse is 418 ns. Finally, the width of the probe pulse and the time delay between the probe pulse and the pump pulse are tailored to 624.5 ns as depicted in Fig. 4(b) so that the probe wave can cover the entire $\Delta L$section resulting in a distributed strain measurement.

 

Fig. 4 Plot of typical signal received by PD in Fig. 3. Note that A and C are the received probe pulses reflected by FRM2 and FRM1, respectively. While, B and D are the received pump pulses reflected by FRM2 and FRM1, respectively. (a) Probe pulse (95 ns); (b) Probe pulse (624.5 ns) for the length difference $\Delta L=$62.45 m.

Download Full Size | PDF

3. Experimental results and discussions

Two FMHs (FMH1 and FMH2) based on air-backed mandrel are used to be tested by the proposed pulsed-probe DPP-BOTDA system. The cross section of the hydrophone is schematically shown in Fig. 5. Four optical fiber coils, forming the sensor arm of a Michelson hydrophone, were fixed on the pedestal. Each coil with two fiber layers is nearly of the same length. This ‘push-pull’ configuration has been widely used because it not only enhances the sensitivity but also cancels the acceleration response. The reference fiber is bonded to the mandrel and prevented from the outside acoustic pressure. The sensor fiber is bonded to the elastic tube so that the fiber would be in state of the plane strain.

 

Fig. 5 Cross section of the fiber Michelson hydrophone.

Download Full Size | PDF

The first hydrophone (FMH1) with a 2.12-m reference arm and a 64.57-m sensor arm is used to test the strain induced by the environmental temperature. The probe pulse of 624.5 ns is applied in the DPP-BOTDA system, where the probe pulse width corresponds to the length difference of 62.45 m of the two arms. As shown in Fig. 6, the BFS distribution of the $\Delta L$section is measured by the proposed pulsed-probe DPP-BOTDA system with a spatial resolution of 5 cm (8/8.5 ns) at 20°C. The fluctuation of the BFS over the $\Delta L$section can be clearly distinguished that could be caused by the residual stress induced by various disturbance factors during the fiber drawing and coating process or the uneven stress as winding the fiber to the mandrel [21]. A zoom-in view of the segment at 40 m~42 m is plotted in the inset. It is clearly seen that a spatial period of 0.56 m equivalent to the spatial period of the fiber spool is emerged, which means that the sensor fiber should be subjected to uneven stress as winding the sensor fiber to the spool. Consequently, since the BFS is sensitive to the strain and temperature variation, the proposed pulsed-probe DPP-BOTDA system can be used to realize the distributed strain measurement of the sensor fiber. Noted that single mode fibers (SMF) are used in the reference arm and the sensor arm of the two FMHs.

 

Fig. 6 The measured BFS distribution of the $\Delta L=$62.45 m section

Download Full Size | PDF

Then, the distributed BFS measurements of the senor arm of FMH1 is tested as the temperature is increased from −40°C to 80°C with a step of 20°C under the standard atmospheric pressure, and the results are shown in Fig. 7(a). As shown in Fig. 7(b), the distributed strain variations (i.e. the temperature-induced strain) of the sensor arm as the temperature increasing from −20°C to 80°C are calculated by subtracting the BFS distribution at −40°C [the black curve in Fig. 7(a)] and the BFS change induced by temperature at loose state, and it can be expressed as

 

Fig. 7 (a) The measured BFS distribution of the $\Delta L=$62.45 m section with the temperature range from −40°C to 80°C, at the standard atmospheric pressure. (b) Strain variation at different temperature relative to −40°C.

Download Full Size | PDF

It can be seen that the strain variation increases with increasing temperature, which could be attributed to several factors. The first factor is the radial strain of the elastic tube due to expansion of the protection layer of epoxy resin so as to consequently increase the tensile stress of fiber drawing; secondly, the radial expansion of the sensing coil induced by each layer fiber also increases the strain of the sensor fiber [24]. It can be clearly distinguished that the strain distribution has a characteristic periodic variation with a period of ~8 m, which equals to the length of one fiber layer. From −20°C to 80°C, the whole strain variation distribution increases with increasing temperature, while most crossover regions exhibit strain-peaks. The existence of the strain-peak means that the crossover regions such as the crossover region c1 (inside the dashed elliptical circle), where fiber crosses from layer to layer in one fiber coil, experience additional stretch as temperature increasing. Nevertheless, the strain of the most crossover regions e.g. the crossover region c2 (inside the dashed elliptical circle) where fiber crosses from coil to coil varies a little bit. The results imply that different disturbance factors during the fiber winding process and temperature increasing can impact on the phase sensitivity of the FMH.

Next, the second hydrophone (FMH2) is used to investigate the hydrostatic pressure property of the sensor arm, where the length of the sensor arm and reference arm are 128.96 m and 2.05 m, respectively. The BFS of the sensor arm is measured at hydrostatic pressures ranging from 0.1 to 10 MPa at room temperature ($25$°C). The distributed strain variations of the sensor arm as the hydrostatic pressures ranging from 1 to 10 MPa at room temperature (25°C) are calculated by subtracting the BFS distribution at 0.1 MPa, and it can be expressed as

As shown in Fig. 8, while the probe pulse of 1269 ns is applied. Eight layers with a spatial period of ~16m are demonstrated, which indicates that the fiber layer winding on the mandrel should be subjected to both the uneven stress when the fiber is winded to the mandrel and the uneven pressure between different fiber layers. As for most of the sensor fiber, the negative strain variation increases as the pressure increases, which means the fiber is considerably compressed. At 10 MPa, the maximum strain is 805$\mu \epsilon$at the position of 109.45 m (marked in Fig. 8) and the minimum strain is −1272$\mu \epsilon$located at the position of 114.35 m (marked in Fig. 8), resulting in a fluctuation of 2077$\mu \epsilon$over the entire fiber.

 

Fig. 8 The measured strain variation distributions of the sensor arm at hydrostatic pressure range from 1 to 10 MPa.

Download Full Size | PDF

Figure 9(a) shows a zoom-in view at the segment of 12~33 m. It can be seen that the negative strain variation over the region of 15~30 m increases as hydrostatic pressure increasing while the crossover regions exhibit considerably different strain sensitivity. We first investigated the hydrostatic pressure characteristics at a typical position (position A marked in Fig. 9(a)) from 1 MPa to 10 MPa, and the results are shown in Fig. 9(b). It can be seen that the strain of this position is sensitive to the hydrostatic pressure changes, with a strain variation difference of 905$\mu \epsilon$in the pressure ranges of 1 to 10 MPa. In the experiment, we use the variation in the BFS from five repeated measurements to characterize the BFS measurement standard deviation. The result shows that the BFS measurement standard deviation is $\delta \text{=}$ ± 1.02 MHz, corresponding to a strain measurement error of ±21.2 µε.

 

Fig. 9 (a) The measured strain variation over the segment of 13~33 m at hydrostatic pressure range from 1 MPa to 10 MPa; (b) Strain variation at the position of 22.25 m versus different hydrostatic pressure, 1-10 MPa.

Download Full Size | PDF

4. Conclusions

A pulsed-probe DPP-BOTDA is proposed to enable online distributed strain measurements of the sensor fiber of a FMH. In the scheme, the probe pulse and pump pulse light are created and launched into the FMH so that the interferences between the output lights reflected from the FMH can be avoided through precisely controlling the width of probe pulse which is equal to the time delay between the two arms of the FMH. Meanwhile, an 8/8.5 ns pump pulse pair is used to realize a 5-cm spatial resolution for the strain measurement. In the experiment, FMH1 with the arm length difference of 62.45 m and FMH2 with arm length difference of 126.9 m are applied in the strain variation distribution measurement, respectively. As a result, we have gained the distributed temperature-induced strain variation and pressure-induced strain variation over the sensor arm, which has a close relation with the design process. We believe that the proposed pulsed-probe DPP-BOTDA system can be used to characterize the design process of the FMH and then to improve its performance and quality.