1. Introduction

Smart structures are being developed with built-in sensors for long-term structural health monitoring [1–3]. The detection of damage and impact loading on aerospace structures has been investigated by conventional sensing techniques, such as strain gages, piezoelectric sensors, etc [4,5]. These techniques usually have some drawbacks including electromagnetic interference, corrosion, and discontinuous operation. However, fiber optic sensors have become a promising candidate for structural health monitoring under harsh environments. They are small in size, immune to electromagnetic interference, resistant to many hazardous chemicals, and capable of distributed sensing through one optical fiber line [4]. A famous device for these sensors is fiber Bragg grating (FBG) which has been recently demonstrated as highly sensitive strain sensors. These FBG-base sensors were suggested using methods such as a mode-splitting resonator [6], a fiber ring cavity [7] or a triangle-wave laser tuning [8], or special fibers such as solid Bragg fiber [9] or polymer optical fiber [10]. FBG sensors are very practical but they are localized or point sensors which cannot detect distributed strain.

On the other hand, Brillouin optical correlation domain analysis (BOCDA) using the Brillouin scattering effect in fiber optic sensors makes it possible to measure the fully distributed strain with high spatial resolution through a sensing optical fiber line [11,12]. Brillouin scattering is a three-wave interaction between an incident photon, a backscattered photon, and an acoustic phonon. Pump and probe lightwaves propagating in opposite directions through an optical fiber interfere with each other, and excite an acoustic wave. The refractive index grating caused by the acoustic wave couples the two lightwaves. This coupling gives rise to optical power transfer from the pump to the probe. The probe lightwave experiences gain through the stimulated Brillouin scattering (SBS) process when the optical frequency difference Δν between the probe and the pump is tuned to the Brillouin frequency shift (ν_B) of the fiber. The Brillouin gain spectrum (BGS) is narrow (30MHz) and can be recovered by sweeping the frequency difference between the two lightwaves [13,14]. One method for measuring the BGS of an optical fiber is Brillouin optical time domain analysis, which uses pulsed lightwaves and a time domain technique. Another method is BOCDA in which continuous pump and probe lightwaves are used and the correlation between two lightwaves to generate SBS in a fiber is synthesized. Therefore, the SBS is generated only in a narrow section along the fiber and precise spatial resolution about the BGS change is obtained [11].

The center frequency of the BGS is shifted in proportion to longitudinal stress or temperature applied to it. The strain or temperature dependence of Brillouin scattering in optical fibers was reported for various optical fiber materials and structure conditions [15,16]. However, it is hard to measure the impact trace, that is, the residual strain in an optical fiber that remains after removing the external impact applied to the structure. In some studies, including our previous one, discrete optical devices on an optical fiber line, such as fiber Bragg gratings (FBGs) or long period gratings, have been used for measuring residual strain [17,18]. However, the optical fiber itself has never been used for detecting residual strain in a discrete or distributed way. This is because silica-based optical fibers do not have plastic deformation. To overcome this, Aluminum (Al) foil was used to package the optical fiber, causing the fiber to have the property of plastic deformation in this paper. Our Al-packaging fabrication method is very simple compared to other methods, which involve optical fibers co-braided with braided composites or optical fibers material-thin-foil-deposited by nano process equipment [19].

In this paper, residual strain in an Al-packaged optical fiber is measured using a BOCDA measurement system. In the system, the interaction between a probe and a pump generated from the same light source induces a correlation peak in a specific portion of the optical fiber with 2 cm spatial resolution. The peaks in three-dimensional Brillouin gain graphs as a function of the frequency difference between a probe and a pump were quantitatively measured for applied tensile stresses. Al thin layers were bonded to optical fibers, and the fibers of one, two, and three Al layers were compared using this measurement system. Tensile stress was applied up to 7000 με and peak frequency shifts in the Brillouin gain spectra showed residual strain for each tensile stress. For reliability, residual strain was repeatedly measured for up to nine months after the application of tensile stress. Finally, a composite material plate equipped with an Al-packaged optical fiber sensor was hammered for real impact situation and the residual strain in the plate was detected. This experiment verifies the possibility of detecting residual strain in structures, including aerospace structures.

2. Principles and experimental setup

The experimental setup of a BODCA measurement system is shown in Fig. 1 (a). One high-power distributed feedback laser diode (DFB LD) was used for lightwave generation of both the probe and pump in the system. When only a 330 mA current was applied to a DC current port, the LD produced an output of 37 mW with a peak wavelength of 1553.3 nm. A sinusoidal wave of 8 MHz ($f_m$) frequency was connected to an AC current port of the DFB LD for direct modulation together with the DC current. This frequency-modulated (FM) continuous lightwave caused the interaction of the probe and the pump to induce the correlation peak in a specific portion of the optical fiber. The width of this correlation peak decides the spatial resolution (Δz) of the measurement. The spatial resolution is theoretically given by $\Delta z=({\upsilon }_g\cdot \Delta {\upsilon }_B)/(2\pi \cdot f_m\cdot \Delta f)$, where $\Delta {\upsilon }_B$ is the linewidth of the intrinsic BGS, $\Delta f$is the FM amplitude, and $f_m$ is the FM frequency. The correlation peak occurs periodically with interval $d_m={\upsilon }_g/2f_m$ along the optical fiber [11]. The FM frequency was changed between 8 MHz and 9 MHz and the Brillouin scattering gain spectrum was measured at each position of the optical fiber. The FM amplitude was 6.2 GHz, which was chosen for 2 cm spatial resolution. The modulated lightwave was divided by a 3 dB coupler and was launched into two different directions as a probe and a pump. The probe wave was modulated again by a LiNbO3 intensity Mach-Zehender (MZ) modulator supplied with a microwave of frequency ${\upsilon }_B$. This modulation caused two sidebands to appear, while the main band was suppressed by a DC bias voltage. The lower frequency sideband of the output was used as a probe with a frequency difference of ${\upsilon }_B$as compared to the pump. This is graphically explained in Fig. 1(a), where the probe and a pump are presented as solid and dashed lines at a frequency axis. The optical spectra were monitored by a 10% tap coupler. At the correlation peaks, the pump and probe maintained a frequency difference of around 11 GHz, which was necessary to maintain SBS. A pump light was chopped by another MZ modulator with a maximal extinction ratio of 20 dB for high detection sensitivity. A polarization controller (PC) was used before each modulator for low insertion loss. After passing through an optical isolator the probe was launched into an optical fiber under test. A chopped pump was amplified by an erbium-doped fiber amplifier (EDFA) with an output of 18 dBm and was launched into a circulator. The pump and probe interacted in a tested fiber, and the interaction between lightwaves and an acoustic wave transferred the pump power into the probe power. This power transfer caused an SBS gain. In order to clearly avoid a zeroth-order correlation peak (which means there is no path difference between a probe and a pump), a 60 m long delay line was inserted in the pump path. Additionally, the signals could not propagate to opposite fiber paths because of an isolator and a circulator. Another role of the circulator is to make a probe lightwave amplified by SBS to be inserted into an FBG, removing the higher frequency sideband in the probe lightwave as shown in the graph next to an FBG in the figure.

 

Fig. 1 A schematic of the experimental setup with a BOCDA measurement system (a), the optical fiber under test (b) and optical spectra of a probe before and after an FBG (c).

Download Full Size | PDF

A probe of a single side band was detected by a photo-detector (PD) and converted to an electrical signal. The small SBS gain in this measurement required a lock-in-amplifier (LIA) to electrically amplify the converted signal. Pump light chopping was requested for the operation of the LIA, and a chopping frequency for a MZ modulator was synchronized to the LIA reference frequency. A function generator for FM modulation and chopping, and a LIA were connected to a data acquisition computer for automatically measuring the SBS gain spectra. Both ends of the fiber under test were fusion-spliced to an isolator and a circulator in the measurement set, as show in Fig. 1(b). This test fiber was 2.5 m long and was spliced at both ends to a 10 cm dispersion shifted fiber (DSF), which was used as a position reference. The frequency shift of a Brillouin gain peak in the DSF is 10.52 GHz, which is significantly different from that of the single mode optical fiber (SMF) (10.84 GHz). The middle of the measured fiber (15 cm) was fixed to a tension device and tensile stress was applied. An FBG was used to remove the higher frequency sideband from the probe lightwave. The FBG had a 3 dB reflection bandwidth of 0.2 nm, and its center wavelength was stably controllable within more than 0.5 nm. The reflectivity was higher than 25 dB. Figure 1(c) shows the optical spectra of a probe before and after an FBG. The dotted line is the spectrum before the FBG and shows a highly diminished center wavelength band by a modulator DC bias voltage, and two sidebands. The full width at half maximum (FWHMs) of these wavelength bands were less than 1 MHz before frequency modulation and became 6.2 GHz by modulation chirping after modulation. The solid line is the spectrum after the FBG and is asymmetric. The shorter wavelength band (the higher frequency sideband) was rejected up to −30 dB and signal distortion caused by due to beating noise between the two sidebands was minimized.

3. Experimental results

The tension experiment was first applied to a standard SMF (SSMF). The polymer coating of the fiber was removed and both ends of 15 cm length were glued to moving stages. Figure 2 shows the Brillouin frequency shifts obtained from the measured Brillouin gain spectra when the fiber was extended by 0, 2500, 5000, and 7500 με. The fiber broke at a tensile strength around 10000 με. At each tension, a BGS was measured as a function of distance and the gain peak frequency was shifted by applied tension. The blue circles on the graph corresponded to the measured Brillouin frequency shifts, and the dotted line is a fit to the data. The relation between tensile stress and the Brillouin frequency shift in SSMF is linear with a slope of 0.049 MHz/με. This figure completely agrees with the previously reported results that used a Brillouin optical time domain analysis technique [15]. This means that value of the tensile stress can be obtained by measuring a BGS. However, the frequency shift from the gain spectrum reverted to its original condition after the stress was removed. This reversion to the original state is because the optical fiber does not undergo plastic deformations.

 

Fig. 2 The Brillouin frequency shift of a bare optical bare fiber as a function of applied tension.

Download Full Size | PDF

An SSMF covered with thin Al foil is suggested for measuring the residual strain after removing applied tension. The Al foil used was 18 μm thick. An SSMF was placed between the folded foil and they were glued together using the bonding adhesive CC-33A (Cyano-acrylate base cement, KYOWA). When multiple layers of Al were needed, another piece of Al foil was consecutively glued to the Al-foil-covered fiber. The polymer coating of the fiber was not removed for bare fiber protection and packaging practicality. The Al-foil-packaged fiber had a width of 5 mm flat. This Al foil fiber was then fixed to the tension test equipment, which was composed of fixed and controllable moving stages. Tensions of 1000, 3000, and 5000 με were applied and removed from the Al foil fiber in order, and a BGS was measured at each step. The Brillouin frequency shift obtained from these spectra is summarized in Fig. 3. In this experiment, two layers of Al foil were glued to a fiber and three sets of the Al-foil-packaged fiber were repeatedly tested for acquiring deviation values. Seven circles on a dotted line were measuring points for these test fibers. For each value of tension on the horizontal axis, the upper point and the lower point correspond to the applied and removed tension conditions, respectively. The Brillouin frequency shift had a slope of 0.035 MHz/με with respect to applied tension, which is 73% of a bare fiber without Al foil layers. A dotted line for a bare fiber is also shown in Fig. 3 for comparison. It is estimated that this reduction in shift resulted from the facts that the bonding adhesive partially absorbed the applied tension and that the fiber polymer coating partially slid on the surface of the bare fiber. After removing the tensile stress, it is very clear that there is residual strain in the Al-packaged optical fiber. The slope of the frequency shift for the residual strain is 0.024 MHz/με, which is more than two-thirds of that for the case before removing the stress. In this figure the deviation of the frequency shift increases with applied tension and has maximally a 7% for 5000 με tension.

 

Fig. 3 The Brillouin frequency shift of optical fibers packaged with two Al layers as a function of applied tension.

Download Full Size | PDF

Figure 4 shows three-dimensional (3D) graphs of Brillouin gain as a function of fiber distance and frequency difference between the pump and probe with 5000 με of applied tension. In the graphs, Brillouin gains are given as continuous colors from blue (lowest gain) to red (highest gain) and, hence, the Brillouin frequency with the highest gain can be observed at each position. The frequency shift at a given position corresponds to the magnitude of the applied tension. The tension was initially applied to an optical fiber without an Al foil package and the gain graph is depicted in Fig. 4(b) while the tension was held fixed. The 15 cm segment under tension is easily distinguished from the sides that have no tension. Next, tension was applied to an optical fiber with two Al-foil layers. Figure 4(a) shows the gain graph after the applied tension was removed. The frequency shifts in the middle segment clearly show that residual strain in an Al-foil fiber was successfully measured by Brillouin gain. The residual strain linearly increased from 1000 με to 5000 με, whereas it became much smaller under 1000 με tension. When the tension was around 7000 με, an Al foil fiber was broken. In addition, the frequency shift for an Al-packaged fiber continuously changed in the area under stress, unlike in the case for the optical fiber without the Al packaging, where the change was discontinuous. It is estimated that the continuous change in the frequency shift came from non-uniform applied tension that was transferred through the Al-package. It is noted that when these tensions were randomly applied to an Al-packaged fiber, the strain caused by the highest stress, not that caused by the latest tension, was detected as the residual strain. This result is practical considering the fact that a bigger external impact is more critical for structure safety.

 

Fig. 4 3D graphs of Brillouin gain in optical fibers with (a) and without Al foil package (b) as a function of the fiber distance and frequency difference with 5000 με of applied tension.

Download Full Size | PDF

In order to investigate the effect as a function of the number of Al foil layers, optical fibers with one, two, and three Al-foil layers are compared in Fig. 5. A cross sectional photograph of an Al-foil-packaged fiber is presented as an inset of each graph and the fiber cladding and polymer coating areas can clearly be seen. The distinct shiny lines above the circular fiber are the layers of Al foil, and the number of Al layers is seen in each photograph. The space between the layers is the bonding adhesive. For each value of the tension, the upper point is a Brillouin frequency shift measured while the tension was applied and the lower point is another measurement after the tension was removed. The fiber with one layer had the largest residual strain slope and the fiber with three layers had a smaller slope than the fiber with two layers. These relationships between the Brillouin frequency shift and tension are summarized in Table 1 in detail. The percentage given in parentheses of Table 1 is the value relative to a bare optical fiber without an Al foil layer. We repeated the experiments under the same conditions three times, and their means and standard deviations are included in Table 1. The Brillouin frequency shift caused by the applied tension decreased from 0.038 MHz/με to 0.022 MHz/με as the number of Al foil layers increased. The inverse relationship between the number of Al-foil layers and the frequency shift confirms that Al foil layers having many micro-fold lines partially absorbed the applied tension and transferred the diminished tension to the optical fiber. This happened because the Al-packaged fiber was glued on only one side to the tension stage, not on both sides. As the number of Al foil layers increased, the deviation of the frequency shift increased. When the applied tension was removed, the frequency shift remained from 68% for three Al foil layers up to 84% for one Al foil layer. The deviation was also much smaller with one layer than with three layers. Figure 6 shows three-dimensional graphs of the Brillouin gain as a function of fiber distance and of the frequency difference between the pump and probe for three Al layers (a) and one Al layer (b). The asymmetry of Brillouin gain at both sides of fiber position to the peak frequency confirms that the applied tension was not uniformly transferred along the entire fiber distance. However, there was a large difference in the frequency shift of the peak gain between the two graphs, and this means that their residual strains can be successfully measured and quantitated by using an Al-foil fiber.

 

Fig. 5 The Brillouin frequency shift as a function of applied tension when one (a), two (b) and three (c) Al-foil layers were used for optical fiber packaging.

Download Full Size | PDF

Table 1. The Brillouin frequency shift and residual strains for different applied tensions, where the percentage in parentheses is a relative value to that of a bare optical fiber (without the Al foil layer).

View Table

 

Fig. 6 3D graphs of Brillouin gain as a function of the fiber distance and frequency difference between a pump and a probe in optical fibers with three Al layers (a) and one Al layer (b).

Download Full Size | PDF

As a reliability test, the residual strain was measured in a one-Al-layer fiber that had experienced 5000 με of tensile stress five months ago. A graph of Brillouin frequency shift as a function of fiber distance is shown in Fig. 7(b), where the dotted line shows the measured values on the first day of tensile stress and the solid line shows the measured values of the same fiber five months later. There is no difference in the peak Brillouin frequency shift between them except for a 4% reduction which is in the deviation range of 5.6% shown Table 1. One visible change is that the frequency shift slopes at both edges of 15 cm stress range. The slope is vertical and this means that the stressed area can be completely separated from the non-stressed area. This is confirmed by a 3D graph of Brillouin frequency gain shown in Fig. 7(a). The peak of high gain in the middle area is isolated from both sides and is more similar to a 3D graph for the bare optical fiber shown in Fig. 4(b). This change is estimated to originate because the residual strain difference between the micro-fold lines of the Al foil was balanced for a long time and the strain distribution became more uniform. Hence, this graph closely resembles that of a bare optical fiber, shown in Fig. 4(b), and not that of an Al-foil-packaged fiber on the first fabrication day, shown in Fig. 6(b). The same measurement was conducted again after nine months and the peak Brillouin frequency shift is shown in Fig. 7(c) along with previous measurements. It shows that there is no degradation of the residual strain in an Al-packaged optical fiber after nine months.

 

Fig. 7 3D graphs of Brillouin gain (a) and Brillouin frequency shift measurement (b) in a one-Al-layer fiber after five months and a Brillouin frequency shift change in Al-packaged optical fiber for nine months (c).

Download Full Size | PDF

This Al-packaged optical fiber was applied to a square plate of composite materials to verify the memory effect of an external impact. The plate was composed of carbon-fiber-reinforced polymer (CFRP) and had a structure of [0deg2/+45deg2/-45deg2/90deg2]s. 30 cm of an Al-packed fiber was spliced in the middle region for the entire 4 m of the fiber length and was glued to the composite plate as shown in Fig. 8(a). The plate was strongly clamped to a heavy metal block. A hammer equipped with a pressure sensor was used to give an impact to the plate. The applied impact is shown as a force diagram in Fig. 8(b) and was about 10 J. The impact trace on the surface of the plate could be faintly seen and is depicted by a big arrow in Fig. 8(a), where a red line indicates that an optical fiber is hidden under Al layers. The impact position was chosen to be close to the fiber for definite residual strain. Brillouin frequency shifts were measured before and after the hammer impact. Figure 8(c) shows the difference between measured Brillouin frequency shifts due to the impact. A sharp peak appeared at the impact position of the optical fiber and its positive value suggests that the impact area was expanded for very short impact time. The frequency base line out of the peak area has tiny steps and it came from the fact that the 4 m optical fiber was composed of three slightly different SMF types for the position reference. The magnified peak in the inset of Fig. 8(c) shows 2 cm spatial resolution that matches the value calculated by BOCDA technology. These results together suggest the possibility that Al-packaged optical fiber can detect impact applied to structures in a distributed and quantitative way long after impact.

 

Fig. 8 A plate of composite materials equipped with the Al-packaged optical fiber (a), a force diagram for a hammer (b), the Brillouin frequency shift difference after impact on the plate using a hammer (c).

Download Full Size | PDF

4. Conclusion

Residual strains in Al-packaged optical fibers were quantitatively measured using a BOCDA system after the fibers were subjected to various tensile stresses. In the BOCDA measurement system, a high-power DFB LD with a peak wavelength of 1553.3 nm was directly frequency-modulated with an 8 MHz frequency and 6.2 GHz modulation amplitude. The interaction of a probe and a pump lightwave generated from the LD induced a correlation peak in a specific portion of an optical fiber with 2 cm of spatial resolution. An optical fiber was packaged with thin Al foil with a thickness of 18 um and the Al foil fiber had a flat width of 5 mm. Optical fibers with one, two, and three Al foil layers were compared for their ability to detect Brillouin scattering gain. Tensile stress was applied up to 7000 με and peak frequency shifts of the Brillouin gain spectra showed residual strain for each applied tension. The slope of the Brillouin frequency shift with respect to tensile stress changed from 0.038 MHz/με to 0.022 MHz/με as the number of Al foil layers increased. The relationship between the number of Al-foil layers and the frequency shift confirms that the Al foil layers having many micro-fold lines partially absorbed the applied tension and transferred the diminished tension to the optical fiber. When the applied tension was removed, the frequency shift remained 68% for three Al foil layers up to 84% for one Al foil layer. The deviation was also much smaller with one layer as compared with that with three layers. This residual strain was repeatedly measured up to nine months later as a reliability test, and there was no change except for a 4% reduction, which was within the standard deviation range. Finally, a composite material plate equipped with an Al-packaged optical fiber sensor was impacted with a hammer, and the residual strain in the plate was successfully detected. We suggest that the Al-packaged optical fiber can be adapted as a residual strain sensor to detect impact in a distributed and quantitative way for smart structures, including aerospace structures, even though a long time has elapsed since the impact.