Open Access
Add to BibSonomyAbstract
In this work, we present an all fiber-optics based multi-parameter structure health monitoring system, which is able to monitor strain, temperature, crack and thickness of metal structures. This system is composed of two optical fibers, one for laser-acoustic excitation and the other for acoustic detection. A nano-second 1064 nm pulse laser was used for acoustic excitation and a 2 mm fiber Bragg grating was used to detect the acoustic vibration. The feasibility of this system was demonstrated on an aluminum test piece by the monitoring of the temperature, strain and thickness changes, as well as the appearance of an artificial crack. The multiplexing capability of this system was also preliminarily demonstrated.
© 2016 Optical Society of America
1. Introduction
Active ultrasonic non-destructive evaluation (NDE) has been a widely used technique for many structure health monitoring and material characterization applications. This is usually achieved by the generation of acoustic vibration on a test piece and the analysis of the reflected or transmitted acoustic signal with piezoelectric transducers (PZTs). However, using PZTs brings some drawbacks to the monitoring system including poor survivability in harsh environment, bulky size, complexity in distributed monitoring, high cost and susceptibility to electromagnetic interference (EMI).
Driven by the needs for an ultrasonic NDE system that is more compact, multiplexable, and capable of working in relatively harsh environments, researchers have studied the possibilities of using fiber optic technologies in ultrasonic NDE [1–7]. The basic idea of an optical fiber-based ultrasonic NDE system is using optical fibers to replace the PZTs in the traditional NDE system as broadband ultrasonic sources [1, 4, 5, 8] or broadband detectors [3, 7]. There was one work to replace all PZTs in the ultrasonic NDE system [6]. However, the system was not multiplexable due to the use of an extrinsic Fabry-Perot interferometer (EFPI) type fiber optic sensor. In addition, only qualitative response to artificial crack was demonstrated in that work. To elevate this technology toward industrial applications, in this paper, we present the design, fabrication and demonstration of a multiplexable all fiber-optics based multi-parameter structure health monitoring (AFO-SHM) system.
In our design, an AFO-SHM network is made of cascaded sensing nodes. A single AFO-SHM node is made of an acoustic generation unit and an acoustic detection unit, as shown in Fig. 1. The acoustic generation unit is a micro structure fabricated in a multimode fiber, which serves to scatter light out of the optical fiber. The scattered light generates acoustic vibration on the surrounding material through the thermal-elastic effect. A consistent acoustic emission strength at different nodes can be achieved by the gradual increasing of the scattering ratio of each node. The acoustic detection unit is a fiber Bragg grating (FBG) in a single-mode fiber, which detects the dynamic strain on the optical fiber. The serially connected FBGs in an AFO-SHM network are fabricated with different center wavelengths. To obtain the acoustic signal of a certain sensing node, the wavelength of the probe light will be tuned to the full width half maximum (FWHM) wavelength of the FBG in that node. By switching the probe light wavelength through all the FBGs in the network, multiplexed acoustic detection can be achieved.
It is a common practice in the fiber optics sensing field to detect strain or temperature with an FBG [9]. However, a single FBG cannot be used to detect simultaneously changing temperature and strain due to its cross sensitivity to both parameters [10]. In our system, we solve this cross-sensitivity issue by determining temperature through the measurement of acoustic velocities. The temperature induced FBG spectrum shift is then compensated so the measurement of the FBG shift gives the true value of strain. Besides temperature and strain, other parameters representing the condition of the workpiece can be monitored by the analysis of the acoustic signature changes.
2. System design and operation procedure
The schematic of a multiplexed AFO-SHM system is shown in Fig. 2. The acoustic detection fiber used was a Corning SMF-28e standard fiber, and the acoustic excitation fiber was a 105/125 step-index multimode fiber. The laser used for acoustic excitation was a flash lamp pumped 1064nm Nd:YAG free space laser with 10 Hz repetition rate and 50ns pulse width. The laser pulse energy was 150uJ. An optical sensing interrogator was used for fast spectrum acquisition, and for monitoring the spectra of all FBGs in the sensing chain as shown in Fig. 2(b). After obtaining the system spectrum, the spectrum location of the target FBG in the current cycle can be read. A C-band tunable laser was used as the probe light, the wavelength of which was tuned to the FWHM wavelength of the target FBG. A balanced photodetector was used to measure the fluctuation of the reflected optical power from the target FBG, which is related to the acoustic signal strength. A 50:50 fiber coupler was employed before the photo-detector to provide a reference arm, which eliminated the DC component in the measurement and allowed the detector to work at the highest possible gain. A high-speed oscilloscope with 2 GHz bandwidth was used to record the acoustic signal detected by the photo-detector. The system was switched between the FBG spectrum acquisition cycle and the acoustic signal acquisition cycle by a MEMS optical switch. The parameters of our interest can be monitored by the analysis of the obtained spectrum information and the acoustic signature. All the instruments in the system were controlled and synchronized by a PC through USB and Ethernet ports.
Fig. 2 (a) System schematic of an AFO-SHM system; (b) Sample spectrum of an AFO-SHM system containing 5 nodes.
3. AFO-SHM node installation and acoustic signature analysis
To demonstrate the multi-parameter monitoring capability of this AFO-SHM system, a quarter-inch thick 7075 aluminum plate was used as the device under test (DUT). A single AFO-SHM node was surface-attached to the aluminum plate as shown in Fig. 3(a). The FBG section was attached by an epoxy while the acoustic generation unit was attached with an epoxy-graphite mixture. The cured epoxy was semi-transparent and the added graphite helped increase laser absorption. The micro structure used in this acoustic generation unit was a special splicing point fabricated with a thermal fusion splicer as shown in Fig. 3(b). This structure can be easily fabricated by moving the two fiber ends to different heights while applying arc to splice them, and the light scattering ratio can be controlled by using different offset height. The light scattering ratio on this node was 40%.
Fig. 3 (a) AFO-SHM node installed on the tested aluminum plate; (b) optical fiber special splicing structure; (c) pre-installation test with EFPI fiber sensor and direct laser illumination.
To find the optimized location to mount the AFO-SHM unit on the DUT, two sets of pre-installation tests were performed. In the first set of the pre-installation tests, an EFPI fiber sensor and direct laser illumination were used to generate and record the acoustic signal. Figure 3(c) shows the schematic for this test. A multimode fiber transmitting the excitation laser pulses was positioned to directly illuminate the surface of the DUT and generate pulsed acoustic vibrations. The polished end of a single-mode fiber was placed near the reflective surface of the DUT, forming a Fabry-Perot interferometer (FPI) that monitors the normal displacement of the surface. With this setup, we were able to monitor the acoustic signature of the DUT while changing the locations of the acoustic excitation and detection. In these tests, the acoustic generation point and the acoustic detection point were placed about 20 mm to the nearest edge of the DUT, and the distance between themselves (source-receiver distance) was linearly increased. In the second test set, the FPI in the first test set was replaced with a surface attached FBG. The distance to the nearest edge was also 20 mm.
To better analyze the acoustic signatures obtained in these two sets of experiment, the acoustic velocities for different elastic modes are given in the following Table 1.
Table 1. Propagation speed of the elastic waves in 7075 aluminum
The acoustic signatures obtained in these two test sets are shown in Fig. 4(a) and 4(b) separately. The EFPI sensor only detects the normal surface displacement, making itself less sensitive to body waves directly come from the acoustic source because their particle movement directions are almost parallel to the surface. The FBG detects acoustic wave by its sensitivity to strain, thus is more sensitive to body waves compared to the EFPI sensor. Because of this difference, more peaks can be observed in the second set of acoustic signatures. In Figs. 4(a)-4(b), three distinguishable peaks are found in the acoustic signatures. The locations of these peaks in the time domain are recorded by a peak finding algorithm and used in the following curve fitting formula (Eq. (1)) to determine their propagation velocities and numbers of reflection experienced.
Fig. 4 (a) Acoustic signatures obtained with EFPI sensor and direct laser surface illumination, the shift of each curve represents the source-receiver distance; (b) Acoustic signatures obtained with FBG sensor and direct laser surface illumination, the shift of each curve represents the source-receiver distance; (c) The final acoustic signature of the AFO-SHM unit and its related acoustic signatures from the two pre-installation tests.
In this equation, t is the peak location in time domain, L is the source-receiver distance, D is the distance from the source-receiver line to the presumed reflection surface, V is the velocity of a certain elastic mode, andis a needed term mainly to compensate two possible errors: 1) the error in the recorded source-receiver distance; 2) the difference between the peak location in time domain and the actual travel time of the wavefront of a certain acoustic pulse.
The results of the curve fitting are shown in Table 2. By comparing the curve fitting results to the expected values, these three peaks are identified to be the directly propagating surface Rayleigh wave (referred to as direct R-wave), the side reflected surface Rayleigh wave (referred to as reflected R-wave), the directly propagating pressure wave (referred to as direct P-wave). An “In-direct wave zone” is marked in Fig. 4(a) because it is after the slowest direct wave (direct R-wave). Inside this zone, the area before the fastest possible side reflection wave (side reflected pressure wave) is marked as the “Bottom reflection zone” in Fig. 4(b).
Table 2. Curve fitting results of the three peaks and the expected values
Based on the findings in the pre-installation tests, the final orientation of the AFO-SHM unit on the DUT was set to be 20.5 mm away from the nearest edge and the source-receiver distance was chosen to be 7 mm to obtain sufficient separation between peaks. We found that the reflected R-wave was not very obvious in the second test possibly due to its overlapping with other multi-reflection body waves. By slightly changing the to-edge distance from 20 mm to 20.5 mm, this peak can be recognized in the final acoustic signature. The final AFO-SHM acoustic signature was shown as the green curve in Fig. 4(c), the red curve and the violet curve are the data from the pre-installation tests with 7 mm source-receiver distance and 20 mm to-edge distance. In the following multi-parameter monitoring demonstration, the direct P-wave (referred to as #T1) and the reflected R-wave (referred to as #T2) will be used for temperature monitoring while the largest peak in the “Bottom reflection zone” (peak #B) will be used for thickness monitoring.
4. Multi-parameter monitoring demonstration of a single AFO-SHM node
In this section, we present the monitoring of temperature, strain, thickness of the DUT and initiation of cracks on the 6.35 mm thick aluminum plate.
4.1 Strain monitoring
To apply strain on the sample, the sample was placed on a mechanical vice and the compression force was exerted by the jaws of the vice along the fiber direction. A dial indicator was used to record the relative displacement of the two jaws. We assume that the deformation of the steel vice jaws was small due to its much higher stiffness compared to the thin aluminum plate, so this relative displacement is equal to the compression of the aluminum sample. The ratio of this compression over the total length of the sample was calculated and used as an approximated reference for the strain. The relationship between the calculated strain and the FBG peak shift is showed in Fig. 5.
Fig. 5 (a) FBG spectrum monitoring during the strain change process; (b) FBG peak location shifts caused by the strain changes.
A linear curve fitting of the FBG peak location data gives a strain sensitivity at 0.101 nm/millistrain. The following Eq. (2) shows the theoretical relationship between the FBG peak location shift and the applied strain along the fiber axis [12]. is the FBG peak wavelength which is around 1560 nm in our experiment, n is the effective refractive index of the optical fiber which is about 1.4682 at 1550nm for the Corning SMF-28e we used, is the Poisson’s ratio which is 0.17 for silica, p11 and p12 are the components in the strain-optic tensor which are 0.1 and 0.285 respectively for silica [13].
This calculation yields a theoretical strain sensitivity of about 1.2 nm/millistrain. The difference between this theoretical value and the linear curve fitting result can be used to calculate the strain transfer ratio from the host material to the FBG, which is around 0.101/1.2 = 8.4%. In the prior research [14], the strain transfer ratio was found to be increasing with the increase of the adhesive length and to ranging from 56% to 82% for adhesive length from 5 cm to 15 cm. Considering that the adhesive length is only 3.5 mm in our experiment, we believe 8.4% strain transfer ratio is reasonable.4.2 Thickness monitoring
During the thickness monitoring demonstration, the sample was milled off by about 10% of the initial thickness from the bottom by 4 incremental steps. The acoustic signatures obtained at these thicknesses are shown in Fig. 6. We can see that the temperature indication peak #T1 and #T2 did not shift during the thickness changing process. This is because that these two peaks are acoustic waves traveling along the surface and have no interaction with the bottom of the sample. The thickness indication peak #B, shifted almost linearly during the thickness changes. The total travel length for peak #B can be calculated by , where L is the source-receiver distance, t is the thickness of the sample and N is an integer representing how many rounds the wave gets reflected before it reaches the receiver. When the relative change of the thickness is small, the travel length difference may be approximately governed by Eq. (3), which explains the linear relationship between the peak shift and the thickness change.
Fig. 6 (a) Acoustic signatures in the thickness monitoring test; (b) peak #T1 center shift; (c) peak #B center shift; (d) peak #T2 center shift.
4.3 Temperature monitoring
In the temperature monitoring demonstration, the sample was placed inside a temperature control furnace while the temperature was raised step by step and was stabilized for about 1 hour at each step before a measurement was taken. Figure 7 shows the time domain shifts of the three signature peaks due to the temperature induced acoustic velocity changes. The linear curve fittings give the peak shift versus temperature response rates for these monitored peaks. Using these rates and the temporal locations of these peaks, the acoustic velocity change rates can also be calculated. The results are presented in Table 3.
Fig. 7 (a) Acoustic signatures in the temperature monitoring test; (b) peak #T1 center shift; (c) peak #B center shift; (d) peak #T2 center shift.
Table 3. Peak location associated response rates to temperature change
The FBG spectra were also recorded during the temperature changing process and are shown in Fig. 8(a). The FBG peak locations are shown in Fig. 8(b) and a linear curve fitting shows that its response rate to temperature change is about 13.4 pm/°C in this temperature range.
Fig. 8 (a) FBG spectrum monitoring during the temperature change process; (b) FBG peak location shifts caused by the temperature changes.
4.4 Crack detection
To demonstrate the capability of the system to actively detect a crack, a slot with about 1 mm width, 3 mm depth and 1cm length was machined on the sample 1cm away from the AFO-SHM node as shown in Fig. 9(a). The acoustic signature evolution is shown in Fig. 9(b). Comparing the acoustic signature with/without the crack, it can be seen that several new peaks were generated from the reflections of the crack. In ultrasonic flaw detection, the generally accepted lower limit of detection for a small flaw is one-half wavelength. Because of the short laser pulse width (50 ns) used in the system, the generated acoustic pulse width is in the sub-millimeter range. Therefore, we believe this method should be responsive to cracks with sizes in the millimeter range and above.
Fig. 9 (a) AFO-SHM sample with a machined slot on it; (b) Comparison of the original acoustic signature and acoustic signature with a machined slot on the sample.
To summarize, we have demonstrated the FBG spectrum change and acoustic signature variation for applied strain on the sample, applied temperature change, thickness change and crack. For temperature monitoring, peak #T1 and peak #T2 in the acoustic signature will be used because their peak locations are not affected by the thickness change. Peak #B will be used for thickness monitoring and its temperature induced shift should always be compensated by the shifts of peak #T1 and #T2 according to their temperature response rates. As for strain monitoring, the FBG peak shift will be used after its temperature induced shift being compensated using the current temperature value and its temperature response rate acquired in section 4.3. In future practical applications for real time crack detection, the differential of the acoustic signature should be updated continuously to reveal the appearance of crack-related new peaks.
5. Signal acquisition demonstration of a multiplexed AFO-SHM system
This section presents a preliminary demonstration of the multiplexing capability of the AFO-SHM system. The design of a multiplexed AFO-SHM system has been shown in Fig. 2 previously. Figure 10(a) shows an AFO-SHM network testing unit constructed with two AFO-SHM nodes. The length of the optical fibers connecting these two nodes was about 1 meter. Node #1 was attached to the sample piece with a simulated crack, with an FBG wavelength of 1560 nm. Sample #2 had a similar geometry to sample #1 but had no crack on it, and its FBG wavelength was around 1530 nm. The spectrum containing these two FBG peaks is shown in Fig. 10(b). The optical scattering ratio in node #1 was about 40% and the scattering ratio in node #2 was about 60%. This scattering ratio together with 150injection laser pulse energy resulted in around 55 μJ energy in both nodes so the acoustic signature intensities were close as can be seen in Figs. 9(c) and 9(d).
Fig. 10 (a) AFO-SHM network sample made of two AFO-SHM nodes; (b) The system spectrum consisting of two FBG reflection peaks; (c) Acoustic signature of node #1 which has a crack on the DUT; (d) Acoustic signature of node #2.
6. Conclusion
In conclusion, a multiplexable all-fiber-optics structure health monitoring system was presented. In the demonstration of this system, the temperature, strain, thickness changes of an aluminum sample were quantitatively associated with the FBG spectrum shift and the peak shifts in its acoustic signature. The occurrence of an artificial crack was related to the appearance of new peaks in the acoustic signature. The multiplexing capability was also preliminarily demonstrated by the acquisition of spectrum information and acoustic signatures from an AFO-SHM network containing two cascaded nodes. Despite that the concept of AFO-SHM has been brought up for several years, to our best knowledge, this is the first experimental realization and demonstration of such a system.
The demonstrated temperature range is from 24°C to 60°C, which is the allowed working temperature of the epoxy used in the AFO-SHM node. By changing the epoxy to a high temperature adhesive such as a ceramic adhesive, this system is expected to be able to reach much higher temperatures.
Disclaimer
This Letter was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference hereinto any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
Funding
This work is sponsored by the National Energy Technology Lab (NETL) at the U.S. Department of Energy (DOE) under contract DE-FE0007405.
References and links
1. E. Biagi, M. Brenci, S. Fontani, L. Masotti, and M. Pieraccini, “Photoacoustic generation: Optical fiber ultrasonic sources for non-destructive evaluation and clinical diagnosis,” Opt. Rev. 4(4), 481–483 (1997). [CrossRef]
2. P. A. Fomitchov, A. K. Kromine, and S. Krishnaswamy, “Photoacoustic probes for nondestructive testing and biomedical applications,” Appl. Opt. 41(22), 4451–4459 (2002). [CrossRef] [PubMed]
3. E. Biagi, S. Cerbai, P. Gambacciani, and L. Masotti, “Fiber optic broadband ultrasonic probe,” in Sensors, 2008 IEEE(2008), pp. 363–366.
4. J. Tian, Q. Zhang, and M. Han, “Distributed fiber-optic laser-ultrasound generation based on ghost-mode of tilted fiber Bragg gratings,” Opt. Express 21(5), 6109–6114 (2013). [CrossRef] [PubMed]
5. X. Zou, N. Wu, Y. Tian, and X. Wang, “Broadband miniature fiber optic ultrasound generator,” Opt. Express 22(15), 18119–18127 (2014). [CrossRef] [PubMed]
6. V. Kochergin, K. Flanagan, Z. Shi, M. Pedrick, B. Baldwin, T. Plaisted, B. Yellampelle, E. Kochergin, and L. Vicari, “All-fiber optic ultrasonic structural health monitoring system,” in SPIE Smart Structures and Materials + Nondestructive Evaluation and Health Monitoring(International Society for Optics and Photonics, 2009), pp. 72923D–72928.
7. H. Guo, G. Xiao, N. Mrad, and J. Yao, “Fiber optic sensors for structural health monitoring of air platforms,” Sensors (Basel) 12(4), 3687–3705 (2011). [CrossRef] [PubMed]
8. E. Biagi, F. Margheri, and D. Menichelli, “Efficient laser-ultrasound generation by using heavily absorbing films as targets,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 48(6), 1669–1680 (2001). [CrossRef] [PubMed]
9. R. Kashyap, Fiber Bragg Gratings (Academic, 1999).
10. Y. Zhao and Y. Liao, “Discrimination methods and demodulation techniques for fiber Bragg grating sensors,” Opt. Lasers Eng. 41(1), 1–18 (2004). [CrossRef]
11. A. D. McKie, J. W. Wagner, J. B. Spicer, and J. B. Deaton Jr., “Dual-beam interferometer for the accurate determination of surface-wave velocity,” Appl. Opt. 30(28), 4034–4039 (1991). [CrossRef] [PubMed]
12. G. Wild and S. Hinckley, “Acousto-ultrasonic optical fiber sensors: Overview and state-of-the-art,” IEEE Sens. J. 8(7), 1184–1193 (2008). [CrossRef]
13. K. Vedam, “The elastic and photoelastic constants of fused quartz,” Phys. Rev. 78(4), 472–473 (1950). [CrossRef]
14. S.-C. Her and C.-Y. Huang, “The effects of adhesive and bonding length on the strain transfer of optical fiber sensors,” Appl. Sci. 6(1), 13 (2016). [CrossRef]
