sensors Article
Fibre Bragg Gratings in Embedded Microstructured Optical Fibres Allow Distinguishing between Symmetric and Anti-Symmetric Lamb Waves in Carbon Fibre Reinforced Composites Ben De Pauw 1,2, *, Sidney Goossens 1,2 , Thomas Geernaert 1,2 , Dimitrios Habas 3 , Hugo Thienpont 1,2 and Francis Berghmans 1,2 1
2 3
*
Department of Applied Physics and Photonics (TONA), Vrije Universiteit Brussel, Brussels Photonics (B-PHOT), Pleinlaan 2, 1050 Brussels, Belgium; [email protected] (S.G.); [email protected] (T.G.); [email protected] (H.T.); [email protected] (F.B.) Flanders Make, Oude Diestersebaan 133, 3920 Lommel, Belgium Hellenic Aerospace Industry, Engineering Research Design and Development Directorate, 32009 Shimatari, Greece; [email protected] Correspondence: [email protected]; Tel.: +32-2-629-1381
Received: 31 July 2017; Accepted: 19 August 2017; Published: 24 August 2017
Abstract: Conventional contact sensors used for Lamb wave-based ultrasonic inspection, such as piezo-electric transducers, measure omnidirectional strain and do not allow distinguishing between fundamental symmetric and anti-symmetric modes. In this paper, we show that the use of a single fibre Bragg grating created in a dedicated microstructured optical fibre allows one to directly make the distinction between these fundamental Lamb wave modes. This feature stems from the different sensitivities of the microstructured fibre to axial and transverse strain. We fabricated carbon fibre-reinforced polymer panels equipped with embedded microstructured optical fibre sensors and experimentally demonstrated the strain waves associated with the propagating Lamb waves in both the axial and transverse directions of the optical fibre. Keywords: fibre Bragg gratings; Lamb waves; CFRP; impact damages
1. Introduction Lamb wave-based ultrasonic inspection for damage identification in engineering structures is considered to be an attractive alternative for conventional and time-consuming point-by-point ultrasonic inspection techniques, mainly because Lamb waves can travel long distances across the structure and therefore allow for inspecting fairly large areas. Such Lamb waves can be excited and detected in various manners depending on the size and material of the structure under test [1–3]. In the aerospace industry, for example, the most common technique to both excite and detect Lamb waves in plate-like composite components uses piezo-electric transducers (PZTs). Such transducers feature a small size and low power consumption. The main disadvantages of PZTs in Lamb wave-based inspection are twofold. First, each PZT requires being individually connected, which calls for dealing with a multitude of wires when using arrays of transducers. Second, PZTs exhibit unselective sensitivity in the sense that Lamb waves propagating along different directions in the plane of a panel are all picked up with the same sensitivity. Critical directions, corresponding to, for example, a defect in the panel, are recorded simultaneously and with the same sensitivity as every other direction. It is thus challenging to identify the result of a defect in a panel on Lamb waves propagating in this structure and, therefore, it is difficult to locate and assess the size of the defect. Optical fibre Bragg gratings (FBGs) have already been proposed to detect Lamb waves in order to cope with the aforementioned Sensors 2017, 17, 1948; doi:10.3390/s17091948
www.mdpi.com/journal/sensors
Sensors 2017, 17, 1948
2 of 9
shortcomings [3–6]. Multiple FBGs can easily be multiplexed on a single optical fibre, which drastically decreases the amount of wiring when using an array of such sensors. Furthermore, and owing to the geometry of the FBG, the response to Lamb waves depends on the angle between the fibre and the Lamb wave’s propagation direction. Finally, the small size of the optical fibre allows these sensors to be embedded between the laminar layers of a composite structure in a minimally intrusive manner, avoiding the need to attach bulky and disturbing sensors to the surface of a structure [7]. Lamb waves can be categorised into two groups based on the associated propagating perturbation of the composite with respect to the mid-plane of a panel, i.e., in symmetric and anti-symmetric modes. These modes are denoted as Si and Ai , respectively, where i refers to the order of the mode. The number of symmetric and anti-symmetric modes that can propagate along a plate-like structure under test depends on the excitation frequency [3,8]. Owing to their low dispersion and attenuation, the fundamental symmetric S0 and anti-symmetric A0 modes are frequently considered suitable for inspecting the structure for the presence of defects. Since they have different mode shapes, S0 and A0 modes are sensitive to different defects in the structure: the S0 wave is extensional in nature and thus more sensitive to defects inside the structure under test, whilst the A0 is flexural and thus more sensitive to surface cracks, as an example. Both modes also have a different frequency-dependent propagation velocities. Consequently, both types of waves are of interest for identifying and locating damage in composites. However, properly distinguishing between A0 and S0 waves in the sensor responses remains very difficult in ultrasonic inspection using both conventional FBGs and PZTs [3,9–11]. In this paper, we demonstrate that a sensor consisting of a single FBG created in a specialty microstructured optical fibre (so-called MOFBG) can be used to discern between the A0 and S0 waves. To do so, we used MOFBG sensors embedded within the composite panel to measure the Lamb waves, while until now mostly surface mounted FBGs have been used for this purpose. Our paper is structured as follows. Section 2 describes the unique measurement strategy that MOFBG sensors can offer to record Lamb waves. We discuss how the mounting procedure and experimental setup help to distinguish between the A0 and S0 waves. The subsequent Section 3 deals with the experimental results obtained and discusses how these can be applied when detecting impact damage using Lamb waves. Section 4 closes our paper with a summary and conclusions. 2. MOFBG Sensors and Measurement Strategy MOFBGs can exhibit peculiar characteristics that cannot be achieved using conventional optical fibre technology. More specifically, the design flexibility of such microstructured fibres (MOFs) allows for the development of sensors that exhibit selective sensitivities to e.g., axial strain, transverse strain, or even shear strain, whilst being negligibly cross-sensitive to temperature changes. Examples of MOF-based sensors have been reviewed in several research papers [12–20]. In this paper, we use a specific microstructured fibre design which we refer to as a ‘butterfly’ MOF [17–20]. Figure 1a shows an SEM image of the cross-section of the butterfly MOF [18]. The peculiar microstructure induces a high level of birefringence in the optical fibre. When this MOF is equipped with an FBG, the latter reflects two Bragg wavelengths corresponding to the two orthogonal polarised propagation modes. The Bragg wavelengths (or peaks) will shift with the same amount when an axial load is applied to the optical fibre, but will move in opposite ways when a transverse load (parallel to axis ‘S’) is applied. The absolute and relative position of the Bragg peaks therefore encodes axial strain as well as transverse strain, respectively, into the reflection spectrum of the MOFBG. This principle is summarised in Figure 1b.
Sensors 2017, 17, 1948
3 of 9
Sensors 2017, 17, 1948
3 of 9
Sensors 2017, 17, 1948
3 of 9
Figure 1. (a) SEM image of the cross-section of the fabricated butterfly microstructured fibre (MOF);
Figure 1. (a) SEM image of the cross-section of the fabricated butterfly microstructured fibre (MOF); (b) Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to (b) Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to both peaks moving in the same way, whilst transverse strain (applied parallel to axis ‘S’) corresponds both Figure peaks moving inimage the same way, whilst transverse strain (applied parallel to axis fibre ‘S’) corresponds 1. (a) moving SEM of the cross-section of the fabricated butterfly microstructured (MOF); to the peaks in opposite directions [18]. to the(b) peaks moving in opposite directions [18]. Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to both moving in the same whilst transverse strainand (applied parallelstrain to axis can ‘S’) corresponds The peaks selective sensitivity of away, single MOFBG to axial transverse be exploited to to the peaks moving in opposite directions [18]. differentiate between the S0 and A0 modes. This stems from the nature of the induced strain
The selective sensitivity of a single MOFBG to axial and transverse strain can be exploited to distribution (or strainSmode shape) in the axial and from transverse directions ofinduced a plate-like structure, differentiate between This stems the nature of the distribution 0 and A The selectivethe sensitivity of0 amodes. single to axial and transverse strain canassociated bestrain exploited to which is different for these modes. OwingMOFBG to the anti-symmetric shape of the A0, the strain (or strain mode shape) in the axial and transverse directions of a plate-like structure, which is different differentiate between the S 0 and A0 modes. This stems from the nature of the induced strain amplitude in the transverse direction is close to zero near the mid-plane of the panel [21]. The for these modes. Owing tomode the anti-symmetric shape the A0 , the associated amplitude in the distribution in the axial and of transverse directions a strain plate-like structure, symmetry of(or thestrain S0 mode, onshape) the other hand, results in an increase of the of strain amplitude at that transverse direction is close to zero near the mid-plane of the panel [21]. The symmetry of the S mode, which is different for these modes. Owing to the anti-symmetric shape of the A 0, the associated strain 0 location. Consequently, an MOFBG embedded for example at a depth of ¼ (or equivalently ¾) of the amplitude the transverse to zero nearthe the ofinthe [21]. The an on the other hand, results in andirection increase ofclose the strain at that location. Consequently, thickness ofina plate-like structure will beisable to pick up amplitude both S0 mid-plane and A0 mode thepanel axial direction, 3 of the strain amplitude at that symmetry of Smode 0 mode, on transverse the hand, in an increase ⁄4 (orasequivalently ⁄4) of2,the MOFBG embedded example at aother depth of 1results thickness of atypical plate-like but mostly thethe S0for in the direction, depicted in Figure which illustrates location. Consequently, an MOFBG embedded for example at a depth of or ¼ (or equivalently ¾) ofSthe strainwill mode the up lateral of a A plate-like structure. At ¼ ¾ of the thickness, the A 0 structure be shapes able toin pick bothdirection the S0 and mode in the axial direction, but mostly the 0 0 mode thickness of a plate-like structure will be able to pick uptransverse both the S0 strain and A0components, mode in the axial direction, wave as well as the S 0 wave will still exhibit some but given the in the transverse direction, as depicted in Figure 2, which illustrates typical strain mode shapes in but mostly thethe S0axial mode in the transverse direction, as 3depicted in Figure 2, which illustrates typical 1 contrast with strain components, the At MOFBG distinguishing between the two ⁄4 or allows ⁄4 of the the lateral direction of a plate-like structure. thickness, the A0 wave asparties. well as the strain mode shapes in the lateral direction of a plate-like structure. At ¼ or ¾ of the thickness, the A0 Notewill thatstill the exhibit actual measured strain amplitude of both the S0 and A 0 modes also depends on the S0 wave some transverse strain components, but given the contrast with the axial wave as well as the S 0 wave will still exhibit some transverse strain components, but given the transfer of the strain induced in the material by the Lamb waves to the optical fibre and on the straincontrast components, MOFBG allows distinguishing between the two parties. Note that the actual with thethe axial strain components, the MOFBG excitation amplitude and attenuation of each mode. allows distinguishing between the two parties. measured strain of both the amplitude S0 and A0ofmodes also depends on the the Note that theamplitude actual measured strain both the S0 and A0 modes alsotransfer dependsofon thestrain transfer of material the strainby induced in the material by the Lamb waves fibre and on the and induced in the the Lamb waves to the optical fibre andtoonthe theoptical excitation amplitude excitation and attenuation of each mode. attenuation of amplitude each mode.
Figure 2. Illustration of strain mode shapes of both S0 and A0 waves in the lateral direction of a platelike structure. The strain induced by an A0 wave close to the mid-thickness tends to be zero, in contrast to the induced strain of the S0 wave, which increases at the same depth. An MOFBG embedded at a Figure 2. Illustration of ¼ strain mode shapes of both S0 and A0 waves theaxial lateral direction of a strain, platedepth of, for example, of the panel thickness, that isSsensitive to in both and transverse Figure 2. Illustration of strain mode shapes of both 0 and A0 waves in the lateral direction of a like structure. The strain induced by an A 0 wave close to the mid-thickness tends to be zero, in contrast A0 modes can exploit this difference to distinguish between S0 and plate-like structure. The strain induced by an A0the wave close to the [18,21]. mid-thickness tends to be zero, at the same depth. An MOFBG embedded at a to the induced strain of the S0 wave, which increases in contrast to the induced strain of the S wave, which increases at the same depth. An MOFBG 0 depth of, for example, ¼ of the panel thickness, that is sensitive to both axial and transverse strain, 1 embedded at a depth of, for example, ⁄ 4 of the panel thickness, that is sensitive to both axial and can exploit this difference to distinguish between the S0 and A0 modes [18,21].
transverse strain, can exploit this difference to distinguish between the S0 and A0 modes [18,21].
Sensors 2017, 17, 1948
4 of 9
To demonstrate the capability of an MOFBG to distinguish between S0 and A0 modes using the To demonstrate the capability of anaMOFBG between S0 and A0 modes using the approach outlined above, we fabricated 250 mm to bydistinguish 25 mm carbon fibre-reinforced polymer (CFRP) approach outlined above, we fabricated a 250 mm by 25 mm carbon fibre-reinforced polymer (CFRP) panel made from Hexcel M21/T800S prepreg, with a lay-up of (45/−45/02/90/0)s. The thickness of each panel Hexcel withthickness a lay-up of of 2.21 (45/− 45/0We of s . The thickness 2 /90/0) ply is made close from to 0.184 mm,M21/T800S resulting inprepreg, an overall mm. embedded an MOFBG each ply is close to 0.184 mm, resulting in an overall thickness of 2.21 mm. We embedded an MOFBG between the third and fourth laminar layers (i.e., at ¼ of the thickness of the panel) and aligned the 1 between the with thirdthe and0°fourth laminar layers (i.e., thickness of thefibres panel) aligned optical fibre orientation of the plies (i.e.,atparallel the carbon in and the plane of the the 4 of thewith ◦ optical fibre with the 0 orientation of the plies (i.e., parallel with the carbon fibres in the plane of the laminate plies) to minimize the effect of the embedding and, in particular, to avoid the creation of laminate plies) to We minimize the effect of the embedding and, inofparticular, avoid the creation of resin pockets [7]. positioned the MOFBG at the half-length the panel to and angularly oriented resin pocketssuch [7]. We the MOFBG at the half-length of theto panel and angularly oriented the MOFBG thatpositioned the peak separation was essentially sensitive the out-of-plane transverse the MOFBG such that the peak separation was essentially sensitive to the out-of-plane transverse load. To do so, we rotated the fibre about its axis (i.e., changed the roll angle) so that the butterfly load. To do so, rotated the fibre about its axis (i.e.,the changed roll so that theTo butterfly axis (referred towe as ‘S’ in Figure 1a) was aligned with desiredthe axis of angle) the CFRP panel. obtain axis (referred to as ‘S’ in Figure 1a) was aligned with the desired axis of the CFRP panel. To obtain the desired orientation of the microstructured optical fibre, we relied on our technique described in the desired[18], orientation of theangular microstructured optical we relied on fibre our technique described Reference which yields accuracies of thefibre, roll angle of the with respect to the in Reference [18], which yields angular accuracies of the roll angle of the fibre with respect to roll the instrumented structure close to ±8°. We have previously determined that such a tolerance on the ◦ instrumented structure close toto±loads 8 . Wealong havethe previously determined a tolerance on the angle guarantees a sensitivity desired CFRP panel that axis such of well beyond 85% of roll the angle guarantees a sensitivity to loads along the desired CFRP panel axis of well beyond 85% the sensitivity achieved when the fibre’s microstructure is perfectly aligned along the desired axisof[19]. sensitivity whenfor thethe fibre’s microstructure is perfectly axis [19]. This is not achieved only sufficient demonstration purpose reportedaligned in this along paper,the butdesired also allows one This is not only sufficient for the demonstration purpose reported in this paper, but also allows one to to envisage its implementation in practical applications. To excite the Lamb waves in the fabricated envisage its implementation in practical applications. To excite the Lamb waves in the fabricated panel, panel, we surface-mounted a disk-shaped PZT of type PIC255 (PI Ceramic GmbH, Lederhose, we surface-mounted a disk-shaped PZT ofextremity. type PIC255 Ceramicremark GmbH,here Lederhose, Germany) Germany) with a diameter of 10 mm on one One(PI important is that the distance with a diameter of 10 mm on one extremity. One important remark here is that the distance between the PZT and the MOFBG is only 10 cm, and is thus typically insufficient for the Abetween 0 and S0 the PZTtoand theseparate MOFBGowing is onlyto10 cm,different and is thus typically insufficient forthe theused A0 and S0 modes to modes fully their propagation velocities (in material). More fully separate owing to their different propagation velocities (in the used material). More specifically, specifically, we observed from the experimentally and numerically determined dispersion curves of we from experimentally and determined ourfrequencysetup that ourobserved setup that thethe delay between the A0numerically and S0 waves was onlydispersion about 18 curves µs. Forofthe the delay between the A S0 waves was only 18 µs.ofFor 0 and the thickness used in this paper, measured phaseabout velocities thethe S0 frequency-thickness and A0 waves were used closein tothis 4.5 paper, the measured phase velocities of the S and A waves were close to 4.5 m/ms and 2.8 m/ms, 0 0 m/ms and 2.8 m/ms, respectively. A schematic of the experimental setup is depicted in Figure 3. We respectively. A PZT schematic the experimental setup is depicted infive-cycle Figure 3. We controlled theaPZT using controlled the usingof a waveform generator that generates tonebursts with Hanning a waveform that generatesThe five-cycle tonebursts withto a Hanning filter applied at a preset filter appliedgenerator at a preset frequency. response of the FBG the impinging Lamb waves was frequency. The response of the FBG to the impinging Lamb waves was measured using the measured using the so-called edge filtering method [22], in which the output wavelength of aso-called tunable edge [22], in which the output wavelength of a tunable is tuned to theofwavelength laser filtering is tunedmethod to the wavelength where the slope of the Bragg peak islaser steepest. A shift the Bragg where the slope of the Bragg peak is steepest. A shift of the Bragg peak resulting from a passing peak resulting from a passing strain wave is then recorded as a change of optical intensity by strain a fast wave is then recorded asPDA20CS). a change ofNote optical by a fast photodetector photodetector (Thorlabs thatintensity the birefringence in the MOFBG(Thorlabs producesPDA20CS). two Bragg Note birefringence in the two(two Bragg peaks (see 1), meaningcan thatbe a peaksthat (seethe Figure 1), meaning thatMOFBG a total ofproduces four slopes ascending andFigure two descending) total of four slopes (two ascending and two descending) can be used. used.
Figure 3. A setup. Figure 3. A schematic schematic of of the the experimental experimental setup.
3. Experimental Results 3. Experimental Results 3.1. MOFBGs between A A0 and S0 Waves 3.1. MOFBGs Allow Allow Distinguishing Distinguishing between 0 and S0 Waves Using the the setup setup described described above, above, we we excited excited the the fundamental fundamental SS0 and and A A0 modes in the fabricated Using 0 0 modes in the fabricated panel by actuating the PZT with a 20 Vpp five-cycle toneburst at 250 kHz with the modulated amplitude panel by actuating the PZT with a 20 Vpp five-cycle toneburst at 250 kHz with the amplitude modulated using a Hanning filter, and we measured the response with the embedded MOFBG. To
Sensors 2017, 17, 1948
5 of 9
2017, 17, 1948 5 of 9 the usingSensors a Hanning filter, and we measured the response with the embedded MOFBG. To improve signal-to-noise ratio, each measurement was repeated 1000 times and averaged. The measured improve the signal-to-noise ratio, each measurement was repeated 1000 times and averaged. The wavelength shifts as a function of time for the four slopes of the MOFBG are shown in Figure 4a. measured wavelength shifts as a function of time for the four slopes of the MOFBG are shown in The average of the shift fourofmeasurements is proportional to thetoaxial strain applied Figure 4a.wavelength The average shift wavelength the four measurements is proportional the axial strain to theapplied MOFBG, while the shift in spectral distance between the first and third slopes or between to the MOFBG, while the shift in spectral distance between the first and third slopes or the second and fourth slopes isfourth proportional the transverse applied to applied the MOFBG. The resulting between the second and slopes isto proportional to thestrain transverse strain to the MOFBG. wavelength shiftswavelength corresponding to the axial (in transverse (in red) (in directions are shown The resulting shifts corresponding to black) the axialand (in black) and transverse red) directions are shown in Figure 4b. Although the sensitivity in the transverse direction is not as high as that in axial in Figure 4b. Although the sensitivity in the transverse direction is not as high as that in the the axial direction [18–20], we observed a clear difference in the obtained signals. In particular, the direction [18–20], we observed a clear difference in the obtained signals. In particular, the signal signal in measured the axialproved direction be more responsive to an impinging 0 wave measured the axialindirection to proved be moretoresponsive to an impinging A0 waveA(because the (because the PZT excitation of A0 is more pronounced), while the signal measured in the transverse PZT excitation of A0 is more pronounced), while the signal measured in the transverse direction was direction was more responsive to the S0 wave. more responsive to the S0 wave.
Figure 4. (a) The measured wavelength shifts for the four slopes of the MOFBG embedded in the
Figure 4. (a) The measured wavelength shifts for the four slopes of the MOFBG embedded in the carbon carbon fibre reinforced polymer panel versus time; (b) Based on the properties of the MOFBG, the fibre reinforced polymer panel versus time; (b) Based on the properties of the MOFBG, the measured measured signals of the four slopes have been converted into axial and transverse wavelength shifts signals of the four slopes have been converted into axial and transverse wavelength shifts that are that are proportional to the respective strains applied to the MOFBG. proportional to the respective strains applied to the MOFBG.
The identification of both types of waves is confirmed by the time-of-arrival of both the wave packets that arrive first welltypes as theof wave packets returning after a reflection at the extremity thewave The identification of as both waves is confirmed by the time-of-arrival of bothofthe panel (see also Section 3.3). The results therefore evidence that a single embedded MOFBG, sensitive packets that arrive first as well as the wave packets returning after a reflection at the extremity of to both axial and transverse strains, allows distinguishing between the A0 and S0 Lamb waves the panel (see also Section 3.3). The results therefore evidence that a single embedded MOFBG, propagating through the CFRP panel.
sensitive to both axial and transverse strains, allows distinguishing between the A0 and S0 Lamb waves propagating through the CFRPAllow panel. 3.2. Selection of Preferential Directions Filtering out Undesired Edge Reflections The of directional sensitivity of Allow the MOFBG canout bring an additional advantage to Lamb wave3.2. Selection Preferential Directions Filtering Undesired Edge Reflections
based ultrasonic inspection. In a geometry similar to the CFRP panel explained in the previous The directional sensitivity of the MOFBG can bring an advantage to LambMOFBG wave-based section, the response of a PZT sensor surface mounted atadditional the position of the embedded ultrasonic inspection. In a geometry similar to the CFRP panel explained in the previous section, would be significantly influenced by the multitude of reflections. Such a situation would make it very challenging thesurface A0 and Smounted 0 waves from PZTposition recordings, a PZT is sensitive to all the response oftoa distinguish PZT sensor at the of since the embedded MOFBG would incident directions and the Lamb waves arrive at the Such sensoraquasi-simultaneous. The key be significantly influenced byreflected the multitude of reflections. situation would make it very feature of is that to two perpendicular (one axial to all challenging toMOFBGs distinguish thethey A0 are andessentially S0 wavessensitive from PZT recordings, since directions a PZT is sensitive and one transverse to the optical fibre), which allowed us to choose, by carefully angularly orienting incident directions and the reflected Lamb waves arrive at the sensor quasi-simultaneous. The key the MOFBG, to filter out part of the edge reflections. To demonstrate this, we fabricated a second feature of MOFBGs is that they are essentially sensitive to two perpendicular directions (one axial and CFRP panel with the same geometry as before and with the MOFBG embedded at the same location. one transverse tohowever, the optical which allowed us to choose, by carefully In this panel, we fibre), angularly oriented the MOFBG (i.e., adjusted the roll angularly angle of theorienting fibre to the MOFBG, to filter out part of the edge reflections. To demonstrate this, we fabricated a second align the axis ‘S’ of the butterfly structure) so that to the peak separation is essentially sensitive to theCFRP paneltransverse with the in-plane same geometry before the andmeasurement with the MOFBG embedded at the same location. load. Weas repeated procedure and we separated again the axialIn this panel, however, we angularly the MOFBG (i.e., adjusted roll angle of the fibre to The align the load and transverse loads, oriented as performed for the previous panel.the Figure 5 shows results.
axis ‘S’ of the butterfly structure) so that to the peak separation is essentially sensitive to the transverse in-plane load. We repeated the measurement procedure and we separated again the axial load and
Sensors 2017, 17, 1948
6 of 9
transverse loads, as performed for the previous panel. Figure 5 shows the results. The reflections Sensors 2017, 17, 1948 6 of 9 from the long edges are more clearly observed than in the previous composite panel, as shown in reflections from the long edgeson area more clearly in the previous composite panel, a asmore Figure 5a, using a measurement single slope.observed Similarthan to the previous panel, we noticed shown in Figure 5a, using a measurement on a single slope. Similar to the previous panel, we noticed significant response to the S0 wave than to the A0 mode in the transverse direction and vice versa in a more significant response to the S0 wave than to the A0 mode in the transverse direction and vice the axial direction. Therefore, carefully aligning the butterfly structure by adjusting the roll angle of versa in the axial direction. Therefore, carefully aligning the butterfly structure by adjusting the roll the MOFBG during mounting allows for the selection or elimination of certain directions of strain and, angle of the MOFBG during mounting allows for the selection or elimination of certain directions of hence, potentially accounts for undesired effects resulting from Lamb wave reflections the edges strain and, hence, potentially accounts for undesired effects resulting from Lamb wave from reflections of a panel under test. from the edges of a panel under test.
Figure 5. (a) A measurement of a single slope of a MOFBG in a similar CFRP panel, where the MOFBG Figure 5. (a) A measurement of a single slope of a MOFBG in a similar CFRP panel, where the MOFBG is oriented to be sensitive to axial load and transverse in-plane load; (b) The measured responses of is oriented to be sensitive to axial load and transverse in-plane load; (b) The measured responses of the the MOFBG converted into axial and transverse loads. MOFBG converted into axial and transverse loads.
3.3. Consequences for Impact Damage Detection in CFRP
3.3. Consequences for Impact Damage Detection in CFRP
The demonstrated ability to resolve the two components of Lamb waves is also beneficial when
Lambdemonstrated waves are used to detect damagethe in two a structure. In most involving The ability to resolve components of experiments Lamb waves is also Lamb beneficial wave-based damage detection, the signals of an undamaged structure are compared to the signals when Lamb waves are used to detect damage in a structure. In most experiments involvingofLamb the damaged structure [23–25].the Damage in the structure will cause propagating wavesto to the reflect wave-based damage detection, signals of an undamaged structure are Lamb compared signals and refract, yielding an alteration in the recorded Lamb waves. In general, one can distinguish of the damaged structure [23–25]. Damage in the structure will cause propagating Lamb waves to between two situations. reflect and refract, yielding an alteration in the recorded Lamb waves. In general, one can distinguish First, the damage can be on the direct path between the actuator and the sensor. In this case, the between two situations. fraction of the Lamb waves that are influenced (because of reflection, refraction, attenuation, etc.) will First, theenergy damage can beinon direct path between theup actuator and the In this case, alter the contained thethe wave packets that are picked by the sensor [25].sensor. As an example the fraction of the Lamb waves that are influenced (because of reflection, refraction, attenuation, of such a situation, we experimentally mimicked an impact damage by applying a dot of synthetic etc.) will alter the energywith contained in of theabout wave packets are picked up byinthe sensor rubber adhesive a diameter 5 mm to thethat CFRP panel described Section 3.1, [25]. and weAs an recorded Lamb waves withwe the experimentally embedded MOFBG. We showan theimpact result indamage Figure 6, by in which we have example of such a situation, mimicked applying a dot of plotted the wavelength shiftsa corresponding to strains into the axial andpanel transverse directions. The 3.1, synthetic rubber adhesive with diameter of about 5 mm the CFRP described in Section dashed lines Lamb in the waves figure indicate pristine situation, the solid lines correspond toin the and we recorded with thethe embedded MOFBG.while We show the result in Figure 6, which damaged situation. Since the simulated damage is located on the direct path between the PZT we have plotted the wavelength shifts corresponding to strains in the axial and transverse directions. actuator and the MOFBG sensor, a fraction of the Lamb waves will be attenuated, yielding less The dashed lines in the figure indicate the pristine situation, while the solid lines correspond to the energetic strain waves to be transmitted and thus smaller recorded wavelength shifts. The latter can damaged situation. Since damage isdirections located on the the direct PZT be clearly seen in boththe thesimulated axial and transverse when firstpath wavebetween packets the arrive at actuator the and the MOFBG sensor, a fraction of the Lamb waves will be attenuated, yielding less energetic MOFBG. In addition, due to the additional attenuation which is possibly accompanied with partialstrain waves to beconversion transmitted andAthus recorded shifts. The latter can be amplitude clearly seen in mode from 0 to Ssmaller 0, the second wavewavelength packet shows a significantly larger both compared the axial and transverse directions when the first wave packets arrive atsituation, the MOFBG. In addition, to the undamaged state in the transverse direction. In the described we observed anthe increase of the amplitude of which the second wave packet of more than a factor ofmode 2. Recall, however, from due to additional attenuation is possibly accompanied with partial conversion for the used wave setup packet the excitation A0 mode islarger more pronounced compared to 0 mode. A0 tothat S0 , the second showsofa the significantly amplitude compared tothe theSundamaged Nevertheless, our approach illustrates the potential benefits of using MOFBGs as sensors in Lamb state in the transverse direction. In the described situation, we observed an increase of the amplitude of the second wave packet of more than a factor of 2. Recall, however, that for the used setup the excitation of the A0 mode is more pronounced compared to the S0 mode. Nevertheless, our approach
Sensors 2017, 17, 1948 Sensors 2017, 17, 1948
7 of 9 7 of 9
illustrates the17,potential benefits of using as sensors in between Lamb wave-based Sensors 2017, 1948 of 9 wave-based damage detection owing toMOFBGs their ability to discern the strains damage recordeddetection in7 the owing to their ability to discern between the strains recorded in the axial and transverse directions. axial and transverse directions. wave-based damage detection owing to their ability to discern between the strains recorded in the axial and transverse directions.
Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged situation for the sample described in Section 3.1. situation for the sample described in Section 3.1. Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged situation sample can described Section 3.1. path between the actuator and sensors. In such a Second, for thethe damage be offin the direct
Second,Lamb the damage be off the direct path between thewave actuator and sensors. a situation, situation, waves can altered from the damage will yield packets arriving In at such the sensor at Second, the damage can be off the direct path between the actuator and sensors. In suchtimes. a Lamb wavestimes. altered the damage willtoyield wave packets arriving at of thewave sensor at different different In from this case, the ability determine the time-of-arrival packets is crucial situation, Lamb waves altered from the damage will yield wave packets arriving at the sensor at In [23,24]. this case, thetoability to determine the of time-of-arrival wave packets is crucialdetection, [23,24]. we Thus, Thus, illustrate the potential the MOFBGs of in time-of-arrival damage different In this case, theMOFBGs ability to in determine theLamb time-of-arrival of waveinpackets is crucial to experimentally illustratetimes. the potential of the time-of-arrival damage detection, we determined the dispersion curve of the waves generated theexperimentally CFRP panel [23,24].also Thus, to on illustrate the potential of of the in time-of-arrival damage detection, weon which relies accurate measurements theMOFBGs time-of-arrival wavepanel packets. Recall the determined the dispersion curve of the Lamb waves generated in of thethe CFRP which alsothat relies experimentally determined the dispersion curve of the Lamb waves generated in the CFRP panel distance between the actuator and sensor is only cm, which is typically insufficient for the A0 andthe accurate measurements of the time-of-arrival of the10wave packets. Recall that the distance between which also on accurate measurements of the time-of-arrival ofdots the wave packets. phase Recall velocity that the S 0 waves torelies separate. The result shownisintypically Figure 7,insufficient in which thefor actuator and sensor is only 10 cm,iswhich therepresent A0 and Sthe waves to separate. 0 distance between the observed actuator and sensor is only 10 cm, which is typically insufficient for the A0 and determined from the time-of-arrival for a certain excitation frequency. We determined The result is shown in Figure 7, in which the dots represent the phase velocity determined the from S0 waves to separate. is shown in as Figure in the dots wave represent the The phase velocity of bothThe theresult first packet well 7, as thewhich first reflected packet. solid lines of thetime-of-arrival observed time-of-arrival for wave a certain excitation frequency. We determined the time-of-arrival determined from time-of-arrival for a certain excitation frequency. We determined the in Figure showthe theobserved dispersion calculated material and the panel both the first7wave packet as well ascurves the first reflectedusing wavethe packet. Theparameters solid lines in Figure 7 show time-of-arrival of both the first wave packet as well as the first reflected wave packet. The solid lines (detailed incalculated Section 2).using We conclude that there is goodand agreement between the calculated thegeometry dispersion curves the calculated material parameters the panel geometry (detailed in in Figure 7 show the as dispersion curves usingexceeding the material and thelowest panel and observed values, we observed no discrepancies 10% parameters except the second Section 2). We conclude that there is good agreement between the calculated and observed values, geometry (detailed intested Section We conclude good agreement between the calculated frequency-thickness (at 2). 0.1105 MHz·mm)that for there the Ais 0 mode, which differed by 14%. Using the as and we observed no discrepancies exceeding 10% except the second lowestexcept frequency-thickness tested observedvelocities, values, as exceeding the second propagation wewe canobserved calculateno thediscrepancies distance required for the10% five-cycle wave packetslowest of A0 (atfrequency-thickness 0.1105 MHz·mm) for the A mode, which differed by 14%. Using the propagation velocities, we 0 tested 0.1105 MHz·mm) for the A0 mode, which differed by 14%. Using thecan and S0 to separate, which is (at about 18 cm for the frequency-thickness of 0.221 MHz·mm. calculate the distance required for the five-cycle wave packets of A and S to separate, which is 0 propagation velocities, we can calculate the distance required for 0the five-cycle wave packets ofabout A0 18 and cm for the frequency-thickness MHz mm. S0 to separate, which is aboutof180.221 cm for the·frequency-thickness of 0.221 MHz·mm.
Figure 7. The data points represent the measurement dispersion curve, determined using the MOFBG a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated Figure The datapoints pointsrepresent representthe the measurement measurement dispersion using thethe MOFBG dispersion curves. Figure 7. 7. The data dispersioncurve, curve,determined determined using MOFBG a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated dispersion curves. dispersion curves.
Sensors 2017, 17, 1948
8 of 9
4. Conclusions In conclusion, we demonstrated, for the first time to our knowledge, the use of MOFBG-based sensors embedded within a composite panel to measure Lamb waves. Until now, mostly surface-mounted FBGs have been used for this purpose. The MOFBGs used in this paper exhibit selective sensitivity to axial and transverse strains, which allows distinguishing between the two fundamental Lamb wave modes. The ability of our sensing scheme to successfully discern between the two fundamental Lamb wave modes offers perspectives for detecting higher order modes as well. In addition, the orientation of the MOFBG (by adjusting the roll angle and aligning the fibre axis ‘S’) allows for the selection of preferential directions of strain. In this manner, directions corresponding to critical locations in a structure can be promoted while undesired reflections impinging from other directions can be impeded. Furthermore, we illustrated in two scenarios the ability to discern the axial and transverse directions with the intention to improve and simplify Lamb wave-based damage detection with the damage located either on or off the direct path between the actuator and sensor. We showed that the ability to make the distinction between the axial and transverse strain directions is clearly beneficial in the former case, as the distinction between the S0 and A0 waves allowed for the individually evaluation of additional losses and for the observation of partial mode conversion. The latter situation was exemplified by determining the time-of-arrival and associated dispersion of the two fundamental Lamb wave modes for a range of frequencies. Despite the small dimensions of the tested panel, we were able to determine the dispersion curve using an embedded MOFBG with an accuracy essentially exceeding 90%. The results summarised above evidence the added value of using MOFBG sensors in Lamb wave detection-based applications. Acknowledgments: This work was partially supported by the Joint Technology Initiative Cleansky 2 project “SHERLOC—Structural HEalth Monitoring, Manufacturing and Repair Technologies for Life Management Of Composite Fuselage”, funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 314768. This work was also partially supported by the IWT-SBO Project ‘Self Sensing Composites—SSC’ with contract No. 120024. Thomas Geernaert is a post-doctoral fellow of the Research Foundation Flanders (FWO). B-PHOT acknowledges the Vrije Universiteit Brussel’s Methusalem programme as well as the Hercules programme of the FWO. Author Contributions: Ben De Pauw and Sidney Goossens conceived, designed, and performed the experiments. Ben De Pauw wrote the main part of the paper. Dimitrios Habas fabricated the instrumented panels with the support of Sidney Goossens, and both contributed in the interpretation of the results. Thomas Geernaert, Hugo Thienpont, and Francis Berghmans initiated the research and helped in interpreting the results, and in revising and writing the paper. Conflicts of Interest: The authors declare no conflict of interest.
References 1.
2. 3. 4. 5.
6. 7.
Zhao, X.; Gao, H.; Zhang, G.; Ayhan, B.; Yan, F.; Kwan, C.; Rose, J.L. Active health monitoring of an aircraft wing with embedded piezoelectric sensor/actuator network: I. Defect detection, localization and growth monitoring. Smart Mater. Struct. 2007, 16. [CrossRef] Paget, C.A.; Levin, K.; Delebarre, C. Actuation performance of embedded piezoceramic transducer in mechanically loaded composites. Smart Mater. Struct. 2002, 11. [CrossRef] Su, Z.; Lin, Y.; Ye, L. Guided Lamb waves for identification of damage in composite structures: A review. J. Sound Vib. 2006, 295. [CrossRef] Betz, D.C.; Thursby, G.; Culshaw, B.; Staszewski, W.J. Advanced layout of a fiber Bragg grating strain gauge rosette. J. Lightwave Technol. 2006, 24. [CrossRef] Takeda, N.; Okabe, Y.; Kuwahara, J.; Kojima, S.; Ogisu, T. Development of smart composite structures with small-diameter fibre Bragg grating sensors for damage detection: Quantitative evaluation of delamination length in CFRP-laminates using Lamb wave sensing. Compos. Sci. Technol. 2005, 65. [CrossRef] Tsuda, H.; Toyama, N.; Urabe, K.; Takatsubo, J. Impact damage detection in CFRP using fiber Bragg gratings. Smart Mater. Struct. 2004, 13. [CrossRef] Takeda, S.; Okabe, Y.; Takeda, N. Delamination detection in CFRP laminates with embedded small-diameter fiber Bragg grating sensors. Compos. Part A 2002, 33. [CrossRef]
Sensors 2017, 17, 1948
8. 9. 10. 11. 12. 13. 14. 15. 16. 17.
18.
19.
20.
21. 22. 23. 24. 25.
9 of 9
Su, Z.; Ye, L. Identification of Damage Using Lamb Waves; Springer: Hong Kong, China, 2009. Wu, W.; Yu, H.; Chen, W. Indentation responses of piezoelectric layered half-space. Smart Mater. Struct. 2013, 22. [CrossRef] Hayashi, T.; Kawashima, K. Single Mode Extraction from Multiple Modes of Lamb Wave and Its Application to Defect Detection. J. Solid Mech. Mater. Eng. 2003, 46. [CrossRef] Zhang, H.-Y.; Yu, J.-B. Piezoelectric transducer parameter selection for exciting a single mode from multiple modes of Lamb waves. Chin. Phys. B 2011, 20. [CrossRef] Frazao, O.; Santos, J.; Araujo, F.; Ferreira, L. Optical sensing with photonic crystal fibers. Laser Photonics Rev. 2008, 2. [CrossRef] Canning, J. Properties of Specialist Fibres and Bragg Gratings for Optical Fiber Sensors. J. Sens. 2009, 871580. [CrossRef] Canning, J.; Groothoff, N.; Cook, K.; Pohl, A.; Holdsworth, J.; Bandyopadhyay, S.; Stevenson, M. Grating Writing in Structured Optical Fibres. Laser Chem. 2008, 239417. [CrossRef] Canning, J. Fibre Gratings & Devices for Sensors & Lasers. Lasers Photonics Rev. 2008, 2. [CrossRef] Pinto, A.; Lopez-Amo, M. Photonic Crystal Fibers for Sensing Applications. J. Sens. 2009, 598178. [CrossRef] Berghmans, F.; Geernaert, T.; Sonnenfeld, C.; Sulejmani, S.; Thienpont, H. Microstructured Optical Fibre-Based Sensors for Structural Health Monitoring Applications, Optofluidics, Sensors and Actuators in Microstructured Optical Fibres; Woodhead Publishing: Cambridge, UK, 2015; p. 139. Sonnenfeld, C.; Luyckx, G.; Sulejmani, S.; Geernaert, T.; Eve, S.; Gomina, M.; Chah, K.; Mergo, P.; Urbanczyk, W.; Thienpont, H.; et al. Microstructured optical fiber Bragg grating as an internal three-dimensional strain sensor for composite laminates. Smart Mater. Struct. 2015, 24. [CrossRef] Sonnenfeld, C.; Sulejmani, S.; Geernaert, T.; Eve, S.; Lammens, N.; Luyckx, G.; Voet, E.; Degrieck, J.; Urbanczyk, W.; Mergo, P.; et al. Microstructured optical fibre sensors embedded in laminate composite for smart material applications. Sensors 2011, 11, 2566. [CrossRef] [PubMed] Sulejmani, S.; Sonnenfeld, C.; Geernaert, T.; Luyckx, G.; van Hemelrijck, D.; Mergo, P.; Urbanczyk, W.; Chah, K.; Caucheteur, C.; Mégret, P.; et al. Shear stress sensing with Bragg grating-based sensors in microstructured optical fibres. Opt. Express 2013, 21, 20404. [CrossRef] [PubMed] Rose, J. Ultrasonic Waves in Solid Media; Cambridge University Press: Cambridge, UK, 1999. Cusano, A.; Cutolo, A. Fibre Bragg Grating Sensors: Recent Advancements, Industrial Applications and Market Exploitation; Bentham Science Publishers: Valencia, Spain, 2014. Sharif-Khodaei, Z.; Ghajari, M.; Aliabadi, M. Determination of impact location on composite stiffened panels. Smart Mater. Struct. 2012, 21, 105026. [CrossRef] Sharif-Khodaei, Z.; Aliabadi, M. Assessment of delay-and-sum algorithms for damage detection in aluminium and composite plates. Smart Mater. Struct. 2014, 23. [CrossRef] Lambinet, F.; Sharif-Khodaei, Z.; Aliabadi, M. Damage Detection in Composite Skin Stiffener with Hybrid PZT-FO SHM System. Key Eng. Mater. 2017, in press. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Fibre Bragg Gratings in Embedded Microstructured Optical Fibres Allow Distinguishing between Symmetric and Anti-Symmetric Lamb Waves in Carbon Fibre Reinforced Composites Ben De Pauw 1,2, *, Sidney Goossens 1,2 , Thomas Geernaert 1,2 , Dimitrios Habas 3 , Hugo Thienpont 1,2 and Francis Berghmans 1,2 1
2 3
*
Department of Applied Physics and Photonics (TONA), Vrije Universiteit Brussel, Brussels Photonics (B-PHOT), Pleinlaan 2, 1050 Brussels, Belgium; [email protected] (S.G.); [email protected] (T.G.); [email protected] (H.T.); [email protected] (F.B.) Flanders Make, Oude Diestersebaan 133, 3920 Lommel, Belgium Hellenic Aerospace Industry, Engineering Research Design and Development Directorate, 32009 Shimatari, Greece; [email protected] Correspondence: [email protected]; Tel.: +32-2-629-1381
Received: 31 July 2017; Accepted: 19 August 2017; Published: 24 August 2017
Abstract: Conventional contact sensors used for Lamb wave-based ultrasonic inspection, such as piezo-electric transducers, measure omnidirectional strain and do not allow distinguishing between fundamental symmetric and anti-symmetric modes. In this paper, we show that the use of a single fibre Bragg grating created in a dedicated microstructured optical fibre allows one to directly make the distinction between these fundamental Lamb wave modes. This feature stems from the different sensitivities of the microstructured fibre to axial and transverse strain. We fabricated carbon fibre-reinforced polymer panels equipped with embedded microstructured optical fibre sensors and experimentally demonstrated the strain waves associated with the propagating Lamb waves in both the axial and transverse directions of the optical fibre. Keywords: fibre Bragg gratings; Lamb waves; CFRP; impact damages
1. Introduction Lamb wave-based ultrasonic inspection for damage identification in engineering structures is considered to be an attractive alternative for conventional and time-consuming point-by-point ultrasonic inspection techniques, mainly because Lamb waves can travel long distances across the structure and therefore allow for inspecting fairly large areas. Such Lamb waves can be excited and detected in various manners depending on the size and material of the structure under test [1–3]. In the aerospace industry, for example, the most common technique to both excite and detect Lamb waves in plate-like composite components uses piezo-electric transducers (PZTs). Such transducers feature a small size and low power consumption. The main disadvantages of PZTs in Lamb wave-based inspection are twofold. First, each PZT requires being individually connected, which calls for dealing with a multitude of wires when using arrays of transducers. Second, PZTs exhibit unselective sensitivity in the sense that Lamb waves propagating along different directions in the plane of a panel are all picked up with the same sensitivity. Critical directions, corresponding to, for example, a defect in the panel, are recorded simultaneously and with the same sensitivity as every other direction. It is thus challenging to identify the result of a defect in a panel on Lamb waves propagating in this structure and, therefore, it is difficult to locate and assess the size of the defect. Optical fibre Bragg gratings (FBGs) have already been proposed to detect Lamb waves in order to cope with the aforementioned Sensors 2017, 17, 1948; doi:10.3390/s17091948
www.mdpi.com/journal/sensors
Sensors 2017, 17, 1948
2 of 9
shortcomings [3–6]. Multiple FBGs can easily be multiplexed on a single optical fibre, which drastically decreases the amount of wiring when using an array of such sensors. Furthermore, and owing to the geometry of the FBG, the response to Lamb waves depends on the angle between the fibre and the Lamb wave’s propagation direction. Finally, the small size of the optical fibre allows these sensors to be embedded between the laminar layers of a composite structure in a minimally intrusive manner, avoiding the need to attach bulky and disturbing sensors to the surface of a structure [7]. Lamb waves can be categorised into two groups based on the associated propagating perturbation of the composite with respect to the mid-plane of a panel, i.e., in symmetric and anti-symmetric modes. These modes are denoted as Si and Ai , respectively, where i refers to the order of the mode. The number of symmetric and anti-symmetric modes that can propagate along a plate-like structure under test depends on the excitation frequency [3,8]. Owing to their low dispersion and attenuation, the fundamental symmetric S0 and anti-symmetric A0 modes are frequently considered suitable for inspecting the structure for the presence of defects. Since they have different mode shapes, S0 and A0 modes are sensitive to different defects in the structure: the S0 wave is extensional in nature and thus more sensitive to defects inside the structure under test, whilst the A0 is flexural and thus more sensitive to surface cracks, as an example. Both modes also have a different frequency-dependent propagation velocities. Consequently, both types of waves are of interest for identifying and locating damage in composites. However, properly distinguishing between A0 and S0 waves in the sensor responses remains very difficult in ultrasonic inspection using both conventional FBGs and PZTs [3,9–11]. In this paper, we demonstrate that a sensor consisting of a single FBG created in a specialty microstructured optical fibre (so-called MOFBG) can be used to discern between the A0 and S0 waves. To do so, we used MOFBG sensors embedded within the composite panel to measure the Lamb waves, while until now mostly surface mounted FBGs have been used for this purpose. Our paper is structured as follows. Section 2 describes the unique measurement strategy that MOFBG sensors can offer to record Lamb waves. We discuss how the mounting procedure and experimental setup help to distinguish between the A0 and S0 waves. The subsequent Section 3 deals with the experimental results obtained and discusses how these can be applied when detecting impact damage using Lamb waves. Section 4 closes our paper with a summary and conclusions. 2. MOFBG Sensors and Measurement Strategy MOFBGs can exhibit peculiar characteristics that cannot be achieved using conventional optical fibre technology. More specifically, the design flexibility of such microstructured fibres (MOFs) allows for the development of sensors that exhibit selective sensitivities to e.g., axial strain, transverse strain, or even shear strain, whilst being negligibly cross-sensitive to temperature changes. Examples of MOF-based sensors have been reviewed in several research papers [12–20]. In this paper, we use a specific microstructured fibre design which we refer to as a ‘butterfly’ MOF [17–20]. Figure 1a shows an SEM image of the cross-section of the butterfly MOF [18]. The peculiar microstructure induces a high level of birefringence in the optical fibre. When this MOF is equipped with an FBG, the latter reflects two Bragg wavelengths corresponding to the two orthogonal polarised propagation modes. The Bragg wavelengths (or peaks) will shift with the same amount when an axial load is applied to the optical fibre, but will move in opposite ways when a transverse load (parallel to axis ‘S’) is applied. The absolute and relative position of the Bragg peaks therefore encodes axial strain as well as transverse strain, respectively, into the reflection spectrum of the MOFBG. This principle is summarised in Figure 1b.
Sensors 2017, 17, 1948
3 of 9
Sensors 2017, 17, 1948
3 of 9
Sensors 2017, 17, 1948
3 of 9
Figure 1. (a) SEM image of the cross-section of the fabricated butterfly microstructured fibre (MOF);
Figure 1. (a) SEM image of the cross-section of the fabricated butterfly microstructured fibre (MOF); (b) Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to (b) Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to both peaks moving in the same way, whilst transverse strain (applied parallel to axis ‘S’) corresponds both Figure peaks moving inimage the same way, whilst transverse strain (applied parallel to axis fibre ‘S’) corresponds 1. (a) moving SEM of the cross-section of the fabricated butterfly microstructured (MOF); to the peaks in opposite directions [18]. to the(b) peaks moving in opposite directions [18]. Response of a fibre Bragg grating (FBG) created in the butterfly MOF: axial strain corresponds to both moving in the same whilst transverse strainand (applied parallelstrain to axis can ‘S’) corresponds The peaks selective sensitivity of away, single MOFBG to axial transverse be exploited to to the peaks moving in opposite directions [18]. differentiate between the S0 and A0 modes. This stems from the nature of the induced strain
The selective sensitivity of a single MOFBG to axial and transverse strain can be exploited to distribution (or strainSmode shape) in the axial and from transverse directions ofinduced a plate-like structure, differentiate between This stems the nature of the distribution 0 and A The selectivethe sensitivity of0 amodes. single to axial and transverse strain canassociated bestrain exploited to which is different for these modes. OwingMOFBG to the anti-symmetric shape of the A0, the strain (or strain mode shape) in the axial and transverse directions of a plate-like structure, which is different differentiate between the S 0 and A0 modes. This stems from the nature of the induced strain amplitude in the transverse direction is close to zero near the mid-plane of the panel [21]. The for these modes. Owing tomode the anti-symmetric shape the A0 , the associated amplitude in the distribution in the axial and of transverse directions a strain plate-like structure, symmetry of(or thestrain S0 mode, onshape) the other hand, results in an increase of the of strain amplitude at that transverse direction is close to zero near the mid-plane of the panel [21]. The symmetry of the S mode, which is different for these modes. Owing to the anti-symmetric shape of the A 0, the associated strain 0 location. Consequently, an MOFBG embedded for example at a depth of ¼ (or equivalently ¾) of the amplitude the transverse to zero nearthe the ofinthe [21]. The an on the other hand, results in andirection increase ofclose the strain at that location. Consequently, thickness ofina plate-like structure will beisable to pick up amplitude both S0 mid-plane and A0 mode thepanel axial direction, 3 of the strain amplitude at that symmetry of Smode 0 mode, on transverse the hand, in an increase ⁄4 (orasequivalently ⁄4) of2,the MOFBG embedded example at aother depth of 1results thickness of atypical plate-like but mostly thethe S0for in the direction, depicted in Figure which illustrates location. Consequently, an MOFBG embedded for example at a depth of or ¼ (or equivalently ¾) ofSthe strainwill mode the up lateral of a A plate-like structure. At ¼ ¾ of the thickness, the A 0 structure be shapes able toin pick bothdirection the S0 and mode in the axial direction, but mostly the 0 0 mode thickness of a plate-like structure will be able to pick uptransverse both the S0 strain and A0components, mode in the axial direction, wave as well as the S 0 wave will still exhibit some but given the in the transverse direction, as depicted in Figure 2, which illustrates typical strain mode shapes in but mostly thethe S0axial mode in the transverse direction, as 3depicted in Figure 2, which illustrates typical 1 contrast with strain components, the At MOFBG distinguishing between the two ⁄4 or allows ⁄4 of the the lateral direction of a plate-like structure. thickness, the A0 wave asparties. well as the strain mode shapes in the lateral direction of a plate-like structure. At ¼ or ¾ of the thickness, the A0 Notewill thatstill the exhibit actual measured strain amplitude of both the S0 and A 0 modes also depends on the S0 wave some transverse strain components, but given the contrast with the axial wave as well as the S 0 wave will still exhibit some transverse strain components, but given the transfer of the strain induced in the material by the Lamb waves to the optical fibre and on the straincontrast components, MOFBG allows distinguishing between the two parties. Note that the actual with thethe axial strain components, the MOFBG excitation amplitude and attenuation of each mode. allows distinguishing between the two parties. measured strain of both the amplitude S0 and A0ofmodes also depends on the the Note that theamplitude actual measured strain both the S0 and A0 modes alsotransfer dependsofon thestrain transfer of material the strainby induced in the material by the Lamb waves fibre and on the and induced in the the Lamb waves to the optical fibre andtoonthe theoptical excitation amplitude excitation and attenuation of each mode. attenuation of amplitude each mode.
Figure 2. Illustration of strain mode shapes of both S0 and A0 waves in the lateral direction of a platelike structure. The strain induced by an A0 wave close to the mid-thickness tends to be zero, in contrast to the induced strain of the S0 wave, which increases at the same depth. An MOFBG embedded at a Figure 2. Illustration of ¼ strain mode shapes of both S0 and A0 waves theaxial lateral direction of a strain, platedepth of, for example, of the panel thickness, that isSsensitive to in both and transverse Figure 2. Illustration of strain mode shapes of both 0 and A0 waves in the lateral direction of a like structure. The strain induced by an A 0 wave close to the mid-thickness tends to be zero, in contrast A0 modes can exploit this difference to distinguish between S0 and plate-like structure. The strain induced by an A0the wave close to the [18,21]. mid-thickness tends to be zero, at the same depth. An MOFBG embedded at a to the induced strain of the S0 wave, which increases in contrast to the induced strain of the S wave, which increases at the same depth. An MOFBG 0 depth of, for example, ¼ of the panel thickness, that is sensitive to both axial and transverse strain, 1 embedded at a depth of, for example, ⁄ 4 of the panel thickness, that is sensitive to both axial and can exploit this difference to distinguish between the S0 and A0 modes [18,21].
transverse strain, can exploit this difference to distinguish between the S0 and A0 modes [18,21].
Sensors 2017, 17, 1948
4 of 9
To demonstrate the capability of an MOFBG to distinguish between S0 and A0 modes using the To demonstrate the capability of anaMOFBG between S0 and A0 modes using the approach outlined above, we fabricated 250 mm to bydistinguish 25 mm carbon fibre-reinforced polymer (CFRP) approach outlined above, we fabricated a 250 mm by 25 mm carbon fibre-reinforced polymer (CFRP) panel made from Hexcel M21/T800S prepreg, with a lay-up of (45/−45/02/90/0)s. The thickness of each panel Hexcel withthickness a lay-up of of 2.21 (45/− 45/0We of s . The thickness 2 /90/0) ply is made close from to 0.184 mm,M21/T800S resulting inprepreg, an overall mm. embedded an MOFBG each ply is close to 0.184 mm, resulting in an overall thickness of 2.21 mm. We embedded an MOFBG between the third and fourth laminar layers (i.e., at ¼ of the thickness of the panel) and aligned the 1 between the with thirdthe and0°fourth laminar layers (i.e., thickness of thefibres panel) aligned optical fibre orientation of the plies (i.e.,atparallel the carbon in and the plane of the the 4 of thewith ◦ optical fibre with the 0 orientation of the plies (i.e., parallel with the carbon fibres in the plane of the laminate plies) to minimize the effect of the embedding and, in particular, to avoid the creation of laminate plies) to We minimize the effect of the embedding and, inofparticular, avoid the creation of resin pockets [7]. positioned the MOFBG at the half-length the panel to and angularly oriented resin pocketssuch [7]. We the MOFBG at the half-length of theto panel and angularly oriented the MOFBG thatpositioned the peak separation was essentially sensitive the out-of-plane transverse the MOFBG such that the peak separation was essentially sensitive to the out-of-plane transverse load. To do so, we rotated the fibre about its axis (i.e., changed the roll angle) so that the butterfly load. To do so, rotated the fibre about its axis (i.e.,the changed roll so that theTo butterfly axis (referred towe as ‘S’ in Figure 1a) was aligned with desiredthe axis of angle) the CFRP panel. obtain axis (referred to as ‘S’ in Figure 1a) was aligned with the desired axis of the CFRP panel. To obtain the desired orientation of the microstructured optical fibre, we relied on our technique described in the desired[18], orientation of theangular microstructured optical we relied on fibre our technique described Reference which yields accuracies of thefibre, roll angle of the with respect to the in Reference [18], which yields angular accuracies of the roll angle of the fibre with respect to roll the instrumented structure close to ±8°. We have previously determined that such a tolerance on the ◦ instrumented structure close toto±loads 8 . Wealong havethe previously determined a tolerance on the angle guarantees a sensitivity desired CFRP panel that axis such of well beyond 85% of roll the angle guarantees a sensitivity to loads along the desired CFRP panel axis of well beyond 85% the sensitivity achieved when the fibre’s microstructure is perfectly aligned along the desired axisof[19]. sensitivity whenfor thethe fibre’s microstructure is perfectly axis [19]. This is not achieved only sufficient demonstration purpose reportedaligned in this along paper,the butdesired also allows one This is not only sufficient for the demonstration purpose reported in this paper, but also allows one to to envisage its implementation in practical applications. To excite the Lamb waves in the fabricated envisage its implementation in practical applications. To excite the Lamb waves in the fabricated panel, panel, we surface-mounted a disk-shaped PZT of type PIC255 (PI Ceramic GmbH, Lederhose, we surface-mounted a disk-shaped PZT ofextremity. type PIC255 Ceramicremark GmbH,here Lederhose, Germany) Germany) with a diameter of 10 mm on one One(PI important is that the distance with a diameter of 10 mm on one extremity. One important remark here is that the distance between the PZT and the MOFBG is only 10 cm, and is thus typically insufficient for the Abetween 0 and S0 the PZTtoand theseparate MOFBGowing is onlyto10 cm,different and is thus typically insufficient forthe theused A0 and S0 modes to modes fully their propagation velocities (in material). More fully separate owing to their different propagation velocities (in the used material). More specifically, specifically, we observed from the experimentally and numerically determined dispersion curves of we from experimentally and determined ourfrequencysetup that ourobserved setup that thethe delay between the A0numerically and S0 waves was onlydispersion about 18 curves µs. Forofthe the delay between the A S0 waves was only 18 µs.ofFor 0 and the thickness used in this paper, measured phaseabout velocities thethe S0 frequency-thickness and A0 waves were used closein tothis 4.5 paper, the measured phase velocities of the S and A waves were close to 4.5 m/ms and 2.8 m/ms, 0 0 m/ms and 2.8 m/ms, respectively. A schematic of the experimental setup is depicted in Figure 3. We respectively. A PZT schematic the experimental setup is depicted infive-cycle Figure 3. We controlled theaPZT using controlled the usingof a waveform generator that generates tonebursts with Hanning a waveform that generatesThe five-cycle tonebursts withto a Hanning filter applied at a preset filter appliedgenerator at a preset frequency. response of the FBG the impinging Lamb waves was frequency. The response of the FBG to the impinging Lamb waves was measured using the measured using the so-called edge filtering method [22], in which the output wavelength of aso-called tunable edge [22], in which the output wavelength of a tunable is tuned to theofwavelength laser filtering is tunedmethod to the wavelength where the slope of the Bragg peak islaser steepest. A shift the Bragg where the slope of the Bragg peak is steepest. A shift of the Bragg peak resulting from a passing peak resulting from a passing strain wave is then recorded as a change of optical intensity by strain a fast wave is then recorded asPDA20CS). a change ofNote optical by a fast photodetector photodetector (Thorlabs thatintensity the birefringence in the MOFBG(Thorlabs producesPDA20CS). two Bragg Note birefringence in the two(two Bragg peaks (see 1), meaningcan thatbe a peaksthat (seethe Figure 1), meaning thatMOFBG a total ofproduces four slopes ascending andFigure two descending) total of four slopes (two ascending and two descending) can be used. used.
Figure 3. A setup. Figure 3. A schematic schematic of of the the experimental experimental setup.
3. Experimental Results 3. Experimental Results 3.1. MOFBGs between A A0 and S0 Waves 3.1. MOFBGs Allow Allow Distinguishing Distinguishing between 0 and S0 Waves Using the the setup setup described described above, above, we we excited excited the the fundamental fundamental SS0 and and A A0 modes in the fabricated Using 0 0 modes in the fabricated panel by actuating the PZT with a 20 Vpp five-cycle toneburst at 250 kHz with the modulated amplitude panel by actuating the PZT with a 20 Vpp five-cycle toneburst at 250 kHz with the amplitude modulated using a Hanning filter, and we measured the response with the embedded MOFBG. To
Sensors 2017, 17, 1948
5 of 9
2017, 17, 1948 5 of 9 the usingSensors a Hanning filter, and we measured the response with the embedded MOFBG. To improve signal-to-noise ratio, each measurement was repeated 1000 times and averaged. The measured improve the signal-to-noise ratio, each measurement was repeated 1000 times and averaged. The wavelength shifts as a function of time for the four slopes of the MOFBG are shown in Figure 4a. measured wavelength shifts as a function of time for the four slopes of the MOFBG are shown in The average of the shift fourofmeasurements is proportional to thetoaxial strain applied Figure 4a.wavelength The average shift wavelength the four measurements is proportional the axial strain to theapplied MOFBG, while the shift in spectral distance between the first and third slopes or between to the MOFBG, while the shift in spectral distance between the first and third slopes or the second and fourth slopes isfourth proportional the transverse applied to applied the MOFBG. The resulting between the second and slopes isto proportional to thestrain transverse strain to the MOFBG. wavelength shiftswavelength corresponding to the axial (in transverse (in red) (in directions are shown The resulting shifts corresponding to black) the axialand (in black) and transverse red) directions are shown in Figure 4b. Although the sensitivity in the transverse direction is not as high as that in axial in Figure 4b. Although the sensitivity in the transverse direction is not as high as that in the the axial direction [18–20], we observed a clear difference in the obtained signals. In particular, the direction [18–20], we observed a clear difference in the obtained signals. In particular, the signal signal in measured the axialproved direction be more responsive to an impinging 0 wave measured the axialindirection to proved be moretoresponsive to an impinging A0 waveA(because the (because the PZT excitation of A0 is more pronounced), while the signal measured in the transverse PZT excitation of A0 is more pronounced), while the signal measured in the transverse direction was direction was more responsive to the S0 wave. more responsive to the S0 wave.
Figure 4. (a) The measured wavelength shifts for the four slopes of the MOFBG embedded in the
Figure 4. (a) The measured wavelength shifts for the four slopes of the MOFBG embedded in the carbon carbon fibre reinforced polymer panel versus time; (b) Based on the properties of the MOFBG, the fibre reinforced polymer panel versus time; (b) Based on the properties of the MOFBG, the measured measured signals of the four slopes have been converted into axial and transverse wavelength shifts signals of the four slopes have been converted into axial and transverse wavelength shifts that are that are proportional to the respective strains applied to the MOFBG. proportional to the respective strains applied to the MOFBG.
The identification of both types of waves is confirmed by the time-of-arrival of both the wave packets that arrive first welltypes as theof wave packets returning after a reflection at the extremity thewave The identification of as both waves is confirmed by the time-of-arrival of bothofthe panel (see also Section 3.3). The results therefore evidence that a single embedded MOFBG, sensitive packets that arrive first as well as the wave packets returning after a reflection at the extremity of to both axial and transverse strains, allows distinguishing between the A0 and S0 Lamb waves the panel (see also Section 3.3). The results therefore evidence that a single embedded MOFBG, propagating through the CFRP panel.
sensitive to both axial and transverse strains, allows distinguishing between the A0 and S0 Lamb waves propagating through the CFRPAllow panel. 3.2. Selection of Preferential Directions Filtering out Undesired Edge Reflections The of directional sensitivity of Allow the MOFBG canout bring an additional advantage to Lamb wave3.2. Selection Preferential Directions Filtering Undesired Edge Reflections
based ultrasonic inspection. In a geometry similar to the CFRP panel explained in the previous The directional sensitivity of the MOFBG can bring an advantage to LambMOFBG wave-based section, the response of a PZT sensor surface mounted atadditional the position of the embedded ultrasonic inspection. In a geometry similar to the CFRP panel explained in the previous section, would be significantly influenced by the multitude of reflections. Such a situation would make it very challenging thesurface A0 and Smounted 0 waves from PZTposition recordings, a PZT is sensitive to all the response oftoa distinguish PZT sensor at the of since the embedded MOFBG would incident directions and the Lamb waves arrive at the Such sensoraquasi-simultaneous. The key be significantly influenced byreflected the multitude of reflections. situation would make it very feature of is that to two perpendicular (one axial to all challenging toMOFBGs distinguish thethey A0 are andessentially S0 wavessensitive from PZT recordings, since directions a PZT is sensitive and one transverse to the optical fibre), which allowed us to choose, by carefully angularly orienting incident directions and the reflected Lamb waves arrive at the sensor quasi-simultaneous. The key the MOFBG, to filter out part of the edge reflections. To demonstrate this, we fabricated a second feature of MOFBGs is that they are essentially sensitive to two perpendicular directions (one axial and CFRP panel with the same geometry as before and with the MOFBG embedded at the same location. one transverse tohowever, the optical which allowed us to choose, by carefully In this panel, we fibre), angularly oriented the MOFBG (i.e., adjusted the roll angularly angle of theorienting fibre to the MOFBG, to filter out part of the edge reflections. To demonstrate this, we fabricated a second align the axis ‘S’ of the butterfly structure) so that to the peak separation is essentially sensitive to theCFRP paneltransverse with the in-plane same geometry before the andmeasurement with the MOFBG embedded at the same location. load. Weas repeated procedure and we separated again the axialIn this panel, however, we angularly the MOFBG (i.e., adjusted roll angle of the fibre to The align the load and transverse loads, oriented as performed for the previous panel.the Figure 5 shows results.
axis ‘S’ of the butterfly structure) so that to the peak separation is essentially sensitive to the transverse in-plane load. We repeated the measurement procedure and we separated again the axial load and
Sensors 2017, 17, 1948
6 of 9
transverse loads, as performed for the previous panel. Figure 5 shows the results. The reflections Sensors 2017, 17, 1948 6 of 9 from the long edges are more clearly observed than in the previous composite panel, as shown in reflections from the long edgeson area more clearly in the previous composite panel, a asmore Figure 5a, using a measurement single slope.observed Similarthan to the previous panel, we noticed shown in Figure 5a, using a measurement on a single slope. Similar to the previous panel, we noticed significant response to the S0 wave than to the A0 mode in the transverse direction and vice versa in a more significant response to the S0 wave than to the A0 mode in the transverse direction and vice the axial direction. Therefore, carefully aligning the butterfly structure by adjusting the roll angle of versa in the axial direction. Therefore, carefully aligning the butterfly structure by adjusting the roll the MOFBG during mounting allows for the selection or elimination of certain directions of strain and, angle of the MOFBG during mounting allows for the selection or elimination of certain directions of hence, potentially accounts for undesired effects resulting from Lamb wave reflections the edges strain and, hence, potentially accounts for undesired effects resulting from Lamb wave from reflections of a panel under test. from the edges of a panel under test.
Figure 5. (a) A measurement of a single slope of a MOFBG in a similar CFRP panel, where the MOFBG Figure 5. (a) A measurement of a single slope of a MOFBG in a similar CFRP panel, where the MOFBG is oriented to be sensitive to axial load and transverse in-plane load; (b) The measured responses of is oriented to be sensitive to axial load and transverse in-plane load; (b) The measured responses of the the MOFBG converted into axial and transverse loads. MOFBG converted into axial and transverse loads.
3.3. Consequences for Impact Damage Detection in CFRP
3.3. Consequences for Impact Damage Detection in CFRP
The demonstrated ability to resolve the two components of Lamb waves is also beneficial when
Lambdemonstrated waves are used to detect damagethe in two a structure. In most involving The ability to resolve components of experiments Lamb waves is also Lamb beneficial wave-based damage detection, the signals of an undamaged structure are compared to the signals when Lamb waves are used to detect damage in a structure. In most experiments involvingofLamb the damaged structure [23–25].the Damage in the structure will cause propagating wavesto to the reflect wave-based damage detection, signals of an undamaged structure are Lamb compared signals and refract, yielding an alteration in the recorded Lamb waves. In general, one can distinguish of the damaged structure [23–25]. Damage in the structure will cause propagating Lamb waves to between two situations. reflect and refract, yielding an alteration in the recorded Lamb waves. In general, one can distinguish First, the damage can be on the direct path between the actuator and the sensor. In this case, the between two situations. fraction of the Lamb waves that are influenced (because of reflection, refraction, attenuation, etc.) will First, theenergy damage can beinon direct path between theup actuator and the In this case, alter the contained thethe wave packets that are picked by the sensor [25].sensor. As an example the fraction of the Lamb waves that are influenced (because of reflection, refraction, attenuation, of such a situation, we experimentally mimicked an impact damage by applying a dot of synthetic etc.) will alter the energywith contained in of theabout wave packets are picked up byinthe sensor rubber adhesive a diameter 5 mm to thethat CFRP panel described Section 3.1, [25]. and weAs an recorded Lamb waves withwe the experimentally embedded MOFBG. We showan theimpact result indamage Figure 6, by in which we have example of such a situation, mimicked applying a dot of plotted the wavelength shiftsa corresponding to strains into the axial andpanel transverse directions. The 3.1, synthetic rubber adhesive with diameter of about 5 mm the CFRP described in Section dashed lines Lamb in the waves figure indicate pristine situation, the solid lines correspond toin the and we recorded with thethe embedded MOFBG.while We show the result in Figure 6, which damaged situation. Since the simulated damage is located on the direct path between the PZT we have plotted the wavelength shifts corresponding to strains in the axial and transverse directions. actuator and the MOFBG sensor, a fraction of the Lamb waves will be attenuated, yielding less The dashed lines in the figure indicate the pristine situation, while the solid lines correspond to the energetic strain waves to be transmitted and thus smaller recorded wavelength shifts. The latter can damaged situation. Since damage isdirections located on the the direct PZT be clearly seen in boththe thesimulated axial and transverse when firstpath wavebetween packets the arrive at actuator the and the MOFBG sensor, a fraction of the Lamb waves will be attenuated, yielding less energetic MOFBG. In addition, due to the additional attenuation which is possibly accompanied with partialstrain waves to beconversion transmitted andAthus recorded shifts. The latter can be amplitude clearly seen in mode from 0 to Ssmaller 0, the second wavewavelength packet shows a significantly larger both compared the axial and transverse directions when the first wave packets arrive atsituation, the MOFBG. In addition, to the undamaged state in the transverse direction. In the described we observed anthe increase of the amplitude of which the second wave packet of more than a factor ofmode 2. Recall, however, from due to additional attenuation is possibly accompanied with partial conversion for the used wave setup packet the excitation A0 mode islarger more pronounced compared to 0 mode. A0 tothat S0 , the second showsofa the significantly amplitude compared tothe theSundamaged Nevertheless, our approach illustrates the potential benefits of using MOFBGs as sensors in Lamb state in the transverse direction. In the described situation, we observed an increase of the amplitude of the second wave packet of more than a factor of 2. Recall, however, that for the used setup the excitation of the A0 mode is more pronounced compared to the S0 mode. Nevertheless, our approach
Sensors 2017, 17, 1948 Sensors 2017, 17, 1948
7 of 9 7 of 9
illustrates the17,potential benefits of using as sensors in between Lamb wave-based Sensors 2017, 1948 of 9 wave-based damage detection owing toMOFBGs their ability to discern the strains damage recordeddetection in7 the owing to their ability to discern between the strains recorded in the axial and transverse directions. axial and transverse directions. wave-based damage detection owing to their ability to discern between the strains recorded in the axial and transverse directions.
Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged situation for the sample described in Section 3.1. situation for the sample described in Section 3.1. Figure 6. Wavelength shifts corresponding to axial and transverse strains for a pristine and damaged situation sample can described Section 3.1. path between the actuator and sensors. In such a Second, for thethe damage be offin the direct
Second,Lamb the damage be off the direct path between thewave actuator and sensors. a situation, situation, waves can altered from the damage will yield packets arriving In at such the sensor at Second, the damage can be off the direct path between the actuator and sensors. In suchtimes. a Lamb wavestimes. altered the damage willtoyield wave packets arriving at of thewave sensor at different different In from this case, the ability determine the time-of-arrival packets is crucial situation, Lamb waves altered from the damage will yield wave packets arriving at the sensor at In [23,24]. this case, thetoability to determine the of time-of-arrival wave packets is crucialdetection, [23,24]. we Thus, Thus, illustrate the potential the MOFBGs of in time-of-arrival damage different In this case, theMOFBGs ability to in determine theLamb time-of-arrival of waveinpackets is crucial to experimentally illustratetimes. the potential of the time-of-arrival damage detection, we determined the dispersion curve of the waves generated theexperimentally CFRP panel [23,24].also Thus, to on illustrate the potential of of the in time-of-arrival damage detection, weon which relies accurate measurements theMOFBGs time-of-arrival wavepanel packets. Recall the determined the dispersion curve of the Lamb waves generated in of thethe CFRP which alsothat relies experimentally determined the dispersion curve of the Lamb waves generated in the CFRP panel distance between the actuator and sensor is only cm, which is typically insufficient for the A0 andthe accurate measurements of the time-of-arrival of the10wave packets. Recall that the distance between which also on accurate measurements of the time-of-arrival ofdots the wave packets. phase Recall velocity that the S 0 waves torelies separate. The result shownisintypically Figure 7,insufficient in which thefor actuator and sensor is only 10 cm,iswhich therepresent A0 and Sthe waves to separate. 0 distance between the observed actuator and sensor is only 10 cm, which is typically insufficient for the A0 and determined from the time-of-arrival for a certain excitation frequency. We determined The result is shown in Figure 7, in which the dots represent the phase velocity determined the from S0 waves to separate. is shown in as Figure in the dots wave represent the The phase velocity of bothThe theresult first packet well 7, as thewhich first reflected packet. solid lines of thetime-of-arrival observed time-of-arrival for wave a certain excitation frequency. We determined the time-of-arrival determined from time-of-arrival for a certain excitation frequency. We determined the in Figure showthe theobserved dispersion calculated material and the panel both the first7wave packet as well ascurves the first reflectedusing wavethe packet. Theparameters solid lines in Figure 7 show time-of-arrival of both the first wave packet as well as the first reflected wave packet. The solid lines (detailed incalculated Section 2).using We conclude that there is goodand agreement between the calculated thegeometry dispersion curves the calculated material parameters the panel geometry (detailed in in Figure 7 show the as dispersion curves usingexceeding the material and thelowest panel and observed values, we observed no discrepancies 10% parameters except the second Section 2). We conclude that there is good agreement between the calculated and observed values, geometry (detailed intested Section We conclude good agreement between the calculated frequency-thickness (at 2). 0.1105 MHz·mm)that for there the Ais 0 mode, which differed by 14%. Using the as and we observed no discrepancies exceeding 10% except the second lowestexcept frequency-thickness tested observedvelocities, values, as exceeding the second propagation wewe canobserved calculateno thediscrepancies distance required for the10% five-cycle wave packetslowest of A0 (atfrequency-thickness 0.1105 MHz·mm) for the A mode, which differed by 14%. Using the propagation velocities, we 0 tested 0.1105 MHz·mm) for the A0 mode, which differed by 14%. Using thecan and S0 to separate, which is (at about 18 cm for the frequency-thickness of 0.221 MHz·mm. calculate the distance required for the five-cycle wave packets of A and S to separate, which is 0 propagation velocities, we can calculate the distance required for 0the five-cycle wave packets ofabout A0 18 and cm for the frequency-thickness MHz mm. S0 to separate, which is aboutof180.221 cm for the·frequency-thickness of 0.221 MHz·mm.
Figure 7. The data points represent the measurement dispersion curve, determined using the MOFBG a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated Figure The datapoints pointsrepresent representthe the measurement measurement dispersion using thethe MOFBG dispersion curves. Figure 7. 7. The data dispersioncurve, curve,determined determined using MOFBG a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated a Lamb wave sensor in the small, slender CFRP panels. The solid lines indicate the calculated dispersion curves. dispersion curves.
Sensors 2017, 17, 1948
8 of 9
4. Conclusions In conclusion, we demonstrated, for the first time to our knowledge, the use of MOFBG-based sensors embedded within a composite panel to measure Lamb waves. Until now, mostly surface-mounted FBGs have been used for this purpose. The MOFBGs used in this paper exhibit selective sensitivity to axial and transverse strains, which allows distinguishing between the two fundamental Lamb wave modes. The ability of our sensing scheme to successfully discern between the two fundamental Lamb wave modes offers perspectives for detecting higher order modes as well. In addition, the orientation of the MOFBG (by adjusting the roll angle and aligning the fibre axis ‘S’) allows for the selection of preferential directions of strain. In this manner, directions corresponding to critical locations in a structure can be promoted while undesired reflections impinging from other directions can be impeded. Furthermore, we illustrated in two scenarios the ability to discern the axial and transverse directions with the intention to improve and simplify Lamb wave-based damage detection with the damage located either on or off the direct path between the actuator and sensor. We showed that the ability to make the distinction between the axial and transverse strain directions is clearly beneficial in the former case, as the distinction between the S0 and A0 waves allowed for the individually evaluation of additional losses and for the observation of partial mode conversion. The latter situation was exemplified by determining the time-of-arrival and associated dispersion of the two fundamental Lamb wave modes for a range of frequencies. Despite the small dimensions of the tested panel, we were able to determine the dispersion curve using an embedded MOFBG with an accuracy essentially exceeding 90%. The results summarised above evidence the added value of using MOFBG sensors in Lamb wave detection-based applications. Acknowledgments: This work was partially supported by the Joint Technology Initiative Cleansky 2 project “SHERLOC—Structural HEalth Monitoring, Manufacturing and Repair Technologies for Life Management Of Composite Fuselage”, funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 314768. This work was also partially supported by the IWT-SBO Project ‘Self Sensing Composites—SSC’ with contract No. 120024. Thomas Geernaert is a post-doctoral fellow of the Research Foundation Flanders (FWO). B-PHOT acknowledges the Vrije Universiteit Brussel’s Methusalem programme as well as the Hercules programme of the FWO. Author Contributions: Ben De Pauw and Sidney Goossens conceived, designed, and performed the experiments. Ben De Pauw wrote the main part of the paper. Dimitrios Habas fabricated the instrumented panels with the support of Sidney Goossens, and both contributed in the interpretation of the results. Thomas Geernaert, Hugo Thienpont, and Francis Berghmans initiated the research and helped in interpreting the results, and in revising and writing the paper. Conflicts of Interest: The authors declare no conflict of interest.
References 1.
2. 3. 4. 5.
6. 7.
Zhao, X.; Gao, H.; Zhang, G.; Ayhan, B.; Yan, F.; Kwan, C.; Rose, J.L. Active health monitoring of an aircraft wing with embedded piezoelectric sensor/actuator network: I. Defect detection, localization and growth monitoring. Smart Mater. Struct. 2007, 16. [CrossRef] Paget, C.A.; Levin, K.; Delebarre, C. Actuation performance of embedded piezoceramic transducer in mechanically loaded composites. Smart Mater. Struct. 2002, 11. [CrossRef] Su, Z.; Lin, Y.; Ye, L. Guided Lamb waves for identification of damage in composite structures: A review. J. Sound Vib. 2006, 295. [CrossRef] Betz, D.C.; Thursby, G.; Culshaw, B.; Staszewski, W.J. Advanced layout of a fiber Bragg grating strain gauge rosette. J. Lightwave Technol. 2006, 24. [CrossRef] Takeda, N.; Okabe, Y.; Kuwahara, J.; Kojima, S.; Ogisu, T. Development of smart composite structures with small-diameter fibre Bragg grating sensors for damage detection: Quantitative evaluation of delamination length in CFRP-laminates using Lamb wave sensing. Compos. Sci. Technol. 2005, 65. [CrossRef] Tsuda, H.; Toyama, N.; Urabe, K.; Takatsubo, J. Impact damage detection in CFRP using fiber Bragg gratings. Smart Mater. Struct. 2004, 13. [CrossRef] Takeda, S.; Okabe, Y.; Takeda, N. Delamination detection in CFRP laminates with embedded small-diameter fiber Bragg grating sensors. Compos. Part A 2002, 33. [CrossRef]
Sensors 2017, 17, 1948
8. 9. 10. 11. 12. 13. 14. 15. 16. 17.
18.
19.
20.
21. 22. 23. 24. 25.
9 of 9
Su, Z.; Ye, L. Identification of Damage Using Lamb Waves; Springer: Hong Kong, China, 2009. Wu, W.; Yu, H.; Chen, W. Indentation responses of piezoelectric layered half-space. Smart Mater. Struct. 2013, 22. [CrossRef] Hayashi, T.; Kawashima, K. Single Mode Extraction from Multiple Modes of Lamb Wave and Its Application to Defect Detection. J. Solid Mech. Mater. Eng. 2003, 46. [CrossRef] Zhang, H.-Y.; Yu, J.-B. Piezoelectric transducer parameter selection for exciting a single mode from multiple modes of Lamb waves. Chin. Phys. B 2011, 20. [CrossRef] Frazao, O.; Santos, J.; Araujo, F.; Ferreira, L. Optical sensing with photonic crystal fibers. Laser Photonics Rev. 2008, 2. [CrossRef] Canning, J. Properties of Specialist Fibres and Bragg Gratings for Optical Fiber Sensors. J. Sens. 2009, 871580. [CrossRef] Canning, J.; Groothoff, N.; Cook, K.; Pohl, A.; Holdsworth, J.; Bandyopadhyay, S.; Stevenson, M. Grating Writing in Structured Optical Fibres. Laser Chem. 2008, 239417. [CrossRef] Canning, J. Fibre Gratings & Devices for Sensors & Lasers. Lasers Photonics Rev. 2008, 2. [CrossRef] Pinto, A.; Lopez-Amo, M. Photonic Crystal Fibers for Sensing Applications. J. Sens. 2009, 598178. [CrossRef] Berghmans, F.; Geernaert, T.; Sonnenfeld, C.; Sulejmani, S.; Thienpont, H. Microstructured Optical Fibre-Based Sensors for Structural Health Monitoring Applications, Optofluidics, Sensors and Actuators in Microstructured Optical Fibres; Woodhead Publishing: Cambridge, UK, 2015; p. 139. Sonnenfeld, C.; Luyckx, G.; Sulejmani, S.; Geernaert, T.; Eve, S.; Gomina, M.; Chah, K.; Mergo, P.; Urbanczyk, W.; Thienpont, H.; et al. Microstructured optical fiber Bragg grating as an internal three-dimensional strain sensor for composite laminates. Smart Mater. Struct. 2015, 24. [CrossRef] Sonnenfeld, C.; Sulejmani, S.; Geernaert, T.; Eve, S.; Lammens, N.; Luyckx, G.; Voet, E.; Degrieck, J.; Urbanczyk, W.; Mergo, P.; et al. Microstructured optical fibre sensors embedded in laminate composite for smart material applications. Sensors 2011, 11, 2566. [CrossRef] [PubMed] Sulejmani, S.; Sonnenfeld, C.; Geernaert, T.; Luyckx, G.; van Hemelrijck, D.; Mergo, P.; Urbanczyk, W.; Chah, K.; Caucheteur, C.; Mégret, P.; et al. Shear stress sensing with Bragg grating-based sensors in microstructured optical fibres. Opt. Express 2013, 21, 20404. [CrossRef] [PubMed] Rose, J. Ultrasonic Waves in Solid Media; Cambridge University Press: Cambridge, UK, 1999. Cusano, A.; Cutolo, A. Fibre Bragg Grating Sensors: Recent Advancements, Industrial Applications and Market Exploitation; Bentham Science Publishers: Valencia, Spain, 2014. Sharif-Khodaei, Z.; Ghajari, M.; Aliabadi, M. Determination of impact location on composite stiffened panels. Smart Mater. Struct. 2012, 21, 105026. [CrossRef] Sharif-Khodaei, Z.; Aliabadi, M. Assessment of delay-and-sum algorithms for damage detection in aluminium and composite plates. Smart Mater. Struct. 2014, 23. [CrossRef] Lambinet, F.; Sharif-Khodaei, Z.; Aliabadi, M. Damage Detection in Composite Skin Stiffener with Hybrid PZT-FO SHM System. Key Eng. Mater. 2017, in press. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
