Enhancement of directional sensitivity of magnetostrictive phased array sensor using a circular comb-shaped nickel patch

A magnetostrictive patch transducer (MPT) or magnetostrictive sensor (MsS) is an ultrasonic guided wave-based structural health monitoring tool alternative to well-known piezoelectric transducers. The MPT/MsS assessing the strain-induced magnetic property variation in ferromagnetic materials has been widely used for the nondestructive ultrasonic testing in waveguides such as cables, pipes, and plates used in engineering fields.^1,2 Various MPTs have been developed by specially configuring the sensing coils and biasing magnets for specific applications of the guided wave techniques.² Thin nickel, iron-cobalt alloy, and iron-gallium alloy (Galfenol) patches may be served as magnetostrictive materials which should be firmly bonded onto or into a waveguide structure for the guided Lamb wave (GLW) testing.^2–4 Our previous work has theoretically and experimentally demonstrated the advantages of the use of comb-shaped nickel patches in the practice of directional sensing operation of an omnidirectional MPT.⁵ The circular magnetic circuit device of the omnidirectional MPT system had no directional sensing preference due to its design concept of a pancake type sensing coil and cylindrical biasing magnet installations.⁶ The magnetic shape anisotropy feature in the consequence of the high-aspect-ratio comb finger formation to the nickel patch induced the directional sensitivity and sensing performance enhancement of the MPT for detecting GLWs in a metallic plate. Recently, Kim et al. performed extensive efforts to investigate the effects of the slits (machined gaps) presence in a circular nickel patch of the omnidirectional MPT on the transduction performance increase.⁷ 

In general, a phased array sensor developed for the GLW technique is based on piezoelectric materials due to their attractive actuation and sensing performance by the piezoelectric effect. Numerous studies have been conducted by using the piezoelectric-based phased array sensor to identify the structural integrity of metallic and composite panel structures.^8–13 There were no directional preferences in the individual sensing parts of the traditional phased array sensors. The directional sensing aspect of the array sensor was only based on the phased array signal processing algorithms. Salas and Cesnik devised a composite long-range variable-direction emitting radar (CLoVER) transducer as a phased array sensor, providing the directional GLW actuation and sensing capability.¹⁴ However, the phased array image resolution from the CLoVER transducer would be restricted by the total numbers of the individual sensors in the CLoVER transducer.

In this work, we present the development of a magnetostrictive phased array sensor (MPAS) based on a single magnetostrictive patch and investigate the improvement of the directional sensitivity and sensing performance of the MPAS by using a circular comb-shaped magnetostrictive patch machined from a polycrystalline nickel sheet. Some research groups developed phased sensor array systems using thin magnetostrictive strips/patches for the guided wave testing of pipes or plates.^15,16 However, their sensor arrays entirely surround the circumferential direction of pipes or occupy a relatively large area in plates. The presented MPAS utilizes one nickel patch with a diameter of 1” and a hexagon magnetic circuit device with a radius of 0.5” containing six sensing coils in order to inspect a thin aluminum plate with a dimension of 24” × 24”. Although the MPAS contains a fixed number of the sensing coils, the sensing positions in the nickel patch can be extended by simply altering the orientation of the magnetic circuit device. The MPAS’s useful feature provides higher resolution image results than the regular phased array sensor with a fixed number of sensors. To analyze GLW signals acquired by the MPAS, the wavenumber filtering technique is applied for the phased array signal processing approach.¹⁷ The MPAS using a nickel disc patch is also examined to highlight the improved GLW sensing effects on the practical use of the circular comb-shaped patch.

The developed MPAS contains two main components of a magnetostrictive nickel patch and a magnetic circuit device enclosing six sensing coils and vertically polarized cylindrical permanent magnets (biasing magnets). The magnetic circuit device is a detachable device, while the nickel patch should be permanently attached to the surface of a plate under GLW inspection. Two nickel patches with an overall disc shape (Ø1” × 0.02”) were prepared from a 99.5% nickel sheet (from Alfa Aesar). The circular comb-shaped nickel patch with 24 comb fingers was machined from an annular nickel patch by using a slow speed diamond saw. The angular difference of the individual comb fingers was 15° and its width and length were around 0.1” and 0.25”, respectively. As shown in FIG. 1, the nickel patches were firmly bonded with M-Bond 200 kit onto the top surface of a 2024-T3 aluminum plate (24” × 24” × 0.04”) and the patches were positioned at symmetrical locations of the plate in order to compare the associated directional sensing features based on GLW signal measurement and analysis data. Piezoelectric (PZT) discs (PSI-5A4E from Piezo Systems, Inc.) with a diameter of 0.25” were bonded at two different locations and used to generate omnidirectional GLWs in the plate structure. The individual PZT actuators were equally distant from the nickel patches. The MPAS was used as a passive sensor to capture GLWs initiated from the PZT actuator and traveling in the plate. To mitigate the wave reflections from the boundaries of the plate, which make difficult to analyze GLW signals, tacky tapes served as a damping material were mounted at the four edges of the plate’s bottom surface, which are not shown in the FIG. 1.

Although the nickel patches were permanently bonded to the host plate structure, the magnetic circuit device was a mobile instrument attached to the nickel patch by the magnetic force of the embedded biasing magnets. The six individual sensing coils were arranged in a hexagon with a radius of 0.5” as shown in FIG. 2(a) and the coil design parameters of the sensing coil are given in TABLE I. The detailed configurations of the components inside of the magnetic circuit device can be found in the 3D CAD drawing of FIG. 2(b). The distance between the center of the sensing coil and the center of the biasing magnet was 0.375” and the location information of the sensing coils was utilized for the phased array signal processing. Two collocated neodymium magnets with a dimension of Ø0.25” × 0.125” (D42-N52 from K&J Magnetics, Inc.) were used as the biasing magnets. A prototype of the hexagon magnetic circuit device was constructed as shown in FIG. 2(c). The coil bobbins and housing components in the device were fabricated by using an Eden 350 3D printer from Stratasys, Ltd. The sensing coils were wound on the printed bobbins by 150 turns with the 32 AWG magnet wires. The lift-off gap between the sensing coil and the surface-mounted nickel patch was set to 0.05” maintained by the 3D printed components. While the permanent magnets located at the center of the MPAS generate the static biasing magnetic fields along the radial direction of the nickel patch, the GLWs propagating through the magnetostrictive patch induce the dynamic magnetic fields that produce alternating currents in the sensing coils. The sensing coils of the MPAS detect the rate of change of magnetic property due to the deformed state of the magnetostrictive patch induced by the elastic wave propagation. The detailed sensing mechanisms of the MPAS can be found in our previous work.⁴ Although the magnetic circuit device comprises only six sensing coils, the sensing points may be multiplied as altering the orientation of the magnetic circuit device. For example, as shown in the FIG. 2(a), the total sensing points of the MPAS can be doubled if the sensing coils’ orientations are changed with a 30° angular direction in counter-clockwise manner.

Using National Instrument (NI) DAQ system with NI USB-6259 and a developed LabVIEW program, a 4.5-cycle Hanning windowed toneburst at a given excitation frequency was generated and linearly amplified by an EPA-104 linear amplifier (Piezo System, Inc.) before being sent to the PZT actuator to generate GLWs in the plate. The GLW signals captured by the MPAS were then amplified by a preamplifier prior to storing in the NI DAQ system. Four ultrasonic excitation frequency cases (60, 70, 80, and 90 kHz) were investigated to evaluate the sensing performance and directional sensitivity of the two MPASs. The excitation frequency range was selected to acquire relatively large A0 mode signals (the fundamental anti-symmetric mode) of GLWs. The input toneburst amplitude for the PZT actuation was ±80 V after the linear amplifier. In this work, the GLWs generated by the piezoelectric effect of the PZT actuator are captured by the magnetostriction effect of the MPAS by the induced strain related to the GLW propagation over the nickel patch.

The difference of magnetic flux density associated with the biasing magnetic field between the two nickel patches with different geometrical shapes was evaluated by COMSOL Multiphysics software as shown in FIG. 3. Only one quadrant of the model geometry was simulated to determine the nickel patches’ magnetic flux density characteristics, assuming symmetrical characteristics of the modelling components. The simulation results demonstrate that the magnetic flux density of the nickel disc patch gradually decreases along the radial direction of the patch. On the other hand, the magnetic flux density of the circular comb-shaped nickel patch is concentrated at around the starting area of the comb fingers and progressively decreases along the comb finger directions. The demagnetization factor strongly depending on the geometrical shapes of a magnetic material mainly results in the different magnetic flux density features shown in the FIG. 3. The demagnetization factor dramatically decreases as the aspect ratio of the nickel patch increases.¹⁸ This means that creating comb fingers in the nickel patch reduces the demagnetization factor and produces easy magnetization axis along the comb fingers’ directions. From the results, it is expected that such magnetic shape anisotropy effect of the circular comb-shaped nickel patch may improve the overall sensing performance of the associated MPAS due to the presence of the high-aspect-ratio fingers allowing to diminish the demagnetization factor along the radial direction of the patch. For instance, when the incoming GLWs travel along a certain comb finger direction, the strain induced by the elastic wave propagation generates the reorientation of magnetic domains in the magnetostrictive patch and the randomly oriented magnetic domains are preferably rotated and aligned along the comb finger direction since it is the easy magnetization axis in the nickel patch.

The MPAS enables to capture the GLW signals at six predetermined locations within the nickel patch. As changing the orientation of the magnetic circuit device, the associated MPAS is capable of additionally detecting the GLW signals at the altered locations of the nickel patch. FIG. 4 shows sample GLW signals corresponding to 70 kHz excitation frequency acquired from four distinct locations in the nickel patches. The GLW signals relative to $S1′$ and $S2′$ sensing positions were measured by using the S1 and S2 sensing coils with the rotated orientation of 30° counter-clockwise direction (see the FIG. 2(a)). The first waveforms before 0.05 ms are the cross-talk signals between the actuator and sensors (sensing coils) and should be ignored for the GLW signal processing since the cross-talk signals are almost identical for the individual sensors of the MPAS regarding the same PZT actuator. The blue solid and red dotted lines denote the GLW signals based on the disc and comb-shaped nickel patches, respectively. The highlighted waveforms in the FIG. 4 are the first arrival A0 mode signals of the GLWs from the PZT actuator directly to the MPAS’s individual sensors. And the other waveforms after the direct A0 mode signals are the GLW reflection signals bounced from the boundaries of the plate geometry. Because of the short GLW travel distance between the PZT2 actuator and the MPASs, the direct A0 mode signal arrives much earlier than the case of the PZT1 actuator located 14” apart from the MPASs. For both PZT actuator cases, the S1 sensing coil of the MPAS was aligned to the PZT actuator to demonstrate the GLW signal change effects such as phase shift and amplitude change depending on the sensing location variation. As expected, the highest responsive GLW signals were captured as the orientation of the sensing coil was precisely aligned to the PZT actuator. Also, the FIG. 4(a) displays that the GLW reflection signals of the MPAS using the nickel comb patch show greater responses than the disc patch case. On the other hand, the direct A0 mode signals of the MPAS using the nickel comb patch shown in the FIG. 4(b) appear to have larger waveforms than the disc patch case.

The directional sensing performance of the MPAS with the individual nickel patches was initially evaluated by determining the peak amplitude of the GLW signal measurement data obtained from 12 evenly distributed locations within the nickel patch by using the original and rotated sensing coils of the MPAS. FIG. 5 shows the directional sensitivity results of the MPAS using the disc and comb-shaped nickel patches according to the various excitation frequencies. The curves fits matching the measurement data was determined by the built-in interpolation function in MATLAB. First, we observed that the MPAS using the disc patch exhibited the symmetrical directional sensitivity along the PZT actuator and its opposite directions, while the MPAS using the comb-shaped patch showed apparent directional sensitivity to the PZT actuator direction. This characteristic of the nickel disc-based MPAS can be found for all other frequency cases. Also, the directional sensitivity of the nickel comb-based MPAS becomes broader along the opposite direction of the PZT actuator as the excitation frequency increases. And its directional sensitivity along the PZT actuator direction shows insignificant variation for the increase of the excitation frequency. Further research should be conducted to clearly understand the broadening effect of the incoming GLW signals along the opposite direction of the PZT actuator in the case of the nickel comb patch. Most importantly, the FIG. 5 validates that the nickel comb-based MPAS appears to have the highest sensitivity to the incoming GLW A0 mode when the both orientations of the sensing coil and the comb fingers are aligned to the GLW actuation source. In contrary, the lowest sensitivity of the nickel comb-based MPAS is observed when the both orientations of the sensing coil and the comb fingers are aligned perpendicular to the GLW actuation source direction.

The wavenumber filtering technique¹⁷ was employed to analyze the GLW signals acquired from the MPASs. To emphasize the difference between the disc- and comb-based MPASs, the phased array image results with a selected color scaling range are displayed in FIG. 6 according to the two PZT actuations. In the figures, the white solid arrow indicates the active PZT actuator direction. The original phased array image with a full color scaling range is also shown in the upper right corner of each figure. The actuation PZT locations were clearly identified by the highest magnitude waveform areas in the array images. There are two shadow waveform areas in the similar time zone corresponding to the main waveform area of the PZT actuator, due to the side lobe effect during the phased array signal processing. Other waveforms in the figures represent the boundary-reflected GLWs and their shadow waveforms. All of the array images show good agreements with the actuated PZT locations. Since the nickel disc and comb patches were attached to the opposite locations with respect to the PZT2 actuator, the array images in FIG. 6(b) and (d) are shown as the mirrored images. The array image results of the MPAS using the comb patch validate its improved sensing performance and directional sensitivity by apparently identifying the GLW actuation sources and its boundary reflections. The array image resolution can be enhanced by increasing the total numbers of sensing locations. As mentioned before, the sensing locations of the MPAS can be multiplied by simply altering the orientation angle of the hexagon magnetic circuit device with the same nickel patches. Note that the directional sensing characteristics of the MPAS using the nickel disc patch was solely rooted on the directionality of the sensing coils in the magnetic circuit device. However, the improved sensing feature and directionality characteristics of the nickel comb-based MPAS were the great combination work of the magnetic shape anisotropy effect of the comb fingers and the predefined directionality of the sensing coils.

This paper presented the development of the magnetostrictive phased array sensor (MPAS) for the ultrasonic guided Lamb wave (GLW) inspection technique. The MPAS based on the circular comb-shaped nickel patch and the hexagon magnetic circuit device enclosing six sensing coils with distinct directional preferences was designed and prototyped in laboratory. The benefits of the presented MPAS were validated by the experimental tests to detect GLWs generated by two PZT actuators in the aluminum plate. The improved sensing performance and directional sensitivity of the nickel comb-based MPAS were demonstrated by comparing the measurement and analysis data with the MPAS using the nickel disc patch. The high magnetic shape anisotropy feature along the individual comb finger directions of the nickel comb patch resulted in the improvement of the proposed MPAS.

The authors would like to thank Dr. Suok-Min Na at University of Maryland for his discussion about the magnetic shape anisotropy of ferromagnetic materials.