18th World Conference on Nondestructive Testing, 16-2 April 212, Durban, South Africa Ultrasonic Guided Wave Testing of Cylindrical Bars Masanari Shoji, Takashi Sawada NTT Energy and Environment Systems Laboratories, Nippon Telegraph and Telephone Corporation, 9-11, Midori-Cho 3-Chome Musashino-Shi, Tokyo 188585, Japan e-mail: shoji.masanari@lab.ntt.co.jp, sawada.takashi@lab.ntt.co.jp Abstract Small-diameter cylindrical bars are fundamental components in civil engineering and architectural structures. In some cases, ultrasonic testing is required from a small side area of the bar, instead of from the end surface. Ultrasonic guided wave testing on 13-mm-diameter steel bars has been investigated experimentally by using piezoelectric transducers (3, 6,12 and 22 khz) attached to the sides of the bars. It is shown that longitudinal and flexural modes can be generated individually by appropriate phase control of voltages applied to the transducers. Each guided wave mode has been verified by both propagating velocity (dispersion curve) and a symmetric property of the displacement of the test surface. Various types of axisymmetric cross-sectional losses (minimum diameter, length in axis direction, and shape) have been made on the bars as artificial reflectors, and the detection performance of the reflectors by means of the guided waves has been evaluated by using up to two or four transducers at each frequency. Keywords: Ultrasonic testing (UT), cylindrical bar, guided wave, longitudinal mode, flexural mode, piezoelectric transducer 1. Introduction Small-diameter cylindrical bars are fundamental components in civil engineering and architectural structures and there are many applications of non-destructive testing of such objects. Various-bolt or tendon inspection is a typical application of ultrasonic nondestructive testing of bars. Thus far, ultrasonic testing of such bars has been studied mostly by using bulk ultrasonic waves and usually conducted from the free end surface of the bars [1]. In some cases, however, ultrasonic testing from a small side area of the bar is required because of structural restrictions. Moreover, ultrasonic testing by means of ultrasonic guided waves could be promising especially for long-range inspections [2]. There have been studies of ultrasonic guided wave testing of bars [3-6], but the guided waves are basically generated in the vicinity of the end surface of bars. There have been several researches of ultrasonic guided wave testing that could be conducted from a side area of bars [7, 8]. However, magnetostrictive transducers (MTs) or electromagnetic acoustic transducers (EMATs) have been utilized to excite guided waves and there have not been many studies on using piezoelectric transducers attached on the sides of small-diameter bars. In general, both piezoelectric transducers and MTs or EMATs have their own advantages and disadvantages. MTs or EMATs can be installed more easily without a coupling medium. Piezoelectric transducers can easily generate large amplitude waves and can detect detailed properties of test surface displacements by using several transducers mounted at different locations in the circumferential direction on the surface. In this context, we have investigated ultrasonic guided wave generation and testing on 13-mm-diameter steel bars by using piezoelectric transducers attached to the sides of the bars.
2. Experimental setup 2.1 Transducers and Wedges The dispersion curves of longitudinal mode (L mode) and flexural mode (F mode) for a 13-mm-diameter cylindrical steel bar are shown in Figs. 1 and 1. In this evaluation, bulk longitudinal and transverse wave velocities were assumed to be 59 m/s and 32 m/s. In this low frequency region, the dispersion curves are rather simple and the group velocities of L(,1) and F(1,1) modes are not so dispersive, less than approximately 1 khz and more than approximately 1 khz, respectively. Since simple dispersion curves make signal interpretation straightforward and a non-dispersive nature has advantages for long-range inspections or high resolution, four frequencies of 3, 6, 12 and 22 khz were selected as the design values of piezoelectric transducers. We made two longitudinal wave transducers for 3 and 22 khz and four longitudinal wave transducers for 6 and 12 khz. These transducers were commonly used for generating both L and F modes and their diameters were 5 mm. Wedges for attaching the transducers to the sides of bars are made of acrylic material and have different designs for L and F modes regarding incident angles. The incident angles of the waves to the bar for generating L(,1) mode waves were determined by Snell s law assuming a plane boundary between the bulk longitudinal wave in the wedges and the L(,1) mode waves. On the other hand, the incident angle for generating F(1,1) mode was selected to be degrees. Since the phase velocities of F(1,1) mode in the bar are lower than the velocity of the bulk longitudinal wave in the wedge of acrylic material, it is impossible to choose the incident angle for F(1,1) mode in the same manner as for L(,1) mode. Including the selection of the incident angles for both modes, it is not straightforward to optimize the design of the transducers and wedges attached to the sides of the bars for generating specific guided waves, because wavelength of the bulk longitudinal wave in the wedge is comparable to or larger than the diameter of the cylindrical bar. Accordingly, the transducers and wedges used in the following experiments are just prototypes, and further examination and improvements are required for practical use. Velocity (m/s) 8 6 4 L(,1) Group L(,2) Phase L(,1) Phase Velocity (m/s) 8 6 4 F(1,2) Phase F(1,1) Group F(1,3) Phase 2 2 L(,2) Group 1 2 3 4 Frequency (khz) F(1,1) Phase F(1,2) Group F(1,3) Group 1 2 3 4 Frequency (khz) Figure 1. Dispersion curves of Longitudinal mode (L mode) and flexural mode (F mode) for perfectly elastic 13-mm-diameter cylindrical bar. Solid and dashed lines indicate group and phase velocities, respectively.
2.2 Test objects with and without axisymmetric cross-sectional losses Test objects were 13-mm-diameter 2.4-m-long cylindrical steel bars having flat end surfaces normal to the axis (SS4 in Japanese Industrial Standard). In addition to an intact bar, we prepared eight bars having different types of cross-sectional losses for evaluating the detection performance of cross-sectional losses by means of guided waves. The losses were axially symmetric and consisted of two minimum diameters (8 and 6 percent, residual percentages of diameter), two lengths (1 mm and 1 mm) in the axis direction and two shapes (rectangular and V notch), as shown in Fig. 2 and Table 1. The locations of the centre of all the cross-sectional losses were 5 mm from the end surface of the bars. 24 mm 5 mm d = 13 mm Length in axis direction Length in axis direction Table 1. Test objects of 13-mm-diameter 2.4-meter-long cylindrical steel bars with axisymmetric cross-sectional losses Test object No. d Length (mm) d = 13 mm Figure 2. Side view of bar with axisymmetric cross-sectional loss (V notch), side view of rectangular notch, (c) side view of V notch and (d) photograph of axisymmetric cross-sectional losses Axisymmetric cross-sectional losses Shape of side view Minimum diameters (residual percentage of diameter, 1 x d /d) Length in axis direction (mm) Distance between end surface and centre of cross-sectional loss (mm) 1 24 Rectangular 6 1 5 2 24 notch 8 1 5 3 24 6 1 5 4 24 8 1 5 5 24 V notch 6 1 5 6 24 8 1 5 7 24 6 1 5 8 24 8 1 5 d (c) d = 13 mm (d) 2.3 Ultrasonic test equipment and operation Figure 3 illustrates a typical experimental setup using two transducers attached to the side of a test object. The excitation signals applied to the transducers by an ultrasonic test instrument were one cycle square-wave pulse of each frequency with the same voltage amplitude. Glycerin paste was used as the coupling medium. The ultrasonic test instrument
was operated using a PC and received signal data was transmitted to the PC. The received signals were averaged for eight measurements. Ultrasonic test instrument Control Data PC Transducers Test object Figure 3. Schematic view of experimental setup 3. Experimental results and discussion 3.1 L(,1) mode generation by using two transducers Figures 4 and 4 present pulse echo results of an intact bar by using the single probe technique with two transducers of 6 and 12 khz, respectively. The two transducers were put on the upper and lower surfaces of the bar with L-mode wedges, as shown in Fig. 5. The distance between the back end surface of the bar and the probe index was determined to be one wavelength of the phase velocity of L(,1) mode at each frequency to avoid signal reduction by reflection from the back end surface. In this measurement, the excitation signals applied to the two transducers were exactly the same (i.e. the phase difference was ) and the pulse echo results were the sum of the two signals received by the two transducers. Since the bar did not have any flaws or cross sectional loss, all echoes were due to reflections from the bar s ends. The wave velocities were evaluated using the time intervals of the multiple reflection echoes and are plotted for all four frequencies in Fig. 6 together with the group velocity dispersion curve of the L(,1) mode. The velocities agree well with the calculated group velocity dispersion curve of the L(,1) mode. Amplitude 1.5 -.5-1 1 2 3 4 5 Amplitude 1.5 -.5-1 1 2 3 4 5 Figure 4. Pulse echo results by using single probe technique with two transducers for 6-kHz L mode and 12-kHz L mode One wavelength Probe index Transducers bar Figure 5. Schematic view of two-transducer arrangement for single probe technique and photograph of 12-kHz transducers and L-mode wedges
8 Velocity (m/s) 6 4 2 1 2 3 4 Frequency (khz) Figure 6. Evaluated velocities (circles) with pulse echo results and calculated group velocity dispersion curve of L(,1) mode In order to examine the surface displacements of the waves, we also evaluated the pulse echo signals received by a single transducer. The measurement was conducted using the double probe technique with four transducers. The arrangement of the four transducers is illustrated in Fig. 7. Waves were generated by two transmitting transducers (Tt1 and Tt2) and received by two receiving transducers (Tr1 and Tr2). The distance between the back end surface of the bar and the receiving transducer s probe index was one wavelength of the phase velocity of L(,1) mode and the excitation signals were applied to Tt1 and Tt2 at the same time (the phase difference was ). Figures 8 and 8 exhibit pulse echo results obtained by Tr1 and Tr2 for 12-kHz L mode, respectively. These results are almost identical. Figures 8(c) and 8(d) respectively represent the results of the sum of the two pulse echo results (Tr1+Tr2) and their difference (Tr1-Tr2). In Fig. 8(c), the signal amplitudes of multiple reflection echoes have increased by a factor of about 2 compared with those in Fig. 8 or One wavelength Receiving transducers (Tr1 and Tr2) Transmitting Transducers (Tt1 and Tt2) (c) Figure 7. Four-probe arrangement for double probe technique Figure 8. Pulse echo results received by Pr1 and Pr2 transducers for 12-kHz L mode, and results of (c) sum (Pr1+Pr2) and (d) difference (Pr1-Pr2) of two pulse echo results bar (d)
8, and in Fig. 8(d) these signals have disappeared almost completely, which indicates that the upper and lower surface displacements of these waves are symmetric with respect to the centre of the bar. This symmetric displacement agrees with that of the L(,1) mode. Consequently, the generation of the L(,1) mode has been verified by both propagating velocity and a symmetric property of the displacement of the test surface. According to the normal mode expansion concept [9, 1], it is expected that symmetric surface tractions by transducers generate symmetric displacement wave mode (L mode), which is consistent with the above results. 3.2 F(1,1) mode generation by using two transducers Figures 9 and 9 show pulse echo results of an intact bar by using the single probe technique with two transducers of 6 and 12 khz, respectively. The two transducers were put on the upper and lower surfaces of the bar with F-mode wedges, as shown in Fig. 5. The distance between the back end surface of the bar and the receiving transducer s probe index was determined to be one wavelength of the phase velocity of F(1,1) mode at each frequency to avoid signal reduction by reflection from the back end surface. In this measurement, excitation signals were applied to the two transducers with the time difference of half a period of each frequency (i.e. the phase difference was π) and the pulse echo results are the difference of the two signals received by the two transducers. The wave velocities were evaluated using the time intervals of the multiple reflection echoes and are plotted for all four frequencies in Fig. 1 together with group velocity dispersion curve of the F(1,1) mode. The velocities agree with the calculated group velocity dispersion curve of the F(1,1) mode. Amplitude Amplitude Figure 9. Pulse echo results by using single probe technique with two transducers for 6-kHz F mode and 12-kHz F mode 8 Velocity (m/s) 6 4 2 1 2 3 4 Frequency (khz) Figure 1. Evaluated velocities (circles) by using pulse echo results and calculated group velocity dispersion curve of F(1,1) mode The pulse echo signals received by a single transducer were also evaluated by using the double probe technique with four transducers. The arrangement of the four transducers was the same as shown in Fig. 7. Waves were generated by two transmitting transducers (Tt1 and Tt2) and received by two receiving transducers (Tr1 and Tr2). The distance between the back end surface of the bar and the receiving transducer s probe index was one wavelength of the
phase velocity of F(1,1) mode and the excitation signals were applied to Tt1 and Tt2 with the time difference of half a period (the phase difference wasπ). Figures 11 and 11 exhibit pulse echo results obtained by Tr1 and Tr2 for 6-kHz F mode, respectively. Figures 11(c) and 11(d) respectively represent the results of the sum of the two pulse echo results (Tr1+Tr2) and their difference (Tr1-Tr2). In Fig. 11(c), the signal amplitudes of the multiple reflection echoes have disappeared almost completely, and in Fig. 11(d) these signals have increased by a factor of about 2 compared with those in Fig. 11 or 11, which indicates that the upper and lower surface displacements of these waves are antisymmetric with respect to the centre of the bar. This antisymmetric displacement agrees with that of the F(1,1) mode. Consequently, The generation of the F(1,1) mode has also been confirmed by both propagating velocity and a symmetric property of the test surface displacement. The F mode generation is consistent with the normal mode expansion concept [9, 1]. (c) Figure 11. Pulse echo results received by Pr1 and Pr2 transducers for 6-kHz F-mode, and results of (c) sum (Pr1+Pr2) and (d) difference (Pr1-Pr2) of two pulse echo results (d) 3.3 Ultrasonic guided wave testing of artificial reflectors In order to assess the potential of ultrasonic testing of bars by means of the L(,1) and F(1,1) modes, we evaluated the detection performance for artificial reflectors with the test Amplitude Amplitude S S F No. 1 S F No. 2 S F No. 3 S F No. 4 F F No. 5 S F No. 6 S No. 7 F No. 8 Figure 12. Pulse echo results of eight test objects by means of 12-kHz L(,1) mode by using double probe technique with four transducers. Numbers indicate test objects. S and F denote signal from artificial reflector and first reflection signal from end surface, respectively.
objects described in Sec. 2.2. Figure 12 represents the pulse echo results of the eight test objects by means of 12-kHz L(,1) mode. The measurements were conducted by using the double probe technique with four transducers. As can be seen in these figures, except for test object No. 8, all the artificial reflectors have been definitely detected as echo signals before the first reflection signal from the end surface. All echo height results by using the single probe technique with two transducers are summarized in Figs. 13 and 14. In these figures, the echo heights have been normalized with respect to the value of test object No. 3 for each mode at each frequency, and no data denotes that definite echoes from the reflector were not 3 khz 6 khz (c) 12 khz 22 khz (d) Figure 13. Echo heights for eight test objects by means of L(,1) mode by using single probe technique with two transducers. Echo heights were normalized with respect to value of test object No. 3 at each frequency. No data indicates that definite echo from cross-sectional loss was not detected. 3 khz 6 khz (c) 12 khz 22 khz (d) Figure 14. Echo heights for eight test objects by means of F(1,1) mode by using single probe technique with two transducers. Echo heights were normalized with respect to value of test object No. 3 at each frequency. No data indicates that definite echo from cross-sectional loss was not detected.
detected [for example, test object No.s 7 and 8 for 6-kHz L(,1) mode in Fig. 13]. The 12-kHz L(,1) mode [Fig. 13(c)] exhibited the highest detection performance since this mode detected all the reflectors except test object No. 8. As can be seen in Figs. 13, 14 and 14(d), the 6-kHz L(,1) mode and 6- and 22-kHz F(1,1) modes detected all the reflectors except test objects No.s 7 and 8. However, it should be noted that no data does not necessarily mean that the reflector cannot be detectable in principle by the guided wave. This is because there are many uncontrolled factors affecting the detection performance. For example, quantitative evaluation of the relation between guided-wave generation efficiency and the detection performance of the cross-sectional losses is a subject for future study. We also examined pulse echo signals by using four transmitting transducers. Figure 15 compares echo heights by using the single probe technique with four and two transducers for 12-kHz L(,1) mode. Because the number of both transmitting and receiving transducers are doubled when using four transducers, the expected ratio of echo heights between four and two transducers is about four. The results shown in Fig. 15 agree with this expectation, which indicates that increasing the number of transducers is effective for increasing the signal-tonoise ratio. Two transducers Four transducers Figure 15. Comparison of echo heights by using single probe technique with two and four transducers for 12-kHz L(,1) mode 4. Conclusions In order to examine the potential of ultrasonic guided wave testing of small-diameter cylindrical bars by using piezoelectric transducers attached to the sides of the bar, we made prototype transducers and wedges for low frequency (3, 6, 12 and 22 khz) guided waves, and experimentally investigated the generation of guided waves and the detection performance of axisymmetric cross-sectional losses on 13-mm-diameter cylindrical steel bars. It has been shown that L(,1) and F(1,1) modes can be generated individually by appropriate phase control of voltages applied to the transducers. Each guided wave mode was verified by both propagating velocity (dispersion curve) and a symmetric property of the displacement of the test surface. Eight types of axisymmetric cross-sectional losses were made on the bars as artificial reflectors, and the detection performance of the reflectors by means of the guided waves was evaluated. It has been found that almost all the artificial reflectors can be detectable if an appropriate guided wave mode and frequency are selected. The results show sufficient potential for ultrasonic guided wave testing of small-diameter cylindrical bars by using piezoelectric transducers attached to the sides of the bars.
Acknowledgements The authors would like to thank K. Hayashi for his technical advice and assistance throughout the experiments. References 1. G. B. Niles, 'In Situ Method of Inspecting Anchor Rods for Section Loss using the Cylindrically Guided Wave Technique', Power Delivery, IEEE Transactions on Vol. 11, No. 3, pp 161-165, July 1996. 2. J. L. Rose, 'A baseline and vision of ultrasonic guided wave inspection potential', J. Pressure Vessel Technol., Vol. 124, pp 273-282, August 22. 3. M. Rucka, 'Experimental and numerical studies of guided wave damage detection in bars with structural discontinuities', Arch. Appl. Mech., Vol. 8, pp 1371-139, 21. 4. M. D. Beard, M. J. S. Lowe and P. Cawley, 'Ultrasonic guided waves for inspection of grouted tendons and bolts', ASCE J. Mat. in Civ. Eng., Vol. 15, pp 212-218, May/June 23. 5. W. B. Na, T. Kundu, and M. R. Ehsani, 'Ultrasonic Guided Waves for Steel Bar Concrete Interface Testing', Mater. Eval., Vol. 6, pp 437-444, March 22. 6. W. B. Na and T. Kundu, 'A Combination of PZT and EMAT for Interface Inspection', J. Acoust. Soc. Am., Vol. 111, No. 5, pp 2128-2139, May 22. 7. L. Laguerre, J. C. Aime and M. Brissaud, 'Magnetostrictive pulse-echo device for nondestructive evaluation of cylindrical steel materials using longitudinal guided waves', Ultrasonics, Vol. 39, No. 7, pp 53-514, April 22. 8. H. Kwun, S. Y. Kim and G. M. Light, 'Long-Range Guided Wave Inspection of Structures Using the Magnetostrictive Sensor', J. Korean Soc. NDT, Vol. 21, pp 383-39, 21. 9. J. J. Ditri and J. L. Rose, 'Excitation of guided elastic wave modes in hollow cylinders by applied surface tractions', J. Appl. Phys., Vol. 72, No. 7, pp 2589-2597, October 1992. 1. J. Li and J. L. Rose, 'Excitation and propagation of non-axisymmetric guided waves in a hollow cylinder', J. Acoust. Soc. Am., Vol. 19, No. 2, pp 457-464, February 21.