Piezoelectric transducer excitation for guided waves propagation on pipeline with flexural wave modes

9 th European Workshop on Structural Health Monitoring July 10-13, 2018, Manchester, United Kingdom Piezoelectric transducer excitation for guided waves propagation on pipeline with flexural wave modes More info about this article: http://www.ndt.net/?id=23344 Xudong Niu 1,2, Hua-Peng Chen 2 and Hugo R. Marques 3 1 National Structural Innovation Research Centre, TWI Ltd, Granta Park, Cambridge CB21 6AL, UK, x.niu@greenwich.ac.uk 2 Department of Engineering Science, University of Greenwich, Chatham Maritime, Kent ME4 4TB, UK, hp.chen@outlook.com 3 Integrity Management Group, TWI Ltd, Granta Park, Cambridge CB21 6AL, UK, Hugo.Marques@twi.co.uk Abstract Ultrasonic guided wave inspection is one of the non-destructive testing (NDT) techniques available for the structural health monitoring of engineering structures. Compared with other NDT techniques, guided waves can propagate over tens of metres with a relatively high sensitivity to defects in the structure. For pipeline inspections, by using an array of transducers, guide waves enable the identification of the location of defects on the pipe circumference. The general sensitivity range of the operation is 3% - 9% reduction of the cross-sectional area, depending on the signal-to-noise ratio. However, optimisation of guided wave testing method is still a requirement, as the technique is currently subject to a complex analysis due to wide number of guided wave modes generated. This can be done by optimising the transducer array design. In this paper, the behaviour of a single piezoelectric transducer upon excitation in a tubular structure is assessed. This is achieved through a combination of finite element analysis and experimental studies. A group of wave modes are considered from the torsional T(0,1) mode to several flexural wave modes. The behaviour of a single transducer element is evaluated at a specific frequency of 35 khz, as it is the interaction of the generated wave modes with simulated notches. The core objective of the work is to use the information gathered to optimise the design of transducer arrays aimed at exciting the T(0,1) mode with a significant level of mode purity. This will significantly reduce the complexity of guided wave analysis, enhancing effectively the health evaluation of structures, and subsequently reduce the industry maintenance cost. 1. Introduction In NDT techniques, structural integrity evaluation of pipelines by using ultrasonic guided waves is particularly attractive since the outstanding advantages of guided waves is to propagate over tens of metres along pipelines with a high sensitivity and low attenuation under ideal conditions. In general, the operated frequency range of guided Creative Commons CC-BY-NC licence https://creativecommons.org/licenses/by-nc/4.0/

wave (GW) testing is from 20 khz to 100 khz [1], and Mudge and Speck [2] evaluated that a 3% - 9% of cross-sectional area reduction of a pipe wall-thickness can be detected by using guide wave testing. However, the higher level of sensitivity in pipe inspection is still an area of interest for industry. The core objective of guided wave testing is to only generate axisymmetric guided waves for defect detection. It needs a transmitter with 360 degrees uniform loading on the pipe circumference [3]. However, many commercial tools have gaps between their transducers [4]. Therefore, the exisitng technique is subject to a complex analysis due to wide number of non-axisymmetric wave modes generated. This can be solved by optimising transducer design. The wave mode excitation and propagation in pipeline [5] and evaluation of structural integrity for pipe inspection [6,7] have been widely investigated. Gazis [8] investigated the propagation of three axisymmetric wave modes with their groups of nonaxisymmetric wave modes in a tubular structure. A nomenclature of X(m,n) is used to describe these wave modes [9], in which X represents wave mode characteristics, namely L, T and F for longitudinal, torsional and flexural wave modes, respectively; m indicates the circumferential order (m = 1, 2, 3, ), and n indicates the wave mode order of occurrence (n = 1, 2, 3, ). The dispersion curve to describe the phase velocity of wave modes is shown in Fig. 1 [10,11, 12]. The material of a steel pipe includes 219.1 mm (8 inches) outer diameter and 8.18 mm (schedule 40) wall thickness. The selected frequency range is from 0 to 100 khz. As shown in Fig. 1, the fundamental wave modes L(0,1), T(0,1) and L(0,2) are followed by three families of flexural wave modes F(m,1), F(m,2) and F(m,3), respectively. Since the total field in a pipe is the superposition of all the generated wave modes with different phase velocities, the difference in phase velocities of wave modes leads to a variation of the phase match among the different wave modes [13]. Figure 1. Phase velocity dispersion curves for an 8-inch schedule 40 steel pipe In this study, the behaviour of a single piezoelectric transducer upon excitation in a pipe is analysed. The core objective of the work is to use the information gathered to optimise the design of transducer arrays aimed at exciting an axisymmetric wave mode 2

with a significant level of mode purity. Commercial finite element analysis (FEA) software ABAQUS is adopted to simulate guided wave propagation on a 5000 mm long, outer diameter 219.1 mm, wall-thickness 8.18 mm steel pipe. A single transducer is modelled to produce shear traction in the circumferential direction at the top of the pipe to generate torsional type guided waves T(m,1), composed of the foundational torsional wave mode T(0,1) and torsional type flexural wave modes F(m,2). The angular profiles by using polar plots illustrate the displacement distribution around the pipe circumference at the different distances from the transducer location in the axial direction. The sensitivity of defect on the single transducer at a specific frequency of 35 khz is evaluated, as it is the interaction of the generated wave modes with simulated notches. To reduce signal-to-noise ratio, the received signals from 24 equally spaced points in the pipe circumference are superposed to suppress the undesired wave modes. Finally, the experimental validation is undertaken by using a commercially available tool, composing of three circumferential transducer arrays (i.e. collars). Each collar includes 24 piezoelectric transducers with a gap of 33 degrees between start and end transducers. The torsional type guided waves are generated and propagated on a 4450 mm, 8-inch, schedule 40 steel pipe. 2. Theoretical background The theoretical background has been developed for guided wave excitation and propagation in the axial direction of a hollow cylinder. If the structure is assumed as a pipe-like structure with traction-free boundary conditions, the Navier s governing wave equation for guided wave propagation in an elastic medium can be written as [8] 2 2 λ + µ U + µ U = ρ 2 ( ) U t (1) where λ and µ are Lamé constants, t and ρ indicate the time and the density, U 2 represents the displacement vector, represents the three-dimensional Laplace operator. The assumptions of complete solutions for torsional waves are considered as [13] Φ= f() r e e imθ i( kz ωt) H = h r e e m= r imθ i( kz ωt ) r( ) ( 0,1, 2...) H = h () r e e θ θ imθ i( kz ωt ) (2) H = h () r e e z z imθ i( kz ωt ) where Φ represents the dilatational scalar potential and Hξ ( ξ = r, θ, z) represents the displacement fields in the radial, circumferential and axial directions, respectively. The terms of hξ ()( r ξ = r, θ, z) are the corresponding displacement amplitudes composed of 3

Bessel functions in the radial direction; kz represents the wavenumber at the distance in the axial direction, and ω represents the radian frequency and i = 1. In the finite element (FE) modelling and experimental studies, torsional waves T(m,1) are created by a traction loading source in the circumferential direction and they were undertaken by a Hanning windowed tone burst signal, given by 1 2π ft u( t) = [1 cos( )]sin(2 π ft) 2 n c (3) where t is the time, f is the centre frequency of the transmitted signal, and n c is the number of the pulse signal cycles. Figure 2. Guided wave signal excitation for a single transducer: (a) time domain (b) frequency spectrum As shown in Figs. 2(a)-(b), the waveform of the transmitted signal is excited as a 10- cycle Hanning-modulated sinusoidal signal at 35 khz by using a single transducer. The selected number of waveform cycles aims to reduce the pulse signal bandwidth. The centre frequency of 35 khz was chosen as a low frequency signal in the operated frequency range. The excitation signal at 35 khz has been evaluated [14], and it is shown that it has a higher energy distribution in the circumferential direction for pipe inspection in the operation frequencies. 3. Finite element numerical modelling A scheme of finite element simulations is described, as shown in Fig. 3. A 5000 mm long, 8-inch, schedule 40 steel pipe for pipe inspection is modelled and its linear 3 isotropic material is simulated with a mass density ρ = 7932 kg/ m, Young s modulus E = 216.9 GPa and Poisson s ratio υ = 0.2865. A single transducer is placed at 1100 mm away from the left pipe end. The transducer as point source is designed to produce shear traction in the pipe circumferential direction to generate torsional type guided waves at 4

the top of the pipe. The waveform of the pulse signal is excited by a 10-cycle Hanning windowed pulse with the central frequency of 35 khz and bandwidth of ±7 khz. The receiving arrays, composed of 24 uniformly spaced point receivers, are located at 0 mm, 1000 mm and 2000 mm away from the transducer, respectively. The No.1 and No. 7 receivers are located at 90 degrees and 0 degree in the cross-sectional area of the pipe model. Figure 3. FE modelling of an 8-inch schedule 40 steel pipe with a part-circumferential throughthickness notch The three through-thickness notch sizes are simulated in the pipe, i.e. 30 degrees (Case No.1), 60 degrees (Case No.2) and 90 degrees (Case No.3) at the top of the pipe circumference. The notch length is 4 mm in the axial direction, and the location of notches is 3000 mm away from the transducer. The absorbing regions of 600 mm are constructed for avoiding the reflection from the two pipe ends. The maximum length of the elements is selected to be less than one-eighth per wavelength in the axial direction for satisfactory accuracy [5]. Therefore, the mesh distribution is arranged as an average mesh size of 4 mm in the axial direction and 4 elements in the pipe wall-thickness. Meshes are generated with a total of 839,944 Hex elements (Case No.1), 839,888 Hex elements (Case No.2) and 839,832 Hex elements (Case No.3). The impulse time of the transmitted signal is 286 µs and the total time of wave propagation is 2.7 ms. The step time is 200 ns. The angular profiles of the T(m,1) mode group at different axial distances for the point source loading on the 8-inch schedule 40 steel pipe at 35 khz are shown in Figs. 4(a)- (c). It shows that a narrower focused beam is generated upon a partial loading at 90 degrees, when the axial distance is located at the transducer location in Fig. 4(a). Then, the energy spreads out as the propagated distance increases, as shown in Figs. 4(b)-(c). Results show the energy distribution changes more sharply in the circumferential 5

direction in comparison with the result in Fig. 4(a), because the narrower loading length leads a larger number of flexural wave modes. Figure 4. The angular profiles of the T(m,1) mode group at different axial distances for a point source loading on an 8-inch schedule 40 steel pipe at 35 khz On the contrary, an increased loading length on the pipe circumference can be considered to suppress the undesired wave modes. Therefore, the wave modes with high circumferential orders can be generated with small amplitudes. When a group of torsional type wave modes are considered from the torsional fundamental wave mode T(0,1) to several flexural wave modes, the influence of higher order modes can be ignored in the energy distribution of a wave group. Figure 5. Normalised circumferential displacements for the T(m,1) wave modes incident upon a part-circumferential defect at the top of the pipe. Figure 5 shows the normalised circumferential displacements for the T(0,1) and propagation of higher order flexural wave modes in time domain. The results are recorded at the axial distance 2000 mm away from the transducer by the No.1 (90 degrees) and No.7 (0 degree) receivers. It can be clearly seen that the results in Fig. 4(c) and Fig. 5 agree well for the displacement distribution of the transmitted signal around the pipe circumference. Results also show the torsional wave mode T(0,1) propagation and its interaction with higher order flexural wave modes. The circumferential 6

displacements of the reflection from the notch have similar distribution to that of the transmitted signal at No.1 and No.7 receivers because of the notch location. Figure 6. Normalised circumferential displacement amplitudes by summing up the results at 24 receivers in time domain To suppress the flexural wave modes, the received signals at 24 receivers at the transducer location are superposed by post-processing. The results for the three notch sizes are shown in Fig. 6. The wave mode T(0,1) can be easily seen at reflection from a notch, but higher flexural wave modes have not been suppressed completely by superposition of 24 receivers. Results show the normalised amplitude of the circumferential displacements for reflection from a 60-degree notch is higher than that of a 30-degree notch, but it has similar amplitudes to the result for a 90-degree notch. Thus, it indicates that the higher flexural waves affect the results from the three notches. More receiving locations should be considered for future work. 4. Experimental validation Figure 7. Set up of vibrometer pipe experiment with a single transducer 7

To verify the FE simulations, an experimental work was undertaken, as shown in Fig. 7. The torsional type wave modes were excited by a piezoelectric transducer. A 4450 mm, 8-inch, schedule 40 steel pipe without defects was used as specimen. The transducer was placed at the pipe end. A non-scanned 3D laser vibrometer was used to measure 24 receiving points, which are equally spaced around the pipe at 2000 mm away from the transducer. The No.1 and No.7 receivers are located at 82.5 degrees and 352.5 degrees in angular profiles. The impulse signal is a 10-cycle Hanning-modulated sinusoidal signal at 35 khz. Figure 8. Angular profiles of FE and experimental results Figure 8 shows a comparison of normalised circumferential displacements obtained from experimental measurements and finite element simulations by a polar plot. There is a good agreement between the experimental and FEA results. Therefore, the FE modelling for the guided wave excitation and propagation in the pipe is then verified from the related experimental results. Results in Fig. 4(c) and Fig. 8 agree reasonably well for the displacement distribution. 5. Conclusions In this paper, the behaviour of a single piezoelectric transducer upon excitation in a 5000 mm, 8-inch, schedule 40 steel pipe is analysed. This is achieved through a combination of finite element analysis and experimental studies. The FE simulations are verified by experimental measurements. A group of wave modes are considered from the torsional T(0,1) mode to higher flexural wave modes. The behaviour of a single transducer element is evaluated at a specific frequency of 35 khz, as it is the interaction of the generated wave modes with a group of the simulated notches. The angular profiles of the torsional type wave mode group at different axial distances for the point source loading. The energy distribution for the narrower loading length becomes less sharp, when the loading length increases in the circumferential direction. From the information gathered, the design of transducer arrays is optimised to excite the T(0,1) mode with a significant level of mode purity. After post-processing by superposition of 24 receivers, the sensitivity of a single transducer for pipe inspection is evaluated. To 8

further suppress higher flexural wave modes, an increased number of receiver locations is required for future work. Acknowledgements This publication was made possible by the sponsorship and support of TWI Ltd. and University of Greenwich. The work was enabled through, and undertaken at, the National Structural Integrity Research Centre (NSIRC), a postgraduate engineering facility for industry-led research into structural integrity established and managed by TWI through a network of both national and international Universities. The authors would like to thank Dr Wenbo Duan for his contribution to the dispersion curves shown in Fig. 1. Author Contributions: Xudong Niu carried out the three-dimensional numerical analysis in ABAQUS and experimental studies. Hugo R. Marques supervised the experimental work; Xudong Niu and Hua-Peng Chen contributed to the drafting of the manuscript; Hua-Peng Chen and Hugo R. Marques supervised the work in this paper. References [1] P Mudge, Field application of the Teletest long-range ultrasonic testing technique, Insight - Non-Destructive Testing and Condition Monitoring 43(2), pp 74-77, 2001. [2] P Mudge and J Speck, Long-range ultrasonic testing (LRUT) of pipelines and piping, Inspectioneering Journal 10(5), pp 1, 2004. [3] X Niu, H-P Chen and HR Marques, Piezoelectric transducer array optimization through simulation techniques for guided wave testing of cylindrical structures, The 8th ECCOMAS Thematic Conference on Smart Structures and Materials (SMART 2017), pp 424-435, 2017. [4] X Niu, HR Marques and H-P Chen, Sensitivity analysis of guided wave characters for transducer array optimisation on pipeline inspections, The 2017 World Congress on Advances in Structural Engineering and Mechanics (ASEM17), 2017. [5] DN Alleyne, MJS Lowe and P Cawley, The reflection of guided waves from circumferential notches in pipes, Journal of Applied Mechanics 65(3), pp 635-641, 1998. [6] D Alleyne and P Cawley, Long range propagation of Lamb waves in chemical plant pipework, Materials Evaluation 55, pp 504-508, 1997. [7] P Wilcox, M Lowe and P Cawley, The effect of dispersion on long-range inspection using ultrasonic guided waves, NDT & E International 34(1), pp 1-9, 2001. [8] D Gazis, Three dimensional investigation of the propagation of waves in hollow circular cylinder. I. Analytical foundation, J. Acoust. Soc. Am. 31, pp 573-578, 1959. [9] M Silk and K Bainton, The propagation in metal tubing of ultrasonic wave modes equivalent to lamb waves, Ultrasonics 17(1), pp 11-19, 1979. [10] W Duan and R Kirby, A numerical model for the scattering of elastic waves from a non-axisymmetric defect in a pipe, Finite Elements in Analysis and Design 100, 9

pp 28-40, 2015. [11] W Duan, R Kirby and P Mudge, On the scattering of elastic waves from a nonaxisymmetric defect in a coated pipe, Ultrasonic 65, pp 228-241, 2016. [12] W Duan R Kirby and P Mudge, On the scattering of torsional waves from axisymmetric defects in buried pipelines, The Journal of the Acoustical Society of America 141(5), pp 3250-3261, 2017. [13] JL Rose, Ultrasonic Guided Waves in Solid Media, Cambridge: Cambridge University Press, 2014. [14] X Niu, HR. Marques and H-P Chen, Transducer array optimisation for guided wave testing of pipes using finite element numerical simulations and experimental studies, The 8th International Conference on Structural Health Monitoring of Intelligent Infrastructure (SHMII8), 2017. 10