Available online at www.sciencedirect.com
ScienceDirect Procedia Engineering 188 (2017) 148 – 155
6th Asia Pacific Workshop on Structural Health Monitoring, 6th APWSHM
Guided wave propagation and damage detection in composite pipes using piezoelectric sensors Matthieu Gresila,*, Adisorn Poohsaia, Neha Chandaranaa a
i-Composites Lab, School of Materials, University of Manchester, 79 Sackville Street, Manchester M1 3NJ, United Kingdom
Abstract Composite materials have many attractive properties, e.g. light-weight combined with high mechanical strength, and are recognized as an enabling technology for deep-water high pressure high temperature (HPHT) and corrosive environment applications. However, composite still has a barrier to entry in relation to steel due to limited field experience and lack of systematic failure prediction and assurance of in-service performance. Advances in such areas would enable the use of composites in a wider range of applications. Evaluation of in-service prototypes offers a partial solution but is costly. Thus a method that allows constant health monitoring of the composite in real-time and in-situ would be extremely useful. Ultrasonic guided wave-based structural health monitoring (SHM) technology is one of the most prominent options in non-destructive evaluation and testing (NDE/NDT) techniques. To investigate the feasibility of guided wave-based SHM for composite pipes, propagation characteristics of guided waves in an epoxy hybrid carbon/glass fibres pipe are systematically studied using finite element (FE) simulation and experiments. Both axisymmetric modes of propagation, the longitudinal L(0,n) and torsional T(0,n) modes are considered in the simulation process, however only the longitudinal modes can be captured by the piezoelectric sensors. Additionally, the tuning curves experimentally plotted are used to obtain the frequency with the maximum amplitude of each guided mode. Finally, guided waves in the composite pipe are tested with artificial defects (gel coupled coins) to understand the behaviour of guided waves after interaction with the defects. The different condition of artificial defects, e.g., the defect size and defect location are studied to find out the proper condition and limitation of using guided waves to monitor and inspect composite pipes. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-reviewunder under responsibility of organizing the organizing committee of the 6th APWSHM. Peer-review responsibility of the committee of the 6th APWSHM Keywords: Lamb waves; composite pipes; finite element method; piezoelectric sensors; dispersion curves; damage deection
* Corresponding author. Tel.: +44(0)161 306 5744. E-mail address:
[email protected]
1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the 6th APWSHM
doi:10.1016/j.proeng.2017.04.468
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
1. Introduction Many variants of high performance steels have the high strength, stiffness and good performance at high temperatures required for subsea oil and gas pipes. However, corrosion is the significant drawback in the use of steel pipe for offshore deployment, estimated to cost the offshore oil and gas industry more than $1.3 billion every year. Steel corrosion poses a threat which often seriously undermines the structural integrity of oil and gas pipes. Corrosion of steel can be dealt with by a range of techniques such as coating, painting, sacrificial anodes and chemical inhibitors. However, none of these methods is 100% successful, most are expensive and, in the case of inhibitors, they can represent a seriously detrimental environmental threat. Although carbon steel and copper nickel alloy pipe had traditionally been used on offshore platforms, advanced composites were known to be stronger, more resistant to corrosion, and lighter than steel. For example, composite pipe with a 6-inch diameter weighs 4 lb/ft, whereas copper nickel pipe with the same diameter weighs 24 lb/ft. There is, however, obstacles to using composite piping that were related primarily to the lack of test data to support the materials' long-term durability. Advances in such areas would enable the use of composites in a wider range of applications. Evaluation of inservice prototypes offers a partial solution but is costly. Thus a method that allows constant health monitoring of the composite in real-time and in-situ would be extremely useful. Ultrasonic guided wave-based structural health monitoring (SHM) technology is one of the most prominent options in non-destructive evaluation and testing (NDE/NDT) techniques. For the purpose of composite pipe inspection, simulation analysis and experiments of ultrasonic guided waves are required to understand the characteristics of wave propagation along the pipe waveguide and the interaction of the guided wave with the defects in the composite pipes. In solid hollow cylinders, three modes of wave can propagate: (i) longitudinal modes which propagate along the axial direction by a compressional motion, L(0,n); (ii) torsional modes which propagate along the axial direction by shear motion parallel to the circumferential direction, T(0,n); and (iii) flexural modes which propagate along the axis by a flexural motion in a radial direction, F(m,n). The modes of longitudinal and torsional waves in cylindrical structures are equivalent to Lamb wave and SH wave modes in plates, respectively. On the other hand, the flexural mode is the specific mode for a cylindrical structure [1]. The equivalences between guided wave modes in hollow cylinders and Lamb wave modes in plates are described in references [2-4]. The longitudinal and torsional modes are axisymmetric while the flexural mode is non-axisymmetric. The notation for each mode is (m,n) where ‘m’ is a circumferential order (the number of wavelengths around the circumference; m=0 is axisymmetric mode) and ‘n’ is the number of mode shapes. A rapid screening technique for corrosion detection has been studied using ultrasonic guided waves [5]. Moreover, the improvement of the sensitivity and reliability of the guided wave inspection is necessary. Rose et al. [6] showed that several possible defects can be detected correctly due to the sensitivity of different frequencies and modes of guided waves. For multi-layered pipes, most literature was focused on metallic pipes with coating, painting or other layers of material. Ultrasonic guided waves were used to detect the delamination of coating and flaws of coated pipes [7]. Guided waves open the way for SHM on long distance to develop self-sending structures of large dimensions without necessarily increasing the number of sensors and thus without increasing the complexity of the monitoring system. However, guided wave or Lamb wave signals are difficult to characterise due to the multimodal character of the waves. Numerical analysis is used to obtain wave numbers, dispersion curves, etc. Well-known computational approaches were developed to obtain the dispersion curves for multi-layered and axisymmetric structures, namely the transfer matrix method [8] and the global matrix method [9]. Commercial software such as “Disperse” [10] and “Matlab code PCDisp” [11] have been developed to calculate the dispersion curves for plates and cylindrical structures from these matrix methods. However, complex structure (multi-layered, anisotropic) is still a limit for these analyses. This leads to simulation and visualisation using finite element method (FEM) to model numerically the waveguides and wave propagation. FEM is the most common method for analysis wave propagation in the structures due to its simplicity and flexibility for arbitrary geometries and materials with various boundary conditions [12-15]. FE analysis of isotropic hollow cylinders for axisymmetric vibration in comparison with experimental results showed excellent agreement [16]. The dispersion properties of guided wave propagation in isotropic and multi-layered hollow cylindrical waveguides were developed [17, 18]. However, based on the author’s knowledge, most of the literature on guided wave behaviour in non-isotropic hollow cylinders focuses on metallic pipes with coating, painting or other substance layers. There is a lack of studies involving 3D FE simulation of guided wave propagation behaviour
149
150
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
in composite pipes. The purpose of this paper is to characterise the wave propagation behaviour in hybrid carbon/glass fibre composite pipes using commercial FE codes. The simulated wave propagation, dispersion curves and wave signals measured at transducers (sensors/actuators) were carried out and compared with the experimental results of the wave propagation in composite pipes. Moreover, the interaction between guided waves and defects on the surface of the composite pipes is analysed and quantified. 2. Materials presentation and experimental set-up A 1 m composite pipe was manufactured from unidirectional pre-preg of carbon fibres (Toray T700) oriented along the length of the pipe (0° layers) and unidirectional pre-preg of glass fibres (E-class) oriented in the circumferential direction (90° layers). The lay-up order is [0, 90, 0, 90, 0]. The geometry and properties of the composite pipe is shown in Error! Reference source not found.. Table 1. Geometries and properties of the composite pipe from the manufacturer. Internal diameter (ID, mm)
Wall thickness (mm)
Density, ρ (kg/m3)
Young’s modulus, E (GPa)
Shear modulus, G (GPa)
Poisson’s ratio, ν
60.3
1.6
1600
E1 = 90; E2 = 19; E3 = 19
G12 = 4.6; G23 = 4.6; G13 = 4.6
ν 12 = 0.14; ν 23 = 0.2; ν 13 = 0.2
Piezoelectric wafer active sensor (PWAS) discs (PI c255, 10 mm-diameter and 0.5 mm thickness) are used for Lamb wave propagation analysis. The PWAS are bonded, using Cyanoacrylate adhesive, on the composite pipes as shown in Fig. 1a. The experimental acquisition has been performed, with frequency varying from 15 to 750 kHz in steps of 15 kHz, using transmitter piezoelectric wafer active sensors (T-PWAS) as actuator and two PWAS discs as receiver piezoelectric wafer active sensors (R-PWAS) at 0.5 m and 1 m. An 80 V peak-to-peak three count tone burst exciting signals was used. A Tektronix AFG3052C arbitrary signal generator was used to generate the windowed harmonicburst excitation to active sensors with a 10 Hz repetition rate. A Tektronix DPO5034B four channel digital oscilloscope, synchronized with the signal generator, was used to collect the response signals from the receiver PWAS transducers with a sampling rate of 2.5 Ms/s. At each frequency, the wave amplitude and the time of flight (ToF), using the Hilbert transform were collected.
(a)
(b)
Fig. 1. (a) Experimental setup for testing guided wave propagation in composite pipes; (b) 3D FE model of the composite pipes with the locations of excitation (T-PWAS) and sensors (R-PWAS).
The artificial defects are coupled on the surface with an ultrasound gel to interact with the guided waves.The locations of the artificial defects are as follows: (i) Defect A: at 0.25 m from T-PWAS; (ii) Defect B: at 0.75 m from T-PWAS. To test the sensitivity of guided waves interacting with the different size of defects, three different sizes of coins were used for simulating damage on the pipe surface. The details of each coin are as follows: (i) Small-sized
151
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
defect (5-pence coin); Diam.=18 mm; Thickness=1.7 mm; Mass=3.25 g; (ii) Medium-sized defect (1-penny coin); Diam.=20.3 mm; Thickness=1.65 mm; Mass=3.56 g; (iii) Large-sized defect (2-pence coin); Diam.=25.9 mm; Thickness=2.03 mm; Mass=7.12 g. 3. Simulation techniques In wave propagation problems, there are two types of simulation: frequency domain and time domain analysis. The modal analysis in the frequency domain is applied to obtain the Eigen-frequencies and mode shapes of wave propagation in the structures. The dynamic problem analysis in the time domain is used for the visualisation of wave propagation through the structures during the travelling time. First of all, the 2D FE models of the composite pipes in the commercial finite element codes such as ABAQUS, and ANSYS were generated. The 2D models of the pipe wall section were discretised into small sections with an appropriate meshing size, typically 10 to 20 nodes per minimum wavelength [14, 17, 19, 20]. The different modes of propagation; for example, longitudinal, torsional and also flexural (if available) were distinguished by the simulated mode shapes. To calculate the phase velocity and group velocity of each mode of propagation, some essential parameters such as wavelength O , angular wave number k , and angular velocity Z is calculated:
O
L p , k
2S O , Z
2S f
(1)
Where L the length and p the numbers of the wave cycle at that length. After this calculation, the phase and group velocities can be solved by using: C ph
Z k , Cgr
'Z 'k
(2)
To visualise the wave propagation along the composite pipes, the dynamic analysis in time domain is developed. Firstly, the 3D models of pipes with transmitters and receivers is created in cylindrical coordinates using the commercial code ABAQUS, as shown in Fig. 1b. The 3D composite properties are shown in Error! Reference source not found. and the Rayleigh damping coefficients from references [21, 22] are used. The excitation loading was applied as concentrated forces around one end of the pipes to axi-symmetrically generate either longitudinal loading or torsional loading [23]. To obtain accurate results, an adequate integration time step is an important factor to be considered depending on the frequency of interest [14, 19, 20]. The received signals can be extracted at the location of receivers in terms of displacement or velocity against the time. From extracted signals, the numbers of propagating modes passing the receivers at different time periods and the group velocity can be observed and calculated. 4. Results and discussions 4.1. Simulation of guided wave in composite pipes Guided wave propagating in composite pipes is different from that of metallic pipes because multiple unexpected scattering and mode conversion can occur along the layers in the composite materials. In the frequency domain, Eigen frequencies and mode shapes were obtained with 2D-axisymmetric models using ABAQUS and ANSYS which were used to extract the dispersion curves of the longitudinal and torsional modes. These two software were used for comparison purposes only. The mode shapes these guided modes at the frequency below 1 MHz are shown in Error! Reference source not found.a. From the mode shapes and Eigen-frequencies, the dispersion curves were extracted using Eq. (2). The results of the group velocity are shown in Fig. 2 where L(0,n) are longitudinal modes and T(0,n) are torsional modes. The highest velocity is L(0,2) mode around 7500 m/s while L(0,1) and T(0,1) have the same velocity which is near to 1700 m/s between 150 and 600 kHz. Fig. 3a shows image snapshot of overall displacement amplitude of the guided wave pattern in the composite pipes taken at 120-μs when a normalised longitudinal load is applied on the left edge. The group velocity of L(0,1) and L(0,2) is 1600m/s and 7500m/s respectively, which is in good agreement with the dispersion curve calculate using the frequency domain. Fig. 3b shows image snapshot of overall displacement amplitude of the guided wave pattern in the composite pipes taken at 300-μs when a torsional load is applied on the left edge. The group velocity of T(0,1) is
152
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
1600m/s which is in good agreement with the dispersion curve calculate using the frequency domain.
(b)
(a)
Fig. 2. (a) Mode shapes of guided wave propagating in the composite pipes using 2D-axi-symmetric models; L(0,n) are longitudinal modes and T(0,n) are torsional modes; (b) Simulated group velocity dispersion curve of L(0,n) and T(0,m propagating in the composite pipes (ABAQUS in black, and ANSYS in red).
(a)
(b)
Fig. 3. FEM simulation of the guided wave propagation: (a) a longitudinal load is applied - (L(0,1) and L(0,2) at 100 kHz at t=120μs; (b) a torsional load is applied - (L(0,1) and L(0,2) at 100 kHz at t=300μs;
4.2. Experimental validation and tuning curves Fig. 4 shows the PWAS tuning guided wave for L(0,1) and L(0,2) modes. The L(0,2) mode reaches a peak at 255kHz with a magnitude of 6 Vpp and then decreases. L(0,1) mode reaches two peaks for the first R-PWAS (i.e. at 0.5 m) at 35 and 60 kHz with a magnitude of 300 and 370 mVpp, respectively. Moreover, L(0,1) mode reaches one peak for the second R-PWAS (i.e. at 1 m) at 45 kHz with a magnitude of 315 mVpp. This frequency tuning variation is still under investigation, because it is not supposed to be a distance variable. 4.3 Damage detection and quantification The experiments were performed at 30, 45, and 60 kHz which focused on L(0,1) modes and at 255 kHz which used L(0,2) for defect detection.
153
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
Fig. 4. The PWAS tuning guided wave for (a) L(0,1), and (b) L(0,2) modes
From the experiments, a comparison between signals from damaged (with the artificial damage, e.g. gel coupled coin on the surface) and undamaged conditions is used to calculate the changes in measured signals after interaction with the defects. Moreover, the damage index (DI) is calculated to show the differences between damaged and undamaged conditions and the trends of the differences affected by the frequencies of excitation, and defects sizes.
DI
Aundamaged Adamaged Aundamaged
(3)
Where AUndamaged is the amplitude of undamaged signals and ADamaged is the amplitude of damaged signals. Fig. 5 shows that DI decreases with the frequency. This means that the low frequency excitation can be used to detect this artificial defect with more accuracy. The most appropriate frequency for detection is 30 kHz with the L(0,1) mode. For the L(0,2) mode (i.e. at 255 kHz) the DI is very low in all cases. This means that L(0,2) at 255 kHz is not suitable to detect this kind of artificial damage. Moreover, artificial defects A and B can be detected with both receivers even if the defect location is near the excitation and the receiver is 1 m away. 5. Conclusion In this research, the simulation techniques using commercial finite element codes such as ABAQUS and ANSYS to calculate the dispersion curve and to visualise the guided wave propagation were performed on composite pipes in the frequency domain and time domain, respectively. The wave propagation in the time domain analysis with the longitudinal and torsional excitations was applied with the Rayleigh damping to obtain the propagating animation and extracted signals at the receivers in the same way as the condition of real composite pipes. In order to validate the simulation results, a composite pipe with one transmitter (T-PWAS) and two receivers at 0.5 m (1st R-PWAS) and 1 m (2nd R-PWAS) away from the excitation were instrumented to detect defects using guided wave propagation. The PWAS is unable to capture the signals of the torsional guided mode (shear mode), maybe due to the high level of damping in composite pipes. Therefore, only longitudinal modes were compared. Moreover, the plots of the signal amplitude at each frequency, called the tuning curve, were plotted separately for the different guided modes; L(0,1) and L(0,2) to obtain the peak signals of each mode for further study on the application of guided waves in the composite pipe. Guided L(0,1) mode at 30, 45 and 60 kHz have the maximum sensitivity, while the guided L(0,2) mode has the maximum signal amplitude at 255 kHz. With the use of guided waves in the NDT and SHM applications, the interaction between guided waves and the defects on the composite pipe was carried out at the frequencies of the peak signals from the tuning curve. The defects were simulated by coupling different sizes of coin on the top surface of the pipe. The different conditions of the defect were tested to find out the limitations of defect detection using guided waves in the composite pipe. Finally, the results showed that low frequency longitudinal guided modes; L(0,1) can be used to detect the defect with high sensitivity and on relatively large distance. A study on the slowness profile of the guided waves propagating in composite pipe to observe the profile of the
154
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
wave front. Moreover, the interaction between guided waves and real defects on composite pipes e.g. hole, crack, delamination will be carried out to find out the limitations of using guided waves for defect detection in composite pipes.
Fig. 5. Variation of the damage indices (DI) of guided waves at different frequencies interacting with different sizes of 0-degree defect A and B excited by one transmitter – Blue dots are at 1st R-PWAS and Red dots are at 2nd R-PWAS.
Acknowledgements The authors would like to acknowledge the funding and technical support from BP through the BP International Centre for Advanced Materials (BP-ICAM) which made this research possible. The authors also acknowledge financial support from EPSRC (EP/L01680X) through the Materials for Demanding Environments Centre for Doctoral Training.
Matthieu Gresil et al. / Procedia Engineering 188 (2017) 148 – 155
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
[11] [12] [13] [14] [15]
[16] [17] [18] [19] [20] [21] [22] [23]
V. Giurgiutiu, "Stuctural Health Monitoring with Piezoelectric Wafer Active Sensors," in Structural Health Monitoring with Piezoelectric Wafer Active Sensors (Second Edition), ed Oxford: Academic Press, 2014. Z. Su and L. Ye, Identification of Damage Using Lamb Waves vol. 48: Springer London, 2009. J. L. Rose, "Successes and Challenges in Ultrasonic Guided Waves for NDT and SHM," in Proceedings of the National Seminar & Exhibition on Non-Destructive Evaluation, 2009. M. G. Silk and K. F. Bainton, "The propagation in metal tubing of ultrasonic wave modes equivalent to Lamb waves," Ultrasonics, vol. 17, pp. 11-19, 1979/01/01 1979. D. N. Alleyne, B. Pavlakovic, M. J. S. Lowe, and P. Cawley, "Rapid, long range inspection of chemical plant pipework using guided waves," AIP Conference Proceedings, vol. 557, pp. 180-187, 2001. J. L. Rose, Y. Cho, and M. J. Avioli, "Next generation guided wave health monitoring for long range inspection of pipes," Journal of Loss Prevention in the Process Industries, vol. 22, pp. 1010-1015, 11// 2009. M. K. Hinders and J. P. Bingham, "Lamb wave pipe coating disbond detection using the dynamic wavelet fingerprinting technique," AIP Conference Proceedings, vol. 1211, pp. 615-622, 2010. W. T. Thomson, "Transmission of Elastic Waves through a Stratified Solid Medium," Journal of Applied Physics, vol. 21, pp. 89-93, 1950. L. Knopoff, "A matrix method for elastic wave problems," Bulletin of the Seismological Society of America, vol. 54, pp. 431-438, 1964. B. Pavlakovic, M. Lowe, D. Alleyne, and P. Cawley, "Disperse: A General Purpose Program for Creating Dispersion Curves," in Review of Progress in Quantitative Nondestructive Evaluation: Volume 16A, D. O. Thompson and D. E. Chimenti, Eds., ed Boston, MA: Springer US, 1997, pp. 185-192. F. Seco and A. R. Jimenez, Modelling the generation and propagation of ultrasonic signals in cylindrical waveguides Dr. Santos (Ed), InTech, 2012. H. Gravenkamp, C. Song, and J. Prager, "Numerical simulation of ultrasonic guided waves using the Scaled Boundary Finite Element Method," in 2012 IEEE International Ultrasonics Symposium, 2012, pp. 2686-2689. M. Gresil and V. Giurgiutiu, "Time-domain hybrid global--local concept for guided-wave propagation with piezoelectric wafer active sensor," Journal of Intelligent Material Systems and Structures, 2013. M. Gresil, V. Giurgiutiu, Y. Shen, and B. Poddar, "Guidelines for Using the Finite Element Method for Modeling Guided Lamb Wave Propagation in SHM Process," 6th European Workshop on Structural Health Monitoring, Dresden, Germany, July, pp. 3-6, 2012. M. Gresil, Y. Shen, and V. Giurgiutiu, "Benchmark problems for predictive fem simulation of 1-D and 2-D guided waves for structural health monitoring with piezoelectric wafer active sensors," in Review of progress in quantitative nondestructive evaluation, 2012, pp. 1835-1842. G. M. L. Gladwell and U. C. Tahbildar, "Finite element analysis of the axisymmetric vibrations of cylinders," Journal of Sound and Vibration, vol. 22, pp. 143-157, 1972/05/22 1972. Ş. Sorohan, N. Constantin, M. Găvan, and V. Anghel, "Extraction of dispersion curves for waves propagating in free complex waveguides by standard finite element codes," Ultrasonics, vol. 51, pp. 503-515, 5// 2011. M. V. Predoi, "Guided waves dispersion equations for orthotropic multilayered pipes solved using standard finite elements code," Ultrasonics, vol. 54, pp. 1825-1831, 9// 2014. F. Moser, L. J. Jacobs, and J. Qu, "Modeling elastic wave propagation in waveguides with the finite element method," NDT & E International, vol. 32, pp. 225-234, 6// 1999. A. Demma, P. Cawley, M. Lowe, A. G. Roosenbrand, and B. Pavlakovic, "The reflection of guided waves from notches in pipes: a guide for interpreting corrosion measurements," NDT & E International, vol. 37, pp. 167-180, 4// 2004. M. Gresil and V. Giurgiutiu, "Prediction of attenuated guided waves propagation in carbon fiber composites using Rayleigh damping model," Journal of Intelligent Material Systems and Structures, vol. 26, pp. 2151-2169, 2015. M. Gresil and V. Giurgiutiu, "Prediction of attenuated guided wave propagation in carbon fiber composites," in The 19th International Conference on Composite Materials, 2013. T. Furukawa and I. Komura, "Simulation and visualization of guided wave propagation by large-scale 3DFEM," E-Journal of Advanced Maintenance, vol. 3, pp. 92-101, 2011.
155
