Experimental and Finite Element Study for Quantification of Crack in an
Aluminium Plate using Nonlinear Lamb Wave
Nitesh P. Yelve a , Mira Mitra b and Prasanna M. Mujumdar c
Department of Aerospace Engineering, Indian Institute of Technology Bombay, Mumbai 400076, India a nitesh@aero.iitb.ac, b mira@aero.iitb.ac.in, c mujumdar@aero.iitb.ac.in Keywords: Nonlinear Lamb wave, Breathing crack, Higher harmonics, Finite element method Abstract. The present research focuses on detecting and quantifying a transverse crack present in an aluminium plate, using nonlinear response of Lamb wave induced by contact nonlinearity at the breathing crack. The nonlinear response of Lamb wave appears in the form of higher harmonics and is seen in both experimental and Finite Element (FE) simulation studies. Time-frequency analysis is carried out using Continuous Morlet Wavelet Transform (CMWT) for both experimental and simulation results to show that the fundamental and higher harmonics follow the dispersion curves of Lamb wave. The nonlinearity parameter is estimated from the frequency domain results. It is found to be increasing with the increase in crack depth and its trend is similar in both the studies. In summary, in this paper the experimental procedure using PW transducers and the simulation procedure using FE are presented for exploring the nonlinear behaviour of Lamb wave for quantifying the breathing crack. Introduction Structural components used in mechanical, civil, and aerospace applications are often subjected to time varying load. In many cases such loading gives rise to fatigue cracks in the components which ultimately undergo fatigue failure. Thus there is need for non-destructively detecting and quantifying theses fatigue cracks so as to ensure structural safety. Ultrasonic testing using guided waves, typically Lamb wave is a popular method as it offers benefits such as built-in transduction and moderately large inspection ranges [1]. Lamb wave-based methods can be broadly classified into two groups. The first group involves linear methods which are based on the principles of linear ultrasonics and relies on measuring parameters like attenuation, transmission, and reflection coefficients [2] for detecting damage in a material. However the frequency content of the input and output signal is the same [2]. These methods are sensitive to gross defects or opens cracks where there is an effective barrier to transmission, whereas it is less sensitive to small cracks or material degradation. The second group encompasses nonlinear methods, which show change in the frequency content in response to the nonlinearity in the material introduced by the defect itself. This change in the frequency content appears in the form of higher harmonics, sub-harmonics, shift of resonance frequency, and mixed frequency response [2]. These are more sensitive to small cracks, compared to the amplitude and phase change involved in linear ultrasonic techniques. Also, existing literature [3] on damage detection techniques using nonlinear ultrasonics suggest that nonlinear techniques are robust to factors such as complicated geometry or moderate environmental variations like wind and temperature. These capabilities make the nonlinear ultrasonics techniques attractive for field applications. In the present study only higher harmonics generation is considered as a measure of nonlinearity introduced by the damage. There are basically two mechanisms by virtue of which damage introduces nonlinearity in the material [2]. The first mechanism is nonlinear elasticity introduced in the material by the defect; for example distribution of dislocations during fatigue damage progression, introduces nonlinear elasticity in the material continuum. In the second mechanism faces of a crack interact with each
other as wave passes across it and induces local change in the stiffness giving rise to contact nonlinearity. When wave reaches the breathing crack, the compressional part of the wave can penetrate it, but its tensile part cannot penetrate. Thus, after passing across the crack, the wave gets modified nonlinearly and gives rise to higher harmonics. Existing literature [4, 5, 6, 7, 8, 9, 10] on the generation of second harmonic in Lamb wave response mainly deals with the experimental studies where the nonlinearity which causes the second harmonic response is a weak material nonlinearity arising from enhancement of microscopic defects dispersed throughout the material due to fatigue, aging, and plastic deformation. Further as the wave propagates, the second harmonic response is build up due to the cumulative resonance effect. This is possible only in case of specific pairs of resonant Lamb modes at very high frequencies (in MHz range) which have the same phase and same group velocities at the fundamental and second harmonic frequencies. In contrast to this domain of research, the present work deals with stronger nonlinearities arising from damage such as breathing crack, due to relatively larger stiffness changes at the contact interfaces as the crack closes and opens during the wave proapagation across it. In this case, the requirement of resonant Lamb wave modes, as stated above is not a necessary condition. The strength of second harmonic is not due to cumulative resonance, but due to the strong contact nonlinearity. This nonlinearity generates even more number of higher harmonics in Lamb wave even at fairly low excitation fundamental frequencies in kHz range. Also, the literature lack in showing up the work done with nonlinear Lamb wave resulting from contact nonlinearity, for characterizing cracks in the material, both experimentally and using FE technique. The present research focuses on detecting and quantifying a transverse crack present in an aluminium plate. For this purpose nonlinear response of Lamb wave resulting from contact nonlinearity at the breathing crack is used. The nonlinear response of Lamb wave appears in the form of higher harmonics. Firstly, experiments are carried out on a thin aluminium plate having transverse surface crack. Instead of wedge type transducers [4,6], Piezoelectric Wafer (PW) transducers are used for actuating and sensing the Lamb wave. In comparison with wedge transducers, PW transducers are portable, small in size, and cost effective, which make them useful for in situ applications. Next, analysis of the nonlinear interaction of Lamb wave with a crack is carried out using Finite Element (FE) simulation in ANSYS
. The FE modeling of Lamb wave
propagation across a crack, involves intricacy in modeling the breathing crack. The breathing crack must open and close as wave passes through it so as to introduce contact nonlinearity in the wave propagation. To ensure the interaction of crack faces as wave passes across the crack, the crack faces are assigned contact elements. Augmented Lagrangian (AL) algorithm is used for solving the contact problem, as the AL algorithm offers advantages such as reduced ill-conditioning of governing equations and fairly good agreement with the constraints while dealing with the finite penalties [11]. The experiments and FE simulation show the presence of three higher harmonics resulting from the nonlinear interaction of Lamb wave with the crack. Further, time-frequency analysis is carried out using Continuous Morlet Wavelet Transform (CMWT) for both experimental and simulation results to show that the fundamental and higher harmonics follow the dispersion curves of Lamb wave [12]. Nonlinearity parameter is estimated from the frequency domain results. It is found to be increasing with the increase in crack depth and its trend is similar in both the studies. The paper is organized into three sections. The experimentation part is explained first. The following section deals with the FE simulation, and the last section depicts the estimation of nonlinearity parameter . Experimental Study In the experimental work, PW transducers are used to transmit and receive Lamb wave across a transverse surface crack in an aluminium plate. The size of the PW transducers used here is 10 mm 7 mm 0.5 mm and material type is SP-5H. The grade of aluminium material used here is 5052- H32 with modulus of elasticity 70.3 GPa, poissons ratio 0.33 and density 2680 kg/m 3 [13].
Objective here is to observe the effect of nonlinear interaction of Lamb wave with a transverse breathing crack, and thereby quantify the crack. To achieve this, it is required to have plates with different accurate crack depths. It is difficult to produce surface cracks of different controlled depths in a thin plate even if an accelerated fatigue machine is used. Therefore, the test plates for experimental work are constructed by bonding three pieces of aluminium plates together as shown in Fig. 1. The method followed for having crack in the test plate is the same as that adopted by Douka et al [14] for time-frequency analysis of the free vibration response of a beam with crack. Plates 2 and 3 are bonded to plate 1 but not to each other so that the faces of plates 2 and 3 in contact behave like faces of a breathing crack. Care is taken to not to have high compressive stresses at the crack interfaces while fabricating the test plate. Araldite
epoxy adhesive is used for
bonding the plates. Curing is done at the room temperature for 48 hours while plates are clamped together using C-clamps to ensure good bonding of plates. The plan form dimensions of the test plate are 400 mm 200 mm. The thicknesses of the upper two plates in the test plate are same whereas that of lower plate is different, so as to have cracks with different depths as shown in Table 1.
Fig. 1 Experimental setup. Table 1 Plate Configurations Plate configuration number Thickness of plate 1 Thickness of plates 2 and 3 Total plate thickness Crack depth as percentage of total plate thickness [mm] [mm] [mm] [%] 1 2 2 4 50 2 2 1.6 3.6 44 3 2 1 3 33 The experimental setup as shown in Fig. 1 consists of Tektronix AFG 3021B single channel arbitrary function generator, NF BA4825 high speed bipolar amplifier, Tektronix TDS 1002B two channel digital storage oscilloscope, and a computer having online connection with the oscilloscope. The three test plates have different total thicknesses as shown in Table 1. This is because of the constraint on the availability of individual plate thicknesses. However, generation of Lamb wave with similar characteristics in all the three test plates requires a condition to be satisfied that the frequency thickness product must be same in all the three cases. The experiments are carried out at a range of low excitation frequencies (85 kHz to 400 kHz) and at all the frequencies higher harmonics are observed as a result of contact nonlinearity. Three different frequencies, 255 kHz, 283.33 kHz, and 340 kHz are chosen for the experiments so as to have same frequency thickness product, 1,020 kHz mm in all the three test plates as shown in Table 2.
Table 2 Frequencies used in the experiments Plate configuration number Total plate thickness Frequency Frequency thickness product [mm] [kHz] [kHz mm] 1 4 255 1,020 2 3.6 283.33 1,020 3 3 340 1,020 Excitation tone burst used is a 8.5 cycle sine wave windowed by Gaussian function. Gaussian windowing function is used here because it produces a tone burst which has narrow frequency bandwidth with less sidebands. The number of cycles considered for the tone burst is 8.5, because for these many cycles, the Lamb wave modes are seen well separated, the frequency bandwidth is less, and the half cycle in the tone burst brings the peak amplitude at the centre which helps in calculating group velocity. The tone burst with 8.5 cycles and amplification of 300 Volt (peak to peak) supplied to PW actuator is found to possess sufficient energy required for exciting higher harmonics at a range of low frequencies mentioned above. The received Lamb wave signals at the PW sensor are shown by the oscilloscope and are sent to computer for further offline signal processing. Experiments are carried out on the test plates with crack depths 33%, 44%, and 50% of plate thickness respectively and also on a test plate without crack, i.e. having only two plates, upper and lower, of same plan form dimensions, bonded together. The plate without crack is treated as an undamaged test plate. The frequency spectra in case of undamaged test plate and a test plate with the crack depth of 50% are shown in Fig. 2. Three higher harmonics at 510 kHz, 765 kHz, and 1,020 kHz can be seen in Fig. 2, whereas the test plate without crack does not show any higher harmonic. This means that the adhesive bond between the upper and lower plates does not have any contribution to the higher harmonics observed in the experiment. This also obviates the possibility that any instrumentation nonlinearity may be a reason for the appearance of higher harmonics in the response. Therefore it can be concluded that the higher harmonics observed are solely because of nonlinear interaction of the wave with the breathing crack. Similar frequency domain results are obtained in case of the plates with crack depths 33% and 44%.
Fig.1 Experimental results in frequency domain. To further ensure that the frequency contents in the signal, picked up at the PW sensor are corresponding to Lamb wave modes, a time-frequency analysis of the received signal is carried out and the arrival times of particular frequencies are identified. Continuous Morlet Wavelet Transform (CMWT) is used for this purpose with the central frequency parameter 10 to satisfy the admissibility condition [15,16]. CMWT essentially shows the frequency contents of the response signal, at their respective arrival times. Next, the dispersion curves of A 0 and S 0 modes of Lamb
wave, between the group velocity and the frequency for the known plate thickness, are obtained analytically [12]. As the distance between actuator and sensor are known, the arrival times of the Lamb wave modes are obtained from the group velocity and plotted against the frequency. Such a dispersion plot is then overlapped on the wavelet plot using MATLAB
. These results are shown in
Fig. 3a for the test plate without crack, and in Fig. 3b for the test plate with crack depth of 50%. The Morlet wavelet plot in Fig. 3a shows frequency 255 kHz, and in Fig. 3b shows frequencies 255 kHz and 510 kHz, contained in the response signal. They all lie close to the dispersion curves. Thus, from these two figures it can be said that, undamaged test plate shows only A 0 and S 0 modes of Lamb wave at fundamental frequency in the response signal and no higher harmonics, whereas in case of the test plate with 50% crack depth, the frequency 255 kHz corresponds to A 0 and S 0 modes of Lamb wave, and the second harmonic 510 kHz corresponds to S 0 mode of Lamb wave. Third and fourth harmonic frequencies are not visible in Fig. 3b as their amplitudes are feeble compared to the fundamental frequency. Similar time-frequency results are obtained in case of the plates with crack depths 33% and 44%.
(a) Plate without crack. (b) Plate with 50% crack depth. Fig. 3 CMWT of the experimental output data. Finite Element Simulation The challenge faced in the FE simulation of Lamb wave propagation across the breathing crack, in order to study their nonlinear interaction, is to model the crack using efficient contact modeling which can closely approximate the physical contact of crack faces. ANSYS
is used here for FE
simulation. An aluminium plate (grade 5052-H32) with a breathing crack, and PW actuator and sensor for transmitting and receiving Lamb wave, is modeled in the thickness plane with plane strain approximation. This approximation essentially saves computational time without affecting the trend of the results [17]. PLANE82 (8-node quadrilateral) elements are used in the plate region and PW transducers are modeled using PLANE223 (8-node quadrilateral coupled-field) elements. Selection of element size is crucial for FE simulation of wave propagation. The size of the elements must be sufficiently small for there to be atleast 10 nodes per wavelength of the propagating wave [18]. The peak excitation frequency used in the study is 340 kHz and its fourth harmonic, 1,360 kHz is expected to be generated as a result of contact nonlinearity. Therefore the mesh size should be able to capture 1,360 kHz. Only A 0 and S 0 modes of Lamb wave are involved in the analysis and A 0
mode has the smaller wavelength at 1,360 kHz which is 1.847 mm. Accordingly, element size of 0.35 mm is used for the simulation. The plate and PW transducer models are meshed through mapped meshing option which gives uniform distribution of elements throughout the model. The nodes on the upper and the lower faces of the PW transducer models are electrically coupled to form master nodes separately on both the faces. Zero voltage constraint is applied on the master nodes at
the bonded faces of the patches. Excitation voltage is applied at the master node on the top face in case of the PW actuator. In the case of the PW sensor, voltage is measured at the master node on its top face. Newmark algorithm is used here for time integration in FE analysis with = 0.25 and = 0.5 where and are the Newmark parameters. For these values of and the Newmark algorithm is unconditionally stable [18]. In this case, the time step size to be used is based on the trade-off between the desired accuracy and computational efforts. The accuracy in the integrations can be stated in terms of period elongation and amplitude decay [18]. In the FE simulation of Lamb wave propagation, both period elongation and amplitude decay affect the group velocity. Therefore a convergence study is carried out to decide the appropriate time step for solution so that the group velocities of A 0 and S 0 Lamb wave modes obtained through simulation are in good agreement with those obtained analytically while requiring optimal simulation efforts both in terms of time of simulation and space required in the computer. Accordingly, the time step 2 10 7 seconds is used for the simulation. To model the contact nonlinearity at the crack, the two faces are assigned CONTA172 (3-node surface-to-surface contact) and TARGE169 (target segment) elements respectively. The algorithm used here to solve the contact problem is the Augmented Lagrange (AL) method. The AL method is an iterative series of penalty updates to find the contact tractions. It has advantages over other methods, such as more penetration control, better conditioning of governing equations, and satisfaction of constraints with finite penalties [11]. Friction is assumed to be insignificant in the present contact problem. Therefore AL algorithm is implemented here under frictionless condition. The FE analysis is carried out for 8.5 cycles, 300 Volt (peak to peak) Gaussian input signal having central frequencies mentioned in Table 2. The Fourier transform of the signal received at the PW sensor in case of the test plate without damage is shown in Fig. 4, which does not show any higher harmonic. The Fourier transform of the signal received in case of the test plate with the crack depth of 50% of plate thickness is also shown in Fig. 4, which shows three higher harmonics at 510 kHz, 765 kHz, and 1020 kHz. The comparison of arrival times of wave modes with those obtained analytically [12] for A 0 and S 0 modes is carried out using CMWT and is shown in Fig. 5 for the test plate without crack and with 50% crack depth. Similar results are also obtained for the test plates with the crack depths 33% and 44%. Results shown in Figs. 4 and 5 are in accordance with the experimental results. Estimation of Nonlinearity Parameter From the results observed so far, it can be said that breathing crack fabricated in the experiments and modeled in ANSYS
using AL algorithm is capable of causing contact nonlinearity in Lamb
wave propagation. Next task is to quantify the breathing crack. For this purpose, the nonlinearity parameter (defined as A 2 /A 1 2 , where A 1 and A 2 are the amplitudes of fundamental and second harmonics respectively) [19] is estimated from the experimental and simulation frequency domain results. This nonlinearity parameter has been used extensively in the literature [2] dealing with higher harmonics, for quantifying nonlinearity. As crack depth increases the local nonlinearity in the plate increases which eventually causes energy transfer from fundamental harmonic to higher harmonics. This results in increase of amplitudes of higher harmonics with simultaneous decrease of fundamental harmonic amplitude. This is evident from Fig. 6 which shows increasing trend of with increase in crack depth. However, the trend of for experimental results is little deviated in one of the cases. This is because, for estimating only first two harmonics are considered and it works well when only second harmonic is generated through the cumulative effect. It does consider the higher harmonics which also share the nonlinear effect. Therefore, it is required to define a new damage index capable of considering more or rather all the higher harmonics generated, for estimating the severity of damage.
Fig. 4 Simulation results in frequency domain.
(a) Plate without crack. (b) Plate with 50% crack depth. Fig. 5 CMWT of the simulation output data.
Fig. 6 Variation of nonlinearity parameter with the crack depth. Conclusions Experiments carried out on the aluminium test plates with crack using Lamb wave, show presence of three higher harmonics. In the FE simulation, contact nonlinearity is modeled at the breathing crack using contact elements and the contact problem is solved using Augmented Lagrangian algorithm. This simulation effectively shows three higher harmonics in the Lamb wave response.
The time-frequency analysis of the signals received at the sensor in both experiments and FE simulation, carried out using CMWT shows that higher harmonics lie on the dispersion curves. The nonlinearity parameter obtained from experimental and simulation results increases with increase in crack depth. This shows that the experimental and the FE simulation procedures presented in the paper are capable producing contact nonlinearity and thereby higher harmonics in the Lamb wave response. The nonlinearity parameter help in quantifying the breathing crack. It is also seen that is appropriate to use as a damage index when only second harmonic is generated cumulatively. When more number of higher harmonics are produced a comprehensive damage index needs to be used which may consider the effect of all the higher harmonics. References [1] D. Dutta, H. Sohn, K. Harries, and P. Rizzo, A nonlinear acoustic technique for crack detection in metallic structures, Struct. Health Monit. 8(3) (2009) 251-262. [2] K. Jhang, Nonlinear ultrasonic techniques for nondestructive assessment of micro damage in material: A review, Int. J. Precis. Eng. Man. 10(1) (2009) 123-135. [3] A. Hillis, S. Neild, B. Drinkwater, and P. Wilcox, Global crack detection using bispectral analysis, P. Roy. Soc. A-MATH. PHY. 462 (2006) 1515-1530. [4] M. Deng and J. Pei, Assessment of accumulated fatigue damage in solid plates using nonlinear Lamb wave approach, Appl. Phys. Lett. 90 (2007) 121902 1-3. [5] C. Pruell, J. Kim, J. Qu, and L. Jacobs, Evaluation of fatigue damage using nonlinear guided waves, Smart Mater. Struct. 18 (2009) 035003 1-7. [6] C. Pruell, J. Kim, J. Qu, and L. Jacobs, Evaluation of plasticity driven material damage using Lamb waves, Appl. Phys. Lett. 91 (2007) 231911 1-3. [7] H. Xu, C. Xu, S. Zhou, and Y. Hu, Time-frequency analysis for nonlinear Lamb wave signal, Proceedings of CISP09, Tianjin 308 (2009) 201-211. [8] W. Li, Y. Cho, and J. Achenbach, Detection of thermal fatigue in composites by second harmonic Lamb waves, Smart Mater. Struct. 21 (2012) 085019 1-8. [9] M. Deng and J. Yang, Characterization of elastic anisotropy of a solid plate using nonlinear Lamb wave approach, J. Sound Vib. 308 (2007) 201-211. [10] C. Bermes, J. Kim, J. Qu, and L. Jacobs, Nonlinear Lamb waves for the detection of material nonlinearity, Mech. Syst. Signal Pr. 22 (2008) 638-646. [11] J. Simo and T. Laursen, An augmented Lagrangian treatment of contact problems involving friction, Comput. Struct. 42(1) (1992) 97-116. [12] Z. Su, L. Ye, and Y. Lu, Guided Lamb waves for identification of damage in composite structures: A review, J. Sound Vib. 295 (2006) 753-760. [13] ASM Handbook, Properties and selection: Nonferrous alloys and special purpose materials, vol 2, Materials Park, OH: ASM International, 1992. [14] E. Douka and L. Hadjileontiadis, Time-frequency analysis of the free vibration response of a beam with a breathing crack, NDT&E Int. 38 (2005) 3-10. [15] J. Goswami and A. Chan, Fundamentals of Wavelets: Theory, Algorithms, and Applications, John Wiley and Sons, New Jersey, 2011. [16] P. Addison, The Illustrated Wavelet Transform Handbook: Introductory Theory and Applications in Science, Engineering, Medicine and Finance, IOP Publishing Ltd, London, 2002. [17] B. Poddar, C. Bijudas, M. Mitra, and P. Mujumdar, Damage detection in a woven-fabric composite laminate using time-reversed Lamb wave, Struct. Health Monit. 11 (2012) 602-612. [18] K. Bathe, Finite Element Procedures in Engineering Analysis, Prentice-Hall Inc, New Jersey, 1982. [19] K. Jhang, Applications of nonlinear ultrasonics to the NDE of material degradation, IEEE T. Ultrason. Ferr. 47(3) (2000) 540-548.