Phase velocity measurement of dispersive wave modes by Gaussian peak-tracing in the f-k transform domain

Abstract

The measurement of guided wave phase velocity is vital for the successful implementation of the guided wave technique for defect detection and material characterization. In this paper, an algorithm for quantitative phase velocity measurement based on f-k (frequency-wavenumber) transformation with a Gaussian fit peak-tracing is proposed. The versatility of the algorithm is demonstrated using numerically simulated data for a wide range of case studies, including non-dispersive, dispersive, dispersive with viscoelasticity, overlapping modes, and highly attenuative materials. It is further validated utilizing two experimental datasets. The maximum error in measuring phase velocity using experimental and numerical data is less than 0.8% when compared to the analytical value obtained using DISPERSE. The phase velocity values obtained with the proposed f-k transformation with a Gaussian fit peak-tracing algorithm are also compared with the conventional methods and found to be superior. Further, the proposed algorithm is suitable for evaluating the phase velocity of the coexistence of forward and reflected wave modes. Hence, the proposed algorithm is appropriate for measuring phase velocity where separation of modes or reflected signal is difficult to achieve, such as for a smaller specimen or even longer specimens with higher-order faster mode.

A study on the use of fundamental antisymmetric-like guided wave for health monitoring of elastic–viscoelastic bilayer structures

“…The phase velocity of the A0 mode obtained using the peak-amplitude tracing algorithm 41 in both FE simulations and experiments was comparable with the velocity obtained from the theoretical dispersion plot. This validates the FE model.…”

“…The theoretical phase velocity of the A0 mode at 100 kHz from the dispersion curve is around 1553 m/s. A comparison of the A0 mode phase velocity obtained from DISPERSE 33 and the peak-amplitude tracing algorithm 41 is shown in Figure 7.…”

“…In this case, the rubber layer was modelled without considering the viscoelastic properties to differentiate the effect of viscoelasticity on the wave propagation in the bilayer structure. The frequency-wavenumber plots for the bilayer with elastic rubber layer, overlaid with the dispersion curves for only steel and bilayer, and the peak traced using the peakamplitude tracing algorithm, 41 when excited from the steel surface, are presented in Figures 8(b) and (c). In the legends of Figures 8(b) and (c), 'Steel(T) Rubber(R)' indicates that the transmitter is on the steel surface and the receiver is on the rubber surface, and 'Steel(T,R)' indicates that both the transmitter and the receiver are on the steel surface.…”

“…The A0 wave mode was generated in the bilayer structure by giving nodal displacements on the steel surface, as shown in Figure 9(a). The frequency-wavenumber plots for the bilayer with viscoelastic rubber layer, overlaid with the dispersion curves for steel and bilayer and the peak traced using the peak-amplitude tracing algorithm, 41 when excited from the steel surface, are given in Figures 9(b) and (c).…”

Reconstruction of Lamb Wave Dispersion Curves in Different Objects Using Signals Measured at Two Different Distances

“…However, the unusual properties of these waves, such as dispersion phenomenon (the velocity as a function of frequency), infinite number of modes, the convergence of modes, modes interlaced, mode splitting due to edges, and other factors [2,[5][6][7], create complications in the analysis of receiving signals and, at the same time, limit the applicability of such waves. The dispersion phenomenon, that leads to phase and group velocities, is frequency-dependent, and affects the variation of signal amplitude is named as one of the main limitations of such waves [6,[8][9][10]. However, the other problem that complicates the application of the Lamb waves is the infinite number of modes and their convergence.…”