Using a validated finite-element modeling code, buried out-of-plane (two dimensional code (2D)) and in-plane (three dimensional code (3D)) acoustic emission (AE) dipole point sources were operated at different depths below the top surface of a 25.4 mm (1 inch) thick steel plate with large transverse dimensions. The depths ranged from a source centered at 1.25 mm below the top surface to a source centered at the mid-plane. Most of the cases were run with a 2.3 microsecond source rise time. For each depth, the out-of-plane displacements were obtained as a function of time at a series of propagation distances up to 1016 mm (40 inches) from the epicenter position for the out-of-plane sources and up to 381 mm (15 inches) for the in-plane sources. The total time for each signal was from the initiation of the source up to 580 microseconds (2D) and 200 microseconds (3D). Since the displacements were obtained on both the top and bottom surfaces, results representing sources at various depths over the whole plate thickness were available. The modeled signals were examined with no filtering as well as with a 40 kHz high-pass filter or a 100 to 300 kHz bandpass filter. In order to correlate the AE displacement signals with Lamb modes, the relevant group velocity curves were superimposed on wavelet transforms of the signals. Modal regions that carried a significant portion of the AE energy were identified by mode, frequency range and source depth. For sources located near the plate top surface, a Rayleigh wave was observed in the top surface displacement signals. This wave was not present in the signals obtained from the bottom surface or from sources not located near the top surface. Signal arrival times at different propagation distances were obtained from the maximum wavelet transform (WT) coefficients at key frequencies of certain modes. Plots of propagation distance versus arrival times were used to find group velocities for a key frequency/mode combination (102 kHz for either the A0 or S0 modes). These velocities were found to be very close to those obtained from Lamb-wave theory. A method to identify the mode that led to the WT peak at the key frequency was demonstrated for AE events that were detected at four sensors in an array. This method could be automated to process digitized signals to significantly improve source location accuracy in thick plates when non-resonant AE sensors are used.
Citation: J. Acoust. Emiss.
Pub Type: Journals
acoustic emission arrival times, AE dipoles, finite element modeling, Lamb waves, source location, wavelet transform