Hi,
While simulating Lamb waves in a plate using pstdElastic2D, I noticed a difference between theoretical and simulated dispersion. I'm modelling a steel plate of 6.8 mm thickness and expect phase velocities of 1500-2000m/s, source frequency is 60kHz, so the expected wavelength should be around 2.5-3cm. The observed disagreement was while using a mesh grid with step-size dx=dy=0.2 mm (i.e ~125 grid points per wavelength). I have done a convergence test where I reduced the mesh grid step size and I managed to get the results to agree if I went down to a dx=dy=0.1 mm, which is computationally too costly for what I need to do.
Is there an explanation for this behaviour? My impression has been that k-Wave and the pseudospectral method would work also with relatively sparse grids compared to the wavelength.
Best regards
Magnus