Hello Guillermo,

One of the concerns I have with your model is signal aliasing, which happens when the data output frequency is too low for the frequency content of the signal. The input is a 300 kHz periodic signal, which means one period is 3.33e-6 s. The best practice for transient structural analysis is to have about 20 ouputs per period, which means the maximum time step is 1.67e-7 s.  Your plot has labels at 0.5e-4 s and I see the individual output points where the curve makes a sudden change of slope. In Transient Structural, try solving at the suggested maximum time step of 1.67e-7 s.

I like Erik’s suggestion because an Explicit solver may automatically choose a time step of this or smaller.

Another requirement when simulating the transport of high frequency longitudinal waves through a medium is making sure the mesh has enough elements along the wavelength of the signal to accurately represent the sine wave. The wavelength of a sound wave is λ = c / f where c = 5100 m/s is the speed of sound in aluminum and f is 300,000 Hz which means the wavelength is 0.017 m or 17 mm.  The best practice is to have at least 6 quadratic elements or 12 linear elements in one wavelength. Therefore, the maximum edge length in your mesh should be 1.4 mm for linear elements or 2.8 mm for quadratic elements. You may already have that based on the line of code that defines element_size = 0.0025 which I assume is in m.