Convergence to a different maximum
 

I'm working on a problem of acoustic scattering of a periodic pressure pulse from a cylinder using Linearized Euler equation.The spatial discretization is a higher order central difference and Runga-Kutta is used for time marching.I have a numerical filter to add stability and a buffer zone at domain outlet to absorb the outgoing waves.Since the problem is symmetric I use just half the cylinder with a symmetry condition.

I ran my code on different grid sizes and the results do not converge.Once the results become periodic the acoustic pressure is just a sine wave of the initial time period T.My results,for different grid sizes attain periodicity with different peak amplitudes.On a 500*500 grid the peak amplitude was 3e-5,4.5e-5 on a 600*600 grid and 3.4e-5 on a 700*700 grid.There is no dispersion error.

I need some advice on what could be the reason for the observed oscillations.

Re: Convergence to a different maximum
 

What makes you think convergence has to be monotonous? Try 800x800. If you get an amplitude around 4e-5, you're closing in... :)