||March 22, 2011 23:22
Blowing up of spectral method
I am using Fourier spectral method in the spanwise direction in my DNS of boundary layer transition. The nonlinear terms of NS equations are calculated pseudo-spectrally and 3/2 rule have been used for de-aliasing. When I used 8 Fourier modes, the code seems run well, but when I used 16 or more Fourier modes, the code will blow up at certain time stage after long time integration. Any suggestion to solve the problem will be appreciated.