Energy pile up
 

I am doing compressible LES for the decay of turbulence energy(Comte-Bellot and Corrsin experiment). Smagorinsky SGS model is applied and no explicit filter is used. I found the energy piled up in the high frequency area, even I set Fourier coefficients to zero for high frequency at every time step(as advised in "Toward the large-eddy simulation of compressible turbulent flows", JFM vol.238. pp.155-185) . Any suggestion is appreciated.