ali
|
January 5, 2006 06:13 |
spectral methods for DNS and LES
I am working at this time to develope a semi- spectral DNS and LES code to simulate a turbulent channel flow with Reynolds No. Re=180, using a staggred grid arrangement. I have a problem when evaluating all derivatives in stream wise and span wise directions with fourier spectral method which makes my code diverges just after 20-50 iterations. my code FFT going well i tested it to solve the poisson equation in 3D with two periodic dirctions .Also i tested it to solve Burger's equation by spectral methods and i have found the exact solution. but my problem is always the divergence of my code DNS when implementing the pseudo-spectral traitement for all terms in the periodic directions. though i have used the skew-symmetric form for the convective terms (which is stable for spectral calculations(Moin , Kravchenko 1996 )in my code my problem have not been solved. please if any of you have comments or any idea to help me by sending his reponse at : az_matug@yahoo.com i will appreciate his help; thanks
|