physical velocity formulation and porous media
Hello Gurus -
I have a 3D sim with a porous bundle zone. The flow regime is laminar at low Reynolds number. I am using the new physical velocity formulation for the porous media in FLUENT6.1 and have had successful (and rapid) convergence under steady state conditions. However, I am unable to get convergence for a time dependent simulation with fixed timestepping. I'm using the segregated solver with PRESTO! pressure interpolation, PISO P-V coupling, 2nd order momentum discretization and 1st order implicit time discretization. All solver parameters (underrelaxation factors and the like) are defaults. I begin from the steady state solution described above and have taken the timestep down to as low as 0.025 seconds - after one or two successful timesteps the residuals blow up. The size of the timestep doesn't seem to be the issue (although I welcome any "enlightenment" from you). Does anyone else have experience with a similar simulation setup or have some suggestions on what might be the problem? Best regards - Ken |
Re: physical velocity formulation and porous media
May I knwo what system do you simulate?? e.g. liquid-gas, gas-gas or.....
do you grid size is fine enough?? I have tyr porous media with gas-gas, it seem no divergence problem to me... but did you check the concentration of you system?? do its tarry with any experimental or theory result?? |
Re: physical velocity formulation and porous media
Ivan -
It's all liquid - the porous media is in place for pressure drop and fluid flow effects. I am not modeling a mixture or multiphase problem at this time. As for the grid - I achieve rapid convergence with the default solver settings for the steady-state solution so it appears to be fine by that functional metric. I believe the cell skewness and volume limits are ok - its a tet mesh so it ain't perfect, but it does appear adequate. Comparison with experimental data for the steady state systems has been good with past simulations. I am basically looking for the time-dependent simulation to converge to the steady state solution or thereabouts after a fair number of timesteps. All models used are the same but the time-dependent simulation diverges even with a very short timestep. I welcome further comments and thanks for the questions and insight. Regards - KLG |
All times are GMT -4. The time now is 03:57. |