Computation time
Helo,
I still wonder if I am doing something wrong or the computations are really so demanding and time consuming. Can you present here some details about some of your experiences? I start with this: Platform: Win2003, AMD64 3200, 2GB RAM dual PC400 Problem: - unsteady simulation, - rushton turbine (240 rpm) in a 20x20cm tank with 4 baffles, - sliding mesh, - VOF (air-water), - k-epsilon, - PISO, - pressure: Body Force Weighted, - under-relaxation factors - all 0.1, - about 31000 cells of mesh. Time: 1 second is computed in about 10 hours I think its quite long, maybe I make some mistake? |
Re: Computation time
Your settings sound quite good. Typical VOF unsteady application. Anyway your computational time is actually quite large (never did a serious benchmark, but it takes me approximately the same time, 1 second flow time in about 8 hours, for 400k cells model on a Linux machine).
How long does it take you 1 iteration? Which is your time step? How many iterations are required for a time step? Reading back your post: maybe you're URFs are too low...try something like: -pressure: 0.6 -momentum: 0.8 -others: 1 Hope this can help Edi. |
Re: Computation time
1- Check the swap area 2- Check the performance of the microprocessor (Right click on the bottom of the screen, select the task manager then the performance tab)
|
Re: Computation time
Swap file is probably not used as system disposes 2GB RAM memory and fluent takes about 150MB also processor works at 99%.
I found the problem existing rather in simulation settings than in system settings. I changed the under relaxation factors and simulation speed increased 5 times - quite good. I set the urfs very rigorously because at the initial stage the integration bacame unstable - so it's worth to "loose" the urfs a little after the simulation has already passed that troublesome initial stage. reagrds Maciej |
All times are GMT -4. The time now is 01:54. |