i have a simple 2D model which is so similar to BUMP2D (examples in CFX help), a multiphase domain, (free surface)
and i want to make a model of cylinder under seabed, actually i have problem with solver control in cfx-pre,
i don't now how many iterations should be set, what kind a advection scheme, and my biggest problem is
Fluid timescale control.
Here's some tips:
Most free surface simulations are run transient. Convergence is very difficult steady state.
Use any high order advection scheme. Either hybrid with a large blend factor or high resolution.
Use adaptive timestepping, homing in on 3-5 coeff loops per timestep. If convergence is tricky then maybe increase to 4-8 coeff loops per timestep.
As far as iterations needed, if you have the luxury of running it as a batch file and logging out, you can check it periodically and see how it is progressing. I have done this on Windows and Linux/Unix machines. It's a little different in each operating system (and your system admins may have certain restrictions on CPUs or runlevel), but it's not tricky in any case. Once you have an idea of how many iterations your simulation needs, you can take some multiple of that for the max number of iterations to be courteous to other users.
|All times are GMT -4. The time now is 14:42.|