convergence problem
Hi, i am running inhomogeneous free surface model(steady flow). For some runs, residuals of volume fraction stuck at a high level (~1E-2) but other residuals died off to the target quickly. No matter how I change the time scale for vf and global time scale, it didn't converge. Would you please give your comments/advice? thanks for all. Ken
|
Re: convergence problem
In my experience, free surfaces are NEVER steady. There are always wave or ripple effects that are anything but steady-state.
Try running as a transient simulation with 10 coefficent loops max and a smallish time step with a convergence criteria of 1E-5. After the problem starts converging in just a few coefficient loops you can start increasing the time step (better yet use the adaptive algorithm on the volume fraction coefficient loops). Run the simulation long enough that you get a full volume turn over through the model. Hope this helps, Jeff |
Re: convergence problem
I agree with Jeff that free surfacesa re almost never steady. I was also wondering if you have a region in your simulation where there is a change in height or velocity in the free surface. If so, you should choose the timestep to be small enough to `capture' this. For example, if modeling weirm the imestep should be small enough to adequately model the flow where the fluid is fastest.
Don't refine anything until the mass balance is correct, or ou will be wasting time Use second order transient differencing for a noticable speedup in solution. Julian |
Re: convergence problem
The vilume fractions is most unstable solution in the two-phase flows. The developers are recommended to set time step for volume fraction equations in order less tha for other equations.
|
All times are GMT -4. The time now is 02:24. |