3d vof tank fill convergence issue
Hello,
I have been learning about mesh refinement, time stepping, and numerical conditions settings to make 3d vof tank fill problems solve properly with Baram 26.2.4, even with fixed time stepping. I am able to get good convergence and solutions when volume input is high, such as 1 m/s velocity input. However, if I try to reduce the input flow down to something like 1 gpm or 0.014 m/s into a 3 inch pipe, I have to run adaptive stepping, and it take forever, stepping initially at what appears to be about 1e-5 or 1e-6 initially.
Is there something I'm missing, or do I just need a superfine mesh at the inlet plus very small time steps to get this solution to start OK? What do you recommend in general as input fluid flow is small?
Also, on a different topic, if possible in the future, it would be great if you could add a running timer on the Run page under status that shows how long the run has been going in computer time.
Thanks,
Tom
Dear Tom,
When using an adaptive time step, the time step size increases if the velocity decrease. I don’t understand why it’s getting smaller. It seems you used a fixed time step for the 1 m/s condition.
When using an adaptive time step, the time step becomes very small if there are cells of very small size. If those cells are located in regions that aren’t important, it would be best to increase the CFL number or use a fixed time step. If the small cells are in important regions, it would be best to regenerate the mesh.
Another possibility is that abnormally high velocities occur early in the calculation while the flow has not yet converged, causing the time step to decrease. In such cases, increasing the “Max iterations per Time Step” or reducing the relaxation factor may help.
Best regards
I guess I didn't explain well. My primary issue is that, with all numerical settings the same, but just making the inlet velocity lower, such as lowering it from 1 m/s to .1 m/s, the solution won't converge. I'm asking why just making input velocity lower causes convergence issues?
The second issue is that, as you said above, the adaptive stepping method starts out at very small time steps and stays there. I've tried lower and higher CFL, finer meshes (especially at the inlet), and increasing Max Iterations. Sometimes fixed time steps work OK, especially with higher input velocity, but that behaves strangely sometimes as well. I had one problem where a time step of 0.01 sec worked fine, but 0.005 failed.
Getting 3D VOF problems to converge properly and reliably is very hit or miss for me.
Any other suggestions or rules-of-thumb?
Thanks,
Tom
It's hard to say exactly what's causing the issue you mentioned. It doesn't seem to be a common occurrence.
If the number of cells of the mesh is not very large and there are no security concerns, please compress the baramFlow folder and upload it so we can take a look.