[acalado]

Greetings,

I have a transient simulation where for the first period some boundaries are walls, and after that I change them to velocity boundaries with a ramp up increase in the velocity using UDFs.

The ramp up should take the value at the end of the first period and then linearly ramp up with time, however I am having some issues on convergence.

I have tried using small timesteps and lower under-relaxation factors but still with trouble.

Any suggestions are appreciated!

Thanks

[RaiderDoctor]

Hi, I'm afraid I don't understand your setup. Could you give us a more detailed description? Posting an image or two to help explain yourself would really be beneficial.

[acalado]

So basically I have 4 boundaries: 2 of them are open portals (pressure boundaries) and the other 2 are ventilation ducts.

For the first period the ventilation ducts are closed, so they are walls.

After this period, I want to use a UDF for the ventilation ducts such that the velocity is increased (ramp up), with 1 of the portals also being UDF defined (mass flow ramps up), and the other portal is left as a pressure boundary (no change).

However once I set up the new case file with the UDFs and continue the run the simulation quickly diverges.

[RaiderDoctor]

Okay, so now we’re getting somewhere. For the pressure boundary conditions, are they inlets or outlets? I ask because it’s sounding like your divergence might be due to an improper boundary condition.

[acalado]

I'd say on is an inlet, the other an outlet (even though there is a mixed flow of outgoing and incoming fluid).

After that, the previous inlet remains a gauge pressure inlet (positive), and the outlet is a mass flow inlet (negative sign because flow is going out).

However it appears from my surface monitors that the gauge pressure boundary (which does not change from first to second periods) is the one that is diverging.

It says the error is related to the AMG solver: temperature (very low temperatures) which is quite confusing.