[Admo]

Hello everyone,

I recently started to learn CFD and I am currently working on a problem which can be classified as a bouyancy-driven flow (natural convection) in a closed domain. The domain is heated to ~1500 K in the middle but is cooled to ~350 K at the walls. The pressure in the domain is ~1500 bar.

By experimenting and reading the recommendations in the fluent documentation I have come up with a solution technique which seems to give reasonable results:

1. Solve steady state with zero gravity and laminar
2. Set gravity = 9.8e-07 and solve
3. Turn on turbulence and solve steady state
4. Change to transient and run until seem relatively stable
5. Increase gravity to 9.8e-06 and solve transient until seems stable
6. Keep on increasing gravity gradually with one order of magnitude at each step. I will use monitors to see when it seems good to further increase the gravity.

Is there any problem of splitting up the problem into several steps in this way, and especially increasing gravity this slow?

[Zigainer]

I don't see a problem here. Or do you want to make a transient calculation to compare to an experiment or somthing? (want to see .... at 3sec e.g.)
Otherwise this works out. Probaly you can try to increase gravity faster (just to save time), but I think your way is right.

[Admo]

Thanks for the answer Zigainer!

No I am not interested to look at specific times in the heating process, just the final state.
Yes, I can probably increase it a bit faster, but not so much it seems.

When I increase the gravity too much, the flow (and final solution) can change quite dramatically from how it looked in the initial steady state solution. I want the flow directions to change as little as possible in each gravity increase and strive to get a final solution which looks close to the initial steady state solution, just higher velocities.

Is this way to think right? In one way it feels like I am forcing the solution to look like I want it to.

Thanks

[Zigainer]

Originally Posted by Admo

Is this way to think right? In one way it feels like I am forcing the solution to look like I want it to.

Thats what I thought when I heard your answer .... why are you sure that your steady state is correct? Or when you increase the gravity more at the end should be the same solution (probably takes longer) otherwise somethin else is wrong. And with more gravity your solution might be different, but no matter hwo you get to the final, it should look the same!