Natural Convection

[Saint]

Hi,

sorry for asking such basic question in between of complex and sophisticated question; I'm new, but I am trying to learn, I hope you can help me as before, in my last question, the community was very helpful, with special thanks to "kevinarden" memberlist.php?mode=viewprofile&u=7064

I start with a general question, the work flow.
I want to simulate the natural/free convection, over a hot plate/surface.

1. I have simplified my question and geometry, to only air.
2. I have used some meshing program, to mesh it (but the mesh is not good right now (even its not visually symmetric)
3. In Setup: selecting Transient, 100 intervals, with 1 step
4. Starting a new project, I add "K-Epsilon'' equation to the list (since it is difficult for me to add it afterward by hand, or maybe it is very easy but i don't know how!?)
5. Activating these equations: Heat, Navier-Stokes, Results, K-Epsilon (with selecting convection model as computed, in the heat equation tab)
6. Adding, default "air" as the material, with "K-Epsilon" as the viscosity model (in future I want to add thermal compressibility mode as well, but right now, I just try to run something simple)
7. Adding, "body force" so I can activate "Boussinesq" in the Navier-Stokes tab, to implement the effect of gravity, to have a natural convection
8. Initial conditions: 300 for temperature, 0,0,0 for velocity and 10000 for pressure, and 0 for both Kinetic energy and dissipation
9. Boundary Condition: 400 for the hot surface, with none-slip condition
10- Boundary condition, none.slip condition for all other surfaces

Please let me know if I am missing something, or I am doing something extra !?
My plan is to change the viscosity model afterward, see differences and maybe find more accurate results, like going for K-Omega and such and of course the compressibility.

Regardless of the general workflow; my simulation is diverging out of the tolerance so it naturally has been terminated. But I don't know how to diagnose such issues, I have asked different sources and all they said was a list of possible sources for the issues.
But what is the engineering way to approach such problems and fix the problems?
this the sif file (you can also find it in the compressed file):

case.sif

(4.38 KiB) Downloaded 32 times

This is a compress file containing the simulation:
https://file.io/ed5N3KOv9N8l

Thanks again in advance for yur many kindness and help :*

[kevinarden]

I tried to get just the heat equation working first, prior to adding stokes. However, I can't get the heat to conduct.
I re-did the case in 2D and the 2D heat equation looks good.
Not sure why the 3D heat doesn't work, unless it is a mesh quality issue.

heat.sif

(2.79 KiB) Downloaded 33 times

2dbox.zip

(111.65 KiB) Downloaded 37 times

[kevinarden]

The 3d box is changing, it is just changing at a much smaller rate, that is because it is much larger
15 M, by 20M
The 2d box is much smaller, 1 M x 1M
It will take a long time to change the air mass temperature in the big box.
Which leads to a bigger issue for convection, the stokes time steps have to be very small.

2d.PNG

(253.2 KiB) Not downloaded yet

3d.PNG

(103.15 KiB) Not downloaded yet

The 3D box looks good in Steady State

ss.PNG

(106.72 KiB) Not downloaded yet

[Saint]

Thank you so much for the guidance.

- Can you please confirm the steps in the setting up the simulation?; I am worried maybe I am missing something, or maybe I am over doing it.

- So I have to select very small time steps for sake of navier-stokes to catch up. Is there an easy way to differentiate between the time steps? (so the simulation will be lighter)

- again, as a confirmation: having the transient simulation, with bousinesque (in body forces, to put the gravity in effect), along with viscousity and non-slip walls are the correct way to simulate the convection ?

Thank you so much Mr Arden

[kevinarden]

Yes you set up was good. However look at the gravity vector, you had gravity as minus y direction. In your 3D model maybe minus Z?
No slip wall is for stokes, yes.

[kevinarden]

Here is a 2d case similar to yours.

flow.zip

(45.33 KiB) Downloaded 41 times

2d.PNG (19.56 KiB) Viewed 734 times

[Saint]

Thank you so much !
It is good to know that I am on the correct route !

If I am not mistaken, you have not introduced the "viscosity model" and "compressibility model"; (i'm still new in reading sif)?

When I tried to run it again, after you said "I need smaller steps", I even went to "0.000001" step size with "1000" intervals, assuming it will help prevent the divergence, but you have made it with "20" and "0.1" !!
Although my simulation was in 3D, and yours is in 2D, I thought with those small steps and high intervals I must get something....
maybe my understanding is wrong???

Also, I have tried different combinations; in between, when I neglected the viscosity, I had some answers!!!! which logically, directed me toward the viscosity model (or maybe too rough mesh to simulate the viscosity?) (maybe I am missing something in the process of introducing the viscosity )

A good guy, mentioned, that I have to simulate different scenarios, and go, step by step: like starting with K-Epsilon and then going to K-Omega and ... to have some data and prediction, to prevent the divergence he also I might need to introduce closer conditions and data, to the final possible results, to make it easier for the software to converge...... I'm not sure about it, but maybe you agree with him?


Thank you so much again,for the answers and guidances.
sorry for bothering you so much, I am learning and I am under time pressure; it is harder to think straight under pressure.
(most probably you will see more questions from me I hope you and community be supportive as always)

[kevinarden]

Yes I did increase the viscosity variable to get it to converge. Elmer is not very good at turbulent flow. OpenFoam is likely better suited for your problem.
viewtopic.php?f=3&t=6995&hilit=turbulent

[kevinarden]

I try with your mesh, and it diverges right away. You may have a mesh issue. I did in 3D with brick elements, and it will run with high viscosity models.

3d.PNG

(360.72 KiB) Not downloaded yet

[Rich_B]

Hello,

The original link to the simulation files doesn't work, it says 'deleted'. You could try compressing the files into archives. The forum allows up to 1 megabyte per attachment and up to three attachments per post. Posting your current simulation as a working example, including geometry and sif file, would be helpful

Thanks, Rich.