[wj_wanjie]

I simulated a transient natural convection simulation in a closed tank, the fluid is incompressible and its properties are temperature depend(i.e. the Boussinesq is not used). The mesh grid is well and the Y+ is below 1, now the continuity residual is very high(1~10),can anybody give some advise?

[Cyrus69]

hi
Check your BCs , initial values and decrease order of discretization.
I didn't understand why u didnt use boussinsq,it just make it easier to solve it!
I'm also simulating a natural convection in closed domain(my problem is a vacuum flask).maybe u can help me at this!
the flask has 90c water that has air at top of it.idk what BC should I put for top of water.i made it wall but in wall condition there's no option for fluid!
I've sent geometry in below.the white zone is filled with hot water
111.png

[ANTONY A R V]

Hi
Pls mention the initial conditions you have used.

[wj_wanjie]

“i made it wall but in wall condition there's no option for fluid!”，you mean that there is no fluid material option while you are defining the boundary condition? i think if you don't sim ulate the radiation heat transfer in the flask, may be you can set the liquid-air surface as interior boundary condition or the default condition while your import your mesh.

[wj_wanjie]

Originally Posted by ANTONY A R V

Hi
Pls mention the initial conditions you have used.

hello, the attachment is my model and case, it comes from one journal paper, "q" is the heat losses to the environment and the square is full of fluid. The initial conditions is that the fluid is 292 degree(given by the paper),so it is a natural cool down process of the fluid. i think the solution method and boundary conditions as well as the time other settings are the same to the paper(except the initialization that not given in the paper). I changed choose the double precision in the start up page of fluent yesterday and the continuity residual can coverage, but the k residual became divergence( the residual is about 2.0*10-3), i am now trying to find what's wrong with it......

boundary condition.png

mesh.png

[Cyrus69]

Originally Posted by wj_wanjie

“i made it wall but in wall condition there's no option for fluid!”，you mean that there is no fluid material option while you are defining the boundary condition? i think if you don't sim ulate the radiation heat transfer in the flask, may be you can set the liquid-air surface as interior boundary condition or the default condition while your import your mesh.

thanks for answer.no I'm not simulating radiation, it's just cond & conv problem.when i changed the wall BC to interior,it comes divegence in temp.I'm
still working on it.by the way, i opened ur case file. how did u define properties of salt?

[wj_wanjie]

in the material page, you can define any material you like( the material name, the propeties), the correlations of the properties with temperature is given in the paper and when you are defining the correlations, the temperature units should be K rather degree.

[Cyrus69]

Originally Posted by wj_wanjie

in the material page, you can define any material you like( the material name, the propeties), the correlations of the properties with temperature is given in the paper and when you are defining the correlations, the temperature units should be K rather degree.

ok but why u make it "polynomial"? how fluent will calculate the properties?

[wj_wanjie]

In my case, the correlation between the properties and the temperature is given in the paper, and it is polynomial, i think fluent calculates the properties according to the local temperature of every volume.

[ANTONY A R V]

Hi

Pls try to give very some small x ,y and z velocities (like 0.001) as initial condition .

[xicheng]

Hi, have you solved this issue? I faced a same problem when simulating the water natural circulation in a closed domain. I used the K-omega BSL model with temperature-dependent density. The simulation runs well until 1000s and then k is divergence.

Originally Posted by wj_wanjie

hello, the attachment is my model and case, it comes from one journal paper, "q" is the heat losses to the environment and the square is full of fluid. The initial conditions is that the fluid is 292 degree(given by the paper),so it is a natural cool down process of the fluid. i think the solution method and boundary conditions as well as the time other settings are the same to the paper(except the initialization that not given in the paper). I changed choose the double precision in the start up page of fluent yesterday and the continuity residual can coverage, but the k residual became divergence( the residual is about 2.0*10-3), i am now trying to find what's wrong with it......

Attachment 63842

Attachment 63843