CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Please comment on this Results of Natural Convection!!! (http://www.cfd-online.com/Forums/openfoam/105673-please-comment-results-natural-convection.html)

adambarfi August 6, 2012 08:06

Please comment on this Results of Natural Convection!!!
 
1 Attachment(s)
Hi everybody,

I'm solving natural convection. the attached picture is the result of temperature. side walls are isolated (zeroGradient) and bottom wall temperature is at 400K and the ceiling temperature is 300K, the internal temperature is 350K. but I think this result isn't true.

please comment on this result, is it correct? if not, you think from where this result originates?

Thank you

nimasam August 6, 2012 11:09

dear mostafa,

1) whats your substance?
air, water or ?????

2) which solver do you use?

3) whats the direction of gravity?

4) could you put velocity contour and velocity stream pic here too

adambarfi August 7, 2012 04:42

Quote:

Originally Posted by nimasam (Post 375621)
dear mostafa,

1) whats your substance?
air, water or ?????

2) which solver do you use?

3) whats the direction of gravity?

4) could you put velocity contour and velocity stream pic here too

Hi Nima,
1-air, 2-buoyantSimpleFoam 3-y direction (-9.81)

I changed my solver to buoyantBoussinesqSimpleFoam, fortunately, It solved. I guess that problem source was in thermophysicalProperties. this solver doesn't have transport properties dictionary.

but another question:

the side walls are isolated, in blockMeshDict do I use wall type for all of them or I should use wall type for two and empty type for other walls which they are opposite?

thank you

nimasam August 7, 2012 05:52

However
1) buoyantBoussinesqSimpleFoam uses Boussinesq assumption means
(delta {rho})/{rho}<<1
if you have a big temperature difference maybe it is better to use buoyantSimpleFoam

2)it depends what you are going to solve, if you want to simulate heat transfer between to parallel surface, you can use (two empty (frontAndBack) and two symmetry plane (leftAndRight)) but if you want to simulate a box, you should consider all of them as wall, then for sidewall apply zeroGradient for T

adambarfi August 7, 2012 08:01

Quote:

Originally Posted by nimasam (Post 375751)
However
1) buoyantBoussinesqSimpleFoam uses Boussinesq assumption means
(delta {rho})/{rho}<<1
if you have a big temperature difference maybe it is better to use buoyantSimpleFoam

2)it depends what you are going to solve, if you want to simulate heat transfer between to parallel surface, you can use (two empty (frontAndBack) and two symmetry plane (leftAndRight)) but if you want to simulate a box, you should consider all of them as wall, then for sidewall apply zeroGradient for T

Dear Nima,
Thanks a lot for your reply.

If I want to solve my problem with another fluid, in example a viscoelastic fluid, how should I modify the thermophysicalPeroperties (I mean thermoType)? I'm trying to write a viscoelastic solver and according it I want to solve natural convection.

nimasam August 7, 2012 09:29

look at user guide last chapter, it tells you how to modified a thermophysicalPeroperties but i dont know it is appropriate for viscoelastic fluid or not, im not familiar with viscoelastic fluid at all,
but in general for non-newtonian flow, i think you can use buoyantBoussinesqSimpleFoam, and you just need to modify transport model from Newtonian to a non-newtonian type for example pawerlaw with appropriate coefficient in constant/transportproperties

adambarfi August 7, 2012 12:06

2 Attachment(s)
Quote:

Originally Posted by nimasam (Post 375828)
look at user guide last chapter, it tells you how to modified a thermophysicalPeroperties but i dont know it is appropriate for viscoelastic fluid or not, im not familiar with viscoelastic fluid at all,
but in general for non-newtonian flow, i think you can use buoyantBoussinesqSimpleFoam, and you just need to modify transport model from Newtonian to a non-newtonian type for example pawerlaw with appropriate coefficient in constant/transportproperties

Dear Nima,
my trouble is that the non-newtonian solver of OF is just power law, I want PPPT, Gesikes, Oldroyd A and B models. so I must write it and I'm trying.

again I run my model with buoyantSimpleFoam but again the same results obtained. have you any idea? but I think the density of a viscoelastic fluid doesn't depend on Temperature as deep as shear stress.

I attach you the velocity Glyph and a truncated velocity contour.

thanks a lot

nimasam August 7, 2012 13:24

1) i guess those model are implemented in OpenFOAM-1.6.ext or 1.5-dev, so you may want to implement just energy equation to those models

2)you used no slip condition for all your wall, and distribution of velocity seems fine, whats wrong with result?

adambarfi August 7, 2012 14:04

Quote:

Originally Posted by nimasam (Post 375870)
1) i guess those model are implemented in OpenFOAM-1.6.ext or 1.5-dev, so you may want to implement just energy equation to those models

2)you used no slip condition for all your wall, and distribution of velocity seems fine, whats wrong with result?

Dear Nima,
I think the temperature distribution isn't true. it doesn't similar to a natural convection at all. I check it with Fluent results. the temperature distribution according to the boundary conditions seems to be wrong.
as you see in the pictures that attached before, the velocity magnitude is very low, and in fact the convection didn't occur. what is your opinion?

nimasam August 7, 2012 15:53

i dont know exactly, but test a case with these condition
1) high distance between two hot parallel plate (1 m)
2) increase deltaT
3)turn off turbulence
4)use slip condition or symmetryPlane for side walls and empty for frontAndBack
5)use bouyantPressure
i forget to ask you whether the simulation has been converged or not

adambarfi August 7, 2012 15:58

3 Attachment(s)
Quote:

Originally Posted by nimasam (Post 375891)
i dont know exactly, but test a case with these condition
1) high distance between two hot parallel plate (1 m)
2) increase deltaT
3)turn off turbulence
4)use slip condition or symmetryPlane for side walls and empty for frontAndBack
5)use bouyantPressure
i forget to ask you whether the simulation has been converged or not

thank you Nima,
my last solution was converged.
I did what you said. the solution again was converged, and the results had features of natural convection. but I want to solve this in a box and this boundary condition, I think, doesn't true for it.
meanwhile, congratulation for the two Gold and two Silver Medals!!!!!:D


All times are GMT -4. The time now is 12:00.