CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   strange node values @ solid/fluid interface - help (http://www.cfd-online.com/Forums/fluent/49628-strange-node-values-solid-fluid-interface-help.html)

JB October 27, 2008 12:38

strange node values @ solid/fluid interface - help
 
Hey guys,

I'm modeling a heat sink (HS) operating in natural convection. Here is an overview of the model:

Geo + mesh: - one solid volume (al.), - one fluid volume (air), - hex elements in the solid, - tet elements in the fluid, - non-conformal interface between HS and Ambient Air (AA), - heat source at the bottom of the heat sink. - pressure outlet/walls/symetries to close my model (with correct backflow temp.)

Solver set-up: - laminar model, - gradient option: Green-Gauss node-based, - density: boussinesq model (-> steady state), - pressure disc.: body force weighted, - momentum/energy disc.: 2nd order upwind. (all other parameters left to default setup)

First of all, do you see some nonsense in my setup?

Here is my problem:

The 1st run used a quite coarse mesh, and gave a physically realistic solution. When plotting the temperature contours at the HS/AA interface, the facet contour looked great, and the node values contour looked a bit more 'random', but only oscilating by +/- 1 deg.C around facet values. so I thought: no big deal, this might be due to the non conformal interface, let's run the same simulation with a finer mesh and see what happens...

So I ran a second model, with the same setup but a finer mesh. The facet values were still very realistic, but this time, some nodal values on the interface are totally unrealistic (2 nodes in the same cell with a temperature difference of 400K !?)

From Fluent help files, I read that the node values are interpolated from neighbor cell values. But my cell values look ok (no crazy values when reporting min/max temp in the volumes). So what's happening? Is it due to the non-conformal interface?

Another thought was: I'm using the node-based gradient method... so I guess If my node values are wrong, this could affect the validity of my solution? (not only a post-processing issue, if you see what I mean).

For info, the mesh refinement significantly increased the temperature of my heat source (+10%).

Fluent Gurus? Any clue?

Thanks a lot in advance! /JB

JB October 28, 2008 07:18

Re: strange node values @ solid/fluid interface -
 
anyone?

JB November 1, 2008 13:04

Re: strange node values @ solid/fluid interface -
 
up


All times are GMT -4. The time now is 21:48.