[jekmann]

hello,

Can any body please help me to find wrong step in my calculation or meshing:

I've made an empty room's model in which airflow moves from ceiling to bottom/floor. There are 2D and 3D models. The top horizontal face (top horizontal edge for 2D model) was chosen as VELOCITY INLET. The bottom horizontal face (bottom horizontal edge for 2D model) was chosen as OUTFLOW.
The air velocity is constant and equals x=0; y=0; z=-0.5 m/s.
The feature of this model is that the volume of the room consists of 5 volumes with different mesh accuracy for 3D model and 4 areas for 2D model. (pictures 1 and 2)
As I know from Fluent or Gambit guide: all undetermined boundary conditions for faces (3D) or edges (2D) will be determined as a WALL. For that reason,I have determined all internal faces or edges as INTERIOR (picture 3 for 3D).
For calculation I have used standard k-epsilon turbulence model.
But after calculation, I saw strange picture . For 2D model it was O.K. - I have seen unidirectional downward pathlines. (picture 4)
But for 3D I have got strange whirl inside room model. (picture 5)
I used absolutely same parameters and data for both cases (i can give all data, if it will be necessary). I can not understand what is wrong with 3D model. (Mesh check did not show any problems).
Maybe, somebody have had same problem?