[yochule]

hi all...
i need to confirm what's the problem in my set up.
i have two domain: one that´s is the whole inlet, outlet and stator, this domain is stationary. and the second is only around the rotor, this domain is rotating with the rpm of the rotor.
i set up one interface. this interface is only one surface. i set up it in: Interface Models: General Connection. Frame Change: Frozen Rotor.
i put the inlet with 350Pa and the outlet with 0Pa relative.
when i see the results it seems to be wrong, but i don´t know where is the probmem...
can anybody help me....
thanks!

[ghorrocks]

What seems to be wrong?

If you mean the streamlines going wild in the rotating section it is because you need to calculate the streamlines using the "Velocity in Stationary Frame" variable.

Glenn Horrocks

[yochule]

hello glenn.
ok, you're right. but if i plot the velocity in the stationary frame i have streamline leaked (cut down in the interface) and this seems odd. i have this cut near the interface in all plots? (i plot an isosurface of turbulence kinetic energy and see the same).
i expect that the sreamlines have some curvature, but not that have a break. and the streamline in the first face would be the same number that in the back face....
are i'm wrong?

[ghorrocks]

Hi,

I can see the facetting of the inlet and outlet ducts and the mesh is very coarse. You are never going to get anything believable with a mesh as coarse as that. Start refining the mesh and things should behave themselves. Do a sensitivity analysis to determine how fine the mesh needs to be for the accuracy you require.

Anyone - we really need to write a FAQ on the basics of checking mesh, timestep and residuals convergence and sensitivity for these type of basic questions. 90% of the "my simulation is not accurate" questions do not do the basics here - any takers?

Glenn Horrocks

[ckleanth]

Originally Posted by ghorrocks

Anyone - we really need to write a FAQ on the basics of checking mesh, timestep and residuals convergence and sensitivity for these type of basic questions. 90% of the "my simulation is not accurate" questions do not do the basics here - any takers?

Glenn Horrocks

.............

[yochule]

hello guys..
i'm now running a test with a more small mesh...
how can i determine the quality of the mesh before the running? it´s there a tool like in FLUENT to see the best and the worse elements of my mesh?

i make a coarse mesh in the ducts far away of the blades because i think in this place the pressure and velocity gradients will be smooth...but, well i´m now be sure of it...i will post the results..

[yochule]

i have extended the domain and drawing the pathlines, but i still can see a break in the interface with "velocity", when i draw it with "velocity in Stn Frame" some pathlines were missing...it is normal?

[ghorrocks]

The streamlines are behaving as expected. The Velocity variable is in the local frame of reference so when you go through a change of reference frame streamlines drawn with the velocity variable will have a kink. Streamlines drawn using Velocity in Stn Frame are continuous but do hit the blades on occasion as the blades are rotating.

If you don't get this concept I recommend you only use streamlines inside a single domain, and always use the "velocity" variable. Then the streamlines will behave as expected and not hit walls.

Glenn Horrocks

[yochule]

i was thinking about the linepath.... i'm agree about the "velocity" drawn of streamlines. the difference of velocity of the coordinate system of both domain make the streamlines to break.
but, i still don´t understand the picture of "velocity in stationary frame". i think it would be exactly if i see the device in my hands, i expect to see all the lines reach the end. there is not a delta T, so i can't say that the line is cut because the particle reach this position in the end of analysis....
where can i read about this?