Temperature field error in compressible LES
I encountered a serious problem in a high speed gas jet simulation (LES) with sprayFoam.
The jet velocity is 177 m/s with random perturbations added. The temperature at the inlet is set to 600K (!!). The velocity field (attachment) and pressure field are looking good. For the RANS simulation I get reasonable temperature profiles. For the LES (smagorinsky) the temperature in the field remains constant with a non-smooth transition.
Chemistry and so on are switched off.
I tried various fvSchemes (upwind, linear, ..). Also the transsonic option.
My guess is that it has something to do with the thermodynamic settings (same as for aachenBomb-Tutorial).
The only post I found that is dealing with a similar problem is
Can anyone give me a hint of how to solve this problem
Can you explain better what is plotted in the first two pictures (please provide scales)?
Can you provide a picture of the mesh for picture 2?
Some more information about the case:
The jet diameter is 2.2 mm and the cell size at the outlet is about 0.05 mm. An o-block mesh has been used. For the contours the non-interpolated values have been plotted. So it is visible that the mesh is relatively fine at the inlet (velocity field).
The temperature field however is not so so well resolved. I would have expected a temperature field similar to the gas velocity field. The strangest thing is, that the temperature field is static for all time steps.
Hope the case is clear now
From what you say I guess that first picture is velocity and second is temperature on the same cutting plane of the same mesh. In this case I agree with you temperature field looks wired.
If temperature field is the same at every iteration it might be that your fvSolution file is set so that energy is not actually solved.
Can you plot your fvSolution dictionary and a log of few iterations?
Also what solver are you using?
Try to use constant transport properties first (mu and Prandtl) instead of Sutherland as I'm not sure in which cases the formula implemented to estimate alpha (thermal diffusion coefficient) provides reasonable results.
I was able to solve this problem a while ago. It seems to be important for this specific chase to initialise the flow field with a streamwise velocity component u_field>5 m/s (u_jet = 240 m/s). If u_field is in the order of 1 the jet cannot develop well.
Still I think the core is somewhat short in comparison with fluent 13.0.0 .
|All times are GMT -4. The time now is 07:50.|