I am trying to calculate air flow from a big channel to a very narrow channel. The walls are adiabatic. Very simple flow. At the beginning, I assume the air as incompressible fluid and the computation seemed to be well but the pressure drop is clearly too big. Then I put the density as the function of T and P. This time STAR-CD stopped after a couple of steps and the info file indicates that a negative density was found in near the entrance of the narrow channel.

The turbulence switch is on, and normal k-e high Re model was applied.

What could be the probable reason for this kind of error? I am very new to Star-CD and CFD, and really have no idea where the error comes from. Could you guys give me some hint? Thank you very much.

The problem is probably not physically correct. The fact that you had too big a pressure drop incompressible is probably telling you this. When you go compressible (depending on where your pressure reference is) you get into a death spiral whereby the pressure drops to pull more flow, this drops the density, puts up the velocity which will add to the loss which increases the pressure loss required....

You need to do a scoping calc to make sure that the problem set is sensible. Then make sure the BCs are appropriate and the pressure reference set where you want it.

An approach is to run incompressible first then switch over to compressible - however I believe in the restart after changeover it calculates the density compressibly and only then applies the relaxation (I could be wrong I haven't tried it for a while). Or you can start the problem compressible with a very small relaxation factor - i.e. 0.01. You can also try turning off the pressure initialization. The AMG solver option will help propagate a solution faster as well - which may or may not be helpful.

A reasonably large subset of problems that have the run characteristics you describe are badly posed or, if not, quite close to badly posed and thus require help in getting started to find the correct solution.

Thank you very much. I also found the problem and that the wrong scale factor was applied. After correcting the scale factor, every thing is O.K.

Thank you very much!