In calculating the converging-diverging supersonic nozzle with the FINE/Turbo, the NEGATIVE pressure or density in domain comes so often during iterations finally ending up with a crash. Total pressure and temperature with flow angles were given at the inlet boundary, and static pressure was given in the outlet boundary when subsonic. The CFL number has been lowered up to 0.01 but in no use. Anyone with a better idea, please ?

Are you sure of your boundary conditions, i.e. that the static pressure at outlet is lower than the total pressure at inlet ?

Of course. 7 bar pressure at inlet and atmosphere at exit are applied to the impulse turbine stage.

hfhdfd

If you are using the fine turbo upwind scheme this isn't surprising. There are major errors in the implementation.

hi

Just check the operating pressure what you given . if u specified Pr. there just make it zero. while definig inlet Pr. make gauge pr. as 101325 and defaine ur inlet pr. and the outlet pr. must be less than inlet pr.

[Gone]

I have the same problem, I changed turbulence model and cell width, but it didn't work

[colinda1]

Could you please provide more information on your project set-up?
Even if you have also a negative pressure, the cause may be completely different. Launch the simulation before the divergence and open it in CFView™. You may see through iso-surfaces where is the region of unphysical pressure and/ or temperature. Is the flow direction at inlet and outlet correct, the turbulence field looks correct, the initial solution is appropriate... etc.

[Hamidzoka]

Hi;
could you please provide more information about your case e.g. :
- Does the run diverege at early iterations or at further iterations?
- What about boundary conditions?
- Meshing strategy of the domain? Mesh distribution?
- Selected urbulence model and near wall mesh settings?

Check if your case convereges at subsonic conditions.
Note that if the CFL goes down below 1 while the case still diverges, further reduction of CFL does not help. So forget about CFL of 0.01.
Attaching some pictures of your domain and solution results prior to divergence will be quite helpful in finding the proper solution.

Regards

[Gone]

I have solved the problem in my project. I changed the current grid level to 2 2 2 which was 0 0 0 at first, and computation with grid level 2 2 2 converged quickly. At last, I took the result from 2 2 2 grid level as initial solution, finally the computation could go on without negative pressure or density.