high pressure ratio
 

hello, I have a problem with centrifugal compressor boundary conditions setting. I have to check 4 points on compressor characteristics and I have already set boundary conditions for 3 points and I got reasonable results from all of them. But the same settings for point with highest pressure ratio and mass flow doesn't provide results with the same deviation:(. Please, does anybody have an advice??:o

Have you done mesh, time step and convergence sensitivity checks?

No I haven't. Pls could you write me down, where could I find it?? I would be very grateful to you.

All conditions you mentioned are set well for other three cases but the last one doesn't work. And I don't know if it is due to high pressure ratio or anything else. Relative pressure on inlet should be 0 Pa but it's about 90 kPa... That's why all other check variables are out of deviation..

The concept of a sensitivity check is simple. Vary the parameter (convergence tolerance, mesh size, time step size) and see if it gives you the same answer to within a tolerance you are happy with. Make sure you make increments between parameters big enough to be effective - I double or halve the time step size or mesh element length (not the node count!) and change the convergence tolerance by a factor of 10.

The Journal of Fluids Engineering has an editorial policy which goes into more detail:
http://journaltool.asme.org/Template...umAccuracy.pdf

And for the key textbook in this field go for "Computational Fluid Dynamics" by Roache.

And yes, it is quite possible that a simulation set up which works well for one situation does not work well for another. The flow may have entered a new regime and the numerics have to be re-tuned to accurately capture that.

Thank you very much. That is exactly my problem, I have to find new set up which works well for different flow regime. I will keep on trying:)

Could be problem in interface?? First I set it up on diameter 281mm which is 1mm from impeller. Do you think that interface on maybe 301mm could solve this problem??

Why don't you try it and find out for yourself?

I have large mesh and it takes a lot of time to get results. That is why I asked you this question. I thought you have some experience with this type of case... Anyway thx

You have not explained your case in much detail so cannot answer. But in all honesty, working it out for yourself is to best way. That way you know from experience and that is always better.