everybody:

now i have meet a big trouble about divergence appeared in fluent.

I am calculating a 3Dflow field of a compressor.but divergence occured to me ofen.

I use the velovity-inley and pressor-outlet boundary .and i can make sure that this B.C. is right because I have finished a similar calculation about another design of compressor. But this time,divergence puzzles me . How can i get rid of the divergence problem ? I hope your advice!!!!

jieremy.

Hi!

There´s a lot of things that you can try. For example:

-decrease underrelaxation factors (don´t use too small underrelaxation!!!)

- approach your operating conditions in a few steps (if you have inlet velocity 10m/s, try to get converged solution for 2m/s then 4...until you reach 10)

-you can also try to make a new mesh

-be shure to search the forum for other ideas...

best regards

MATEUS

-Are you using Boundary Layers?

-Is it possible you have some negative volumes in your mesh?

-Have you checked your mesh before starting calculation?

Hope this could help,

Best regards,

Jkl.

thanks jkl for your help! But I have not used boundary Layers. actually ,i have finished those steps: checking the mesh,it is ok! selecting segregate solver; selecting k-epsilon model; and solution with second order upwind and small under-relaxation factors; at last defining the corresponding limits.

but those do not work well. what can i do with this problem???

thanks MATEUS! I'll try it!

jieremy

You could try to start your calculation using First order upwind scheme and pass to the Second order scheme later.

Hi

ap

jieremy,

also make sure you use a pressure reconstruction which is consistent with the reconstruction used for the calculation of convective fluxes. I. e., if you use second order upwind, use "second order" for pressure solver.

Greetings, Andrea

thakns ap. I have tried it .But it dosen't work. divergence remains . thanks everybody!!! now I think I have found a "good" way to eliminate divergence . Last night I used the couple solber and made the courant number low , I also seleced "quick "scheme . As result , divergence disappears . But now,I find that convergence dosen't occur, so I will set the courant number larger.

thanks mateus,thanks JKL, thans ap , thanks Andrea Panizza .thanks for your help.

jieremy

thanks Andrea Panizza ,I have done it!