DIverging on 1st iteration
 

Hi All!

What does it mean that my model is diverging in first iteration? all the mesh checks and boundary checks were ok (at least those several ones necessary to run the calculation), the geom. file is created, but... it doesn't want to run - just diverges on the first step (was decreasing the time step to 10e-06)...

Please tell me what information I should add to make my problem more comprehensible for you (in general I study the natural convection around solid in max temperature 323K).

Could it be caused by some cells of veeeery bad quality (I know that I have them)?

Thank you!!!

Re: DIverging on 1st iteration
 

Make sure you have no unused cells(created twice), Also i am not sure why u would like to use such a small time step. I have solved a complex 2-D problem earlier and a time step of around 0.0005 used to work really well. You can also try playing around(seeing how it affects the residuals) with the relaxation factors, particularly pressure.

Re: DIverging on 1st iteration
 

HI,

Yes I know this time step was ridiculously small... it was just to 'play around' ;-)

What is astonishing is that the model explodes already on first iteration no matter what is changed in the settings....

DOuble cells should have been reported during checks, right? The checks were ok....

Re: DIverging on 1st iteration
 

Have you checked the info file? no warrnings? is it a mass conservation which is running sky up?

Try some simple physics. Like laminar flow with incompressible fluid, isothermal.....

so you're sure, that there is no probelm with the geometry, but you have to inspect the physics and numerics.

good luck, matej

Re: DIverging on 1st iteration
 

Make sure you have no collapsed faces. These are not checked during geomwrite!

check all,,facecollapse news

Good luck

Oliver

Re: DIverging on 1st iteration
 

To add to what Oliver said, these checks are also important,

check,cset,,areaface,value,.001,news If you get too many change the tolerance of .001 to something else,

check,cset,,aspect,500,news This check will also tell you if a face is collasped even when it passes the face check.

Also it is good to check for negative volume with a tolerance, check,cset,,negvol,.001,news

How you initialize the domain could also be a factor, but usually it is some bad cells.

Kevin

Re: DIverging on 1st iteration
 

Thanks, thanks, thanks!!!

Exactly: the tolerances were increased, so I was allowed to create the .geom file, but there were some bad couples when the check was performend with the default tolerances. NOw it seems to be all right (after some local corrections of the vertexes)...

a.