Page 1 of 1
A question about saturne3.0
Posted: Tue Sep 16, 2014 3:30 am
by yhuang
Hi, everyone
I don't know how it happen that the number of iterative and residual haven't reach the set value when I use saturne3.0. The set value of iterative and residual are 10000 and E-8 respectively.
And everthing is OK when I use saturne2.0 under the same mesh file.
can you tell me some special places when use saturne3.0
PS I haven't use GUI
Thank you very much!
Yuan huang
Re: A question about saturne3.0
Posted: Tue Sep 16, 2014 4:39 pm
by Yvan Fournier
Hello,
The values displayed have different meanings in 2.0 and 3.0: in 2.0, the residue is that of the last linear system resolution. In 3.0, it is a more global (and more proper) residual. I'm not sure about its exact definition, but I believe the change is mentioned as one minor item (among many) in the Code_Saturne 3.0 presentation from the 2013 user meeting...
In any case, to improve the residual in 3.0, adding "sweeps" (under the GUI, Numerical Parameters/Equation parameters/Scheme/RHS Sweep reconstruction) may be useful.
Regards,
Yvan
Re: A question about saturne3.0
Posted: Wed Sep 17, 2014 1:14 am
by yhuang
Hi, Yvan
Thanks very much, but how to do this without GUI?
In my case, the value of pressure and vilocity are up to E+20 after some iterative .Obviously,ti is wrong.But all of this haven't appear in saturne2.0.
The attachment are some information of listing in saturne3.0
Yuan Huang
Re: A question about saturne3.0
Posted: Wed Sep 17, 2014 1:42 am
by Yvan Fournier
Hello,
Search for "sweep" in the user documentation pdf (I think the option is nswrsm, which you can define in usipsu, but I'll let you check).
Also, could you please post your "listing" files as text files rather than as screenshots ? The screenshots are not very readable, and are less complete anyways (add a ".txt" or ".log" extension if the site does not allow uploads otherwise)
Regards,
Yvan
Re: A question about saturne3.0
Posted: Wed Sep 17, 2014 4:16 pm
by yhuang
Hi Yvan
Thanks very much,this is listing of 2.0.6 and 3.1.4
yhuang
Re: A question about saturne3.0
Posted: Thu Sep 18, 2014 10:52 am
by Yvan Fournier
Hello,
You relaxation factors are not quite the same, and gradient reconstruction options may change.
As the gradient converges badly on 2.0, I assume your mesh is not too pretty.
I am not sure about other aspects, but there may be bugs in 3.1, which is not maintained anymore.
For a "safe" comparison, I recommend using 3.0 (stable) instead of 3.1. If you want the latest features, 3.2 and 3.3 will be maintained until 4.0 is released early 2015 (4.0 beta will be branched in october, so early testing will be possible).
Regards,
Yvan