XANSYS Message: 20637 [Go back to message list]
[bookmark on del.icio.us]
Average Rating: 5 (1 votes)
Subject: Re: Unsymmetric CG error
Author: Anh Truong anh@
Date: 2001-01-30 13:56:00Hi Mark,
You are right about the error being associated with the PGMR solver. I
checked my .dbg file and found that there is a failure in solving the energy
equations with the PGMR solver. I didn't specify the inlet temperature for
the flow of air and was trying to solve for the heat distribution in the
solid model, so that probably was the main reason for the instability (it
could have infinite solutions). I fixed several other potential problems as
suggested by Mark F. and Brad, and I'm running my problem now adiabatically
to get the flow solution, then introduce the energy equation. I think that
this method is faster and makes it easier to see where things go wrong.
Thanks for all your help,
Anh
----- Original Message -----
To:
Sent: Tuesday, January 30, 2001 11:43 AM
> Hi Anh, Mark, & Brad,
> After digging in the code, I've determined this error message is
associated
> with FLOTRAN's PGMR solver, and not the PCG or PCCR solvers.
> Normally, I only use PGMR for solving the TEMP DOF matrix, although it is
an
> available option for the PRES DOF matrix as well.
> Mark Troscinski
> > -----Original Message-----
> > From: Brad Lamirand [SMTP:blamirand@t...]
> > Sent: Monday, January 29, 2001 2:21 PM
> > To: xansys
> > Subject: Re: [xansys] Unsymmetric CG error
> > Anh
> > This isn't actually a unsymmetric CG error. What is happening is that
> > your
> > PreConditioned Conjugant Gradient solver (which is the default when
> > solving
> > incompressible problems) could not reach a converged solution in three
> > solver iterations. Although it is ussually the pressure DOF that is
> > causing
> > this, you should look into your .pfl file to see which DOF(s) are
> > diverging.
> > To fix this, although it could be a myraid of things, you should check
to
> > make sure that your boundry conditions are correct and that you don't
have
> > any unconstrained inlet/outlets.
> > Another thing is if you truely have an ill-conditioned problem is to
> > change
> > the PCCG solver over to the PGMR solver which is more stable and try
> > again.
> > Hope this helps,
> > Brad Lamirand
> > Cooper Turbocompressor
> > ----- Original Message -----
> > From: "Anh Truong"
> > To:
> > Sent: Monday, January 29, 2001 1:40 PM
> > Subject: [xansys] Unsymmetric CG error
> > Hello all,
> > I got this error message when running FLOTRAN:
> > ****************ERROR****
> > Unsymmetric CG iterations did not converge three times. The run is
> > terminated.
> > Does anyone know what an unsymmetric CG error is?
> > Thanks muchly,
> > Anh