Bugs item #1474430, was opened at 2006-04-21 16:12
Message generated for change (Comment added) made by wd15
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=681141&aid=1474430&group_id=118428
Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Daniel Wheeler (wd15)
Assigned to: Nobody/Anonymous (nobody)
Summary: Convection - Diffusion sign problems + LinearCGSSolver
Initial Comment:
The ConvectionTerm has problems getting the sign right depending
on which side the terms appear on. The attached code does not work. It
only works if:
>>> eq = diffTerm + PowerLawConvectionTerm(coeff = convCoeff,
diffusionTerm = diffTerm)
is changed to:
>>> eq = TransientTerm(1e-100) == diffTerm +
PowerLawConvectionTerm(coeff = convCoeff, diffusionTerm = diffTerm)
This is due to an underlying assumption when calulating the Peclet
number about which side the DiffusionTerm is on. Not sure how to fix this
as it stands.
There is also a sperate problem that LinearLUSolver is needed to get the
right answer. This is unrelated to the above problem, must be a problem
with LinearCGSSolver
----------------------------------------------------------------------
>Comment By: Daniel Wheeler (wd15)
Date: 2006-04-21 16:43
Message:
Logged In: YES
user_id=1116094
There is actuallt know problem with ConvectionTerm and DiffusionTer, and the
side of the equation on which they appear. That was a complete fallacy. There
is,
however, still a problem with LinearCGSSolver that needs to be addressed. The
attached code should work correctly when only the solver is changed to
LinearLUSolver.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=681141&aid=1474430&group_id=118428
