|February 9, 2007, 20:50||
an alternative to quadtree cartesian grids?
In cartesian finite volume methods every control volume normally has one adjacent volume on the east, west, north and south side. This leads to equations like:
Ap*Cp+ As*Cs + An*Cn + AW*Cw + Ae*Ce=Qc
for every cell.
In case we would like to refine the grid on certain locations it would be handy to have two neighbour cells on one side. A possible solution is using the cartesian quadtree grids where we have grids on different levels.
Is it possible to use a different approach? We could for instance use the two neighbours directly in the balance equation
This would lead to equations like:
Ap*Cp + As1*Cs1+As2*Cs2 + An*Cn + AW*Cw + Ae*Ce=Qc
if we have two cells on the south side and one cell on the other sides. Is such an approach possible? Or will this lead to numerical problems for instance with a pressure correction method?
Since I have never seen such an approach there must be a problem which is not yet obviously to me.
|Thread||Thread Starter||Forum||Replies||Last Post|
|Time step at coarse grids||D3sT||Main CFD Forum||0||December 6, 2010 12:15|
|structured and unstructured grids||user||Main CFD Forum||6||November 25, 2010 02:14|
|How to partition cartesian grid for MPI in fortran||CF||Main CFD Forum||3||January 17, 2008 05:38|
|A question about dynamic overset grids||JXIA||Main CFD Forum||4||January 10, 2008 04:50|
|Can I define grids motion as a time functions||Darcy||CFX||4||March 29, 2001 21:14|