CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Gradient Discretization CFX vs Fluent (http://www.cfd-online.com/Forums/fluent/70693-gradient-discretization-cfx-vs-fluent.html)

Scott Nordsen December 2, 2009 13:02

Gradient Discretization CFX vs Fluent
 
I also posted this on CFX Forum but perhaps there is a Fluent expert who has looked into this

I am a traditional Fluent user trying to compare CFX runs to Fluent runs.

A few versions ago Fluent added an option under discretization for gradients

1) Green-Gauss Node based
2) Green-Gauss Cell based
3) Least Squares Cell based

I am having trouble finding help on 1) does CFX have similar options 2) what is CFXs default

If a simple answer exists (i.e. CFX only does #2 from the list above) that would be great.

Thanks in advance

sbaffini December 3, 2009 19:50

I don't have experience with CFX or finite element methods but in Fluent those three methods are used for the computation of the gradients in the cell centers. These, in turn, are used for the diffusive fluxes and for any convective scheme different from the 1st order upwind.

The method 1 and 2 are similar as both are based on an approximation of the gauss theorem. As this approximation requires the cell face values, the two methods differ by the way these face values are calculated. In the cell based method they are simply calculated as a simple average from the two cells adjacent to the face. In the node based method the face values are computed by a weighted average of the values in the nodes of the face. These in turn are calculated by an additional weighted average from the surrounding cells.

The node based method should be more accurate than the cell based one (it should be at least 1st order accurate on any grid)


All times are GMT -4. The time now is 10:22.