|April 21, 2012, 13:00||
Grid Reynolds is above 2 but Central Linear convection scheme works !
Join Date: Jan 2010
Blog Entries: 2Rep Power: 8
As you know in linear central scheme of convection term the grid Reynolds (=U*dx/nu) number should be less than 2. If it would be more than 2, the results should be diverged or wrong because of minus in coefficient.
I have simulated a 1*1 m2 lid-driven cavity with the Reynolds number of 10000 (Re=U*1/nu) and 200*200 grids which leads to grid Reynolds in range of 50. it is so far from 2 but results are in good agreement with Ghia (1982) benchmark.
Why central scheme works in this high grid Reynolds number?
Has it any artificial viscosity been added in OpenFOAM?
|April 23, 2012, 11:32||
Join Date: Feb 2010
Posts: 165Rep Power: 9
the restriction of Re_cell <= 2 when using the central differencing scheme (CDS) is a sufficient, not a neccessary stability-criterion! (refer to e.g. Ferziger/Peric if you want to know more)
In regions where the solution is more or less smooth (e.g. streamwise direction in channel flow or your lid driven cavity case), the cell reynolds number can be pretty high without any oscillations to appera. It's not neccessary to introduce numerical diffusion to achieve an acceptable stability behavior using CDS.
|Thread||Thread Starter||Forum||Replies||Last Post|
|AUSM scheme ? Central Scheme||boling||Main CFD Forum||5||September 9, 2009 19:06|
|Compact scheme for nonuniform grid?||Dong||Main CFD Forum||3||July 1, 2008 23:44|
|Upwing vs Central scheme||Ruben||Main CFD Forum||15||August 8, 2005 06:13|
|Best scheme for 2d convection||Ambady Suresh||Main CFD Forum||1||June 14, 2002 12:10|
|High-order scheme and grid||z.zeng||Main CFD Forum||6||October 14, 2000 13:10|