|
[Sponsors] |
January 6, 2015, 06:41 |
Higher order implemneted in SU2
|
#1 |
New Member
Shainath
Join Date: Apr 2011
Posts: 18
Rep Power: 15 |
Hi,
I have worked in cell centred finite volume methods earlier, now I want to try SU2 for my research work. I want to implement a new higher order reconstruction method in SU2, for that I want to clear my doubts about node-centred finite volume scheme implemented in SU2. My doubts are as follows. (Please refer this file for figure) Case 1 -Here, primal grid is made up of equilateral triangles (which you show in all your PPTs and papers) -In this case the line joining cell centres will pass through the edge centres of the edge(of primal grid).. -While reconstructing the variables for higher order accuracy, SU2 reconstruct the variable at the edge centre(of primal grid). -Flux can be obtain at the edge centre easily because it is also a mid point of the edge joining cell centres of primal grid. -Finally, flux calculated at the every edge centre can be added to get the total flux at each node. Case 2 -Here, primal grid is made up of different shape of triangles. -In this case the line joining cell centres will not pass through the edge centres of the edge(of primal grid).. -While reconstructing the variables for higher order accuracy, SU2 reconstruct the variable at the edge centre(of primal grid). -Now there are two separate edges joining edge centre and cell centre of primal grid. -How do you calculate the flux in this situation where reconstructed value is available at nodes of dual grid? Please clarify my doubt. |
|
January 7, 2015, 11:07 |
|
#2 |
Super Moderator
Francisco Palacios
Join Date: Jan 2013
Location: Long Beach, CA
Posts: 404
Rep Power: 15 |
Hi Shainath,
I see what you mean... From the geometrical point of view we compute the dual grid as in case 2. However, after computing both segments touching the edge that connects two points we add them and we never store the normal of all the segment (facets in 3D) that creates the surface associated with the edge. At the end, as you know, SU2 only stores one normal vector (defining the control surface) at each edge. The key subroutine is CPhysicalGeometry::SetControlVolume(CConfig *config, unsigned short action) in geometry_structure.cpp Best, Francisco Palacios SU2 lead developer |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
First Order to Higher Order Blending Factor | NormalVector | FLUENT | 4 | November 13, 2023 08:06 |
2nd order boundary conditions for 2nd order discretization? | quarkz | Main CFD Forum | 30 | December 26, 2011 08:12 |
Solve single but higher order equation by OF 1.6 suffering Problem | alundilong | OpenFOAM Programming & Development | 0 | December 23, 2010 14:53 |
Volume 10 has higher order elements | Duncan | FLUENT | 0 | November 2, 2005 14:52 |
Higher order FVM | Sergey Smirnov | Main CFD Forum | 10 | April 15, 2000 02:49 |