[overture] Re: Godunov methods on on overlapping grids
- From: <smallik2@xxxxxxxx>
- To: overture@xxxxxxxxxxxxx
- Date: Thu, 31 Jan 2008 00:18:11 -0600 (CST)
Hi Bill
If i want to implement a finite volume scheme do you think the following
approach is feasible.
1)Take the vertices of the grid as cell centers?
2)On each grid, at the regular pts(non interpolation points,non ghost pts),
solve Riemann problems for all 4 neighbours.
3)Let the interpolate routine do an explicit/ implicit interpolation to fix
values of interpolation points.
Regards
Srinath
---- Original message ----
>Date: Wed, 30 Jan 2008 12:31:55 -0800
>From: Bill Henshaw <henshaw@xxxxxxxx>
>Subject: [overture] Re: Godunov methods on on overlapping grids
>To: overture@xxxxxxxxxxxxx
>
>Hi Srinath,
>
> The Godunov method in cgcns uses a 5-pt stencil as you note.
>There are two "fringe" values supplied at an interpolation
>boundary. If the grid is not made with 2 layers of interpolation
>points, cgcns will fill in an extra layer of values by extrapolation.
>
>Regards,
> Bill
>
>smallik2@xxxxxxxx wrote:
>> Hi Bill
>>
>> This is a doubt regarding how cns is implemented on the overlapping grid
>> framework.
>> I would like to make an implicit version of cns, so i was reading your
>> article "An adaptive numerical scheme for high speed reactive flows".
>>
>> I don't understand how the intercell flux is computed in the overlap region.
>> For a Godunov scheme it would be necessary for an edge to be shared by only
>> 2 cells. But how is this ensured in the overlap region?
>>
>> Regards
>> Srinath
>>
>
Other related posts:
