sci.physics.computational.fluid-dynamics

mathieu.dutour@gmail.com wrote:

> I am using the finite volume model SUNTANS.
> It is an oceanographic model with a vertical stratification, and the
> problem is how to
> obtain a 2D triangulation from the coastline.
>
> The Temperature, Salinity variable are stored at the vertices of the
> triangulation.
> The u, v velocity vectors are stored at the center of the Delaunay
> circles of
> the triangle.
>
> In general I am searching for books, reference explaining finite
> volume specifically,
> about how such schemes are build. The books I know are kind of evasive
> about
> the finite volume setting, about the grid generation and the
> implementation.
>
> In specific, here is the questions I have:
> ---What should be the constraints satisfied by the triangulation?
> -1- I heard about the angle being at most 90 degrees and in fact 85
> deg
> being preferable.
> -2- One logical constraint would be that the edges should be
> relatively long, i.e.
> if two Delaunay centers are near then I would believe that
> this cause problem
> -3- The minimal angle being small. That is a condition that shows
> up.
> It seems to me that condition 1 is chosen so as to imply condition
> 2. But I may
> be wrong. What would the origin of such conditions? Did I missed
> some conditions?
> ---What methods can be used to generate triangulations.
> -1- the method of adding new vertices at centers of empty circles
> can ensure
> the condition of small angle to be satisfied. This is the
> algorithm used by
> the program triangle.
> -2- there was some work done by Marshall Bern, David Eppstein on
> generation
> of triangulation with maximal angle by circle packings. But
> this does not
> seem to be implemented, or possible to implement, in some
> existing generation
> software.
> In a nutshell:
> -1- What are the constraints needed by a finite volume 2D grid?
> -2- What are the possible algorithms for generating with those
> constraints?
> -3- What existing software, preferably free, implementing such
> algorithms.
>
> Thank you for reading me some far. If you have any help, please send
> me info
> here or in the group. Really, I lack solid references on which to rely
> and I am
> open to any suggestions.
>
> Mathieu
Google is your friend - search for fe mesh generators. For particular
answers to your questions above, see:
1) Resolution issues in numerical models of oceanic and coastal circulation.
Greenberg et al, Continental Shelf Research, 27(2007) 1312-1343.
2) A MATLAB mesh generator, which is free and fairly good, is BatTri.
3) Another meshing package is SMS - comercial and hence $$.

--
G. Seibert
sos eastlinkca