[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
Hi, If you want to get them into gmsh you could probably write your own converter. The gmsh file format is pretty simple and for a lot of problems i have written to it directly rather than using the GUI. The problem here is due to the nature of what you are trying to do which is FEA you need to define your problem in terms of surfaces and volumes where as most mesh generation tools are purely verts and surfaces. You could export 3d models as collada then process the collada models to generate the required information for gmsh. But you may need to get in to the hideous world of mesh deconvolution to simplify meshes before injecting into FEA. Although with some care when generating the model you could stop this being a problem. Although there are nice decomp libs available, see http://codesuppository.blogspot.co.uk/2011/05/hacd-hierarchical-approximate-convex.html for that kind of work. Collada provides a list of verts and triangles on a per object basis, so you could convert those triangles in to surfaces and then map the total volume per object. I just fear if you converted every GL tris into a FEA surface it may get computationally expensive but its very model specific anyway so YMMV and i may just be wrong about this as boat hulls are very non linear shapes anyway so you probably do have to have lots of surfaces to map accurately. But that is where decomp will be very useful, run a high level of decomp initially to set things up, then turn that down as you refine the model and want more precision. If you have not taken gmsh apart yet, all the file format is is a list of indexed verts, a list of indexed lines that list two verts by id, an indexed list of surfaces that list multiple lines by id (clockwise) then holes anticlockwise, then a indexed list of volumes that references surfaces. Regards Robin On 22 October 2012 07:41, tom <tompotts@xxxxxxxxxxxxxxxxxxxxxx> wrote: > Currently I need to get models into gmsh and or Elmer. If I can get them > into gmsh I think I can convert them for Elmer. > As for what I want to achieve - anything! I've put my name down for a 64core > 90Gflops > http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone?ref=home_spotlight > and in a couple of years its conceivable we will have teraflop processing. > Main aim at the moment are stresses on boat hulls and near supersonic > airflows in pulse/ram jets. > Theres everything needed out there but it just doesnt join together that > I've found.... > Tom te tom te tom > > > On 21/10/12 21:40, Robin Cornelius wrote: >> >> What are you trying to achieve, what do you need to import the model into? >> >> Some possible convertors are meshlab, assimp viewer and autodesks fbx >> convertor, but i fear these may all be mesh based formats and probably >> not what you are after? >> >> On 21 October 2012 18:00, tom <tompotts@xxxxxxxxxxxxxxxxxxxxxx> wrote: >>> >>> I'm trying to get some work done in Elmer or gmsh but find the model >>> building a bit slow and error prone. >>> There's wings3d or blender which are a bit quicker at building models but >>> cant convert to engineering grade formats. >>> Has anyone experience of weapons grade work of this sort and can point me >>> to >>> some converters or something? >>> Tom >>> >>> -- >>> The Mailing List for the Devon & Cornwall LUG >>> http://mailman.dclug.org.uk/listinfo/list >>> FAQ: http://www.dcglug.org.uk/listfaq > > > > -- > The Mailing List for the Devon & Cornwall LUG > http://mailman.dclug.org.uk/listinfo/list > FAQ: http://www.dcglug.org.uk/listfaq -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq