## [Libmesh-devel] Multiply Connected 1D Elements

 [Libmesh-devel] Multiply Connected 1D Elements From: Derek Gaston - 2013-08-08 21:58:16 Attachments: Message as HTML ```Ok - we've kinda talked about this before but now I really want to know what's involved... We would like to connect more than one element to the end of a 1D element. i.e we want a "branch" kind of structure. The motivation is 1D structural dynamics (like truss and beam elements). I know that find_neighbors() doesn't quite do the right thing right now - what other gotchas are involved with trying to do something like this? Thanks! Derek ```

 [Libmesh-devel] Multiply Connected 1D Elements From: Derek Gaston - 2013-08-08 21:58:16 Attachments: Message as HTML ```Ok - we've kinda talked about this before but now I really want to know what's involved... We would like to connect more than one element to the end of a 1D element. i.e we want a "branch" kind of structure. The motivation is 1D structural dynamics (like truss and beam elements). I know that find_neighbors() doesn't quite do the right thing right now - what other gotchas are involved with trying to do something like this? Thanks! Derek ```
 Re: [Libmesh-devel] Multiply Connected 1D Elements From: Kirk, Benjamin (JSC-EG311) - 2013-08-08 22:06:13 ```Yeah, important for shell elements too. Consider a plate welded perpendicularly to another plate. Could you model that as quads where you have an element coming out of the page? So an edge is connected to three elements? Find neighbors will fail, but as will the entire notion of one neighbor per side. Not insurmountable, but pretty well ingrained in a lot of places, I think. -Ben On Aug 8, 2013, at 4:58 PM, "Derek Gaston" wrote: > Ok - we've kinda talked about this before but now I really want to know what's involved... > > We would like to connect more than one element to the end of a 1D element. i.e we want a "branch" kind of structure. > > The motivation is 1D structural dynamics (like truss and beam elements). > > I know that find_neighbors() doesn't quite do the right thing right now - what other gotchas are involved with trying to do something like this? > > Thanks! > > Derek > ------------------------------------------------------------------------------ > Get 100% visibility into Java/.NET code with AppDynamics Lite! > It's a free troubleshooting tool designed for production. > Get down to code-level detail for bottlenecks, with <2% overhead. > Download for free and get started troubleshooting in minutes. > http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk > _______________________________________________ > Libmesh-devel mailing list > Libmesh-devel@... > https://lists.sourceforge.net/lists/listinfo/libmesh-devel ```
 Re: [Libmesh-devel] Multiply Connected 1D Elements From: Roy Stogner - 2013-08-08 22:29:44 ```On Thu, 8 Aug 2013, Kirk, Benjamin (JSC-EG311) wrote: > Yeah, important for shell elements too. Consider a plate welded > perpendicularly to another plate. Could you model that as quads > where you have an element coming out of the page? So an edge is > connected to three elements? > > Find neighbors will fail, but as will the entire notion of one > neighbor per side. Not insurmountable, but pretty well ingrained in > a lot of places, I think. Yeah; we're going to need to keep Elem::neighbor(n) for backwards compatibility, but we're going to want some kind of Elem::neighbors(n) too. I'm not even sure what the underlying data structure looks like. Round-robin pointers would work in 1D and on uniform parts of a 2D mesh, but what gets stored how at non-manifold hanging nodes in 2D? --- Roy ```