On 08/01/2014 03:45 PM, Geoff Wright wrote:
Hi Robert,
Thanks for the fast response! The poisson_neuman example is making sense to me now, but when I take a step closer to my real problem, I'm still having some issues with the flux. I attached the mesh, problem definition and a screenshot of the result.
The intention was to have a sphere inside another sphere, with the source defined as a region on the inner sphere, and the sink defined as a region on the outer sphere. All other surfaces are supposed to have zero flux. However, if you look at the flux results, computed similar to before, I get the following:
Source flux: 0.978 Sink flux: -1.592 Block flux: -1.592 Entire Surf: -1.608
I am getting:
Sink flux: -0.994294409461 Source flux: 0.977718087637 Block flux: -1.59176490161 Entire surf: -1.60834122343
Whereas I expected source and sink to be equal and opposite, block and entire surf to be zero.
so Source and Sink are pretty close and opposite, but the other fluxes are not zero. What happens if you refine the mesh so that the element size is uniform?
Btw. for field approx. order 2: (integral order 8):
Sink flux: -1.42052213723 Source flux: 1.60881848323 Block flux: 0.0581843449633 Entire surf: 0.24648069096
and approx. order 3:
Sink flux: -1.68822440062 Source flux: 1.86564560107 Block flux: -0.143661701834 Entire surf: 0.0337594986197
-> there is something fishy.
I tried to check the regions, it seems ok - make sure that there are no holes between sink + source and block - maybe check the surface areas using d_surface term?
Then I would check the topological consistency of the mesh etc. but I assume you made it by gmsh, right? So it should be ok.
That's all I can think of now. Let me know how it goes.
r.