search

thelfer / mfem-mgis

Couple MFEM and MGIS
GNU Lesser General Public License v3.0
11 stars 6 forks source link

Forces computations are wrong by a factor 2 using triangles #47

Closed thelfer closed 2 years ago

thelfer commented 2 years ago

Things are quite strange. The error is exactly a factor 2 using linear triangles but it completly depends on the order used. This is not the case with quadrangles which always give the correct result.

thelfer commented 2 years ago

The origin of the error lies down in the implementation of the computeResultantForceOnBoundary post-processing which only considered the contributions of the elements with one face in common with the considered boundary. For triangles, one should also consider the contributions of elements which has a node in common with the considered behaviour.