"Wrong" normal
Closed this issue · 2 comments
santiagobadia commented
Hi @oriolcg
I think that it is far better to always put the normal before the stress tensor, i.e., n * sigma instead of sigma * n. The reason is that it comes from the integration by parts formula and Gauss theorem, and for non-symmetric second order tensors, the result is different. For instance, if you use a vector Laplacian instead of a Laplacian and you have written your fluxes as grad(u) * n, if you move to vectors, this is an error.
I know that for your fluid, solid, and FSI problems it is not an issue, but I think it is good practise. Could you change the tutorials accordignly? Both the one already in the main branch and the one to the merge?
Thanks!
oriolcg commented
Sure! I can take care of that.