Surface integral part of the viscous term

Hi,

This is a question and not an issue.

fluidity/assemble/Momentum_CG.F90

Line 1034 in b3d0955

! if no no_normal_flow

I see that the surface integral part of the advective term is added whenever the boundary condition allows normal flow. However, I don't see a section in the subroutine that takes care of the surface integral part of the viscous term.

PS: Is the fluidity mailing list still active?

[stephankramer]

The viscous surface integral is indeed missing. So when considering the momentum equation on its own it would indeed be applying the wrong boundary condition. I believe the argument, why this is allowed, is that this gets corrected in the pressure correction. I think you can find more details in the Gresho and Sani "Incompressible Flows and the Finite Element Method" books, there are various subtly different pressure correction methods, where the details are all in the treatment of boundary conditions. The reason the viscous integral is often left out, is that it's actually a little tricky to implement as it involves the integral of a normal gradient over a surface, where this gradient depends on degrees of freedom that are not on the surface, so you can't just restrict it to a lower dimensional integral.

The fluidity mailing list is still active, yes - response time may vary though, depending on the type of question.

thanks.