[yinyuehao]

Hallo everyone:

As i write the code, there are some confuses about the assemble the coefficient A_p in momentum equation.

1. As i use FVM, after applying Gauss' theorem， the boundary of my Control Volume coincide with computational domain boundary. There are some integration point on that boundary part. Should I assemble the coefficient from these integration point? because at this boundary, for example inlet, i will introduce the given velocity. so A_p and A_nb anyway will be covered (for example A_p = 1, A_nb = 0). but A_p will be used in Poinsson equation of pressure. What should i do?

2. The same problem also in Pressure poinsson equation. After Gauss' theorem, I had integration point at boundary of computational domain. if I take dp/dx = 0, seem int{dp/dx dS} = 0. and I do not need do anything? is it right? I think it over and over, it seems i implicit apply some boundary condition when i assemble coefficient. should i do like this?

3. No matter assemble or not, in pressure poinsson equation, should i use original A_p or A_p after introduce boundary like A_p = 1, A_nb = 0?

Thank you very much!
Yin
28.11.2011