GenBC

[Kimberly Stevens]

Hello,

This is related to a question I posed at the end of another thread (viewtopicPhpbb.php?f=188&t=10202&p=0&st ... 78be984ab8) so if someone is still working on replying to that one I'm not trying to rush things; I just didn't want that post to get lost, so I created a new one.

How exactly does the LPN solver and the fluid dynamics simulation (Navier-Stokes solver) interact at the boundary condition? From the tutorial, I thought at each time step the LPN solver obtained either pressure (for a Dirichlet bc) or flow (for a Neumann bc) from the 3D NS solver, then iterated 100 times to solve the system of differential equations specified in by the LPN. However, in the case I mentioned in the previous thread, I printed out P(1), the pressure obtained from the 3D solver for a Dirichlet bc, and it was changing on each of the 100 internal iterations, which confused me, since I thought it came directly from the solver. It seemed to me that no matter what was happening with the LPN solver, P(1) would not change in between time steps. Furthermore, I was explicitly specifying the flow on a dirichlet bc (I specified x_i, rather than f_i), so there should have been no need to iterate at all--each of the 100 iterations should have been the same since there was no differential equation to solve. Can you help me understand what exactly P(1) is and why it would change on the internal iterations of the LPN solver?

Thank you,
Kimberly

[Weiguang Yang]

By default, an RK4 time integration is used to integrate ODE from t to t+dt, for any time t_{0d} between t and t+dt, the 3D pressure "p" is interpolated by solutions P_n and P_{n+1}. Therefore, for each RK point, the corresponding 3D p is not exactly the same. You can find more details in another file GenBC.f which shows the RK procedure or in the following paper.
https://www.sciencedirect.com/science/a ... 9112004202