Hi there

I'm stuck with the use of proper boundary conditions for a channel flow problem where i use SIMPLE algorithm to determine the fluid flow. The flow is incompressible. I have a specified velocity inlet on the left and i impose a fully developed flow on the right. The top and bottom walls have no-slip velocity bcs. When i try to solve this using the SIMPLE alorithm, how should the bcs for the p' equation be given? Should the p' equation (that is a poisson equation of p')be seen as a pure boundary value problem in which case i need o set p' on all the sides or be treated as mixed type problem wherein i can replace the p' bcs in some of the edges by the pressure bc itself? I'm also trying the version that patankar has mentioned in his book where the transport coefficient would become zero on the boundary where velocity is specified. That seems not to work beacuse i'm not sure of the bcs on the other sides especially when you deal with p' equation

Please let me know what i'm not considering or may be i have a bug in implementing the p' on the boundaries.

Thanks Bharath Somayaji

Hi!

In the pressure correction equation the massfluxes stand on the right hand side. So you have to add the inlet massflux to the right hand side as a first boundary condition. The same can you do on the outlet side. The massflux there is not known so you can extrapolate velocities from the boundary nearset cells to outlet and calculate massfluxes from them! At the walls you know that the normal velocity to the walls is zero (no massfluxes through walls!). From this follows for example for a north wall that the AN-coefficients from pressure correction are arbitrary so you have not to calculate the correspondend coefficient at walls. (Be careful: Through this you have now Neuman Boundary Conditions on all sides. Direct solvers can not solve this singular system!!)

With this conditions it should work, good luck!