Hi, I am solving flow in a duct using primitive variable formulation ( not using vorticity stream function).
I am having trouble with specifying boundary conditions.
I know du/dx= 0, once flow is fully developed.
What should be my boundary condition for v and dp/dx and dp/dy.
The flow is in-compressible
At the walls velocity will velocity will be zero
At inlet you can specify the inlet velocity or the mass flow rate
In incompressible flow the pressure is calculated automatically.
Dear Mishra, at the walls you should set v=0 if you have no slip, at the inlet you should set v and dp/dx=dp/dy=0 and at the outlet you should impose the pressure and dv/dx=dv/dy=0.
Regards
Franco
As per the previous answers, you should have u = v = 0 on all solid and stationary boundaries. The velocity profile at any inlet should be whatever suits the problem you are modelling. For the pressure, you have different alternatives. The one I usually use is to linearly (or quadratically) extrapolate the pressure at the boundaries based on the pressure inside the computational domain. I'm assuming here that you are using some kind of pressure correction method each time you step forward in time.
Cheers,
Duncan.
Hello everyone,
Thanks for your answers.
The way I see it , since for incompressible flow, pressure is not required but gradient of pressure is, I dont have any boundary condition with respect to pressure.
I have no slip boundary condition on the walls,
U ( x direction) velocity at the inlet .
I am evaluating dp/dx and dp/dy using a new technique but if i impose dp/dx=0 at the outlet, I am not able to think of anything for dp/dy.
I am getting a parabolic velocity profile generated at the outlet and very small vertical velocity but i dont have correct results for pressure field.
Also, I read patankar book and its mentioned that one needs to have outflow far away and then does not need any boundary condition at the outlet.
But is not that contradictory to the physics of flow?
I have some notes that address how to handle the outflow boundary in 2-D channel flows. The notes deal with implementing the Finite Element Method in Mathematica, but the issues are general to other numerical methods. You can download the relevant PDF or Mathematica notebook at https://sites.google.com/site/ekayasolutionsdiffeqns/ then click on the Finite Element link
@Brian G Higgins - Hi,
I looked over your method and thanks a lot, it solved my problem.
The file which you have solves Stokes flow.
Do you have a solution for viscous in compressible fluid in a channel where convective terms are not neglected.
- Badges
- Science method