Hello,
my project involves laminar steady incompressible fluid flow in cylinders of (slightly) non-circular cross section. I would like to have the developed flow profile along the entire length. Is there any good way to do this? I suppose starting with plug flow at the inlet, then taking the outlet flow profile and setting it as inlet BC and repeating the simulation could work, but I'm not sure how to do that. Would I use SaveScalars? The mesh is structured (extruded gmsh), i.e. identical at boths ends. The walls are no-slip and the desired flow is slightly distorted parabolic.
I'd appreciate any tips.
Developed flow profile as entrance BC?
-
david.bragason
- Posts: 8
- Joined: 13 Jan 2013, 02:40
- Antispam: Yes
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Developed flow profile as entrance BC?
Hi David
If your problem is basically of infinite length then why not take just a short section and apply periodic BCs for the velocity components.
-Peter
If your problem is basically of infinite length then why not take just a short section and apply periodic BCs for the velocity components.
-Peter
-
david.bragason
- Posts: 8
- Joined: 13 Jan 2013, 02:40
- Antispam: Yes
Re: Developed flow profile as entrance BC?
Hi Peter,
thank you for the reply, and for all your work on Elmer!
I haven´t had much success using periodic BCs in this problem, convergence is poor, and when it converges (if the cylinder is just a few elements in thickness), the solution doesn´t look right, compared with what the Hagen-Poiseuille law predicts. I´m assuming that the only way to control the flow in this approach (periodic BCs) is to specify the pressure difference across the cylinder.
As the N-S eq. reduces to a 2D Poisson problem with constant source (= -pressure gradient / viscosity) for developed flow in a tube of arbitrary constant cross-section, wouldn´t the easiest way be to solve the Poisson eq. on the 2D inlet boundary, and extending that solution into the cylinder as a constant velocity field?
To do this, I need to use the cylinder boundary as a body, but I can´t find any hints anywhere on how to do that. If I import my mesh from gmsh into ElmerGUI, ElmerGUI tells me there are no "Surface Bodies". Are they something I need to make from within gmsh? ElmerGUI also has the option "Use as Body" in the BC dialog, but I can´t see any effect of that in the sif file. I would really appreciate a hint on this. Thank you!
thank you for the reply, and for all your work on Elmer!
I haven´t had much success using periodic BCs in this problem, convergence is poor, and when it converges (if the cylinder is just a few elements in thickness), the solution doesn´t look right, compared with what the Hagen-Poiseuille law predicts. I´m assuming that the only way to control the flow in this approach (periodic BCs) is to specify the pressure difference across the cylinder.
As the N-S eq. reduces to a 2D Poisson problem with constant source (= -pressure gradient / viscosity) for developed flow in a tube of arbitrary constant cross-section, wouldn´t the easiest way be to solve the Poisson eq. on the 2D inlet boundary, and extending that solution into the cylinder as a constant velocity field?
To do this, I need to use the cylinder boundary as a body, but I can´t find any hints anywhere on how to do that. If I import my mesh from gmsh into ElmerGUI, ElmerGUI tells me there are no "Surface Bodies". Are they something I need to make from within gmsh? ElmerGUI also has the option "Use as Body" in the BC dialog, but I can´t see any effect of that in the sif file. I would really appreciate a hint on this. Thank you!