CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Boundary priority (http://www.cfd-online.com/Forums/cfx/75840-boundary-priority.html)

Felipe Matos May 6, 2010 14:28

Boundary priority
 
Hi

I'm simulating a simple heat transfer problem: a flat square plate with prescibred Temperature in two oposite faces and adiabatic at the others faces. My doubt is: wich one of these boundaries (Prescribed temperature or Adiabatic) is being used at the interfaces nodes (the nodes that belong to two different boundaries faces)? How can I see it? Can I define a stronger boundary so that I know wich one I'm using?

Thanks

ghorrocks May 6, 2010 21:20

Quote:

Can I define a stronger boundary so that I know wich one I'm using?
I do not know what you mean.

For the control volume which includes both the adiabatic boundary and the fixed temeprature boundary, the adiabatic boundary will apply to that face and the prescribed temperature will apply to that face.

The wall boundaries are not evaluated at the nodes to avoid the precise issue you discuss. Instead the wall boundaries are evaluated at the element faces and therefore the state of all boundaries is clear.

Felipe Matos May 11, 2010 17:42

5 Attachment(s)
Quote:

Originally Posted by ghorrocks (Post 257866)
The wall boundaries are not evaluated at the nodes to avoid the precise issue you discuss. Instead the wall boundaries are evaluated at the element faces and therefore the state of all boundaries is clear.

I'm with a problem that (I think) the boundaries are evalueted at the nodes. The problem is:
- Incompressible, laminar (Re = 100) flow into a pipe
- Prescribed constant velocity at the inlet face (v = 5 cm/s)
- Prescribed pressure at the outlet face (P = 1 atm)
- No slip wall

The first node, at the interface of the inlet and the wall boundary, has the inlet velocity ( v = 5 cm/s ), but the next node in the direction of the flow (a node over the wall) has velocity near zero ( v ~ 10^-14 cm/s): both hybrid values. This is a situation where the inlet boundary is stonger than the wall boundary.

The high velocity at the inlet (compared to the wall velocities) os causing a peak of pressure at the wall near the inlet, i.e., a strong pressure drop at wall in the beginning of the pipe.

Isn't this a case where the boundary is store at the node and not the element faces?

ps.: There are some images that can help understand. They're at the attachments.

Thanks

ghorrocks May 11, 2010 18:36

Again you are incorrect. Read the section on discretisation and solution theory - variables are stored at the nodes and control volumes are built around the nodes. The fluxes for the control volumes are evaluated at the control volume faces (at the integration points), and a boundary can form a control volume face. This means that boundaries are evaluated at the control volume faces, not the nodes. This means there is no strange behaviour at the intersection of two boundary conditions like you suggest.

Felipe Matos May 11, 2010 19:10

What could be the pressure peak problem at the entrance?

ghorrocks May 12, 2010 08:41

Isn't that caused by the assumption of constant velocity over the inlet boundary? It would be eliminated by applying an inlet boundary with something closer to a boundary layer profile.


All times are GMT -4. The time now is 18:13.