Inlet and outlet boudary without wall between

Janshi · February 1, 2012, 17:19

Hello,

I have following model:

A rectangular box with an Inlet and outlet. The sides as well as the top and the bottom are defined as wall.
The inlet as well as the outlet boundary conditions are defined via a velocity profile. The velocity profile is defined normal to the boundary.

Following changes are to be done:

The velocity profile definition is to be changed from a normal to the boundary definition to a definition via cartesian componentns. The velocity profile should be defined angular to the boundary.

I´ve consider following changes in the setup:
Can I change the definition for the sides into inlet and outlet boundaries?
In the attachment I´ve made a primitive sketch of my idea. I´m quite sure that there won´t be any recirculation flow, so could such a case calculated in cfx?
Actually I´m not sure, because I´ve never tried a calculation where no wall exists between the inlet and the outlet of the flow domain.

Thank you

ghorrocks · February 1, 2012, 18:30

It is not a good idea to have an inlet and outlet so close. This looks numerically unstable to me. I would move your boundaries further back to where they are separated and simpler.

Janshi · February 1, 2012, 19:33

Hello glenn,

First, thank you for your fast answer.
I already assumed that there would be problems. I´ve already given it some tries but every calculation blow away after some iterations depending on the timestep I´ve chosen. Usually my Mach number grows ridiculous high (beyond 10) and the the calculation blow away with a fatal error that the linear solver diverged.

I assume that if I define the sides as openings, nothing will change at the problem with numerical stability?

Best regards
Shichun

ghorrocks · February 1, 2012, 19:51

No, that will not improve numerically instability.

You might be able to get it to work, first try a smaller time step.

Why are you doing it like this? What are you trying to do? Are you trying to put the central object under different angles of attack without having to remesh?

Janshi · February 2, 2012, 04:08

Quote:

Originally Posted by ghorrocks

Why are you doing it like this? What are you trying to do? Are you trying to put the central object under different angles of attack without having to remesh?

Yes that´s the idea. For special reasons I´ve not the possibilities to remesh and I have to leave the mesh untouched :|

ghorrocks · February 2, 2012, 05:51

Have you considered a C grid like this http://en.wikibooks.org/wiki/ICEM_CFD ?

They are frequently used in AOA sweeps. You make the circular, top and bottom surfaces inlets and the back face an outlet.

February 1, 2012, 18:30		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	It is not a good idea to have an inlet and outlet so close. This looks numerically unstable to me. I would move your boundaries further back to where they are separated and simpler. Janshi likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Water subcooled boiling	Attesz	CFX	7	January 5, 2013 04:32
ATTENTION! Reliability problems in CFX 5.7	Joseph	CFX	14	April 20, 2010 16:45
steam flow in a pipe driven by a pressure gradient between inlet and outlet	SalvoCalvo	COMSOL	0	March 11, 2010 07:52
VOF Outlet boundary condition in cfd - ace	JM	Main CFD Forum	0	December 15, 2006 09:07
what the result is negatif pressure at inlet	chong chee nan	FLUENT	0	December 29, 2001 06:13

February 1, 2012, 19:33		#3
Janshi New Member Join Date: Aug 2010 Posts: 11 Rep Power: 16	Hello glenn, First, thank you for your fast answer. I already assumed that there would be problems. I´ve already given it some tries but every calculation blow away after some iterations depending on the timestep I´ve chosen. Usually my Mach number grows ridiculous high (beyond 10) and the the calculation blow away with a fatal error that the linear solver diverged. I assume that if I define the sides as openings, nothing will change at the problem with numerical stability? Best regards Shichun

February 1, 2012, 19:51		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	No, that will not improve numerically instability. You might be able to get it to work, first try a smaller time step. Why are you doing it like this? What are you trying to do? Are you trying to put the central object under different angles of attack without having to remesh?

February 2, 2012, 05:51		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Have you considered a C grid like this http://en.wikibooks.org/wiki/ICEM_CFD ? They are frequently used in AOA sweeps. You make the circular, top and bottom surfaces inlets and the back face an outlet.