How far to extend boundary

MachZero · June 8, 2013, 10:11

Hello All,

I am simulating a centrifugal fan to try and predict a fan curve. An experiment was done, so I have built my geometry to try and match the experiment. At the exit of the fan, a 2" tube is connected as the exhaust. This tube exhausts into atmosphere. Since my flow is still incompressible, I ended the domain at the end of this tube and called it a pressure outlet at 0Pa.

One thing that I always struggle to figure out is, should I extend this outlet further? Should I model the room that it is in? If so, how far do I need to extend. If I extend the domain and call all the extended walls pressure outlets, then I get reversed flow continuously, which slows the simulation down tremendously. How are these types of things handled by the pros out there?

Any information or examples would be awesome.

sina.sh · June 9, 2013, 00:49

Dear MachZero
you should extend your region until the effects of up wind boundary condition on down wind BC be negligible. its Try and Error!

Regards

MachZero · June 9, 2013, 10:16

Thank you for your response. Do you have any recommendations on how to check for this? What kind of approach to take? For example, should I extend it until there is almost no pressure gradient at the exit? Should the extension be just an extended tube or should it be a large open section representing a room. I look forward to hearing your thoughts. Thanks!

sina.sh · June 10, 2013, 00:46

Yes. For best solution, it should be larger than tube for example: large box. plz plot a variable (for example Uvelocity) in your main domain and each time -when you change extended domain- check deviation of variable from last one. if no change had been seen so size of extended domain is ok.

June 8, 2013, 10:11	How far to extend boundary	#1
MachZero Member Join Date: Aug 2011 Posts: 89 Rep Power: 15	Hello All, I am simulating a centrifugal fan to try and predict a fan curve. An experiment was done, so I have built my geometry to try and match the experiment. At the exit of the fan, a 2" tube is connected as the exhaust. This tube exhausts into atmosphere. Since my flow is still incompressible, I ended the domain at the end of this tube and called it a pressure outlet at 0Pa. One thing that I always struggle to figure out is, should I extend this outlet further? Should I model the room that it is in? If so, how far do I need to extend. If I extend the domain and call all the extended walls pressure outlets, then I get reversed flow continuously, which slows the simulation down tremendously. How are these types of things handled by the pros out there? Any information or examples would be awesome.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Implementation of boundary conditions for FVM	Tom	Main CFD Forum	7	August 26, 2014 06:58
An error has occurred in cfx5solve:	volo87	CFX	5	June 14, 2013 18:44
Low Mixing time Problem	Mavier	CFX	5	April 29, 2013 01:00
CFX doesn't continue calculation...	mactech001	CFX	6	November 15, 2009 22:25
Boundary conditions?	Tom	Main CFD Forum	0	November 5, 2002 02:54

June 9, 2013, 00:49		#2
sina.sh New Member sina Join Date: Sep 2011 Posts: 28 Rep Power: 15	Dear MachZero you should extend your region until the effects of up wind boundary condition on down wind BC be negligible. its Try and Error! Regards

June 9, 2013, 10:16		#3
MachZero Member Join Date: Aug 2011 Posts: 89 Rep Power: 15	Thank you for your response. Do you have any recommendations on how to check for this? What kind of approach to take? For example, should I extend it until there is almost no pressure gradient at the exit? Should the extension be just an extended tube or should it be a large open section representing a room. I look forward to hearing your thoughts. Thanks!

June 10, 2013, 00:46		#4
sina.sh New Member sina Join Date: Sep 2011 Posts: 28 Rep Power: 15	Yes. For best solution, it should be larger than tube for example: large box. plz plot a variable (for example Uvelocity) in your main domain and each time -when you change extended domain- check deviation of variable from last one. if no change had been seen so size of extended domain is ok.