CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Help--how to define outlet boundary condition? (http://www.cfd-online.com/Forums/fluent/102202-help-how-define-outlet-boundary-condition.html)

gxu2 May 24, 2012 17:12

Help--how to define outlet boundary condition?
 
4 Attachment(s)


Hi,everyone.

I want to simulate 2 D Stokes wave using fluent 13, the result is shown in the picture. In this picture, several waves are good in the first half of flume, but at the second half of the flume, the waves dissipate so rapidly. I don't know why.
I am wondering should I define a pressure profile udf to define the outlet?
Could anybody give some advice? Thanks in advance!

jwillie2000 May 25, 2012 02:45

Hi, Can you throw more light on your problem? Like is the flow turbulent? I would assume no? and did you introduce any disturbance at the inlet in the velocity, etc? Jimmy

gxu2 May 25, 2012 09:47

Quote:

Originally Posted by jwillie2000 (Post 363026)
Hi, Can you throw more light on your problem? Like is the flow turbulent? I would assume no? and did you introduce any disturbance at the inlet in the velocity, etc? Jimmy

Thank you very much for helping me, my friend!

Actually, the flow is laminar flow, without any turbulent at the inlet velocity profile. the boudary conditions are velocity inlet, bottom--no slipping wall, top--no slipping wall, outlet--out flow.

Should I define a pressure udf for the out flow?

Thank you very much again.

jwillie2000 May 25, 2012 10:52

It means u defined a velocity profile at the inlet? Am I getting you right? If yes, then you need to just need to increase the amplitude of the disturbance or flutuation in your velocity profile. The code is probably damping the wave and this is likely numerical. Make sure you do grid dependency test to remove any ambiguity. Jimmy

gxu2 May 25, 2012 11:05

Quote:

Originally Posted by jwillie2000 (Post 363123)
It means u defined a velocity profile at the inlet? Am I getting you right? If yes, then you need to just need to increase the amplitude of the disturbance or flutuation in your velocity profile. The code is probably damping the wave and this is likely numerical. Make sure you do grid dependency test to remove any ambiguity. Jimmy

Yes, I defined a velociy profile for the velociy inlet. but you could see all the photos I uploaded, which could tell that the inlet wave profile is good, but when propagating, the wave profile changes. It seems my velocity inlet udf is right, maybe it does not need to adjust the disturbance or flutuation in my velociy profile. Do you think so?

Thank you, Jimmy!

jwillie2000 May 28, 2012 06:41

Hi, is your profile at the inlet a function? sin or cosine? and if yes, what is the amplitude? I suspect the amplitude is not large enough and so it is being damped? so play with that and see. because you are using the euler euqation, we should not have any damping due to viscous effects. If there is then it is numerical. That is why i am asking about whether you did any grid dependency test? Jimmy

Daniel Tanner May 28, 2012 07:33

Might be worth trying the pressure outlet boundary condition (instead of the outflow condition). The outflow condition may not be suitable for this application (it assumes fully developed conditions at the outlet).

gxu2 May 28, 2012 11:16

Quote:

Originally Posted by jwillie2000 (Post 363417)
Hi, is your profile at the inlet a function? sin or cosine? and if yes, what is the amplitude? I suspect the amplitude is not large enough and so it is being damped? so play with that and see. because you are using the euler euqation, we should not have any damping due to viscous effects. If there is then it is numerical. That is why i am asking about whether you did any grid dependency test? Jimmy

For my case, VOF method is employed. inlet profile cosists of U-velocity, V-velocity, and volume fraction factor function. it is second Stokes wave, all the function is defined according to Stokes Second Theory. I didn't enlarge the amplitude.

One small question, could you explain what is grid dependency test? you mean try different grid meshing?

Thank you very much! Happy Memorial Day!

gxu2 May 28, 2012 11:22

Quote:

Originally Posted by Daniel Tanner (Post 363423)
Might be worth trying the pressure outlet boundary condition (instead of the outflow condition). The outflow condition may not be suitable for this application (it assumes fully developed conditions at the outlet).

Thank you for your advice. I will try it.

jwillie2000 May 28, 2012 11:23

Hi, And thanks for the wishes. About grid dependency test. You have a case where the waves you have in your flow are dissipating quickly than expected. The qauestions is is it physical or non-phsical or numerical? To remove the third case, you try have to do grid dependency test. Like you said you need to vary your mesh resolution till you have no change in your critical variable of interest. It could be the pressure or velocity or temperature in a given section or boundary of your flow. When that happens, you can reliably say that the results you are getting are void of any numerical artifacts and therefore must be due to the physics of the flow. In the case you have right now it would be difficult to say. Good luck! James

ghost82 May 28, 2012 11:26

Quote:

Originally Posted by gxu2 (Post 363447)
For my case, VOF method is employed. inlet profile cosists of U-velocity, V-velocity, and volume fraction factor function. it is second Stokes wave, all the function is defined according to Stokes Second Theory. I didn't enlarge the amplitude.

One small question, could you explain what is grid dependency test? you mean try different grid meshing?

Thank you very much! Happy Memorial Day!

Yes, grid dependency test means trying different meshes: start from a coarse mesh and test other finer meshes, and stop when your results don't change to much from a grid to another.

Daniele

EDIT: ok, jwillie was faster :)

gxu2 May 28, 2012 11:40

Quote:

Originally Posted by jwillie2000 (Post 363449)
Hi, And thanks for the wishes. About grid dependency test. You have a case where the waves you have in your flow are dissipating quickly than expected. The qauestions is is it physical or non-phsical or numerical? To remove the third case, you try have to do grid dependency test. Like you said you need to vary your mesh resolution till you have no change in your critical variable of interest. It could be the pressure or velocity or temperature in a given section or boundary of your flow. When that happens, you can reliably say that the results you are getting are void of any numerical artifacts and therefore must be due to the physics of the flow. In the case you have right now it would be difficult to say. Good luck! James

Please excuse me for my little knowledge about "Physical or non-physical", I appreciate your insightful further explaination. Does it bother you too much?

gxu2 May 28, 2012 11:41

Quote:

Originally Posted by ghost82 (Post 363450)
Yes, grid dependency test means trying different meshes: start from a coarse mesh and test other finer meshes, and stop when your results don't change to much from a grid to another.

Daniele

EDIT: ok, jwillie was faster :)

Thank you for your kind-hearted reply. May God bless you!

jwillie2000 May 28, 2012 13:15

Hi, Not sure if your question is already answered about physical or non-physical? Physical is whether the phenomenon you are observing or your results can be attributed to physics or reality or not. Some of them are numerical or non-physical...meaning it cannot be attributed to physics and sometimes simply does not hold. Hope it helps. Jimmy


All times are GMT -4. The time now is 02:35.