open channel flow-sst model

antonio · December 14, 2010, 04:55

Hi all. I am trying to simulate a free surface flow over a hydraulically rough bed using the sst model. In order to have a decent behavior of this turbulence model I would like to ask how can I assure, in a practical point of view, that I have at least 15 nodes in the boundary layer (value presented in the documentation)?

Apart from the fact that y+ should be more or less equal to 1 is there any other recommendation when creating the mesh?

It would be very helpful to hear the feedback of all.Regards

ghorrocks · December 14, 2010, 05:46

I don't think you can do hydraulically rough flow without wall functions (ie y+>11). You should not integrate to the wall for this type of flow.

So put the first node at about y+=15, and grade out from there with a ratio of between 1.01 and 1.05. Hopefully that will have about the right BL resolution for you.

antonio · December 14, 2010, 07:50

Sorry Sir, but i did not fully understand your explanation.

According to the documentation it says that sst model is based on the k- w model and has the same automatic wall treatment, which consists in switching between a low-Reynolds number formulation at low y+values and a wall function approach at higher y+values. Moreover it says that one can account for the roughness effects by considering a hs+ parameter.

ghorrocks · December 14, 2010, 16:31

You are right, you can model rough walls while integrating to the wall. But it does not sound like a good idea, unless you need detailed information about the boundary layer then it would be best to use wall functions. This means start with a mesh which has the first node at y+=15 to 20 or so.

antonio · December 15, 2010, 04:17

Hi.

That`s precisely my case. Basically I am computing some quantities near the wall whose value at moment is very low. I think that can be caused by a bad separation prediction. That´s the reason why I have changed for sst model.

ghorrocks · December 15, 2010, 04:55

SST can handle the first node at y+>11 just fine. You don't need to integrate to the wall.

antonio · December 15, 2010, 05:27

Hi,more two questions please:

Can you tell me why do you consider that`s not a good idea to integrate to the wall and why should I use the first node at y+=15 (or something like this)?

Do you think that even if i put the first node at y+=15, sst will "catch" the recirculation zones with more accuracy?

Best regards

ghorrocks · December 15, 2010, 17:44

Because a mesh using wall functions and a first node will have a smaller mesh by a factor of at least 15, and probably more like 100. That means 100 times less memory requirements and run time.

If wall functions work, then use them.

December 14, 2010, 04:55	open channel flow-sst model	#1
antonio Senior Member Join Date: Jan 2010 Posts: 110 Rep Power: 16	Hi all. I am trying to simulate a free surface flow over a hydraulically rough bed using the sst model. In order to have a decent behavior of this turbulence model I would like to ask how can I assure, in a practical point of view, that I have at least 15 nodes in the boundary layer (value presented in the documentation)? Apart from the fact that y+ should be more or less equal to 1 is there any other recommendation when creating the mesh? It would be very helpful to hear the feedback of all.Regards

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[GAMBIT] Gambit & 3d open channel flow mesh problem	esgee	ANSYS Meshing & Geometry	7	April 10, 2012 00:51
fluent model open channel	yani	FLUENT	0	October 31, 2010 02:54
Yacht in Open Channel Flow	andreimour	FLUENT	1	October 14, 2010 23:54
Open channel flow	Mehdi Moayeri	FLUENT	0	August 26, 2008 10:55
Inclined Open Channel Flow	Abe Gonzalez	FLUENT	2	April 15, 2008 12:32

December 14, 2010, 05:46		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	I don't think you can do hydraulically rough flow without wall functions (ie y+>11). You should not integrate to the wall for this type of flow. So put the first node at about y+=15, and grade out from there with a ratio of between 1.01 and 1.05. Hopefully that will have about the right BL resolution for you.

December 14, 2010, 07:50		#3
antonio Senior Member Join Date: Jan 2010 Posts: 110 Rep Power: 16	Sorry Sir, but i did not fully understand your explanation. According to the documentation it says that sst model is based on the k- w model and has the same automatic wall treatment, which consists in switching between a low-Reynolds number formulation at low y+values and a wall function approach at higher y+values. Moreover it says that one can account for the roughness effects by considering a hs+ parameter.

December 14, 2010, 16:31		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	You are right, you can model rough walls while integrating to the wall. But it does not sound like a good idea, unless you need detailed information about the boundary layer then it would be best to use wall functions. This means start with a mesh which has the first node at y+=15 to 20 or so.

December 15, 2010, 04:17		#5
antonio Senior Member Join Date: Jan 2010 Posts: 110 Rep Power: 16	Hi. That`s precisely my case. Basically I am computing some quantities near the wall whose value at moment is very low. I think that can be caused by a bad separation prediction. That´s the reason why I have changed for sst model.

December 15, 2010, 04:55		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	SST can handle the first node at y+>11 just fine. You don't need to integrate to the wall.

December 15, 2010, 05:27		#7
antonio Senior Member Join Date: Jan 2010 Posts: 110 Rep Power: 16	Hi,more two questions please: Can you tell me why do you consider that`s not a good idea to integrate to the wall and why should I use the first node at y+=15 (or something like this)? Do you think that even if i put the first node at y+=15, sst will "catch" the recirculation zones with more accuracy? Best regards

December 15, 2010, 17:44		#8
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Because a mesh using wall functions and a first node will have a smaller mesh by a factor of at least 15, and probably more like 100. That means 100 times less memory requirements and run time. If wall functions work, then use them.