Flow around a shallowly submerged cylinder

ruminame06 · October 3, 2011, 16:36

Hello guys i am a student & i am working on a simulation on a Flow around a shallowly submerged cylinder in star ccm+.I want to capture the vortex shading and also the drag & lift cofficient.I have used trimmer mesh for this perpose & volumatric control and trimmer wake refinment to have a better vortex shading.
My Re No. is 4500000 and flow velocity is 3.93 m/s as i want to have a Fr. number 1 just to cross check the result with an exsisting result.
The problem is when the vortex is created at the back of the cylinder it never breakup and goes on.But it should breakup as the exsisting result shows.I think it is difficult to state the problem only by word so i have attached a picture of my vortex shading & my model & the exsisting vortex shading
Eagerly waiting for reply
regards
rumi

lava12005 · October 3, 2011, 20:59

Hmm well I never simulate using free surface but the normal vortex shedding around a cylinder starts at Re ~ 50, so something must be wrong if the wake never breakup.

Are you sure you are running unsteady simulation?
Hmm this may be a stupid question but how many time step have you been running the simulation? At high Re it should break up easily (normal case)..
Check the timestep whether it is small enough?
How about residual? Diverging?

ruminame06 · October 4, 2011, 08:48

yes i am using unsteady simulation(implicite unsteady).
i have ran the simulation for almost 1 day.
& the residual is also diverging.
i have ran the simulation many times altering the conditions(Re number,flow velocity etc)but no improvement on the result the wake never breakup.
Is there any chance that it is happining because of y+ value.Actually as a newbee i am little confuse on y+ value.i have calculated y+ value and found it is 0.06.does it has something to do with my problem???
Where should i put this y+ value in star ccm+?

ruminame06 · October 4, 2011, 12:32

@lava
for a better understanding i am attaching a picture of residual & vortex shading of mine simulation.See there is no breakup in the wake.

lava12005 · October 4, 2011, 21:01

It depends on what Turbulence model that you are using. But y+ = 0.06 is too much I think? With standard k-e model, you can use y+ = 30, even for LES, y+ just below 1 should be enough. You can find more guide about y+ in the manual, so you know what y+ you need to chose when using different turbulent model.

Using prism layer model, you can specify the first grid thickness, so you can control the y+ value.

Maybe try to do more simple case, and try to make it works.
have you done the case where there is only single phase (i.e. just a flow around a cylinder)? If not then you can try it first.

abdul099 · October 5, 2011, 19:37

When using a very low Y+ (around or less than 1), you should have enough prism layers to resolve the boundary layer. Let's say at least 10 to 15.
When you don't want to have that much prism layers, you should keep Y+ above 30.

A few month ago, another guy had a similar problem, although is was no multiphase simulation. The problem is, the setup was too stable, so it was necessary to disturb it a little bit. I changed the inflow direction for some time steps to hit the cylinder in a slightly different angle. After changing back to the original value a perfect vortex shedding could be seen. As you are performing a multiphase simulation, I would suggest to modify the gravity vector for some time steps. That should change the flow over the cylinder as well. Or play around with initialisation values.

Additionally make sure to have a small enough time step and your mesh is fine enough to capture the wake.
Do you simulate also heat exchange? When not, get rid of the energy model, that's just computational effort without any sense.

ruminame06 · October 6, 2011, 15:54

thanx abdul099 for your reply.

Basically i am a student & i am completely new with star ccm+and i am trying to learn it by myself.So if can please tell me how to vary the flow direction by some angle and how to modify the gravity vector that would be a great help for me.I can set the flow direction only to three direction but to alter it by some angle i dont know how to do it and i am little confuse about that to do with gravity vector.So if can illustrate that would be a great help for me

abdul099 · October 6, 2011, 18:58

And where is the problem? You've found the velocity components, so why don't you change them? Changing the components of the flow direction changes the direction, or am I wrong?!? And calculating the angle between two vectors is not that hard. Even better, you don't even need to calculate any specific angle, just change one of the components a little bit.
The gravity vector can be found in the physics continuum, I think in reference values.

October 5, 2011, 19:37		#6
abdul099 Senior Member Join Date: Oct 2009 Location: Germany Posts: 636 Rep Power: 21	When using a very low Y+ (around or less than 1), you should have enough prism layers to resolve the boundary layer. Let's say at least 10 to 15. When you don't want to have that much prism layers, you should keep Y+ above 30. A few month ago, another guy had a similar problem, although is was no multiphase simulation. The problem is, the setup was too stable, so it was necessary to disturb it a little bit. I changed the inflow direction for some time steps to hit the cylinder in a slightly different angle. After changing back to the original value a perfect vortex shedding could be seen. As you are performing a multiphase simulation, I would suggest to modify the gravity vector for some time steps. That should change the flow over the cylinder as well. Or play around with initialisation values. Additionally make sure to have a small enough time step and your mesh is fine enough to capture the wake. Do you simulate also heat exchange? When not, get rid of the energy model, that's just computational effort without any sense. ruminame06 likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
benchmark: flow over a circular cylinder	goodegg	Main CFD Forum	12	January 22, 2013 11:47
1 Was FSI for a flow over a cylinder	vmlxb6	CFX	17	May 16, 2011 02:29
reference calcultion, flow around a cylinder	butch85	Main CFD Forum	3	January 31, 2011 16:10
Moving a Cylinder in cross flow after solving Dynamics equations of motion	maruthamuthu_venkatraman	OpenFOAM	1	November 19, 2009 13:55
Flow over a cylinder	Anna	Main CFD Forum	9	March 24, 2006 14:32

October 3, 2011, 20:59		#2
lava12005 Senior Member KHB Join Date: Aug 2010 Location: Singapore Posts: 118 Rep Power: 15	Hmm well I never simulate using free surface but the normal vortex shedding around a cylinder starts at Re ~ 50, so something must be wrong if the wake never breakup. Are you sure you are running unsteady simulation? Hmm this may be a stupid question but how many time step have you been running the simulation? At high Re it should break up easily (normal case).. Check the timestep whether it is small enough? How about residual? Diverging?

October 4, 2011, 08:48		#3
ruminame06 New Member rumi Join Date: Oct 2011 Posts: 4 Rep Power: 14	yes i am using unsteady simulation(implicite unsteady). i have ran the simulation for almost 1 day. & the residual is also diverging. i have ran the simulation many times altering the conditions(Re number,flow velocity etc)but no improvement on the result the wake never breakup. Is there any chance that it is happining because of y+ value.Actually as a newbee i am little confuse on y+ value.i have calculated y+ value and found it is 0.06.does it has something to do with my problem??? Where should i put this y+ value in star ccm+?

October 4, 2011, 21:01		#5
lava12005 Senior Member KHB Join Date: Aug 2010 Location: Singapore Posts: 118 Rep Power: 15	It depends on what Turbulence model that you are using. But y+ = 0.06 is too much I think? With standard k-e model, you can use y+ = 30, even for LES, y+ just below 1 should be enough. You can find more guide about y+ in the manual, so you know what y+ you need to chose when using different turbulent model. Using prism layer model, you can specify the first grid thickness, so you can control the y+ value. Maybe try to do more simple case, and try to make it works. have you done the case where there is only single phase (i.e. just a flow around a cylinder)? If not then you can try it first.

October 6, 2011, 15:54		#7
ruminame06 New Member rumi Join Date: Oct 2011 Posts: 4 Rep Power: 14	thanx abdul099 for your reply. Basically i am a student & i am completely new with star ccm+and i am trying to learn it by myself.So if can please tell me how to vary the flow direction by some angle and how to modify the gravity vector that would be a great help for me.I can set the flow direction only to three direction but to alter it by some angle i dont know how to do it and i am little confuse about that to do with gravity vector.So if can illustrate that would be a great help for me

October 6, 2011, 18:58		#8
abdul099 Senior Member Join Date: Oct 2009 Location: Germany Posts: 636 Rep Power: 21	And where is the problem? You've found the velocity components, so why don't you change them? Changing the components of the flow direction changes the direction, or am I wrong?!? And calculating the angle between two vectors is not that hard. Even better, you don't even need to calculate any specific angle, just change one of the components a little bit. The gravity vector can be found in the physics continuum, I think in reference values.