No Vortex shedding for flow past cylinder

shashanktiwari619 · December 27, 2016, 01:23

I am trying to simulate a 2D case for flow past cylinder. I have taken the diameter of cylinder as 1m. The case has been set up for a Re=80 at which the flow should exhibit vortex shedding. I get perfect vortex shedding results when I use the Viscous-Laminar Turbulence model but any other model does not yield in vortex shedding. I have fine mesh near the wall for getting refined results. I do understand that k-epsilon and k-omega models are not able to show the shedding as they do not solve near wall flows accurately. But I think I should get vortex shedding results by using SST-k-Omega, but it is not so.

Antanas · December 27, 2016, 02:13

Quote:

Originally Posted by shashanktiwari619

I am trying to simulate a 2D case for flow past cylinder. I have taken the diameter of cylinder as 1m. The case has been set up for a Re=80 at which the flow should exhibit vortex shedding. I get perfect vortex shedding results when I use the Viscous-Laminar Turbulence model but any other model does not yield in vortex shedding. I have fine mesh near the wall for getting refined results. I do understand that k-epsilon and k-omega models are not able to show the shedding as they do not solve near wall flows accurately. But I think I should get vortex shedding results by using SST-k-Omega, but it is not so.

You have laminar flow. Why do you want to use turbulence models?
Moreover, eddy viscosity turbulence models model all turbulent scales and can give you only average info.

Use laminar model and transient solver to get realistic results.

shashanktiwari619 · December 27, 2016, 03:33

I am not able to understand that at Re=80 how the flow can still be laminar if the vortex shedding has already begun?
Is it that just because the axisymmetric wake is still intact at this Re range and the shedding occurs periodically at regular intervals the flow is still laminar?

Sent from my Moto G (4) using CFD Online Forum mobile app

Antanas · December 27, 2016, 04:24

Quote:

Originally Posted by shashanktiwari619

I am not able to understand that at Re=80 how the flow can still be laminar if the vortex shedding has already begun?
Is it that just because the axisymmetric wake is still intact at this Re range and the shedding occurs periodically at regular intervals the flow is still laminar?

Sent from my Moto G (4) using CFD Online Forum mobile app

Based on your Re you should have laminar vortex shedding with laminar (von Karman) vortex street.

Vortex shedding may be laminar with formation of laminar vortex street and turbulent with formation of turbulent wake.

LuckyTran · December 27, 2016, 12:02

Quote:

Originally Posted by shashanktiwari619

I am not able to understand that at Re=80 how the flow can still be laminar if the vortex shedding has already begun?
Is it that just because the axisymmetric wake is still intact at this Re range and the shedding occurs periodically at regular intervals the flow is still laminar?

Sent from my Moto G (4) using CFD Online Forum mobile app

Both laminar and turbulence flows can exhibit vortex shedding which is typically an inviscid instability. Do not confuse the presence of vortices with turbulence. Inviscid vortex shedding occurs at discrete frequencies (i.e. coherent & periodic) whereas turbulence is broadband (i.e. decoherent and chaotic). Turbulence is a flow state in which all the frequencies are unstable.

If the flow is laminar use the laminar flow model and not a turbulence model. Unsteady simulations using laminar or turbulence models can capture inviscid vortex shedding.

kaszt · December 28, 2016, 02:38

Are you using a mesh that is symmetric about the centerline? I've found this same problem when using a symmetric mesh with a uniform velocity inlet condition. The way that I solved it was to introduce a perturbation in the inlet velocity condition. Using OpenFOAM, this would be a small velocity in +y that would only last for a short time (again, using OpenFOAM this was accomplished with the "ramp" function). As an example, this would be changing the inlet velocity condition from u={u_inf, 0} for a 2-D case to u={u_inf, 1} from time=0.5 seconds to time=1 seconds. After introducing this perturbation in the y-direction, you would again change the velocity inlet boundary condition to u={u_inf, 0} (I.e. you would again only have a horizontal component of velocity).

As mentioned previously, you shouldn't need a turbulence model for Re=80 since this is laminar flow. The problem might be that you have a symmetric mesh and you need a perturbation to introduce the vortex street behind your cylinder.

Sent from my Nexus 5X using CFD Online Forum mobile app

SPH_CFD · December 28, 2016, 04:22

Quote:

Originally Posted by shashanktiwari619

I am trying to simulate a 2D case for flow past cylinder. I have taken the diameter of cylinder as 1m. The case has been set up for a Re=80 at which the flow should exhibit vortex shedding. I get perfect vortex shedding results when I use the Viscous-Laminar Turbulence model but any other model does not yield in vortex shedding. I have fine mesh near the wall for getting refined results. I do understand that k-epsilon and k-omega models are not able to show the shedding as they do not solve near wall flows accurately. But I think I should get vortex shedding results by using SST-k-Omega, but it is not so.

I have already done this simulation.
Firstly, please using laminar model
Secondly, run transient and wait very long time. The last time I have the same problem. However if I run enough time, they will have fluctuations and vortex

Sent from my LG-F260S using CFD Online Forum mobile app

shashanktiwari619 · December 28, 2016, 09:35

Thanks @Antanas for pointing the difference between laminar and turbulent vortex shedding w.r.t Re. I ran the simulations again this time for a higher Re (Re=350) using laminar-viscous as well as other turbulence models. Even at this range laminar-viscous model does show vortex shedding which can be easily differentiated as compared to that which was obtained at Re=80. Also for this range the turbulence models SST-k-omega, Scale Adaptive Simulation and the Transition SST Model do predict the turbulent vortex shedding however in case of SST-komega the values of Drag Coefficient which I obtain are really vague (around 24 whereas at Re=350 it should be around 1.3 which I do obtain using other turbulence models). The other problem is using RSM model I am unable to get the turbulent vortices, RSM not being an eddy viscosity model should be able to predict the shedding but it is not so!!

December 27, 2016, 01:23	No Vortex shedding for flow past cylinder	#1
shashanktiwari619 New Member Shashank Join Date: Sep 2014 Posts: 28 Rep Power: 11	I am trying to simulate a 2D case for flow past cylinder. I have taken the diameter of cylinder as 1m. The case has been set up for a Re=80 at which the flow should exhibit vortex shedding. I get perfect vortex shedding results when I use the Viscous-Laminar Turbulence model but any other model does not yield in vortex shedding. I have fine mesh near the wall for getting refined results. I do understand that k-epsilon and k-omega models are not able to show the shedding as they do not solve near wall flows accurately. But I think I should get vortex shedding results by using SST-k-Omega, but it is not so.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Flow Past a Cylinder, the Onset of Vortex Shedding	Nauseouslyquick	Main CFD Forum	3	October 23, 2016 16:46
vortex shedding past a cylinder	dreamwork	FLUENT	5	September 20, 2014 03:20
Cylinder vortex shedding (3D)	Apocolapse	STAR-CCM+	2	April 3, 2014 20:33
no VOrtex Shedding flow past a 2D cylinder	Sufyan	Main CFD Forum	7	November 16, 2012 01:13
LES: vortex shedding past a 2D and 3D circular cylinder	Rocchi Daniele	Main CFD Forum	7	April 9, 1999 12:05

December 27, 2016, 03:33		#3
shashanktiwari619 New Member Shashank Join Date: Sep 2014 Posts: 28 Rep Power: 11	I am not able to understand that at Re=80 how the flow can still be laminar if the vortex shedding has already begun? Is it that just because the axisymmetric wake is still intact at this Re range and the shedding occurs periodically at regular intervals the flow is still laminar? Sent from my Moto G (4) using CFD Online Forum mobile app

December 28, 2016, 02:38		#6
kaszt New Member Katherine Join Date: Feb 2016 Posts: 16 Rep Power: 10	Are you using a mesh that is symmetric about the centerline? I've found this same problem when using a symmetric mesh with a uniform velocity inlet condition. The way that I solved it was to introduce a perturbation in the inlet velocity condition. Using OpenFOAM, this would be a small velocity in +y that would only last for a short time (again, using OpenFOAM this was accomplished with the "ramp" function). As an example, this would be changing the inlet velocity condition from u={u_inf, 0} for a 2-D case to u={u_inf, 1} from time=0.5 seconds to time=1 seconds. After introducing this perturbation in the y-direction, you would again change the velocity inlet boundary condition to u={u_inf, 0} (I.e. you would again only have a horizontal component of velocity). As mentioned previously, you shouldn't need a turbulence model for Re=80 since this is laminar flow. The problem might be that you have a symmetric mesh and you need a perturbation to introduce the vortex street behind your cylinder. Sent from my Nexus 5X using CFD Online Forum mobile app

December 28, 2016, 09:35		#8
shashanktiwari619 New Member Shashank Join Date: Sep 2014 Posts: 28 Rep Power: 11	Thanks @Antanas for pointing the difference between laminar and turbulent vortex shedding w.r.t Re. I ran the simulations again this time for a higher Re (Re=350) using laminar-viscous as well as other turbulence models. Even at this range laminar-viscous model does show vortex shedding which can be easily differentiated as compared to that which was obtained at Re=80. Also for this range the turbulence models SST-k-omega, Scale Adaptive Simulation and the Transition SST Model do predict the turbulent vortex shedding however in case of SST-komega the values of Drag Coefficient which I obtain are really vague (around 24 whereas at Re=350 it should be around 1.3 which I do obtain using other turbulence models). The other problem is using RSM model I am unable to get the turbulent vortices, RSM not being an eddy viscosity model should be able to predict the shedding but it is not so!!