Solver and geometry choose for drag coefficient calculation around circular cylinder at large Re

lin · February 26, 2009, 06:42

Hi,all
I have a question with which solver and geometry to choose for drag coefficient calculation around circular cylinder at large Re(around 10^6).

The question about the solver is whether to choose simpleFoam or turbFoam.

The plus for simpleFoam is that I only want to drag coefficient,one averaged quantity.
The plus for turbFoam is that,this kind of flow seems not to be stationary,so called vortex shedding.

Also I have the problem to choose the geometry.The problem is whether I could only calculate half cylinder and use symmetryplane at the center line, or I have to use the whole cylinder geometry all.

The plus for half cylinder and symmetryplane is that,on average,the drag coefficient should be equal on both side.
The plus for whole cylinder is ,for transient flow,the flow field seems not to be symmetry.

So,all in all,there seems to be four choices for me .

(A) simpleFoam with symmetryplane
(B) simpleFoam with whole cylinder
(C) turbFoam with symmetryplane
(D) turbFoam with whole cylinder

Can any experts shed some light on this naive question? Just choose one and some explanation is welcome.

maruthamuthu_venkatraman · February 26, 2009, 08:28

Hello,
Your question is not at all Naive!

1. AT Re 10^6 you have both tubulent boundary layer and turbulent wakes.
Your seperation will be delayed.

2. Since the flow is unsteady and to monitor the drag convergence with steady oscillation i would recommed u to model the full cylinder with symmtery on the top and bottom domain . This is only if you want to catch the frequency.

3. If you are just interested only in Average drag and not interested in unsteadiness then you shall run a steady state solution using SIMPLEFOAM with any of the low re ke model. It will give u good results.

Hope it helps you ...

Remember to keep Y plus less than 1 if u r using any of the low re ke models .

Good luck

lin · February 26, 2009, 12:59

Hi, Maruthamuthu
Thanks for your information.

I will try the transient solver with full cylinder later,though I am only interest in the averaged Cd.

Last day I have tried simpleFoam with half cylinder and symmetryplane at the centerline using 3 k-epsilon models,getting Cd about,0.69,0.46,2.62,respectively.Very strange. The experimental data should be around 0.25 and there has been some report that could reproduce it with fluent.Though I do not know the details.All the simulation I did,y+ is in the range 30-300.

One question about your post is ,Re around 10^6 is large ,why recommend me to use low Re turbulent model?

maruthamuthu_venkatraman · April 16, 2009, 11:50

Low Re Ke model has an ability to correct the near wall fluid flow behaviour using Viscous damping functions. But you must keep your y+ 1.

If u browse the book of Turbulence modeling by Wilcox under low reynolds number effects, you will get it. Basically High Re ke models doesnt account for viscous corrections. So Low Re ke model is devised to solve such application.

I havent looked at your post. Hence there is a delay. May be you might have found it already

February 26, 2009, 06:42	Hi,all I have a question with	#1
lin Senior Member Hua Zen Join Date: Mar 2009 Posts: 138 Rep Power: 17	Hi,all I have a question with which solver and geometry to choose for drag coefficient calculation around circular cylinder at large Re(around 10^6). The question about the solver is whether to choose simpleFoam or turbFoam. The plus for simpleFoam is that I only want to drag coefficient,one averaged quantity. The plus for turbFoam is that,this kind of flow seems not to be stationary,so called vortex shedding. Also I have the problem to choose the geometry.The problem is whether I could only calculate half cylinder and use symmetryplane at the center line, or I have to use the whole cylinder geometry all. The plus for half cylinder and symmetryplane is that,on average,the drag coefficient should be equal on both side. The plus for whole cylinder is ,for transient flow,the flow field seems not to be symmetry. So,all in all,there seems to be four choices for me . (A) simpleFoam with symmetryplane (B) simpleFoam with whole cylinder (C) turbFoam with symmetryplane (D) turbFoam with whole cylinder Can any experts shed some light on this naive question? Just choose one and some explanation is welcome.

February 26, 2009, 08:28	Hello, Your questio	#2
maruthamuthu_venkatraman Member Maruthamuthu Venkatraman Join Date: Mar 2009 Location: Norway Posts: 80 Rep Power: 17	Hello, Your question is not at all Naive! 1. AT Re 10^6 you have both tubulent boundary layer and turbulent wakes. Your seperation will be delayed. 2. Since the flow is unsteady and to monitor the drag convergence with steady oscillation i would recommed u to model the full cylinder with symmtery on the top and bottom domain . This is only if you want to catch the frequency. 3. If you are just interested only in Average drag and not interested in unsteadiness then you shall run a steady state solution using SIMPLEFOAM with any of the low re ke model. It will give u good results. Hope it helps you ... Remember to keep Y plus less than 1 if u r using any of the low re ke models . Good luck

February 26, 2009, 12:59	Hi, Maruthamuthu Thanks for y	#3
lin Senior Member Hua Zen Join Date: Mar 2009 Posts: 138 Rep Power: 17	Hi, Maruthamuthu Thanks for your information. I will try the transient solver with full cylinder later,though I am only interest in the averaged Cd. Last day I have tried simpleFoam with half cylinder and symmetryplane at the centerline using 3 k-epsilon models,getting Cd about,0.69,0.46,2.62,respectively.Very strange. The experimental data should be around 0.25 and there has been some report that could reproduce it with fluent.Though I do not know the details.All the simulation I did,y+ is in the range 30-300. One question about your post is ,Re around 10^6 is large ,why recommend me to use low Re turbulent model?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
drag coefficient of cylinder	Libellen	FLUENT	10	August 29, 2008 22:56
About drag coefficient of 2D circular cylinder!	Haoyin Shan	FLUENT	0	May 7, 2007 00:50
Drag Coefficient of Flow across a Circular Cylinde	Zhihua Li	FLUENT	3	April 20, 2004 02:01
Circular cylinder drag at high Re	Anton Lyaskin	Main CFD Forum	0	January 14, 2003 05:48
Drag Coefficient(Cd) of circular cylinder	Ram	FLUENT	2	June 4, 2000 15:32

April 16, 2009, 11:50		#4
maruthamuthu_venkatraman Member Maruthamuthu Venkatraman Join Date: Mar 2009 Location: Norway Posts: 80 Rep Power: 17	Low Re Ke model has an ability to correct the near wall fluid flow behaviour using Viscous damping functions. But you must keep your y+ 1. If u browse the book of Turbulence modeling by Wilcox under low reynolds number effects, you will get it. Basically High Re ke models doesnt account for viscous corrections. So Low Re ke model is devised to solve such application. I havent looked at your post. Hence there is a delay. May be you might have found it already