Velocity field problem

feizaghaee · February 15, 2010, 16:50

hi. i work with cfx 12 and i'm trying to simulate a free surface flow in a rectanular channel with ssg turbulance model. but there is something rong with the velocity field. i use homogeneous multiphase flow. the max velocity is about 0.9 y from bottom of channel. i attached a pic of velocity field. does anybody know how can i correct it?

feizaghaee · February 16, 2010, 01:48

does anybody know what's the reason of this problem??????

zandi · February 16, 2010, 04:04

salam
i couldn't undrestand what's problem. could you explain it more. is the place of the highest velocity point wrong or value of it. it changes from zero to max from boundary to near the surface, seems to be right.

feizaghaee · February 16, 2010, 09:01

it's not correct because the max velocity occures at .9 y (depth) from bottom of the channel

ghorrocks · February 16, 2010, 18:28

I assume you mean the problem is the max velocity does not occur at the free surface.

Have you done a mesh sensitivity study?

zandi · February 17, 2010, 06:46

if you mean the water surface and
if you have done mesh study just near the boundary or whole domain, did you ever note the water surface. about mesh adaption, did you ever test it for Air at 25 C.Volume Fraction ?

feizaghaee · February 18, 2010, 02:28

Quote:

Originally Posted by ghorrocks

I assume you mean the problem is the max velocity does not occur at the free surface.

Have you done a mesh sensitivity study?

no i mean the max velocity must occurs at the .7 hight of water from bottom of channel.

feizaghaee · February 18, 2010, 02:51

the channel is toward the Z direction and i use hexa mesh. the hexa meshes have .2 m long in Z direction and in the XY plane have a shape like attached picture. the finer mesh is around the free surface. what do you think of mesh. is it too large

ghorrocks · February 18, 2010, 17:46

I must have said this a thousand times, but here goes again.

The only way to be sure your mesh is OK is by a mesh sensitivity check. Run increasingly fine meshes until the parameter of interest converges to a tolerance you can accept.

You cannot just look at a mesh and know whether it is fine enough without extensive experience in that field of modelling.

So that means that in this case, yes, almost certainly the mesh is too coarse.

feizaghaee · February 19, 2010, 04:52

but i've checked with different size and the results was same. i've worked with .85 m long meshes but the result was exactly as same as meshes with .19 m long. the turbulence did'nt affect the results.

ghorrocks · February 20, 2010, 06:06

What makes the maximum velocity not occur at the top of the heavy fluid? Can you explain what you are modelling a bit?

feizaghaee · February 20, 2010, 13:47

i'm modeling a channel with water and air flow. the secondary flows cause max velocity dosen't occur at top of the flow.

zandi · February 20, 2010, 15:43

how the secondonary flow force the max velocity in z direction occur in 0.7y not in near the surface?not sure but may be it is right for the y or x direction not for the main flow direction.

ghorrocks · February 21, 2010, 18:37

Where do the secondary flows come from? Is the channel curved?

feizaghaee · February 22, 2010, 04:01

i finally understand the reason of this problem. the secondary flows in velocity field are exactly inverse (attached pic). does any body know what's the problem?

ghorrocks · February 22, 2010, 20:35

Is your simulation fully converged?

I assume the secondary flows are coming from the Reynolds Stresses in the RSM model. These are tricky things to get right, you may need to tweak some model constants - if you are brave.

zandi · February 23, 2010, 05:05

Quote:

Originally Posted by feizaghaee

i ....... the secondary flows in velocity field are exactly inverse....

fine
would you please explain it more
I want to learn actually

you mean that your problem was about velocity in x and y direction?

feizaghaee · February 23, 2010, 07:22

Quote:

Originally Posted by ghorrocks

Is your simulation fully converged?

I assume the secondary flows are coming from the Reynolds Stresses in the RSM model. These are tricky things to get right, you may need to tweak some model constants - if you are brave.

yes, they are coming from Reynolds stresses.
what do you mean by "These are tricky things to get right, you may need to tweak some model constants" ?

ghorrocks · February 23, 2010, 17:42

I mean exactly that - secondary flows are tricky to get right. Are you sure RSM will adequately capture it? You may need an LES approach. I don't know, this is not my field.

Zandi - Google search "secondary flow". Wikipedia probably has a good description. That will explain what a secondary flow is.

feizaghaee · February 24, 2010, 04:59

diffrerences between Reynolds stresses result in this scondary flows but CFX calculates these flows inversely i don't know why.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Problem Interface Solid Fluid with wall velocity Solver v12	hills1	CFX	2	October 12, 2009 06:36
Extract velocity field in certain time step to MATLAB	spatialtime	CFX	3	September 9, 2009 19:30
Convergence problem with thermal phase change model under low velocity conditions	pitisrisuk	CFX	0	July 21, 2009 12:21
Streamline superimposed on velocity field	allamarein	FLUENT	0	July 9, 2009 21:51
velocity problem	phy	FLUENT	1	February 8, 2008 18:12

February 16, 2010, 01:48		#2
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	does anybody know what's the reason of this problem??????

February 16, 2010, 04:04		#3
zandi Senior Member Fatema Zandi Goharrizi Join Date: Mar 2009 Posts: 158 Rep Power: 17	salam i couldn't undrestand what's problem. could you explain it more. is the place of the highest velocity point wrong or value of it. it changes from zero to max from boundary to near the surface, seems to be right.

February 16, 2010, 09:01		#4
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	it's not correct because the max velocity occures at .9 y (depth) from bottom of the channel

February 16, 2010, 18:28		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	I assume you mean the problem is the max velocity does not occur at the free surface. Have you done a mesh sensitivity study?

February 17, 2010, 06:46		#6
zandi Senior Member Fatema Zandi Goharrizi Join Date: Mar 2009 Posts: 158 Rep Power: 17	if you mean the water surface and if you have done mesh study just near the boundary or whole domain, did you ever note the water surface. about mesh adaption, did you ever test it for Air at 25 C.Volume Fraction ?

February 18, 2010, 17:46		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	I must have said this a thousand times, but here goes again. The only way to be sure your mesh is OK is by a mesh sensitivity check. Run increasingly fine meshes until the parameter of interest converges to a tolerance you can accept. You cannot just look at a mesh and know whether it is fine enough without extensive experience in that field of modelling. So that means that in this case, yes, almost certainly the mesh is too coarse.

February 19, 2010, 04:52		#10
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	but i've checked with different size and the results was same. i've worked with .85 m long meshes but the result was exactly as same as meshes with .19 m long. the turbulence did'nt affect the results.

February 20, 2010, 06:06		#11
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	What makes the maximum velocity not occur at the top of the heavy fluid? Can you explain what you are modelling a bit?

February 20, 2010, 13:47		#12
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	i'm modeling a channel with water and air flow. the secondary flows cause max velocity dosen't occur at top of the flow.

February 20, 2010, 15:43		#13
zandi Senior Member Fatema Zandi Goharrizi Join Date: Mar 2009 Posts: 158 Rep Power: 17	how the secondonary flow force the max velocity in z direction occur in 0.7y not in near the surface?not sure but may be it is right for the y or x direction not for the main flow direction.

February 21, 2010, 18:37		#14
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Where do the secondary flows come from? Is the channel curved?

February 22, 2010, 20:35		#16
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Is your simulation fully converged? I assume the secondary flows are coming from the Reynolds Stresses in the RSM model. These are tricky things to get right, you may need to tweak some model constants - if you are brave.

February 23, 2010, 17:42		#19
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	I mean exactly that - secondary flows are tricky to get right. Are you sure RSM will adequately capture it? You may need an LES approach. I don't know, this is not my field. Zandi - Google search "secondary flow". Wikipedia probably has a good description. That will explain what a secondary flow is.

February 24, 2010, 04:59		#20
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	diffrerences between Reynolds stresses result in this scondary flows but CFX calculates these flows inversely i don't know why.