[Tajnesaie]

Hi
Is it necessary to check y+ for investigate the secondary flow cell in the openning channel or at all for investigate flow in the openning channel? If it is, how much it must be?
thanks
Tajnesaie

[ghorrocks]

Whether y+ is important or not depends on your flow. For some flows it is critical, for some it is irrelevant. Read the documentation on mesh generation and numerical accuracy and do a sensitivity study in your application to determine whether it is important or not.

[Tajnesaie]

Thanks alot
But in my project after I refinemesh y+ go over 200,whereas befor it y+ was between 15-30.I dont know why occur it?and is it correct or happen mistakes?

Tajnesaie

[ghorrocks]

When you are doing mesh refinement there are two regimes. The one everybody thinks about is the convergent regime where as you refine the mesh everything starts converging on the zero-mesh answer. In this regime everything is well behaved.

The second regime is the non-convergent section. This is where you are so far away from mesh independence that weird things happen between mesh refinements. The mesh is far too coarse and the results are total rubbish. Results will go all over the place and nothing behaves as expected.

I suspect you are in the non-convergence regime and are therefore getting strange convergence behaviour. You should continue with finer meshes until you reach the convergent regime and then things will start behaving.

[Tajnesaie]

Thanks ghorrocks
I do it and hope my problem is solved.
god luck

Tajnesaie

[Tajnesaie]

Hi
Can you tell me wath is your scope of flwo kind? and for witch kind of flow it is critical?
tanks
Tajnesaie

[ghorrocks]

I do not understand what you are asking. Can you say the question again more clearly?

[Tajnesaie]

Hi
My question is about this your reply; (Whether y+ is important or not depends on your flow. For some flows it is critical, for some it is irrelevant. Read the documentation on mesh generation and numerical accuracy and do a sensitivity study in your application to determine whether it is important or not.)
I do'nt know your meaning of flow type!and y+ for which flow is critical?My project is a turbulent flow in the channel.
thanks alot
Tajnesaie

[ghorrocks]

Examples of where it is critical would include modelling the separation off an aerofoil. You need good boundary layer resolution for this to get an accurate answer. An example of where it is not important is a Laval nozzle. If all you want is to get some shock waves then you don't need an accurate boundary layer model.

If you are modelling turbulent flow in a channel it probably is important. If you need the boundary layers resolved accurately then it definitely is. You would need to describe what you are doing in more detail for a more precise answer.

[Tajnesaie]

Hi
g.horrocks
I want to investigate the secondary flow cell (vorticity flow) in a trapezoidal channel,i hope you know it,and for it i need to refine mesh very much and the channel that i model it, is very small size (0.2*0.0905*8 m).For modelling this i give help from toturial 7 of cfx help. Number of element is near 6000000.

Thank you so much
Tajnesaie

[ghorrocks]

At what Re number? Is the flow compressible? Do you need an LES approach?

[Tajnesaie]

at Re=9*10^4 , flow is incompressible, exactly i do'nt know ,the turbulent model that i use is k-e, because my master is not so acquainted with other turbulent model.I do'nt know for this project if it is valid or not.
Realy i need emergency help. with this number of elements (6000000) i accost seriouse problem and exactly i dizzy.

thanks

[zandi]

Originally Posted by ghorrocks

When you are doing mesh refinement there are two regimes. The one everybody thinks about is the convergent regime where as you refine the mesh everything starts converging on the zero-mesh answer. In this regime everything is well behaved.

The second regime is the non-convergent section. This is where you are so far away from mesh independence that weird things happen between mesh refinements. The mesh is far too coarse and the results are total rubbish. Results will go all over the place and nothing behaves as expected.

I suspect you are in the non-convergence regime and are therefore getting strange convergence behaviour. You should continue with finer meshes until you reach the convergent regime and then things will start behaving.

Hi
I am in trouble with y+ too

The main objectives my research were to simulate flow over an ogee spillway by a numerical model and investigate the ability of the model to predict several characteristics of flow. At ten different flow head, discharge and pressure were obtained by physical model.

you can see the ogee spillway look like mine in the part c of picture in this link:
http://www.fao.org/docrep/x5744E/x5744ees.gif

Boundary Condition : The hydraulic head in the inlet is defined as a triangular equation of hydrostatic pressure. In the outlet there is no specific characteristic definition so we use simple outlet with pressure gradient equal zero. The bottom boundary is wall and the top one is an opening that let the air move in and out and the pressure gradient is zero.
In the opening the water volume fraction equation is 0 and air volume fraction is 1, and in the inlet this two equation defined as step equation formula due to water and air volume fractions.
For modeling the problem 2d I used symmetry boundary at two side of 3d domain that have a small thickness

Is y+ important for my case?

Could you help me please
Thank you

[ghorrocks]

at Re=9*10^4 , flow is incompressible

The Re does not determine whether the flow is compressible or not, the Mach number does. You can have a supersonic flow with Re=1. I assume your Mach number is low so compressibility effects are insignificant.

the turbulent model that i use is k-e, because my master is not so acquainted with other turbulent model.I do'nt know for this project if it is valid or not.

Failure to achieve mesh independence is one of the problems of the traditional wall function approach. That is why I recommend you switch to SST.

I am in trouble with y+ too
....
Is y+ important for my case?

I doubt it.

But honestly I get a bit sick of people asking questions like this. You cannot short-cut things and get the answer from some mystic on a forum. YOU HAVE TO DO A SENSITIVITY STUDY AND DETERMINE THIS FOR YOURSELF. In the process of doing the sensitivity study you will determine for yourself what are important parameters and what are not, and your analysis of the thing will be much better focussed on what is truly important in your simulation.

This is a much better approach than trying to simply ask on a forum and hope that you get the answer you want. Don't forget that half the posts on this forum are wrong. Your challenge is to work out which ones are correct and which ones are wrong.

[Tajnesaie]

Hi and thanks a lot
but

at Re=9*10^4 , flow is incompressible

my meaning is in this flow Re=9*10^4 and fluid is incompressible.
Am i allowable to use channel with the largest size (4 equivalent) and verify this model with experimentally model?and why in the largest model convergence is rapidly?
thank you so much

[zandi]

Originally Posted by ghorrocks

I doubt it.

But honestly I get a bit sick of people asking questions like this. You cannot short-cut things and get the answer from some mystic on a forum. YOU HAVE TO DO A SENSITIVITY STUDY AND DETERMINE THIS FOR YOURSELF. In the process of doing the sensitivity study you will determine for yourself what are important parameters and what are not, and your analysis of the thing will be much better focussed on what is truly important in your simulation.

This is a much better approach than trying to simply ask on a forum and hope that you get the answer you want. Don't forget that half the posts on this forum are wrong. Your challenge is to work out which ones are correct and which ones are wrong.

Dear Horrocks
I copied it from my writings and my paper so thanks for advising.!
I have done mesh/grid study for my main case model and reached to an answer. but for another case and in the whole of problem I liked to know if y+ is important or not (I do think It is not, it seems).any way thank you
my Rey NO is changing along the spillway due to velocity if I'm right and the flow is noncompressible.
Again thanks for help