[tengf689]

Hi all:
I have a question about CFX user fortran:
my case is a pipe flow (streamwise ：x, x=0,100mm). I wanna extract velocity at Plane A（x=50mm） and do some modification as inlet boundary conditions (at every time step). How can I do it with user fortran ?
Now i can only use fortran program to deal with datas in one plane. But how to use it in two planes ???? I need some help or some information . Thanks

[ghorrocks]

You might be able to do this in CEL and not require fortran at all. Can you describe what you are trying to do?

Also note that this sort of configuration (linking inlet parameters to a calculated value in the middle of the flow) is not physically realistic and is therefore likely to have convergence difficulties. Hopefully there is a better way of doing it.

[tengf689]

Originally Posted by ghorrocks

You might be able to do this in CEL and not require fortran at all. Can you describe what you are trying to do?

Also note that this sort of configuration (linking inlet parameters to a calculated value in the middle of the flow) is not physically realistic and is therefore likely to have convergence difficulties. Hopefully there is a better way of doing it.

I wanna compute flow in a pipe using LES . As known, it is difficult to give inlet BC's . I read lots of papers and wanna try one of those methods : extract datas at middle of pipe with some modification, and give these datas to inlet at every time step.
How can I use CEL to make it ?? And do you need to view .cfx file ?
Thanks a lot

[ghorrocks]

Can you do it using translational periodic boundaries? Then the coupling from the outlet to the inlet is handled automatically, and you can include things like flow rates and pressure drops. If you want to do anything more complicated you can use source terms.

Translational periodic boundaries are just a standard feature of CFX and do not require any fortran so will be much simpler to implement.

[tengf689]

Originally Posted by ghorrocks

Can you do it using translational periodic boundaries? Then the coupling from the outlet to the inlet is handled automatically, and you can include things like flow rates and pressure drops. If you want to do anything more complicated you can use source terms.

Translational periodic boundaries are just a standard feature of CFX and do not require any fortran so will be much simpler to implement.

I need to extract velocity datas at mid plane and I need to rescale these datas by using a factor ‘alpha’（alpha=0,1）. So i can not use translational periodic boundaries directly.
And can I make it by using source terms?? I have not used it before.

[ghorrocks]

Please explain why you need the mid plane velocity, what alpha does and how that scales the input velocity.

The details of what you intend to do make a big difference in how you implement it, so you need to explain what you are intending to do.

[tengf689]

Originally Posted by ghorrocks

Please explain why you need the mid plane velocity, what alpha does and how that scales the input velocity.

The details of what you intend to do make a big difference in how you implement it, so you need to explain what you are intending to do.

the result of LES is sensitive to the inlet boundary conditions, I need to give more realistic BC's . and i think ,it is good idea to extract the velocity fluctuations at mid plane, and give these fluctuation terms to the inlet .


reference: Generation of Turbulent Inflow Data for Spatially-Developing Boundary Layer Simulations.
the factor in this paper is 'weighting function W(n)'
http://www.iawe.org/Proceedings/CWE2006/MA3-04.pdf

[ghorrocks]

It will be a tricky piece of fortran to implement what these authors did. You are going to have to match up nodes on the plane to nodes on the inlet, extract the fluctuating component and adjust the mean flow to get the correct bulk flow. I am not going to be able to help you much with this as I have never implemented something like that.

I do not understand your first post, I cannot see where you combine the results of two planes in this approach.

The periodic translational boundary approach I mentioned previously is slightly different to this, but very similar in concept. It is far easier to implement in CFX as it will not require any fortran, just a source term to drive the bulk flow at the correct velocity. A suspect this approach would give you results virtually indistinguishable from the approach used by the authors you linked to, when done properly.

[tengf689]

Originally Posted by ghorrocks

It will be a tricky piece of fortran to implement what these authors did. You are going to have to match up nodes on the plane to nodes on the inlet, extract the fluctuating component and adjust the mean flow to get the correct bulk flow. I am not going to be able to help you much with this as I have never implemented something like that.

I do not understand your first post, I cannot see where you combine the results of two planes in this approach.

The periodic translational boundary approach I mentioned previously is slightly different to this, but very similar in concept. It is far easier to implement in CFX as it will not require any fortran, just a source term to drive the bulk flow at the correct velocity. A suspect this approach would give you results virtually indistinguishable from the approach used by the authors you linked to, when done properly.

Thank you for your help ! I will keep trying ,and I hope I can make it soon . Best wishes!!!