[happy]

could any one tell me how I can create Neumann conditions for the velocity , zeta, k & T? i mean make the deravative of them with respect to distance equal to 0? please do not tell me use symmetry condition because it is different.
Regards

[ghorrocks]

Have a look in the documentation for symmetry planes and outlets/openings. They all use different combinations of dV/dN=0.

[happy]

Originally Posted by happy

could any one tell me how I can create Neumann conditions for the velocity , zeta, k & T? i mean make the deravative of them with respect to distance equal to 0? please do not tell me use symmetry condition because it is different.
Regards

In reality, I want to set the deravative of my velocity and temp. at one of my b.C to zero. Does I need to use user_getvar routine? Advise me please
Regards

[ghorrocks]

At an inlet or outlet?

[happy]

Originally Posted by ghorrocks

At an inlet or outlet?

hi Glenn

At the outlet.
Before, now I make this boundary as opening with hydodynamic pressure expresion, but the results is not good enough. so i'm trying to changed to outlet with dv/dn=0.
Regards

[ghorrocks]

I see. For CFX the values used are convected from the internal nodes. Why isn't that good enough? I would have thought that would be more accurate than imposing dV/dN=0.

Before you spend weeks developing your own boundary condition (which is probably less accurate than the one already in CFX, so I do not see how it will help you), the normal response when BCs are causing errors in the simulation is to move the BCs further away from the region of interest so the error from the BC causes less error in the region of interest. Why not simply move the boundaries further out?

[happy]

Originally Posted by ghorrocks

I see. For CFX the values used are convected from the internal nodes. Why isn't that good enough? I would have thought that would be more accurate than imposing dV/dN=0.

Before you spend weeks developing your own boundary condition (which is probably less accurate than the one already in CFX, so I do not see how it will help you), the normal response when BCs are causing errors in the simulation is to move the BCs further away from the region of interest so the error from the BC causes less error in the region of interest. Why not simply move the boundaries further out?

yep,,, that is what I'm working with now;however, I read and attempt to find more realistic B.Cs.
thanks for your input.
Regards

[happy]

[QUOTE=happy;349755]yep,,, that is what I'm working with now;however, I read and attempt to find more realistic B.Cs.
By the way, the top of my domain is also my interst place to see the effect of many parameters on the velocity at the top.
So glenn, what you think the solution for my case.
as I read the this website I thing I need to creare routine with fortran compiler. I searched into ansys.com, but the threads overthere is for create expression to calculate the drevative w.r.t. time only into post stage. i have already three references use the same B.C at the domain top.
Any input can help
My best regards

[ghorrocks]

Yes, if you really want to use Neumann BCs then you better code it in fortran.

I still have no idea what you are modelling and why you need this BC, so my general advice remains that moving the BCs further back from the region of interest and using the existing BCs is far easier and more accurate.