Have a look here.

http://aerojet.engr.ucdavis.edu/fluenthelp/html/udf/node60.htm

I quote the Fluent User's Guide: "There is no diffusion flux across a symmetry plane: the normal gradients of all flow variables are therefore zero at the symmetry plane." We can interpret it as that the symmetry condition takes values that are obtained by solving the transport equations inside your domain, and reflects these outside the domain to form a symmetrical flow. So you cannot set the dissipation rate values at the symmetry boundary since the normal gradient of the dissipation rate would not be zero then. You may rather try the pressure outlet boundary condition. I am not an expert in open channel flows so take this only as a suggestion. If I remember well, the usual way how to model this kind of problem is to use multiphase VOF or Eulerian but I understand that that requires more computational power.

DEFINE_SOURCE works as you described - this macro defines the source term that appears in the transport equations. You cannot use it to set a specific value of the dissipation rate.

I agree with Mr. Kalal. You expressed that you are modelling an open channel flows with the surface as symmetry boundary condition. How the free surface can act as a  symmetry boundary. If your are looking for boundary effects on the free surface then you may have to go for two phase modelling with air or gas as other fluid.

 Dear Concli,

I read the DEFINE_DPM_BC just now, but I guess that it's not applicable for fluid, because fluid is assumed to be continuous. I want to define a reflection  boundary condition for turbulence characteristics(parameters) except for dissipation rate epsilong.

Can I define a new kind of boundary condition by UDF, so that I can define the reflection way by myself? 

Thank you!

Dear Jaafar,

I'm glad that you are interested in it. I want to give the expression of dissipation rate at a symmetry boundary(water surface), proportional to 1/y, which is deduced from its simplified transport equation. 

Dear Mr. Kalal,

Thank you for your answer. How can I set the dissipation rate values(expression) at a certain boundary,which can act as a symmetry plane for all turbulent variables except for  dissipation rate?

Dear Jaafar,

Yes, you got it. I want to simulate the damping effect of water surface on turbulence. Can I realize it by commercial software?

Dear Wu,

The symmetry condition states that there is no gradient of any quantity across the boundary and that the velocity component normal to the boundary is zero. In other words, there is no diffusion transport across the boundary (due to zero gradient) and there is also no convective transport across the boundary (due to the zero velocity). So, if you set any fixed values of the dissipation on the symmetry plane, it won't have any effect on the solution since the information on the boundary does not have any way how to spread inside the domain (neither diffusion nor convection). I would recommend you to read a book about CFD where you could see the whole derivation of the linear equations that are actually solved by a CFD code. A very good book that is easy to understand is "An Introduction to Computational Fluid Dynamics: The Finite Volume Method" by Versteeg and Malalasekera.

As Abdul suggested, check some studies on free-surface flows that have been already done, there are many of them. I do not see the reason why to set a dissipation rate value on the liquid surface, I would expect that to be part of your solution.

Dear Kalal,

Thank you very much for your suggestions!

hello

you can use '' DIFINE_ADJUST'' , it offers for you the possibility to impose the  Derychely condition at the semytery plan, this command adjusts in each iteration the value of the dissipation rate in  this plan , you can consult the ANSYS-Fluent help for information about this udf.