Help with the discretization of the source term!
 

Now I am doing numerical research on the turbulence, using FVM and collocated grids arrangement and Low Re turbulence model. Could you tell me how to discreteize the source term for u- v- k- epslion- equation with non-uniform grids.
I've tried a lot of method, but I can't get correct solution. I don't know why?
I'am very grateful for your advice!

which method did you use.
Finite volume method or finite difference method.
in the finite volume method, it is much easier.

Finite volume method!
I don't know how to deal with the source term correctly!
With non-uniform grids

Dear friends:
Since the FVM is used. the grid points are always placed at the center of the control volume.
We only need to evalue the derivative at the interface by iterpolation. we can caculate all the source term.
Could you please detail which source term trouble you.
Regards

The u- and v- source term!
Actually, I've discretized the source term using FVM and coded it. But the results became divergent sudenly during iteration process. I mean in this step the results is convergent, while it will become divergent next step.
I don't know why?
Should I limit some parameters?
If so, which?
Thank you!
Long for hearing your advice.

which alogirithm did you use. Have you validate your code?
If your code is correct, i think you may try to lower the realxation parameter for u and v.
BY the way, which method did you use to solve the linear algrebra equation.

SIMPLER, collocated grid, non-uniform grid
Yes, I've validated my code: lid-driven cavity flow, natural convection in square cavity and so on.
I use ADI to solve the linear algrebra equation.
If I want to simulate turbulent natural convection in a square cavity, the best relaxation factor ? Could you give me some advice?
Thank you!