Hi everyone,

I am trying to modify the turbulent viscosity that goes into the momentum equation without effecting its value in the turbulence equation. It has been suggested by support that I should write the diffusion terms as a source term to cancel the existing terms and implement my values. For example, if I want to change the viscosity to SCALAR*vist then I should do the follwoing,

(S1U-S2U*UP)*(SCALAR - 1)

I have been trying to implement the above without any success. The solution diverges in any attempt to do so.

Has anyone done anything similar to the above or can you suggest anything else. Re-writing the turbulence as scalar does'nt work( I wasted 3 months trying it only to be told by support that it cant be done... )

Please help me solve this...

me too, I am trying to modify the turbulent viscosity in k-epsilon model , I tried an old version of Star-CD which contains Varcmu.f ( variable C_mu)soubroutine, but actually it does not work at all. I am planning to call the S.CD provider. BUt as u said it seem to be long story(-) cheers

I think that it is dangerous to modify the source term S1U that way, because the source term should have also the contribution of the pressure gradient. Ask support to send you the expression they use for evaluating the source term due to the turbulent viscosity approximation i.e. div [ mu_t * grad (Vel) ]. They will have to take a look at the source code, but it should be feasible. Ask them for the explicit formulation. That will allow you to correct the source term in a an explicit manner (which should be stable) and disable the source term at the same time by multiplying the expression by -1 and adding it to S1U, etc...

Hope this helps...

Hi 4xF,

thank you for your response. Anctually we did go and see someoen at STAR about it, and, armmmm, well it seems like it was'nt an easy job... They gave some kind of 'deferred correction' interms of nodal velocity (Up) and it did'nt work

Sheila

Can you drop me a page's write up how you want to modify the viscosity, i do have a steady state solver, if it is possible i will hard code that in my solver and give you the exe to work on with. You might have to wait 10-15 days as i am in process of transfering code from my old solver (that has k-e and k-omega models) to new code. So if you write it up in sufficient details (what you are exactly trying to do), your problem might be solved otherwise.

my email is

zxaar@yahoo.com

did you get the reply email i wrote to you ????