With the aim of implementing a new model, I have the necessity to modify the terms of the Epsilon transport equation of the realizable K-Epsilon model. More specifically I have to delete the Epsilon transport equation through the substraction of all epsilon source terms in the UDF, through the Epsilon source definition, and then to sobstitute them with others. I've found on the Fluent User's Guide the theoretical definition of such a transport equation, but I was wondering about how the "rate of change" (d(rho*epsilon)/dt) and the "convection term" are actually defined in Fluent in order to delete them via UDF. For the "rate of change" I know that fluent uses kind of time-step derivative, but I don't know how it is expressed.
Hello Sir,
Firstly, your idea is brilliant.
Though I ain't an expertise in coding, with my knowledge and experience I am suggesting you the following logics:
1. Eliminate a variable by simply making it zero. (Since each term in Epsilon transport equation contains epsilon term, by making sure that epsilon is zero at each iteration, probably using UDF or during initialization or making generation. production term of epsilon to zero etc.,)
2. Then you can add additional user defined scalar (UDS) variables - and its corresponding transport equation of your choice.
Are you stuck with fluent? Open source codes like OpenFOAM are often better suited for implementing new methods. Especially If you know the equation already you can write it almost as text....
https://www.sharcnet.ca/Software/Fluent6/html/ug/node993.htm#sec-time-theory
Hello Sir,
If you are still thinking of using Fluent for your computation and UDFs, check the above link for the definition of temporal discretization used in the fluent code.[For both implicit and explicit (both 1st and 2nd order discretization) definitions are clearly mentioned in the above link.]
Hope that is what you are looking for!
Thank you for all your kind answers...and sorry if I'm replying so in late. I'm coming back right now on this project. I think that one possible path for representing/deleting the "rate of change" term is recurring to the finite differences. I still have to understand if the same method is correct for convection term. I'm perfectly aware that open-foam is more "user-friendly" when you have to define a model by your own. In fact I will start very soon using it..still Fluent has really positve aspects which I really appreciate. I.e. in my field of study it seems to be more robust...
What is going to be an additional difficulty, is that, as far as I know, even if you are running a steady simulation (as the ones I'm actually doing), FLUENT does anyway calculate a temporal term of the transport equation. I would really like to get rid of such a term and then replace it with one proper for the k-omega approach...but, first of all, I should know how it is computed. Do you have any clue what this "term" looks like?
For both implicit and explicit formulation, for steady case, in fluent,
"delT is proportional to CFL number".
Does that mean that by simply writing a souce term representing negative of the epsilon equation we can make it zero? Did this work for you in steady case?
This is partially true...I will better explain myself. This path can be valid if you want to replace just some parts of the transport equation (i.e. the source terms). The main problem is that in Fluent, EVEN IF YOU ARE SIMULATING A STEADY CASE, there are some terms and calculations (iteration dependent) which are not accessible to the user and are computed internally by the solver. Right now, this makes the complete subtstitution of a transport equation really hard, if not impossible on Fluent. For this reason, if you want to replace a complete transport equation, it is better to rely on OpenFoam.
- Badges
- Science topic