[CFD-fellow]

Hi
I am modeling cavitaion on naca66(mod).I have got the results from fluent and after time-consuming convergency the results are ok as in Singhal paper. But when I implement the formulation of Fluent(I have attached the formulation) of Singhal model into CFX I get good convergency but results a bit different.
what is the k in the formulation(max(1,sqrt(k))?. In Singhal paper has mentioned it can be turbulent kinetic energy but I think there are something different in Fluent formulation.
so
1- What is k in the formulation?
2- Do u have any experience of Singhal model with udf to get the same results as its default

[ghost82]

Hi,
in Singhal et al cavitation model k is the turbulent kinetic energy.
You can read the full article of mathematical implementation of this model:
"Mathematical Basis and Validation of the Full Cavitation Model"

It is also available on baidu for free if you don't have access to the Journal of fluid engineering, just search for the title on google.

I don't know cfx (never used), I wrote some udfs for different cavitation models in fluent: my advice is to correctly implement the maths (obviously ) and to understand what are the different values.
One of my errors in fluent was related to vapor/volume fractions and to plot the correct vapor volume fractions: Singhal et al model is built with the vapor mass fraction, other models (Zwart-Gerber-Belamri and Schnerr and Sauer) with the vapor volume fraction.
In addition, Singhal et al model takes into account also the incondensable mass fraction, which in fluent is plotted with the secondary (vapor) volume fraction.
Writing an udf for the Singhal model is not so difficult, post in the udf section what you have written and what are your problems.

Daniele

[CFD-fellow]

Thank you Daniele
I have the paper and it has been cited in so many papers. Unfortunately in a lot of papers, they have presented different formulation from Singhal and reference into one paper.Fluent is one of them, it has max(1,sqrt(k)) but nowhere in that paper such a term is not presented in its formulation.
I have checked the math but I think the problem is with numerical procedure as Fluent user guide says it has different numerical procedure. And there should be a reason that it has poor convergency!do u know why?

[ghost82]

Yes, I understand what you mean.
The original model is that on the article (square root (k), not max term); I think the "max" term is introduced for stability purposes.
Take into account also that this model is a semi-empirical model, as it has tuning constants Fvap and Fcond, which you can change as you want.