Hi to everybody,

I carried out a liquid sloshing CFD study with Fluent 6.2 VOF model, and I have a few questions about the model implemented in the solver.

Well, I know that the VOF model in Fluent is actually a variable-density approximation of the "real" VOF model by Hirt and Nichols and I also know that Flow Science (Hirt is one of the founders, if I'm not wrong) claims Flow 3D to be one of the few codes (the only one?) with all the three essential features needed to properly model free surfaces (-a scheme to describe the shape and location of a surface, -an algorithm is required to evolve the shape and location with time, -free surface boundary conditions at the surface).

I'm actually going into details of the problem these days and trying to understand whenever and however this approximation will affect the results of the simulations. So my questions are:

1) Has anyone ever faced such a situation? I would really appreciate if someone would share his own experience and deal it with mine...

2) I think that Fluent's VOF model lacking feature is the ability to apply a boundary condition at the interface between the phases (velocity gradients, etc...), I'm right?

3) A fine mesh can improve the model and make the lack neglectable?

4) When body forces' effects decrease and surface tensions, wall adhesion become pre-eminent is the variable-density model even worse?

5) What codes (both commercial and freeware) have the best multiphase models implemented (VOF or even other techniques)?

Thank you in advance for any contribution,

Edi.

For the question number 5), we recently wrote a short review paper about multiphase models, even though it is not evaluations of commerical codes.

take a look at the paper, http://math.uci.edu/~lowengrb/RESEAR.../enc_final.pdf

Thank you very much. I just took a look at it. Absolutely interesting.

Edi.

Hi Dr. Kim,

It is indeed an informative review paper.

In your classifications, you might consider to include another quite different approach: the lattice Boltzmann method (LBM). Sundaresan¡¯s group are working on the LBM. They made some comparisons with Tryggvason¡¯s front tracking method, see International Journal of Multiphase Flow 29: 109-116 (2003).

Despite the difference in the interface description (either tracking or capturing), there is a direct ideological relation among these methods in the sense that the governing systems are solved on Eulerian grids. The difficulties of multiphase flow simulations are due in part to the singularities (discontinuous pressure and normal derivatives of velocity) at a time-dependent free boundary. At present, the most popular way to deal with the singularities at the interface, particularly in the finite difference context, is the regularization of the delta-function using trigonometric approximation. The interface is no longer sharp but has a finite thickness over a few grid cells even though the interface might be tracked explicitly. However, such an approach results in costly computations because it is at best first-order accurate (in space) and it inherently reduces the accuracy of any formally higher-order scheme when the interfacial force becomes significant.

Therefore, there might be a different way to categorize the interfacial modelling in terms of finite or sharp interface (i.e., low-order or high-order methods). We have also been working on the so-called sharp interface method:

J Comput Phys 187: 255-273 (2003). Int J Numer Meth Fluids 45: 1-19 (2004). Int J Numer Meth Fluids 48: 455-466 (2005).

Hope you may find this information useful.

Congratulations and best regards,

T Chen

Thanks for the comments. The original review paper contains Lattice Gas and Lattice Boltzmann sections, but the requirement of page numbers (Encyclopedia of Math. Phys. by Elsevier) forced us to cut those sections and focus on continumm description of two phase-fluid flows. The longer and detailed version is in preparation now.