Learning Forum

[prebenjs]

I've read the Guides but I can't really give a "clear winner" between the implicit and the explicit VOF scheme, for volume fraction and interface tracking in a structure. (transient)

It is mentioned that explicit provides greater accuracy, but this can be sorted out by using a higher order transient formulation with the implicit and thus it should give a comparable accuracy to the explicit. What are the pros and cons of these, when I want to track the interface with time as well ?

Furthermore, the geo-reconstruct is mentioned as being favourable for coarse mesh. But it doesn't say anything about when modified HRIC or Compressive scheme is favourable (the ones present in the implicit scheme). If mesh is adequately fine, should these method give a comparable picture ?

It should be mentioned that surface tension is involved.

[Rob]

Welcome to multiphase modelling!
From memory geo-recon is the most accurate, but requires a well resolved mesh and smaller time step. This costs cpu effort and time. Compressive is marginally less accurate, but may allow a larger time step, so takes less time to run. Both are the correct answer, the question you then need to consider is how quickly you want the results, and how much cpu have you got?

[prebenjs]

The model should be as computational efficient as possible, without losing "considerable" amount of accuracy, regarding the fraction of the phases. Thus I guess it should be reasonable to go for implicit and disregarding the geo-recon, although surface tracking is quite important as well.

[Rob]

It's a question of understanding the trade offs and purpose of the simulation. We want "good enough" for the task. So for a screening check I may use a fairly coarse mesh and make a few assumptions: for a detailed design I would include far more physics and use a finer mesh.

[DrAmine]

Implicit VOF with compressive scheme and moderate CFL number around 5 to 10 is the way to got. Monitor maximum velocity and assess the need to limit it. We ran cases with even very aggressive cfk for the sake of performance.

[DrAmine]

HRIC is diffusive and we do not use it as we have much better schemes in Ansys CFD solutions.