[AlmostSurelyRob]

Dear All,

I would like to invite some comments on the ability of interFoam to simulate stratified flow in a pipeline.

My objective is to use interFoam together with cyclic boundary conditions. To validate my results I specify the flow rate and check the pressure gradient. There is a number of semi-theoretical correlations like Taitel & Dukler (1976) so I can compare with these results.

I've done mesh sensitivity studies, but in the range of values I tried there was no significant effect on pressure drop or shape of the interface.

There is however, a big change when I play with turbulence models. I have tried various models: kEpsilon, kOmega, LaunderSharmaKE, LamBremhorstKE and some very preliminary calculations with LES on a different mesh.

Convergence, interface shape, pressure drop, velocity profiles all seem to be very sensitive to these choices but to this day I haven't really got any satisfactory results. There are some "success stories" with Launder-Sharma model in the literature for channel flow.

I can give more details on the setting and results if anybody is interested. I am probably overlooking something so I would value your comments and criticism. Please, let me know if you have any thoughts or experience with similar problems.

[AlmostSurelyRob]

Sorry for putting a post again, but there is just one remarkably odd thing which I'd like to report. I've just imposed a zero mass flow rate for both phases and initialised the domain with zero velocity. With laminar model I immediately get large velocity vectors in transverse direction. All of these artefacts appear at the interface.