[basecase]

Hi all - I'd be grateful if anyone could share their knowledge/understanding of application of PhaseScalarTransport (PST) together with dynamicMesh. I've read the posts for PST and it seems to present challenges in several areas.

I appreciate earlier posts by Randolph and others on related topics.

I'm running OF9 - I have a case involving motion (dynamicMesh) with two immiscible fluids (oil and water) together with local introduction of a (third) water miscible fluid (I'll call it brine) to the aqueous phase - all contained in a "sloshing" tank. The third fluid is should remain in the lower aqueous phase and disperse therin. OF Solver is interFoam and the PST is applied as function contained in controlDict. The models runs smoothly and motion of container and fluids look good. However - a simple mass balance on the scalar (solute/brine in the aqueous phase) or integration on the scalar show that it is not realistically conserved - it decays quite quickly with time.

I've adjusted all control variables and settings across many sensible ranges (and nonsensical ranges - as one does) and OF runs sweetly every time - but the scalar always decays to a greater or lesser - completely unacceptable extent. I've scoured the code (with my limited knowledge) and found no hint that the PST function may not be compatible with dynamicMesh - but I guess that is a suspicion. There is a dearth of examples for PST although those for scalarTransport - cited as a reference point for PST in the online source code docs, are available.

I was slightly uneasy with the OF method used to normalise the scalar in the single constrained (water) phase - not because I thought it wouldn't give a result - more that the result obtained might not properly reflect scalar transport in that constrained phase.

Short story for my questions - 1) am I expecting too much of OF 9? 2) is there an inherent weakness or incompatibility of OF9 for PST with dynamicMesh? 3) Has anyone else successfully run such PST+dynamicMesh models in OF9? 4) If there is some inherent weakness in the OF 9 code - is this addressed in more recent versions of OF?

From reading the challenges of phaseScalarTransport in these forums and elsewhere - I feel it would be a major service to the community (and an even greater service to me ) if some kind, skilled person could produce as few nice examples of PST in challenging simulations.

Thanks for reading this far. I would appreciate to hear not only of success stories but from those who have also failed to find success with PST function. For sure I learn at least as much from my failures as from my (few) successes.

To those that give their time to development of OF code and to those helping we mere mortals to understand and apply that code - my thanks, it is hugely appreciated.


Seeking enlightenment - always

basecase

[basecase]

A conversation with myself - nothing new there




I moved on to run the (same) OF9 models in OF10 and thankfully these seem to run smoothly and I'm happy to report, conserve the scalar in the constrained system described. As far as I can tell (with my limited coding skills) the problem in OF9 is associated with the interpretation of the field word definition in the function section of controlDict - it seems the code will complain if the word field is not present in that function section - but when it is, the code cares not what the field label is. A scalar field of banana will work as will no value assigned at all (a blank). The same is not so for OF10.




I've not delved further into the code other than 'diffing' the .C and .H files for OF9 vs. OF10 which points the way to further investigation - had I the time and cared not that I'd lose the will to live. I would welcome a more skilled person than I to provide enlightenment.




Short story - probably well advised to take the easy route if you're running models like these and use OF10 rather than OF9.



Despite that all looks rosy for now in OF10 - I am still uncomfortable with the approach used to constrain the scalar - based on this I just can't quite convince myself the dispersion of the scalar is realistically (ok - wrong word - this is a model after all) modelled. Again those brighter than I are invited to put me straight.


I'd also welcome those using 'the other' OF to share experiences of PST together with dynamic mesh based models.


Thanks - and Happy Christmas to all.