[kobelak]

Hi All,

I have been using multiphaseEulerFoam (OF 2.3.1) to study the evolution of the bubble size distribution in a bubble column. I recently found out that the solver may be surprisingly sensitive to the names given to phases.

I am working on a bubble column case where I defined a water phase and n=7 air phases. The air phases have the same characteristics (rho,nu,...) but different diameters and different names then. I named them as "air01", "air02", "air03",....."air07" with increasing diameter.
I ran my simulation. Phase "air05" showed weird values of velocity, two or three times higher then those ones of the other phases: a sort of "outlier" behaviour. After some struggling, I ran the simulation with the SAME set-up, just changing the name "air05" in "airWhatever": the values of velocity of "airWhatever" were normal!

To investigate the reason, I looked at the log files of the two simulations. I noticed that some differences already arose when the drag models for the phases were selected ("Selecting dragModel for phase...."): I used in both simulations a type "blended" for each (air0X water) interface, but the drag model for the different phases were called in a different order.
Therefore I changed the code in order to output the variable "dragCoeffi" (defined in pEqn.H) which stores the drag coefficients for a given phase.
Running again the two simulations, I realized that the solver calculated the same drag coefficients, but they were differently associated to the phases . In particular, in the simulation with "airWhatever" phase, the order was correct, while in the other simulation, the coefficient of phase "air01" was assigned to "air05" and viceversa. Then I could assume that this was the reason of the wrong velocites of "air05" in the first simulation.

To confirm the speculation, I also ran the damBreak4phase tutorial. I changed the name "mercury" in "smercury". Again, the drag Models were called in a different order and the log files presented slightly different values of time steps, Courant number...even though I would have expected the simulations being identical.

The questions are then the following: where am I wrong? Has anyone experienced anything similar? Is it a sort of bug?

Looking forward for some help. Thanks in advance.

Cheers,
Danilo