Turbulence modelling for bubble column in twoPhaseEulerFoam

hester · February 9, 2015, 14:39

Hello Everyone,

I'm currently investigating a bubble column (water-air, after Deen) with twoPhaseEulerFoam in OF 2.3.0. But I'm unsure about the turbulence modelling.

Should I define a turbulence model for both phases in const/turbulenceProperties.* for my RANS simulation? I found cases in this forum, where the dispersed phase was modelled as laminar and the continuous phase as turbulent (e.g. with kEpsilon). Is this dependent on the velocities or something else?

I also found the mixtureKEpsilon modell in twoPhaseEulerFoam which looks promising. Is this a better way to model such multiphase flows?

During my investigation I found that modelling the dispersed phase as laminar and the continuous phase as turbulent (kEpsilon) works but yields unphysical results in the end. When I use mixtureKEpsilon for both phases I get unphysical backflow at the outlet. But when I set turbulence off for the dispersed phase, the simulation works with mixtureKEpsilon and the results look good too. But the simulation output shows nether a k- nor an epsilon-equation beeing solved (also turbulence is still on for the continuous phase).

I hope someone could explain to me what the best way is (if there’s any)!

Regards,
hester

GerhardHolzinger · February 10, 2015, 08:58

If you include the liquid surface into your simulation, you are can use mixtureKEpsilon or you use no preconditioning for the turbulence solver. Using kEpsilon for the liquid phase, will bring the alpha.water=0 values above the liquid surface into your discretized equations. Preconditioning will fail on them.

I can't say much on the accuracy, but I played a bit around and noticed the behaviour described above .

hester · February 10, 2015, 12:42

Hello Gerhard,

thank you for your answer. Sorry for not beeing clear in my initial post. I`m not including the liquid surface in my simulation, because I found it difficult to get a working simulation this way. So the mixturekEpsilon shouldn`t be used in my case, right? Or including the liquid surface would be the best way? I`m new to multiphase simulation and still trying to figure it out .

regards,
hester

February 9, 2015, 14:39	Turbulence modelling for bubble column in twoPhaseEulerFoam	#1
hester Member Join Date: May 2014 Location: Germany Posts: 32 Rep Power: 12	Hello Everyone, I'm currently investigating a bubble column (water-air, after Deen) with twoPhaseEulerFoam in OF 2.3.0. But I'm unsure about the turbulence modelling. Should I define a turbulence model for both phases in const/turbulenceProperties.* for my RANS simulation? I found cases in this forum, where the dispersed phase was modelled as laminar and the continuous phase as turbulent (e.g. with kEpsilon). Is this dependent on the velocities or something else? I also found the mixtureKEpsilon modell in twoPhaseEulerFoam which looks promising. Is this a better way to model such multiphase flows? During my investigation I found that modelling the dispersed phase as laminar and the continuous phase as turbulent (kEpsilon) works but yields unphysical results in the end. When I use mixtureKEpsilon for both phases I get unphysical backflow at the outlet. But when I set turbulence off for the dispersed phase, the simulation works with mixtureKEpsilon and the results look good too. But the simulation output shows nether a k- nor an epsilon-equation beeing solved (also turbulence is still on for the continuous phase). I hope someone could explain to me what the best way is (if there’s any)! Regards, hester

February 10, 2015, 08:58		#2
GerhardHolzinger Senior Member Gerhard Holzinger Join Date: Feb 2012 Location: Austria Posts: 342 Rep Power: 28	If you include the liquid surface into your simulation, you are can use mixtureKEpsilon or you use no preconditioning for the turbulence solver. Using kEpsilon for the liquid phase, will bring the alpha.water=0 values above the liquid surface into your discretized equations. Preconditioning will fail on them. I can't say much on the accuracy, but I played a bit around and noticed the behaviour described above . ehsan_am86 and Dipsomaniac like this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
twoPhaseEulerFoam Simulations of bubble column	vishal3	OpenFOAM	1	July 7, 2014 05:12
Turbulence modelling	kluni	Main CFD Forum	6	December 1, 2009 17:46
k-ep turbulence modelling	allan thomson	Main CFD Forum	5	February 20, 2000 21:39
Question about Low Re turbulence modelling.	ghlee	Main CFD Forum	1	May 28, 1999 20:29
Multigrid Method in Turbulence Modelling	Bipin Lokhande	Main CFD Forum	1	November 11, 1998 20:20

February 10, 2015, 12:42		#3
hester Member Join Date: May 2014 Location: Germany Posts: 32 Rep Power: 12	Hello Gerhard, thank you for your answer. Sorry for not beeing clear in my initial post. I`m not including the liquid surface in my simulation, because I found it difficult to get a working simulation this way. So the mixturekEpsilon shouldn`t be used in my case, right? Or including the liquid surface would be the best way? I`m new to multiphase simulation and still trying to figure it out . regards, hester