Confluence and mixing of two rivers

Magicano · May 21, 2012, 06:19

Hello,

I want to study the mixing of two rivers which have different velocities and densities.

I made a 3D mesh with IcemCFD : rectangular T junction between two channels. I assume that flows are turbulent, and I guess that gravity has to be enabled to view the density effects

Does anyone know how could I parameter Fluent, which model I could use, if I have to use two-phases model ?
Is free-surface useful to view the mixing phenomenon ?

I thank you all for your help !

johnp · May 22, 2012, 16:10

Hi, I'm not sure I can completely answer your questions. But here's a start:

1. flow will almost certainly be turbulent. The only laminar open channel flow that comes to mind is a thin sheet flow.

2. If you know the location of the free-surface, you can include it as a symmetry boundary. This is the computationally least expensive option. However, this may limit your ability to predict some 3D flow structures, e.g. secondary currents.

3. If you do not know the free-surface location and want to use a symmetry boundary condition, you can model your reach with a 1D hydrodynamic model, e.g. HEC-RAS. Keep in mind that a 1D code will give a constant water surface elevation at each cross-section, so you cannot reproduce super elevation (if your channel is curved).

4. The VOF multiphase model can be used to determine the free-surface location. This is computationally much more expensive than the fixed-lid option.

On a side note, have you looked at the papers by S. Lane and colleagues on CFD modeling of confluences? These may be useful. Good luck and it's nice to see someone else looking at river flows with FLUENT,

john

Magicano · May 22, 2012, 19:47

Hi,

thanks for answering,
indeed I didn't read any of their papers, I will for sure, thank you.

The computational cost won't be a problem. The fact is that the purpose of my study is to describe the confluence phenomenon, as realistic as it is, and identify its main phenomena. I use as an experimental validation the confluence between the Rio Negro and the Amazon, a very visual example because of the obvious difference of colour between the two rivers. (perfectly visible on google earth !)

So I try to find on simulations the same order of length needed by the river to be fully mixed. I assume that density difference (heavier river will fall down the lighter one) is as important as the shear or the molecular diffusion (which is insignificant) in the process of mixing, that is why I think that we need to give a 3D-description.

When you talk about VOF, I guess that for you, phase 1 is the air and phase 2 is water, but is it possible, in order to focus on mixing phenomenon, to use VOF model with water 1 at density rho1 as phase 1, and water 2, coming from channel 2 at density rho2 as phase 2 ? and then calculate the volume of fluid of phase 1 in the mixed channel as an indicator of mixing ?

In this case, symmetry condition would be used for the free surface, unless I try a 3 phases model...

Anyway, because I just want to globally identify the mixing process and the part taken by velocity difference and density difference, I doubt that a precise description of the free surface level would be useful, whereas secondary currents seem to be. What do you think about it ?

johnp · May 23, 2012, 10:50

Quote:

Originally Posted by Magicano

When you talk about VOF, I guess that for you, phase 1 is the air and phase 2 is water, but is it possible, in order to focus on mixing phenomenon, to use VOF model with water 1 at density rho1 as phase 1, and water 2, coming from channel 2 at density rho2 as phase 2 ? and then calculate the volume of fluid of phase 1 in the mixed channel as an indicator of mixing ?

Yes, typically with VOF modeling of open channel flows, one phase is liquid and the other is gas. You are looking at the mixing of two liquids, though. Why not just specify two fluids (liquids) with different density and viscosity?

Quote:

Originally Posted by Magicano

Anyway, because I just want to globally identify the mixing process and the part taken by velocity difference and density difference, I doubt that a precise description of the free surface level would be useful, whereas secondary currents seem to be. What do you think about it ?

if this is the case, I would stay away from the VOF. This sounds like an interesting problem. The papers I referred to previously deal with confluences in gravel streams. Quite a difference in scale!

stephanc · June 29, 2012, 04:18

Hello,
I came upon this thread as I am also looking to model river flow in confluences with use of FLuent. My focus also lies on the specific flow fields in the confluence.
Therefore it would be great to exchange some experience and knowledge.
Where you seem to have density currents as a special condition , i will try to look into the effects of vegetation on the stream patterns.

I am new to fluent and cfd modelling so i expect this to be a challenging first step.
Any advice, link or paper would be of great use!
Thanks in advance!

pedroxramos · August 5, 2014, 06:34

Hello.

Do you know some papers about this?

May 21, 2012, 06:19	Confluence and mixing of two rivers	#1
Magicano New Member Join Date: Apr 2012 Posts: 3 Rep Power: 14	Hello, I want to study the mixing of two rivers which have different velocities and densities. I made a 3D mesh with IcemCFD : rectangular T junction between two channels. I assume that flows are turbulent, and I guess that gravity has to be enabled to view the density effects Does anyone know how could I parameter Fluent, which model I could use, if I have to use two-phases model ? Is free-surface useful to view the mixing phenomenon ? I thank you all for your help !

May 22, 2012, 16:10		#2
johnp Member john Join Date: Aug 2010 Location: pasadena, va, usa Posts: 36 Rep Power: 15	Hi, I'm not sure I can completely answer your questions. But here's a start: 1. flow will almost certainly be turbulent. The only laminar open channel flow that comes to mind is a thin sheet flow. 2. If you know the location of the free-surface, you can include it as a symmetry boundary. This is the computationally least expensive option. However, this may limit your ability to predict some 3D flow structures, e.g. secondary currents. 3. If you do not know the free-surface location and want to use a symmetry boundary condition, you can model your reach with a 1D hydrodynamic model, e.g. HEC-RAS. Keep in mind that a 1D code will give a constant water surface elevation at each cross-section, so you cannot reproduce super elevation (if your channel is curved). 4. The VOF multiphase model can be used to determine the free-surface location. This is computationally much more expensive than the fixed-lid option. On a side note, have you looked at the papers by S. Lane and colleagues on CFD modeling of confluences? These may be useful. Good luck and it's nice to see someone else looking at river flows with FLUENT, john

May 22, 2012, 19:47		#3
Magicano New Member Join Date: Apr 2012 Posts: 3 Rep Power: 14	Hi, thanks for answering, indeed I didn't read any of their papers, I will for sure, thank you. The computational cost won't be a problem. The fact is that the purpose of my study is to describe the confluence phenomenon, as realistic as it is, and identify its main phenomena. I use as an experimental validation the confluence between the Rio Negro and the Amazon, a very visual example because of the obvious difference of colour between the two rivers. (perfectly visible on google earth !) So I try to find on simulations the same order of length needed by the river to be fully mixed. I assume that density difference (heavier river will fall down the lighter one) is as important as the shear or the molecular diffusion (which is insignificant) in the process of mixing, that is why I think that we need to give a 3D-description. When you talk about VOF, I guess that for you, phase 1 is the air and phase 2 is water, but is it possible, in order to focus on mixing phenomenon, to use VOF model with water 1 at density rho1 as phase 1, and water 2, coming from channel 2 at density rho2 as phase 2 ? and then calculate the volume of fluid of phase 1 in the mixed channel as an indicator of mixing ? In this case, symmetry condition would be used for the free surface, unless I try a 3 phases model... Anyway, because I just want to globally identify the mixing process and the part taken by velocity difference and density difference, I doubt that a precise description of the free surface level would be useful, whereas secondary currents seem to be. What do you think about it ?

June 29, 2012, 04:18		#5
stephanc New Member Join Date: Jun 2012 Posts: 1 Rep Power: 0	Hello, I came upon this thread as I am also looking to model river flow in confluences with use of FLuent. My focus also lies on the specific flow fields in the confluence. Therefore it would be great to exchange some experience and knowledge. Where you seem to have density currents as a special condition , i will try to look into the effects of vegetation on the stream patterns. I am new to fluent and cfd modelling so i expect this to be a challenging first step. Any advice, link or paper would be of great use! Thanks in advance!

August 5, 2014, 06:34		#6
pedroxramos Member Pedro Ramos Join Date: Mar 2012 Location: Belgium Posts: 81 Rep Power: 14	Hello. Do you know some papers about this?