CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Why average volume fraction greater than 1? (attached files) (https://www.cfd-online.com/Forums/cfx/109661-why-average-volume-fraction-greater-than-1-attached-files.html)

TitusOfei November 23, 2012 02:55

Why average volume fraction greater than 1? (attached files)
 
3 Attachment(s)
Hi all CFX users

I have encountered some CFX post results which seem to me unrealistic.
I am simulating two-phase flow (particles-liquid) through a horizontal concentric annular wellbore at steady state condition. I am using 'particle transport solids' as the morphology for the particle

The particles are injected into the annulus at a constant rate and random. I also considered both one-way and full coupling to model the transport of the particle.

I want to determine the concentration of particles in the annulus. So I created a plane and polyline at the axial center of the annular section. I used the polyline to create a chart of average cuttings(particle) concentration verses axial distance. The results are over unity (1) which is unrealistic. What could be the error here?

I have attached more plots to see how the particles are formed along the axial length of the wellbore.

Your contribution to this problem is welcome. Further information required will be made available.

Many thanks

ghorrocks November 23, 2012 05:20

This is because you are using particle tracking in a regime it is not suited for, well at least not with the sub-models you are using. Particle tracking does not account for maximum packing and particle to particle interactions, so the volume fraction can go as high as it likes. Also note that lagrangian particle tracking is only good for low particle volume fractions - you obviously have high volume fractions so I would suggest the particle tracking model is not a suitable model for this application.

You need to include a packing model to make it more realistic. I would forget about particle tracking and move to eularian particle tracking where there are a few models which can handle dense volume fractions and maximum particle packing.

An alternate approach would be to couple with a DEM software which can handle the particle packing - but this would be additional software and all the fun and games of coupling software together.

TitusOfei November 23, 2012 09:12

Quote:

Originally Posted by ghorrocks (Post 393770)
This is because you are using particle tracking in a regime it is not suited for, well at least not with the sub-models you are using. Particle tracking does not account for maximum packing and particle to particle interactions, so the volume fraction can go as high as it likes. Also note that lagrangian particle tracking is only good for low particle volume fractions - you obviously have high volume fractions so I would suggest the particle tracking model is not a suitable model for this application.

You need to include a packing model to make it more realistic. I would forget about particle tracking and move to eularian particle tracking where there are a few models which can handle dense volume fractions and maximum particle packing.

An alternate approach would be to couple with a DEM software which can handle the particle packing - but this would be additional software and all the fun and games of coupling software together.

Hi Glenn, thanks for your quick response to my problem. Its very useful information.
I can understand from your reply that i need to go for Eulerian-particle tracking. My understanding from the manual (cfx_mod) is that, for Eulerian-particle model, the liquid phase should have a morphorlogy as continuous fluid while the solid phase as dispersed solid. Is it the same as you are suggesting?

This approach is seeking for the volume fraction of each phase at the inlet boundary condition. I'm quite in a dilemma since my aim is to determine the volume fraction of solids at the end of the simulation run.

A clear description of my problem: (drilling process): fluid (water) flows through the inlet of annular pipe at a constant rate and flows out. Cuttings (solids) are also injected at a constant rate through the inlet of the annular pipe and flows out. What is the concentration (volume fraction) of cuttings at the end of the simulation run. I'm considering steady state.

Once again, many thanks for your patience

Titus

ghorrocks November 25, 2012 05:02

Quote:

continuous fluid while the solid phase as dispersed solid. Is it the same as you are suggesting?
That appears suitable for your model.

Quote:

This approach is seeking for the volume fraction of each phase at the inlet boundary condition.
This is just the same BC as what you applied in the particle trakcing model, only you are specifying it in a Eularian framework not a Lagrangian one. The Eularian model is not asking for anything new.

Based on your initial model the cuttings are much heavier than water and quickly fall out of the flow and form a bed on the bottom. Is the physically realistic?

TitusOfei November 25, 2012 10:26

Quote:

Originally Posted by ghorrocks (Post 394025)
That appears suitable for your model.



This is just the same BC as what you applied in the particle trakcing model, only you are specifying it in a Eularian framework not a Lagrangian one. The Eularian model is not asking for anything new.

Based on your initial model the cuttings are much heavier than water and quickly fall out of the flow and form a bed on the bottom. Is the physically realistic?

Thanks Glenn for your suggestion.

Yes, in my initial model the cuttings are much heavier than the water. This is physically realistic in drilling (oil and gas) operations. This model is a horizontal wellbore where the drilled cuttings settle to form a bed due to gravity effect as it travels from upstream(downhole) to downstream(surface).

I will try your suggestion and give you feedback.
Once again, many thanks for your time

Titus

ghorrocks November 25, 2012 17:37

In this case then things like scour, settling and packing are important. This will be a challenege in CFX.... Definitely try Eularian approaches, but you might need to move to a CFD code with an actual scour model. Flow3D has one, but am not sure how applicable it is to this model.


All times are GMT -4. The time now is 08:35.