Transient mode simulation (VoF model) mesh independence study problem

yulu · September 18, 2016, 07:02

Dear All

I'm simulating a case with two fluids flowing in a pipe and try to capture the interracial instabilities of the two fluids. I've been using VoF model, Laminar, transient mode to calculate and so far it produced some results that are qualitatively quite well.

However when I was doing mesh independence study, I kept increasing mesh density and I was getting different results each time! I judge the results by looking at the flow features after the same flow time. So I was wondering if the problem is I still have not reach a dense enough mesh or because the instabilities at the interfaces come from numerical calculations, I would never get same results when changing mesh settings?

In this case, how should I proceed mesh independence study??

Many thanks for any suggestions.

mome · September 18, 2016, 11:43

Hi, if you are calculating a transient solution, are you resolving unsteady features I assume? Are you comparing these structures at the same moment in time? If you do and still get significant changes, I'd say you did not yet achieve a mesh-independent solution for your type of flow. What about the average flow, would that be ok? The crucial point is if these changes have noticealbe impact on your quantity of interest. Can you try to quantify the mesh dependence you observe?
Cheers, Mo

yulu · September 18, 2016, 12:10

Quote:

Originally Posted by mome

Hi, if you are calculating a transient solution, are you resolving unsteady features I assume? Are you comparing these structures at the same moment in time? If you do and still get significant changes, I'd say you did not yet achieve a mesh-independent solution for your type of flow. What about the average flow, would that be ok? The crucial point is if these changes have noticealbe impact on your quantity of interest. Can you try to quantify the mesh dependence you observe?
Cheers, Mo

Thanks Mo
Yes I am calculating unsteady time-dependent flow features and yes I am comparing results at exactly the same flow time when conducting mesh independent study.
May I ask what's the best way to quantify mesh dependence ? At the moment I'm basically comparing flow features (structures) and I do observe differences when increasing mesh density. Please have a look attached a bit summary about the mesh I used (cross-section) and the flow features I was getting (contours of one phase). Many thanks.

mome · September 18, 2016, 12:35

Hi, unfortunately, I'm not an expert for unsteady VoF solutions, so I fear there are other people who can help you better quantifying that in this case..
Although I don't know what quantity you are displaying the contour of, I'm guessing the restrictions represent some location of dilute second phase or so.. How do they form? If their formation is connected to some other quantity, you could compare their location and that other quantity. E.g. if they originate from the developping section of the pipe, you may have a chance to see differing boundary profiles up there or something like that. I think fixing a bias problem for pressure or velocity gradients is easier than for such structures?

LuckyTran · September 20, 2016, 18:30

I don't think these results are bad ones, namely you have shown that it is not mesh independent.

The VoF model is terrible at keeping track of interfaces, the VoF assumption is to completely smear the interface in the first place!

Check your discretization schemes. Maybe you have more numerical instabilities as the mesh gets finer.

I think you had a phenomenally great idea to compare flow structures at the same time! Great work!

September 18, 2016, 07:02	Transient mode simulation (VoF model) mesh independence study problem	#1
yulu New Member Yu Lu Join Date: Jul 2015 Location: England Posts: 26 Rep Power: 11	Dear All I'm simulating a case with two fluids flowing in a pipe and try to capture the interracial instabilities of the two fluids. I've been using VoF model, Laminar, transient mode to calculate and so far it produced some results that are qualitatively quite well. However when I was doing mesh independence study, I kept increasing mesh density and I was getting different results each time! I judge the results by looking at the flow features after the same flow time. So I was wondering if the problem is I still have not reach a dense enough mesh or because the instabilities at the interfaces come from numerical calculations, I would never get same results when changing mesh settings? In this case, how should I proceed mesh independence study?? Many thanks for any suggestions.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
decomposePar problem: Cell 0contains face labels out of range	vaina74	OpenFOAM Pre-Processing	37	July 20, 2020 06:38
Mesh not converging as expected in a FSI problem	amrbekhit	CFX	5	September 8, 2015 03:36
[snappyHexMesh] No layers in a small gap	bobburnquist	OpenFOAM Meshing & Mesh Conversion	6	August 26, 2015 10:38
How to study pressure drop of continous phase in VOF model	sajeesh	FLUENT	4	February 5, 2014 23:01
Turbine: Transient Simulation Problem	John C.	CFX	7	December 5, 2011 09:31

September 18, 2016, 11:43		#2
mome Member Join Date: May 2014 Posts: 30 Rep Power: 12	Hi, if you are calculating a transient solution, are you resolving unsteady features I assume? Are you comparing these structures at the same moment in time? If you do and still get significant changes, I'd say you did not yet achieve a mesh-independent solution for your type of flow. What about the average flow, would that be ok? The crucial point is if these changes have noticealbe impact on your quantity of interest. Can you try to quantify the mesh dependence you observe? Cheers, Mo

September 18, 2016, 12:35		#4
mome Member Join Date: May 2014 Posts: 30 Rep Power: 12	Hi, unfortunately, I'm not an expert for unsteady VoF solutions, so I fear there are other people who can help you better quantifying that in this case.. Although I don't know what quantity you are displaying the contour of, I'm guessing the restrictions represent some location of dilute second phase or so.. How do they form? If their formation is connected to some other quantity, you could compare their location and that other quantity. E.g. if they originate from the developping section of the pipe, you may have a chance to see differing boundary profiles up there or something like that. I think fixing a bias problem for pressure or velocity gradients is easier than for such structures?

September 20, 2016, 18:30		#5
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	I don't think these results are bad ones, namely you have shown that it is not mesh independent. The VoF model is terrible at keeping track of interfaces, the VoF assumption is to completely smear the interface in the first place! Check your discretization schemes. Maybe you have more numerical instabilities as the mesh gets finer. I think you had a phenomenally great idea to compare flow structures at the same time! Great work!