In what case will GGI bring the accurate results for unmatched nodes?

lnk · July 24, 2012, 06:09

Hi, everyone.

Thanks for your time to look at this question. I'm thinking about using GGI for the unmatched nodes interface between fluid-fluid domain for my case. May I ask in what case will GGI bring accurate results? ( I searched for this at the tutorial but didn't find any answer.

)

Best regards and many thanks again,
lnk

ghorrocks · July 24, 2012, 19:33

Yes, GGIs are quite accurate. It is very unusual for a GGI to introduce enough error into a simulation to be a problem.

lnk · July 25, 2012, 02:19

Quote:

Originally Posted by ghorrocks

Yes, GGIs are quite accurate. It is very unusual for a GGI to introduce enough error into a simulation to be a problem.

But you can look at the comments at this link about GGI. There should be some cases that GGI bring low accuracy results. May I ask how to know GGI works well or not?

http://www.cfd-online.com/Forums/ans...eration-2.html

ghorrocks · July 25, 2012, 02:26

You should compare this to a known accurate or analytical answer. If it has some distortion adjacent to the interface but is accurate away from it then that is fine for most users. Also I think you will find the distortion is amplified by the way the contours are drawn in CFD-Post.

So "low accuracy" is dependant on what you are looking for accuracy in. If you need it to get the correct pressure drop across something you will almost certainly find it OK. If you need smooth temperature contours in a CHT body then it might not be so appropriate.

But sweeping comments like "GGIs are low accuracy" are not helpful. But there are cases when they will cause problems.

flotus1 · July 25, 2012, 04:32

I posted this comment in the context of a tedious hexa-mesh generation in a difficult geometry.
I just thought it would be a pity to spend several days creating a block-stuctured mesh and then using an interface to sew together non-matching parts of the mesh.

lnk · July 25, 2012, 12:14

Quote:

Originally Posted by ghorrocks

You should compare this to a known accurate or analytical answer. If it has some distortion adjacent to the interface but is accurate away from it then that is fine for most users. Also I think you will find the distortion is amplified by the way the contours are drawn in CFD-Post.

So "low accuracy" is dependant on what you are looking for accuracy in. If you need it to get the correct pressure drop across something you will almost certainly find it OK. If you need smooth temperature contours in a CHT body then it might not be so appropriate.

But sweeping comments like "GGIs are low accuracy" are not helpful. But there are cases when they will cause problems.

So if I'm only looking for pressure drop, I can always use GGI without considering the inaccuracy it might bring. Right?

ghorrocks · July 25, 2012, 18:42

Quote:

So if I'm only looking for pressure drop, I can always use GGI without considering the inaccuracy it might bring. Right?

No, checking the GGI interface for accuracy in your specific case is always a good idea. If you do not trust it then you should check it.

bratzinger · August 2, 2012, 10:06

Hi

got a similar problems in a GGI-Interface in one of my last simulations and there were also unsteady contour lines for e.g. (total-)pressure at a different location

-> can i assume that if these unsteady contour lines only appear at one "very ugly to mesh geo-part" and the unsteadyness isn't that sever, the error in my solution isn't that significant?

somewhere else in this forum i read about the difference in using massFlowAve()@... and areaAve()@... . what i also noticed is there is a noticeable difference between e.g. massFlowAve()@REGION:... and massFlowAve()@Plane... (<- created plane in cfx-post=exactly the same area as REGION:...)

->anyone knows where this comes from?

my advice is to always use different ways to analyze your results. also create monitor points through different expressions and check if they change over time(steady state sim) ...
(sometimes i get gigantic pressuredrop fluctuations over my fluid-domain when i evaluate with massFlowAve()@REGION:... - even if the end result via massFlowAve()@Plane is very plausible)

ghorrocks · August 2, 2012, 20:28

It is always best to test things you are suspicious of for accuracy in your case. So I would compare a GGI simulation versus one without the GGI and see if the accuracy is acceptable. If generating the non-GGI mesh is impossible (that is why you put the GGI in there in the first place

) then dream up a simple little test case which uses the same physics and fundamental flow, but is simple enough to run with and without the GGI. If the GGI is acceptable in the simple test case then it will be OK for the real simulation.

July 24, 2012, 06:09	In what case will GGI bring the accurate results for unmatched nodes?	#1
lnk Senior Member lnk Join Date: Feb 2011 Location: Switzerland Posts: 118 Rep Power: 15	Hi, everyone. Thanks for your time to look at this question. I'm thinking about using GGI for the unmatched nodes interface between fluid-fluid domain for my case. May I ask in what case will GGI bring accurate results? ( I searched for this at the tutorial but didn't find any answer. ) Best regards and many thanks again, lnk

August 2, 2012, 10:06		#8
bratzinger New Member Join Date: Jul 2012 Location: Germany Posts: 23 Rep Power: 13	Hi got a similar problems in a GGI-Interface in one of my last simulations and there were also unsteady contour lines for e.g. (total-)pressure at a different location -> can i assume that if these unsteady contour lines only appear at one "very ugly to mesh geo-part" and the unsteadyness isn't that sever, the error in my solution isn't that significant? somewhere else in this forum i read about the difference in using massFlowAve()@... and areaAve()@... . what i also noticed is there is a noticeable difference between e.g. massFlowAve()@REGION:... and massFlowAve()@Plane... (<- created plane in cfx-post=exactly the same area as REGION:...) ->anyone knows where this comes from? my advice is to always use different ways to analyze your results. also create monitor points through different expressions and check if they change over time(steady state sim) ... (sometimes i get gigantic pressuredrop fluctuations over my fluid-domain when i evaluate with massFlowAve()@REGION:... - even if the end result via massFlowAve()@Plane is very plausible)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Is Playstation 3 cluster suitable for CFD work	hsieh	OpenFOAM	9	August 16, 2015 14:53
Performance of GGI case in parallel	hannes	OpenFOAM Running, Solving & CFD	26	August 3, 2011 03:07
Test Case results for Semi-Cylinder	shankar1221989	Main CFD Forum	0	June 24, 2009 06:35
Modeling the flow in a fully axisymmetric case %e2%80%93 results with no peripheral symmetry	evgenia	OpenFOAM Running, Solving & CFD	0	February 14, 2008 10:23
Free surface boudary conditions with SOLA-VOF	Fan	Main CFD Forum	10	September 9, 2006 12:24

July 24, 2012, 19:33		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Yes, GGIs are quite accurate. It is very unusual for a GGI to introduce enough error into a simulation to be a problem.

July 25, 2012, 02:26		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	You should compare this to a known accurate or analytical answer. If it has some distortion adjacent to the interface but is accurate away from it then that is fine for most users. Also I think you will find the distortion is amplified by the way the contours are drawn in CFD-Post. So "low accuracy" is dependant on what you are looking for accuracy in. If you need it to get the correct pressure drop across something you will almost certainly find it OK. If you need smooth temperature contours in a CHT body then it might not be so appropriate. But sweeping comments like "GGIs are low accuracy" are not helpful. But there are cases when they will cause problems.

July 25, 2012, 04:32		#5
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	I posted this comment in the context of a tedious hexa-mesh generation in a difficult geometry. I just thought it would be a pity to spend several days creating a block-stuctured mesh and then using an interface to sew together non-matching parts of the mesh.

August 2, 2012, 20:28		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	It is always best to test things you are suspicious of for accuracy in your case. So I would compare a GGI simulation versus one without the GGI and see if the accuracy is acceptable. If generating the non-GGI mesh is impossible (that is why you put the GGI in there in the first place ) then dream up a simple little test case which uses the same physics and fundamental flow, but is simple enough to run with and without the GGI. If the GGI is acceptable in the simple test case then it will be OK for the real simulation.