CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Different result in CFX and fluent for mass trans.? is segregated better? (http://www.cfd-online.com/Forums/cfx/107173-different-result-cfx-fluent-mass-trans-segregated-better.html)

ftab September 19, 2012 12:05

Different result in CFX and fluent for mass trans.? is segregated better?
 
Hi all,

What I am trying to do is the mass transfer from a metal stent inside arterial wall and lumen of the artery.
The problem I am facing is that I am not getting similar results to a reference paper which has used fluent to simulate. What I am missing is the wash out of the mass from the stent to the downstream.
I have done the simulation in Fluent and even after 400 iteration I can clearly see that.(refer to this pic)

http://www.flickr.com/photos/87439749@N04/8005726854/

But in CFX, I cannot capture it even turning off the converged flow solver and increasing the "Time Scale Factor" for the AV equation and reaching to 10-7 for its residuals.
http://www.flickr.com/photos/87439749@N04/8005727457/

First I was thinking it is because the outlet BC is treated differently with Fluent and CFX, so I extended the outlet to minimize its effect.
I can see that it is mainly because of difference between segregated solver of Fluent and coupled solver of CFX, which suggests that segerated solver is the way to go.
I still like to do the simulation in CFX due to some reasons. Could you help me suggesting how to make CFX results closer to already validated Fluent results and how to capture those streaks of concentration shown in fluent image?

It seems that the mass is not diffused in CFX case to downstream, and it takes ages for it to happen. BUT CFX and Fluent should give similar results at the end, right??!!

ghorrocks September 19, 2012 18:34

Quote:

I can see that it is mainly because of difference between segregated solver of Fluent and coupled solver of CFX, which suggests that segerated solver is the way to go.
A well converged simulation in segregated or coupled solver should be the same.

Your images are not visible so I cannot see them, but there is bound to be a set up problem with your simulation. You have not described anything which CFX cannot solve accurately.

ftab September 19, 2012 20:41

Thanks Ghurruks for your prompt reply as always. The links are working as I have checked now. I can also email them if you wanna see them. The set ups are absolutely identical and has been checked several times. Your professional comments are highly appreciated

ftab September 20, 2012 05:49

Copy pasting the link in internet browser works as I checked!

ftab September 20, 2012 05:58

I do not know whay my omages are not shown, here is the Flickr link for fluent simulation:
http://www.flickr.com/photos/87439749@N04/8005726854/
http://www.flickr.com/photos/87439749@N04/8005726854/

and CFX:
http://www.flickr.com/photos/87439749@N04/8005727457/
http://www.flickr.com/photos/87439749@N04/8005727457/

ghorrocks September 20, 2012 07:21

The problem has nothing to do with the outlet. Either your mass transfer model is not working in CFX, or you have not run the simulation long enough (in physical time) for the plume to develop.

I would run a little benchmark case with a simple geometry and the mass transfer to make sure it works. If you have an analytical or experimental result to compare to that is even better.

Then I would run it long enough to ensure the flow and mass transfer is fully developed. This may not be captured in the residuals - I would add the imbalances as well.

ftab September 27, 2012 05:59

Quote:

Originally Posted by ghorrocks (Post 382740)
The problem has nothing to do with the outlet. Either your mass transfer model is not working in CFX, or you have not run the simulation long enough (in physical time) for the plume to develop.

I would run a little benchmark case with a simple geometry and the mass transfer to make sure it works. If you have an analytical or experimental result to compare to that is even better.

Then I would run it long enough to ensure the flow and mass transfer is fully developed. This may not be captured in the residuals - I would add the imbalances as well.


Dear Glenn,
Regarding your comment:
- I checked the imbalances which are in the order of 1e-8 or lower.
-I could not find any analytic solution or experiment to resemble steady solution, which is of my interest. Transient would have its own issues which I prefer to avoid.

-BUT checking the results following your comments, I witnessed something very weird.
The velocity contour in axial mid-plane is obviously wrong for cfx. Check the image:
http://www.flickr.com/photos/87439749@N04/8029236690/
http://www.flickr.com/photos/87439749@N04/8029236690/

And it is despite the fact that flow is converged under criteria of 1e-7.
The velocity in Fluent looks reasonable:
http://www.flickr.com/photos/87439749@N04/8029236656/
http://www.flickr.com/photos/87439749@N04/8029236656/

Every single setting for the CFX and Fluent is checked and is identical.
Suprisingly, when I set Newtonian fluid setting (instead of present Non-newtonian Carreau model) the velocity contours become OK.
http://www.flickr.com/photos/87439749@N04/8029240958/
http://www.flickr.com/photos/87439749@N04/8029240958/

I am just puzzled.

Your professional comment is highly appreciated.

ghorrocks September 27, 2012 07:57

Quote:

I could not find any analytic solution or experiment
I don't think you tried very hard. Here is a few:
* Square box, LHS set to AV=1 and RHS AV=0. No flow and diffusion. Does the AV evenly go from 1 to 0 from left to right.
* Flow in a simple duct, with a point defined as AV=1 with a convection model. Does the AV convect with the flow?

So this is a non-Newtonian model? Then forget all your intuition of what it should look like. I have no idea what the flow should look like so cannot say whether CFX or Fluent is wrong. When you model the Newtonian flow in CFX it looks good, but who knows what the non-Newtonian model will do?


All times are GMT -4. The time now is 01:49.