CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Non-parabolic profile for fully-developed flow (http://www.cfd-online.com/Forums/fluent/33261-non-parabolic-profile-fully-developed-flow.html)

Adrian Lee March 17, 2004 11:39

Non-parabolic profile for fully-developed flow
 
I am using Fluent to investigate biomedical flows in coronary arteries and comparing computational results with experimental measurements.

I am experiencing some difficulties in obtaining comparable velocity contours from Fluent. I have used an in-house code to solve for the same geometry and flow conditions and it produced a much better agreement with my experiment. However, it is desirable to continue using Fluent as there is more functionality and ease of use.

I am concerned that for the straight tube inlet length that I have, Fluent gives me a blunt-paraboloid type velocity profile across the tube section about 15 diameters downstream of a velocity inlet when I would expect an exact parabolic profile for fully-developed flow. I think this would affect the velocity profiles of sections further downstream. Also, there is a significant diffusion effect as the velocities are significantly reduced in magnitude even when looking at the straight tube section.

The mesh used was using a spacing of 0.2 with 4 layers of prismatic elements with a growth rate of 1.2 and a first layer thickness of 0.05. I have a total of about 1.3 million elements in the geometry. I guess that I could flood the geometry with much more elements, but that would be computationally expensive and not desirable.

I understand that using a hexagonal mesh would greatly improve the solution I would get but with these artery-type geometries, GAMBIT only allows me to build hybrid meshes with tetrahedral and prismatic wall layers.

I have used the double precision solver with the default segregated scheme with residuals up to 1e-10 for the velocities and 1e-9 for the continuity.

Can somebocy help?

Thanks!

Ugo March 18, 2004 06:14

Re: Non-parabolic profile for fully-developed flow
 
Hi Adrian; Can I suggest you to check if the Y+ you obtained with your grid is reasonable for the turbulence model you adopted?


All times are GMT -4. The time now is 19:38.