[thinktank]

hii..

continuing from a previous post...http://www.cfd-online.com/Forums/ans...onditions.html

I have now been able to get a 2d model of steady state poissuelle flow working, however I am still having a few problems, not in the implementation but about the results I am getting.

To start off I used a relatively dense mesh


These are the values returned from a quality check, so I think the mesh is ok:

Maximum cell squish = 1.20754e-001
Maximum cell skewness = 2.60630e-001
Maximum aspect ratio = 3.28009e+000

I used water as the fluid medium and applied a gauge pressure of 2 Pa at the inlet and a guage pressure of 0 Pa at the inlet.

I solved the problem for as many iterations as necessary for convergence. For this I set the max number of iterations to 1000, and for all the cases I ran, the problem always converged properly within 500 iterations max.

Next I plotted the x velocity profile at the boundary, however surprisingly I started finding that even for this simple case I obtained different values of the outlet velocity, when I used different schemes. The difference was quite significant and not negligible. These are the output velocities for the different schemes, the schemes used are mentioned at the bottom of each picture:







I also plotted the gauge pressure (static pressure at the inlet). It turned out that even the gauge pressure at the inlet doesnt remain fixed at the boundary condition I assigned(= 2 Pa). It deviates a lot, however the deviation is the same for almost all the schemes I checked



and this is the case file and the mesh file I used, if anybody is interested...

http://www.box.net/shared/0rd1cn1gtz

1)So this leads me to the question, why are these happening? Am I doing something wrong?

2)If it is natural to get different values of the velocity profile with different schemes, how do I know which one will give the accurate result? Is there any guideline I can follow to choose the best scheme?

3)what is the meaning of reference depth, length and area which are supposed to be assigned. I did check the manual, however it doesnt say clearly what it is.

As you can see, I am a beginner in fluent, and I would really appreciate any help from more experienced users at this forum.

thanks a lot

[thecfduser]

Hi
1-U are using very small mesh, so use double precision
2-did u verify that ure solution is fully developped????? did u verify that is independant of the mesh size????? If u refine the mesh, u will get nearby results (refine more for the case with first order upwind)
3-there is no universal scheme. Maybve QUICK is the most used since it has low diffusivity, high accurcy.....but all depends on the simulation itself.