CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   beginner: how to change flow behaviour? (

yvar July 26, 2011 10:24

beginner: how to change flow behaviour?

I am beginner with ansys cfx (v12) and have the next question

My challenge is a big water tank with some thin internal walls (it can be any configuration). Inlet is big hole from one side of the tank. Outlet is a small hole from the other side.
BCs: inlet - opening with some relative pressure; outlet - mass flow rate. Also rotating periodicity is used.
Flow velocity in main volume is slow, but high in the outlet

My problem is:
Solution gives flow trajectories with a lot of local swirlings. Flow moves not in straight direction (as seems to be logical in my model), but turns and flows in any other directions.
BUT I saw CFD analysis of such unit where flow trajectories were straight, uniform and logical...
I tried different turbulence models, remeshed model with smaller size but the pattern did not change significantly. Also tried change boundary conditions - no success..

So the questions are: 1. Can type of elements give so significant difference in flow behaviour? i mean if to remesh model with hexa, not tethra
2. I tried all given in CFX turbulence models, but without any additional controls.. all by default. Could you advice which turbulence model is more suitable for such task?

hope here is enough info.. thanks in advance for any advice!


lindner July 26, 2011 15:56

hi yvar,
Have you tried running with upwind scheme instead of high resolution? The upwind scheme will simplify your flow (it should not be used for final results) and maybe you will see something different.


yvar July 26, 2011 16:12

hi Lindner,
No, I haven't... Thanks, I'll try this scheme!

ghorrocks July 26, 2011 18:54

Upwinding will dampen the wobbly bits but also put so much dissipation in the model that your results are inaccurate. It might give you a good initial condition for a simulation with high order discretisation to get the accurate answer.

Can you post some images of what you are seeing?

yvar July 27, 2011 05:29

1 Attachment(s)
please find img of one of the results.. It is simple design where inlet (opening) is distributed evenly along outer cylindrical surface, outlet (mass flow) is a hole in center.

I saw cfd of the similar design with the same flow rate. Unfortunately I dont have that picture, but unlike my results, there was flow straight and uniform from outside to center. And i only know that overall dimensions and mass flow in that analysis was the same, I dont have more info...

I am not sure that model was correct, but my results also look strange as to me..

Upwind scheme did not change the pattern significantly
Scheme "specific blend factor 1" and high resolution for turbulence numerics is still running

some more info about attached results (high resolution, required RMS 1e-4)
| Average Scale Information |
Global Length = 1.5834E+00
Minimum Extent = 9.0300E-01
Maximum Extent = 3.3790E+00
Velocity = 2.3506E-03
Advection Time = 6.7361E+02
Reynolds Number = 2.3844E+03
| Equation | Loop # | Peak Residual | Final Residual |
| U-Mom-bottom | 3 | 4.27507E-02 | 5.78801E-03 |
| V-Mom-bottom | 3 | 2.45484E-01 | 9.89888E-03 |
| W-Mom-bottom | 3 | 1.25698E-02 | 4.74994E-03 |
| P-Mass-bottom | 1 | 6.72428E-03 | 2.95278E-08 |
| K-TurbKE-bottom | 1 | 5.22813E-01 | 2.01961E-03 |
| E-Diss.K-bottom | 1 | 9.28826E-01 | 9.74163E-04 |

Now I have no idea except remeshing the model with hexa... Automatic tethra gives too heavy model..

ghorrocks July 27, 2011 06:07

The SST turbulence model is to one to use unless you have a good reason not to.

It looks like you are far from converged. Here is some tips:

yvar July 28, 2011 14:54

Thanks for help!
seems I ve got needed results..

All times are GMT -4. The time now is 21:25.