CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Flow ariound sphere (

PetrK June 27, 2006 01:47

Flow ariound sphere
Hi all , I have serious problems with convergence, so I'd like to ask you fo some hints. I'm trying to study flowing around a very buig sphere (11 m radius) and compare its drag with therory. The problem is, that forces on this sphere is fluctuating and i can't get a precious value. Could you help me with it?

My current settings - unstructured mesh (Stadart CFX-mesh generator) - 24 BL layers, y+ < 1.25 - ref pressure - 1 atm (standart atm. conditions) - Inlet speed 12 - 40 m/s - SST turb mode (i'm trying the transitioan model also) - Air ideal gas - Transident flow, Timestep 15s - EXPERT PARAMETERS:

mg solver option = 2

pressure diffusion scheme = 1

relax mass = 1

stage reverse flow robustness = t

tef numerics option = 1

wallscale diffusion scheme = 2

wallscale relaxation factor = 1

Bigger timestem does not help, the k-e turb bule ends with a error. Does the RNG model hepl? OR LES?

Thanks in advance.

Johnson June 27, 2006 02:28

Re: Flow ariound sphere
Are you running steady-state or transient? If you are in the subcritical regime you can expect large scale vortex shedding which will induce flutating forces over time.


PetrK June 27, 2006 07:46

Re: Flow ariound sphere
I'm running the transident flow. So If I understand clearly it is impossible to get non-fluctuating force, as the value of the force is a function of time. I'm right? If so, what is your suggest timestep and number of the coeff loops to get proper results? It is related to the Courant Number? Thanks

Robin June 27, 2006 19:44

Re: Flow ariound sphere
First of all, if you don't want transient results, don't run transient. If run steady state instead, you might be able to get a steady solution with a large enough timestep. On an 11 m sphere, the timestep may be quite large, on the order minutes.

If you're sticking with the transient solution, calculate time average values instead.

Finally, don't mess with the expert parameters unless specifically instructed to do so.

Regards, Robin

BigBrother June 28, 2006 04:16

Re: Flow ariound sphere
You can try to check mesh aspect ratio. Once I had problems with convergence because aspect ratio of prism in BL was much more than thuosand.

PetrK June 28, 2006 06:22

Re: Flow ariound sphere
Thanks, both of you... please could you tell me, how to calculate time average values? It would be good if I can't get the steady solutiont. Thanks

PetrK June 28, 2006 06:32

Re: Flow ariound sphere
I'm appologice to you for this silly quetion. It is really nessesary to use search engine before asking. I'm sorry again.

Julian July 16, 2006 04:31

Flow ariound sphere
Hi there! I try to do some simulation with cfx of the flow over the sphere. But when I get the results using the post of CFX the current lines apears as laminar flow, however the Cd is correct compear with experimental test. If I get the current lines correct (separation on 120 ) the Cd is incorrect. Staionary, r = 0.1 m, SST, V = 5 m/s; eddy lenght scale = 0.1; Physical Time step Ok, mesh = ok. Any idea??

All times are GMT -4. The time now is 22:31.