CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   simulation results for k-w model and SST model (http://www.cfd-online.com/Forums/cfx/24127-simulation-results-k-w-model-sst-model.html)

Li June 15, 2007 05:01

simulation results for k-w model and SST model
 
Hi, guys. I am learning CFX10.0 and now I encountered a problem. My case is simple with the air supply from the left top and air discharge from the right bottom, but I got weild simulation results of streamline and velocity distribution. I mean I found a reference in which he got similar results of k-w model with that of k-e model and different results from SST model. But I got similar results of k-w model with that of SST modle. I don't know why, would you help me to look at it and give me some suggestions. Thank you very much! (1)streamline of k-w model http://imageshack.dk//viewimage.php?...e/M2k94389.jpg (2)streamline of SST model http://imageshack.dk//viewimage.php?...e/tYn97524.jpg (3)velocity distribution at x=3m http://imageshack.dk//viewimage.php?...e/alh97900.jpg


Gert-Jan June 17, 2007 16:25

Re: simulation results for k-w model and SST model
 
Need more info:

- How does your grid look like? Is it appropriate for the models you use?

- Reynolds number?

- What do you specify for turbulence at the inlet?

- How well is it converged? Should not be of any problem if your grid is ok. Do you have monitor points? Imbalance?

Gert-Jan

www.bunova.nl


Li June 18, 2007 02:51

Re: simulation results for k-w model and SST model
 
Gert-Jan, thank you very much for your response. The grid file is here,http://imageshack.dk//viewimage.php?...e/IVK49154.png

Reynolds number is about 5000, and I set the turbulence intensity as 4% and turbulent length scale as 0.0168 which is one-tenth of the inlet height. Actually, I set the residual type as RMS, residual target as 1E-05. Finally, the monitor residual graph shew it arrived at the residual target. But I didn't set monitor points.

With kind regards, Li

Michael B June 18, 2007 02:51

Re: simulation results for k-w model and SST model
 
I've noticed a velocity at the inlet of 0.5 m/s. Are your calculation laminar or turbulent?

Li June 18, 2007 02:54

Re: simulation results for k-w model and SST model
 
Michael, thank you very much for your response. The Reynolds number is about 5000. I am just doing some practice to learn CFX through comparing the simulation results with the measurements. And my supervisor hopes me to compare the simulation result for different turbulent models with the measurements.

Jens June 20, 2007 01:21

Re: simulation results for k-w model and SST model
 
Hi,

I can seen taht you are working on the Annex 20 case.

You should expect quite good results.

First check you mesh. It seems like that the mesh is not very good. Take a look at the profile. You can seen a large jump in the jet core velocity. The core of the flow much resolved.

The turbulence models can be sensitive to the mesh close to the walls. Furthermore check the convergence.

Have you ensure that the residual are below 10-4 and the imbalance is small ???

Take look at the following homepage: http://www.cfd-benchmarks.com/

It has a lot on benchmarking for this kind of flow etc.

Regards

Jens

Li June 20, 2007 03:49

Re: simulation results for k-w model and SST model
 
Hi Jens. Thank you very much for your response. Yeah, I am doing the simulation according to Annex 20. Maybe there is a problem with the mesh file. But I am just wondering it would cause the streamline so different? I also asked one of my friend to simulate it with k-w model in Phoenix, he got the similar streamline as that of k-e model even he adopted very coarse mesh.

Kumar June 29, 2007 04:19

Re: simulation results for k-w model and SST model
 
K-w solves well for near wall flows compared to k-w. SST is just an addition of turb shear stress to k-w,but it will use k-w near wall and k-e away from wall. Again wall functions like std and scalable solve for y+ below 20 with less efforts than automatic which requires arbitrary fine mesh. But automatic wall functions will capture laminar sublayer, and the log layer if you have around 10 nodes in that region.

So it's of no use if u r using sst/k-w and auto wall functions with coarse grid and comparison will give u no idea abt their inequalities with k-e.

Kumar.


All times are GMT -4. The time now is 13:21.