CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Seperated Flow and Turbulence modelling flat plate (

Simon Mizzi December 15, 2003 18:06

Seperated Flow and Turbulence modelling flat plate
Hi, I am currently on my final year project concerning the Hydrodynamic design of an unmanned underwater vehicle usind CFD techniques. I use FLOTRAN as a software package, I know you guys are more into fluent and phoenics but I have a question which could interest you and at the same time solve a dilemma I have had for the past three months. out of all the turbulence models avaialbe which is the best suited for separated flow?I did a test on a flat plate using a varaiton of angle of attack from 0 to 16 degrees for each model. the zero equation model obviously did not cater for stall. the others showed some stall charaterstics however at different angles of attack for a constant reynolds number. I did a flat plate because i wanted something simple in the beginning and was sure to find readily available data on it. however i was wrong i did not find a single databook on flat plate or something similar in classical theory books.I must remark that all models are in fair agreement for the attached flow part.

Is there someone who can enlighten me?

Thanks Simon

Jonas Bredberg December 16, 2003 02:26

Re: Seperated Flow and Turbulence modelling flat p

If you would like to try the behaviour of turbulence models in separated flows, you may be luckier with a diffusor, rather than a single angular flat plate. However don't be surprised if the models don't capture either the separation point or the separation lenght, and if they don't separate for the right angles (say for angles more than 7degrees). Another test case would be a backward-facing step (BFS)flow, for which I know that there exists excellent DNS-data. Which model to use? Use at least a two-equation model, and go with a newer one. Popular models are: Menters k-omega SST and Durbins k-eps-v2-f models, both have some additional physics. If these are not available use k-eps RNG, or a newer k-omega model. I won't bother to try a RSM, as it is critical to resolve the near-wall turbulence to capture the separation point - and these models are usually less accurate in this respect. Good luck with your project,

Regards Jonas

All times are GMT -4. The time now is 06:51.