CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   MiLES vs No-model (https://www.cfd-online.com/Forums/main/9913-miles-vs-no-model.html)

K. S. Chang September 20, 2005 20:59

MiLES vs No-model
 
Hello~ everyone. I'm a Ph. D. candidate majoring in CFD. My research topic is LES and DES. I have some questions about MiLES. I know that MiLES use numerical dissipation implicitly instead of SGS visocity model. Upwind schemes(or flux limiting scheme) are used. In this case, I think that the amount of dissipation is different according to order of scheme. Is it correct ? And what is difference between MiLES and Laminar simulation using upwind scheme ?? Only that is Reynolds number ??

Also I saw no-model simulation paper(of kuwahara <- I'm not sure) in turbulent flow regions. Isn't it laminar simulation ?? I don't know exactly difference between no-model simulation and laminar simulation.

If you know that, please let me know.

Thanks in advance.


Harry Fulmer September 20, 2005 22:40

Re: MiLES vs No-model
 
Get in touch with Prof Paul Tucker at Swansea university in Wales. He's done some MILES stuff, he might know.

Lionel Larchevêque September 21, 2005 10:34

Re: MiLES vs No-model
 
Hi,

- Regarding equations and numerical schemes, MiLES and laminar computations (with an upwind scheme) are identical. However, for MiLES computation mesh and timestep have to be set so as to allow sustained turbulence at locations where the flow has to be turbulent. This implies that classical LES requirements in term of x_i^+ and CFL number have to be fulfilled.

- The amount of numerical dissipation is indeed reduced if high-order schemes are used. It can be seen when looking at the (complex) effective wavenumber of the scheme whose non-zero imaginary part is the the wavenumber distribution of dissipation. Notice that similarly to optimized centered schemes (Tam, ...), for a given stencil you can reduce the formal accuracy in Taylor's sense so as to obtain an optimized dissipation distribution : as low as possible over the wider possible range and with a sharp transition to high levels for the higher wavenumbers. You can even design the scheme to mimic the effect of a given subgrid model based on the turbulent viscosity concept.

- I'm not sure what 'no model' with (partly) turbulent flow means. If the numerical scheme is dissipative, it's some kind of MiLES. On the contrary, if the scheme is described as non-dissipative (centered?), all the turbulent regions should be fully resolved otherwise the computations would most probably blow-up. Consequently, my guess is that either the computation is a DNS one or there are some hidden sources of dissipation somewhere in the code (temporal scheme?).

Regards

Lionel

K. S. Chang September 21, 2005 21:13

Re: MiLES vs No-model
 
Hello~ Dr. Lionel Larcheveque. I studied your paper about cavity flow simulation. Your research is so much helpful to me.

Thank you for kind explanation. I can understand the concept of MiLES for your help.

Thanks again.

zxaar September 21, 2005 22:45

Re: MiLES vs No-model
 
can you give me some paper to read about MiLES, i tried searching on google but could not get any ( my email is zxaar@yahoo.com )

K. S. Chang September 22, 2005 22:51

Re: MiLES vs No-model
 
Hi~ zxaar ! You can find using the keyword "Monotonically integrated LES " in google or other search engine. And download some papers.

Best regards.

Lionel Larchevêque September 23, 2005 09:59

Re: MiLES vs No-model
 
Hi,

Have a look at AIAA papers and article by Fureby and Grinstein (AIAA J. 37(5) 1999, JCP 181(1) 2002, Computing in Science and Engineering 6(2) 2004, ...).

For comparisons between MILES and LES, you can also read Garnier et al. JCP 153(2), 1999 (homogeneous turbulence) and Larcheveque et al. Ph. Fluids 15(1), 2003 (cavity flows).

Hope this helps

Lionel

zxaar September 25, 2005 21:17

Re: MiLES vs No-model
 
Thanks to both of you, i have downloaded the papers. will try to understand this.


All times are GMT -4. The time now is 13:40.