Re: [Gfs-users] Turbulence model

[Tao, Y. <y....@ja...>]

Hi Popinet,
I read your tangaroa paper today and read the related  example Turbulent
air flow around RV Tangaroa.
I find that in the paper there is very little contents related to the
LES you used in the simulation. The example on gerris website does not
include LES.

I do not know whether my request is appropriate, Is it possible for you
to post an example of the simulation described in your tangaroa paper, I
mean including LES?

Cheers,
Tao, Ye

On Sun, 2010-06-06 at 00:45 +0200, Stephane Popinet wrote:

> Hi Martin, Kristjan,
> 
> Just a few comments on this.
> 
> > a) The solution becomes unstable, error growing rapidly until simulation
> > crashes.
> 
> I believe that what you are thinking of is that some advection schemes
> (central finite differences for example) rely on some amount of
> explicit viscosity for stability. The second-order upwind advection
> scheme in Gerris has its own built-in viscosity (as any upwdind scheme
> does) so does not rely on explicit viscosity for stability.
> 
> > b) If the solution is stable it will be incorrect. Just how incorrect
> > depends on what you're trying to simulate.
> 
> I agree if what you are trying to achieve is DNS:
> 
> http://en.wikipedia.org/wiki/Direct_numerical_simulation
> 
> however I believe that what Martin has in mind is some turbulence modelling.
> 
> Martin, if you haven't done so already you should read this paper:
> 
> http://gfs.sf.net/tangaroa.pdf
> 
> where we explain how we use a MILES approach to solve for large-scale
> turbulent flows without an explicit LES model.
> 
> > I'd say that simulating a Re = 1E5 flow is not feasible without a turbulence
> > model. There are approaches where no explicit model is used and you instead
> > rely on numerical dissipation. Whether this is appropriate depends on the
> > problem and can only be tested by comparison with a fully-resolved
> > simulation or experiment.
> 
> Yes, I agree. Generally there are two subproblems two consider: bulk
> turbulence and boundary layers. In many cases, a different boundary
> layer model is required. This is particularly so for 'aerodynamic
> bodies' where boundary layer description (and separation) can be
> critical to getting drag/lift correctly. "Bluff bodies" such as the
> tangaroa flow in the paper above are somewhat easier because boundary
> layer separation is triggered (even at coarse resolution) by the
> geometry (i.e. the sharp edges).
> 
> Overall, a difficult problem....
> 
> good luck
> 
> Stephane
> 
> ------------------------------------------------------------------------------
> ThinkGeek and WIRED's GeekDad team up for the Ultimate 
> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the 
> lucky parental unit.  See the prize list and enter to win: 
> http://p.sf.net/sfu/thinkgeek-promo
> _______________________________________________
> Gfs-users mailing list
> Gfs...@li...
> https://lists.sourceforge.net/lists/listinfo/gfs-users