Hi Manuel, I think the difficulty with your approach is knowing what would be e.g. the correct energy decay rate for the coefficients. For traditional energy spectrum of turbulent flows, we usually rely on the -5/3 slope as reference, but it is not clear how this would translate to coefficient space. Maybe if you had a DNS, than you could compare the under-resolved solution on a same mesh but different polynomial order? But I would have to think more about this. I noticed you mentioned two papers of mine, but not the one I discuss SVV, may this one is more useful to you: http://www.sciencedirect.com/science/article/pii/S0021999115008256 Cheers, Rodrigo . On Mon, Dec 12, 2016 at 10:49 PM, Manuel F. Mejía De Alba < manu2958@gmail.com> wrote:
Hi Rodrigo and Nektar team,
I am working with the Pasquetti's paper but I have to say that the paper is not conclusive, actually, they leave the problem open, they said:
[Choosing “optimal values” of the SVV parameters is not a trivial task, but maybe much easier than with the classical LES, since here the aim is not to adjust the parameters of a SGS model. Our best results have been obtained for the largest values of "SVVCutoffRatio" and the smallest of " SVVDiffCoeff". ]
So, my simple idea is made an analysis of the "energy" archived on each approximation mode. I mean if we are using a nummodes of 10, I want to know how if the values of the coefficient (1:10) to represent that "energy ". I think ( I have to study more to completely understand your work) it would be similar to your work on "Linear dispersion–diffusion analysis and its application to under-resolved turbulence simulations using discontinuous Galerkin spectral/hp methods" and "On the eddy-resolving capability of high-order discontinuous Galerkin approaches to implicit LES / under-resolved DNS of Euler turbulence" but without the representation over Fourier series, instead of that I want to use the polynomials representation. Is it sound crazy idea ?
Please let me know your thought about it and Thanks in advance.
Regards,
On 7 December 2016 at 06:47, Rodrigo Moura <r.moura13@imperial.ac.uk> wrote:
Hi Manuel,
I am not sure if I understand your question, but please keep in mind the following. If you want to produce a standard energy spectrum typical of turbulent flows, there is no direct functionality in Nektar for this. You would have to probe the solution in space and perform the required Fourier transforms with an auxiliary software, e.g. Matlab. Alternatively you could probe a single point in space over a given time span and rely on Taylor's frozen flow hypothesis to produce a spectrum. The latter is of course the only alternative in case you don't have a homogeneous direction in your flow.
I suggest you take a look on a 2005 study by Richard Pasquetti, named "Spectral vanishing viscosity method for LES: sensitivity to the SVV control parameters", which seems to address the questions you are interested in.
Kind regards, Rodrigo .
On Thu, Dec 1, 2016 at 2:46 AM, Manuel F. Mejía De Alba < manu2958@gmail.com> wrote:
Dear Nektar users and team,
I'm working on the correct "guess" to parameters SVVCutoffRatio and SVVDiffCoeff of the model SVV LES (SpectralVanishingViscosity). I trying to determine the influence of that values on the Kinetic energy spectra. So, I need to get the values of the Kinetic energy as a function of the k (where k = 1:NUMMODES).
Using the filter "Modal energy" (user-guide 3.4.9) it is possible to obtain the integral over all the velocity not with of the modes k. It is possible but just with one-dimensional homogeneous expansion and I wanna plot the data on a general model.
May anyone gives me a hand with this, any help is appreciated.
Thanks in advance,
--- Manuel Felipe Mejía De Alba.
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- --- Manuel Felipe Mejía De Alba.