[cfdmms]

Why explicitly filtered LES give better results than implicitly filtered LES?

[FMDenaro]

hello,
I suggest searching for similar posts

[cfdmms]

Originally Posted by FMDenaro

hello,
I suggest searching for similar posts

Hi,
I actually want to know, how explicit filtering effects in terms of reducing the grid sensitivity over the implicit filter? In cases with explicit filtering or the filter width that is higher than the grid size that used as filter in implicit filtering, the high frequency motions are efficiently damped down and the numerical error decreases. Since explicit filtering damped down high frequencies (comparatively small scale eddies) and an increase of the model length scale also decreases the numerical error which ultimately gives better numerical solution compared to implicit filtering.But is this also responsible for reducing the grid sensitivity issue?

[FMDenaro]

Consider h the computational grid size, Delta_ex the characteristic lenght of the explicit filter and Delta_im the characteristic lenght of the implicit filter.
The implicit filtering is due to two effects, the grid cut-off (Pi/h) summed to the numerical transfer function induced by the discretization of the method you chose.
The explicit filtering is chosen by the user in its shape and characteristic lenght.

In practice Delta_im = q*h, where q >1 is some constant depending on the discretization. Conversely, Delta_ex= n*h, n being a parameter you can control.
You can see the reducing h, Delta_im is implicitly reduced (q = constant) whilst for explicit filterin you can take Delta_ex constant as n can be increased. Therefore, for the explicit filtering you can produce a grid-independent LES solution, for the implicit filtering you tend to the DNS one.

[juliom]

Professor;
Why the transfer function are so important in the filtering analysis. Everything I have read so far are based on the function transfers. I have not found any explanation to this fact.!!

[FMDenaro]

Originally Posted by juliom

Professor;
Why the transfer function are so important in the filtering analysis. Everything I have read so far are based on the function transfers. I have not found any explanation to this fact.!!

filtering is better defined in the wavenumber space, therefore the transfer functions are what effectively act on the spectral content.
Give a look here http://www.eng.utah.edu/~rstoll/LES/.../Lecture04.pdf

but I suggest reading the state-of-the-art books on LES

[juliom]

Thank you very much professor... A very interesting LES course. I have been reading Pope's Book. But, there are a lot of stuffs I need to learn and understand..!!

Thank you very much for the material.

[kuzma169]

Hello there everyone . Can you help me? How I can in Fluent, CFX STAR-CCM and other CFD code to make solving with iLes turbulence method. Whilst there are no explicitly included iLes models in them. Please answer , thank you

[FMDenaro]

Originally Posted by kuzma169

Hello there everyone . Can you help me? How I can in Fluent, CFX STAR-CCM and other CFD code to make solving with iLes turbulence method. Whilst there are no explicitly included iLes models in them. Please answer , thank you

ILES is based on the idea that the local truncation error of the scheme acts implicitly as an addy viscosity SGS model. Therefore, you need to set laminar condition and use some upwind scheme or a central bounded scheme.

[kuzma169]

Thank you for help