open TELEMAC-MASCARET The mathematically superior suite of solvers

Telemac3d applies isotropic viscosity in its coding for 3d smagorinsky turbulence. Considering relative grid sizes in estuary and river models, there can be a difference of orders of magnitude between the horzontal and vertical grid sizes. It would appear that this should be considered in the calculation of the viscosities. This approach would seem to create vertical viscosities much too high. Can anisotropic viscosity be applied in telemac3d.

Based on this question, the unsv3d variable is the inverse volume used to calculate a viscosity. The calculation done for gradients required by smagorinsky produces a result that is effectively multiplied by the element volume? So divide the result by the volume will reduce the value to just the gradients? Then a horizontal scale based on the surface area (square rooted), and a vertical scale based on z levels can be used to set a horizontal and vertical mixing lengths for the viscosity calculation? This method is similar to that used in other models.

Comments and advice appreciated

Hi,
maybe I miss the point but in 3D the Smagorinsky model is applied for the horizontal and e.g. the Mixing length model for the vertical which in the end gives anisotropic viscosity values for the horizontal and thr vertical, respectively? In my view it wouldn't make sense to apply a Smagorinsky type model for the vertical, since the vertical mixing length, based on the logarithmic velocity profile and the parabolic viscosity profile and furthermore tunable via a lot of available measurements (damping functions etc.) is quite well defined?


Best regards,
Clemens

Where wind driven flows dominate eg coastal lakes, is a parabolic vertical profile still appropriate? 3D smagorinsky vertical appears to be used successfully in atmospheric modelling.

The telemac3d code does support a 3d smagorinsky treatment and this is where it creates an isotropic viscosity. And it is based on a mixing length defined by the cube root of the element volume. This does not appear appropriate where the vertical and horizontal length scales are orders of magnitude different.

If a 3d smagorinsky treatment is inappropriate then why is it included...

Hi,

ok, I didn't know that Telemac-3D supports 3D Smagorinsky treatment. So I agree, considering the sometimes very different horizontal and vertical length scales, it would make more sense to use the z-levels for the calculation of the vertical mixing lengths. However I don't know how such a model works near a wall.

What I meant:
in river flows usually we use the the Mixing length parabolic model (in Telemac-3D named Nezu and Nakagawa), however there are other Mixing length models available which maybe are more suited to your case.

Best regards,
Clemens