[Marvin]

Dear all,

A small question concerning Large Eddy Simulation and the implementation of the dynamic Smagorinsky subgrid-scale model in Fluent (Lilly's version).

In this procedure, the value of the Smagorinsky coefficient (C_s) is clipped to [0; 0.23].

For the lower bound, it's actually the square of C_s (and hence the subgrid-scale viscosity) which is kept positive to avoid numerical instabilities.

What about the upper bound? Why is it needed to keep C_s below a maximum value? Is it also due to numerical issues? And why this particular value (0.23)?

Any idea or comment is welcome. Thanks!