i wish to convert my laminar code to les one, i have a small doubt, when i calculate mu_sgs, does it remain constant during the sub iterations of that time step, or do i calculate this quantity on each iteration.

A mon avis, it depends on a model. If you use Smogarinsky's model, then in time step mu_sgs remains constant (it is obvious from it's form – gradients of velocities which are constant in each time step for the final iteration). If mu_sgs is inside the iteration process for a single time-step then it changes with iterations, asymptotically reaching the steady state in a single time-step. You can also try filters for small-scale wave perturbations (it's better than mu_sgs models) but here you'll have to use finite-difference schemes. Regards.

thanks Nick, this clears my doubt, i am trying to implement smogorinsky model only, and yes the gradients will remain constant during the time step and hence mu_sgs should also remain constant. thanks again.