ABL/grid/model

Alain BASTIDE

Dear StarCD users,

I try to study flow around square cyninder with a commercial code (k-epsilon) ..... and i've got some problems.

This square cylinder is in atmospheric bounary layer and i found some correlations for neutal ABL :

wind speed V=u_tau/kappa*ln(z/z0) Turbulence Intensity I=1/ln(z/z0) Lenght Scale = f(z,zg,z0,zc) (Harris JWEIA 1986) Re = 4.10^4

First, I see that lot of authors use It= constant and Lenght Scale = constant on vertical profil at boundary layer. Witch modelisation is good? Correlation or constant?

Second, i see that lot of authors use power law : U=Uref(z/ref)^alpha but in the book "Boundary layer Meteorology (Stull)" log law is used to modelise wind speed profil. I know that log law is equal to power law for Re~10^5 and alpha=1/7 but witch profil is good to have got a good modelisation?

Third, Witch grid near wall???? With log law, V=0 for z=z0, i want to use a two-layer turbulence model but how can i generate mesh? Betheen z=0 and z=z0 and for z>=z0 ??? Witch k and epsilon i can take for 0<=z<=z0?

Thanks a lot for your answers ......