Thanks for your answers, this
 

Thanks for your answers, this sheds some light on my questions.

I'm just wondering a basic question about finite volume method. When you're talking about divScheme, it means discretisation of convective term (hope I'm right until then!). When looking at the finite volume equations, there is no derivative to compute in the convective term, which is a sum of fluxes at each interface of a particular cell.
Then, what I understand is that the point of divScheme is to reconstruct the fluxes by interpolation.
So is it right to consider divSchemes as interpolation schemes?

Still about the cylinder case, I find a bit strange the strain tensor components I get using sigmaComponents. For example, along the horizontal streamline from the inlet to the stagnation point, I extracted the sigmas. Sigmaxx increases near the stagnation point, but sigmayy remains zero (I'm using a conventional frame of reference, x is horizontal, y is vertical, wind is horizontal).So this seems to violate mass conservation (I'm using incompressible solver icoFoam).
Has anyone already noticed something strange about sigmayy?

Thanks very much, and sorry if I'm telling rubbish!

Seb