It is reasonable to suppose that viscosity is locally constant in compressible NS?
 

Well, I opened a thread about my project months ago and I am starting again from scratch.

It's a flat plate model of a solar collector for water heating. I want to study the movement of the air (ideal gas approximation) inside by solving the full set of compressible navier stokes.

Now I want to do things well because last time I had convergence problems (I had to put too much relaxation between iterations with a time step so little that it was impossible to get to the steady solution).

I'm going with an implicit method and hybrid differentiation (switching between upwind and central difference depending on the peclet number).

I am using a staggered grid (not necessary because odd-even decoupling shouldn't appear in compressible flows as i have the ideal gas law, but I think it feels better the pressure).

My question now is... can I suppose that the viscosity is locally constant? I mean, I am using a correlation for viscosity but I want to neglect every single term where viscosity derivatives appear. It would simplify things, and it means a shorter code, faster,... Just think as a rectangle where the upper side is cold and the lower side is hot so you have free convection.

Thanks in advance.

viscosity is generally assumed to be a function of the temperature https://en.wikipedia.org/wiki/Temper...quid_viscosity

...for small variation you could assume it constant