Question: Strain rate is basically derived from grad(U). In a simulation, grad(U) is calculated at the grid scale using some scheme, let's say leastSquares. I use a non-Newtonian model for fluid viscosity (i.e. mu = mu(strainRate) a function dependent on grad(U)). Does that make the viscosity value in the cell grid sensitive? In general the physical value of viscosity in such systems comes from a measurement, typically carried out with grad(U) changing over << 1mm. Would that mean, the mu(strainRate) should also vary over a similar resolution? Is there some kind of sanity check I can put on this? Or any engineers in the house who studied this at school?