As far as I know, percolation theory makes very efficient predictions for filler particles whose aspect ratio is much higher than 1, i.e. are not spheres. When the particles look like spheres, predictions are not very accurate as the percolation threshold becomes highly sensitive to many experimental parameters: dispersion state, surface tension and viscosity of the polymer, and so on. When particles are poorly elongated or poorly flattened, I prefer effective-medium theory (EMT), which is much more realistic. In the case of spheres, the threshold is simply 1/3. You can find several papers comparing percolation and effective-medium theories on my RG profile. You will see that the agreement between both is excellent only for high aspect ratios, whether the filler particles are flakes or rods. But metal particles are rather spherical, so better using EMT.

Alain

I totally agree with Alain (it not the first time, it not the last time !)

Thank you Alain for a very good answer.