I think Ed Fowler has a method by which he estimates grain size on his bevels using ferric chloride as a stain. I might have misunderstood what I read but I think this is what he does. You should be able to measure/estimate grain size using optical microscopy. i.e. 1000X should be able to discern grain sizes well below 10. Your model is untested re: grain size. You need experimental evidence to support your claims.