I'm taking down some notes as I read this, so please bear with me.
On page 6, you state:
It can be argued that an exponential function the the data for
the pilsner malt grist better than a linear function.
I assume you mean the data for pilsner malt is better fit by an exponential function rather than a linear one. It's just a typo.
I've only read through the paper once so far, and one of the key things I take away from it is the difference in mash acidity for roasted vs. crystal-type malts. That suggests to me that when estimating salt additions via Palmer's methods, greater emphasis should be placed on the relative proportions of of crystal or roasted malt, rather than simply the estimated color of the wort. Any comments on whether the ranges of RA Palmer recommends for a given mash correlate with your acidity data?
Your work appears to provide a framework for more closely estimating mash pH via the summation of each grain's contribution to mash acidity, rather than by relying on the color alone. However, what doesn't seem clear to me yet is whether that increased accuracy represents a real improvement for the home brewer, or if Palmer's approximations are good enough.
I can sense a lot more math coming up ahead of my next brew. Guess I need a new probe for my pH meter, too.