In 2012 we (33) did math using the old JVN sheet (before battery voltage calculation was added) assuming 11v instead of 12 or 13.2v (this was how we compensated for the voltage drop in acceleration). The model did not correlate well and we did not select the distances well for time-to-distance either, but it wasn’t all the models fault - the CAD for that robot had 8 drive motors, 2x CIM 2x RS550 per side, with the intent to run 3 of the 4 and decide which 3 later, but we only ran 2 CIM. Before CMP we did extensive data capture on the practice robot driving long straight lines with 3 gear ratios (AM shifter low, AM shifter high 2.56:1, AM shifter really high 4:1). Final drive was changed for 2.56:1 and 4:1 final drives so the real 2.56:1 was somewhere in the middle (I think theoretical free speed went from 18fps down to 14 or 13fps - from memory we swapped a 22t with an 18t final drive sprocket). As part of this we determined that our speed loss constant was actually much higher than 81%, we were achieving about 90% of our design speed, and we later exceeded 95% on designs with fewer gearing stages and wheels.
Based on all of that data, I wrote the voltage simulation and tweaked some of the constants a bit and the data matched really well. This is where the battery resistance came from.