Hi AndrewJ,
MetroMPG contacted me through the Instructables site and asked if I would comment here since I created the spreadsheet you're using. Unfortunately, the lower your Cd is, the less accurate the spreadsheet is since the contribution of Crr dominates the results. I punched in your numbers and the data comes out almost linear which means there isn't enough change in drag force during the test to accurately isolate the contribution of Cd from that of Crr.
The easiest way to improve results would be to determine Crr from a different test and enter that value directly. Then run solver again, changing only the Cd value to match the model to the data.
To measure Crr directly you can use a spring scale (50lb capacity should be enough). Measure the force required to pull your car at a constant speed (a walking pace) on level ground. Crr = Frr/M/g where Frr is the force in Newtons (1lb = 4.45 Newtons), M is vehicle mass in kg, and g is 9.81. Measure the force to pull the car forward in one direction, then turn the car around and measure the force pulling forward in the other direction. Use the average force to calculate Crr. Drive the car for a while first to warm up the drivetrain.
Hope that helps.
