Another article desribing EPA's methodology whereby they run a series of very rigorous dynamometer tests and then *arbitrarily* reduce the final range number by 30%. This article didn't discuss how they compensate for different different vehicle aerodynamic characteristics.
Here, aero is briefly touched upon:
Edit: one of the commenters in the above article posted this:
This is how it works: The car is driven to about 75mph, put in neutral, and allowed to coast on a level road. The speed/distance is measured over time, and using F=ma, you can figure out the force imparted to the car from all sources (air drag, road loss, etc) as a function of speed.
During the efficiency test, the dyno then applies a resistance force to the tire that equals the sum of acceleration force (at the test cycle’s current acceleration) and the force from the above measurements (at the test cycle’s current speed). There’s also a correction to not double-apply any losses (i.e. rolling losses from being on the dyno).
It’s a great test if done right. However, if manufacturers cheat with the data put into the dyno, they’ll get a BS result.