Even before the advent of partially- or fully-electric cars, it was becoming increasingly apparent that the old fuel economy metric of miles per gallon isn't as useful for measuring energy consumption in vehicles as when it was first codified in the original Corporate Average Fuel Economy standard in the 1970s. That is due in part to the proliferation of new fuels--E85, LPG, LNG, CNG, methanol, and hydrogen--but also because expressing the relationship between distance and volume in this way obscured the diminishing returns to higher levels of fuel economy. As a Wall St. Journal column earlier this week put it, adding electricity into the mpg mix, "risks giving consumers inaccurate information about the financial and environmental costs of driving." But if we need a new metric, what should it measure?

I've been interested in this issue for some time, and GM's recent announcement that its new Volt plug-in hybrid achieves 230 mpg in city driving prompted some further thought. I don't doubt the accuracy of that figure or the thought that GM's engineers put into bridging this new vehicle type into a system that was designed when the average US fuel economy was 13.1 mpg and unleaded gasoline was the newest fuel around. Yet all this figure tells us is how much liquid fuel the car's generator would consume over a carefully-chosen driving interval, completely ignoring the electricity--with its cost and consequences--required to deliver that result. Nissan's Twittered riposte that it's new Leaf electric car gets 367 mpg is even less useful, because the assumptions behind it are not clear--and might just ignore some basic engineering realities.

Without access to Nissan's calculation, I can only guess at how they might have arrived at it by backing into it. (Skip this if you hate numbers.) Start with the fact that each gallon of petroleum gasoline (without ethanol) carries 115,000 BTUs of energy. At an official conversion of 3,412 BTUs per kilowatt-hour (kWh), that equates to 33.7 kWh per gallon, so 367 mpg implies that the Leaf would go nearly 11 miles per kWh. That's pretty amazing by itself, considering that the Volt is generally expected to go between 4 and 6 miles per kWh. It also suggests that the Leaf would be using less than half of its 24 kWh Lithium Ion battery pack to deliver its advertised 100 mile range. But even if this is all correct, there's a basic problem with the calculation; in the real world it can take a lot more than 3,412 BTUs of primary energy to generate one kWh of electricity, depending on how you do it. If the power source is surplus wind, solar or nuclear power that wasn't already being used to displace power generated from fossil fuels, the BTUs required could be effectively zero. Otherwise, for power generated from coal or natural gas they would range between 6,000-12,000 BTU/kWh. Even assuming a relatively conservative 8,000 BTU/kWh for the natural gas turbines that provide the incremental power supply for many markets, the resulting equivalent mpg falls from 367 to 156 mpg. But that still doesn't tell us enough, in my estimation.

The problem here is the existence of a variety of perspectives on vehicle energy efficiency with competing information needs. From the standpoint of energy policy, we are most concerned about annual oil consumption and greenhouse gas emissions. We already have a new federal mileage standard that is set in terms of grams of CO2-equivalent per mile, which gets at the latter issue. The EPA's current mpg methodology based on liquid fuels comes close to addressing the former, though the increasing contribution of biofuels renders it suspect. Unfortunately, any standard or metric that treats non-petroleum energy as essentially free seems certain to result in colossal unintended consequences, as non-oil energy sources ramp up. The engineer in me would argue strongly for something like the MPGe calculation used for the Automotive X-Prize, comparing all the energy delivered to the car in any form with how far the car went. However, from a consumer perspective that still seems overly complex and opaque. While I would certainly prefer the inverted form of fuel economy--gallons per 100 miles--to our current mpg, it's hard to beat miles per dollar as a means of comparing how much it will cost the average driver to operate any of these new cars.

Money is the common denominator for most of the things we consume, so why shouldn't it be for vehicle energy, as well? At current pump prices, an average American passenger car goes about 9.5 miles per dollar (mp$), while a Prius-type hybrid approaches 20 mp$. If we factor in electricity at the national average retail price of $0.11/kWh, then the Chevrolet Volt would deliver something in the vicinity of 30 mp$, if I've correctly understood how they arrived at their 230 mpg figure, while the Leaf might yield as much as 99 mp$--though my natural skepticism about its unofficial claims leads me to suspect it would be closer to 45 mp$. Of course, when you have to pay $5,000-10,000 extra for a battery pack, you'd certainly hope the operating cost per mile would be a lot lower than for a conventional car. And that's precisely the kind of comparison that a truly useful fuel economy metric should facilitate.

In the near term, the EPA should continue its work on adapting the familiar mpg metric to a new world of more diverse vehicle technologies, but for the longer term it ought to convene other government agencies, car and fuel companies, universities, and consumer groups for the purpose of developing a new and more helpful set of metrics that would tell consumers what they need to know about costs and consequences as the car fleet undergoes its long transition toward an uncertain destination.