EPA Fuel Economy as a Function of Vehicle Parameters

Another semester is over and once again, I can’t stop thinking about systems design. Specifically, as I did last semester, I find myself gathering data and trying to apply analysis techniques from aerospace systems design to try and learn something about the physical limits of fuel efficiency in cars.
 
This time, I started with the EPA’s current list of battery electric (BEV), plug-in hybrid (PHEV), and hybrid cars (HEV)—a comprehensive accounting of fuel economy ratings for several hundred vehicles—to try and get an idea of what the current state of alternative-fuel cars looks like. You can find these at www.fueleconomy.gov. For this analysis, I limited data to MY2025 cars but these included all 2025 BEV, PHEV, and HEV with EPA ratings in the US today with a few exceptions (cars for which I was unable to find other information, such as the Ferrari SF90 and 296, which are plug-in hybrids but for which Ferrari does not publish data such as curb weight). I also excluded 48V “mild hybrids” from this study.
 
As before, I went into this analysis not knowing exactly what I would find. As Carlo Rovelli writes in Helgoland (2020), “If you don’t ask questions, you learn nothing.”
 
Failed Attempts
 
First, I tried charting MPG and MPGe of these cars as a function of vehicle footprint. For Corporate Average Fuel Economy (CAFE) ratings, footprint is defined as the area enclosed by the vehicle’s track (width from centerline to centerline of front or rear tires, along the axle) multiplied by wheelbase (length from center of front wheel to center of rear wheel, along the body).

An example of a small footprint. I took this picture outside a coffee shop across the street from the Jakobikirche, Hamburg (Germany). J.S. Bach auditioned for the post of organist at this church in 1720.

The information is readily available, but it did take a while to collect and calculate it, and all for a result that may as well be a Jackson Pollock painting:

Putting the “scatter” in “scatterplot.”

As you can see, there is (maybe?) a slight downward trend in efficiency as a function of footprint but the effect is not pronounced. I suspect this has to do with a peculiarity of the calculation: performance variants often have wider tires and thus narrower tracks (since the calculation is based on the distance centerline-to-centerline), giving a smaller footprint for the (often) less efficient trim.
 
Next, I decided to plot efficiency as a function of drag area but ran into another problem. There just isn’t enough published data on aerodynamic drag for these cars out there. What little I could find from reliable sources resulted in this chart, for HEV and PHEV:

We can see a downward trend. However, without enough information to flesh it out it’s hard to determine how reliable it is or how much scatter there is in the data (which, based on the Mercedes S580e at the lowermost left, may be significant).
 
Success
 
There is one metric that is readily available, easy to collect, and requires no calculations or estimation: curb weight. Manufacturers almost without exception include this under the specifications of new cars, often broken down by trim. So, it was pretty easy to find this for every car on the list—BEV, HEV, or PHEV. What’s more, the trends this showed were robust and surprising.
 
First up, BEV. As one would expect, there is a pretty clear relationship between efficiency and weight for each of the three EPA ratings:

However, there is still a fair amount of scatter around the main grouping. This may have to do with different battery charge and discharge efficiencies, motor efficiencies, tires (some manufacturers co-develop tires specific to their EV range with tire producers, while others use off-the-shelf tires), or aerodynamic drag. For example, the most efficient car on the highway and combined plots is the Lucid Air, which is moderately heavy but has a very low drag coefficient.
 
Overall, though, this trend seems reliable for BEV. How does it compare to HEV?

Well, well, well. The groupings here are so tight and clear that we’re able to tell the trend is quadratic just by looking at these plots (the two outliers are the Lamborghini Revuelto and Corvette E-Ray, which are not typical hybrids). And this doesn’t really change if we add PHEV, which are overwhelmingly just variants of HEV with added battery capacity (and weight):

The PHEV figures here are running on gas alone. If we compare the plug-in economy of these cars with BEV, it looks like this:

A couple of quirks show up clearly in the data here. Some PHEV still turn on the gas engine if the throttle is depressed past a certain point, and even more of them don’t have enough battery capacity to complete the EPA test cycles on electric power alone and have to use some gasoline. Both of these show up as decreased efficiency of PHEV compared to BEV. This does not show up when PHEV running on gas alone are plotted with HEV since both groups are on an even footing, and the HEV are generally more efficient than their heavier counterparts.
 
Discussion
 
Does this tell us anything useful? I think so.
 
First, I think these results show that weight can be a good predictor of efficiency. This shouldn’t be surprising for several reasons. Weight scales pretty well with the overall size of a car when looking at one group (BEV, PHEV, HEV) alone. Physically larger cars in any group tend to weigh more than physically smaller cars with the same drivetrain (note that this is not true across drivetrains, where BEV tend to be several hundred pounds heavier than HEV of the same physical size). Also, the energy a car uses has to balance three “sinks”: work required to overcome aerodynamic drag, work required to overcome rolling drag, and kinetic energy (I’m leaving out potential energy since that doesn’t figure into EPA testing). Two out of these three are proportional to vehicle mass. What’s more, even in these three vehicle categories which are all differentiated from conventional gas cars by their ability to recapture kinetic energy through regenerative braking, the kinetic energy “sink” appears to still dominate and thus the effect of vehicle weight on efficiency.
 
Second, these data show that future improvements in vehicle efficiency can be driven by reducing weight. Especially in hybrids, efficiency scales clearly and directly with weight—regardless of wheel size, aerodynamic drag, body style (crossover, SUV, truck, sedan, minivan…), or other fundamental aspects of these vehicles. That is surprising, but at the same time it shows a clear path forward to improving the efficiency of our personal vehicle fleet: make them lighter.
 
The clear implication of this realization is that, as home modifiers with a goal of improving the fuel efficiency of our personal cars, it might be worth our while to investigate weight reduction. I’ve experimented with this in the past—the biggest reduction so far (~40 lb) coming from removal of the rear seat and replacement with a camping platform built out of lightweight poplar—and this summer I’ll spend some time thinking about it and modifying what I can.