Real MPG Part 2: It’s Report Card Time
Motor Trend’s Real MPG Program At 100 Cars: Some Better, Some Worse, And Several Surprises
When we commenced the Motor Trend Real MPG program six months ago, we weren't exactly sure where its path would wind, but we knew we'd eventually wind up here -- with a stack of mileage data from 100 cars and trucks in front of us. This is a milestone at which to plot some graphs, rub our chins, and ask ourselves what all we've learned. As it turns out, lots.
Before presenting our high-octane insights, allow me to run the highlight reel of reasons why we've gotten here. Lights, please.
For starters, the mileage measurements you see performed by the media are dubious at best. As a group (including us), we've attempted just about every cockamamie scheme imaginable, from weighing the entire car before and after a mileage run (laughably imprecise) to data-logging the fuel tanks' calibrated drainage via the OBDII port (OK, that one was my idea). All these were patently clumsy side steps around the most obvious approach: inserting a high-quality (expensive) turbine-type fuel-flow meter somewhere in the fuel line. But installing these is easier said than done, kemo sabe. And while rearranging the fuel lines of press-pool cars (which get rapidly swapped among media outlets) has its comic possibilities, it would also draw the ire of manufacturers and attract many unsmiling lawyers (or worse, smiling lawyers).
Enter our partner, Emissions Analytics, which has created a brilliant solution that's precise and noninvasive (and complex and expensive, which is why we didn't come up with it). Way back in the January issue, we presented a fairly comprehensive explanation of how it works. Here's their one-sentence recap: We calculate the fuel's burn rate by capturing the exhaust pipe's gaseous emittance; measure its flow rate, analyze a sample of it to determine its concentrations of CO and CO2; and then walk back the math to figure out -- every single second, mind you -- how much gasoline is being burned to produce these numbers.
The good news is that no modifications to the car were required. The bad news was all the wrinkles we had to iron out, including learning to check the entire exhaust system for leaks, fabricating a zillion (OK, about 25) individualized tailpipe adaptors, keeping the instrumentation running night and day to maintain its sensor's stability, and, critically, establishing a supply of chemically consistent fuels.
For this, we turned to our fuel partner, Chevron, which has brewed 15,000 gallons of regular, premium, and diesel fuels specifically for this project. Why the bother? Gasoline's energy density can vary by about 4 percent among stations, so a program like this had better start with an absolutely known fuel source. Our fuels (of deliberately typical blends) are held at Chevron's Richmond, California, refinery and delivered in barrels to another partner, Dion & Sons in Long Beach, which decants these into 5-gallon batches for delivery to our crew. Chevron periodically tests samples from both Richmond and Long Beach to monitor the fuel's condition.
As critical is EA's proprietary, 88-mile driving loop. Capturing a rich assortment of real-world driving circumstances, it's folded with repeated segments (to check consistency), punctuated with two-direction laps (for wind compensation), and spotted with episodes to specifically illuminate air-conditioning load. How about traffic? Gridlock that's beyond our software's compensation flags a repeated test. But typical traffic variations are actually welcome, as they provide insight into how a car responds to them (as we'll see shortly). And let me point out that the route, an agnostic collection of city and highway elements (later weighted 55/45 percent as per, but not a clone of, the EPA cycle), provides a consistent, balanced setting for real-world mileage-measuring. That's unlike typical magazine long-term testing, which is often colored by an individual driver's commute route and driving style, or an unpredictable mix of multiple drivers.
Multiply that 88-mile loop by 100 cars (with perhaps 15 percent repeated for verification) and you'll see we've driven more than 10,000 miles chasing insights about real-world mileage. So what have we learned? Let me get my pointer.
In the first graph, you can see that our cars' Real MPG city and highway results are fundamentally clustering around their EPA versions, with nearly as many above the "perfect agreement" 45-degree lines as below. The average difference between the car's EPA and Real MPG results is 0.4 percent -- pretty darn close. Nevertheless, they rarely coincide, instead appearing as clouds (at its extremes, 18 percent better and 17 percent worse), and therein reside our insights (graphs 1 and 2).
That scatter you just saw isn't as random as it looks, though. If we average each manufacturer's Real MPG results, some brands appear to be trending better than their official numbers, others, worse (graph 3). Don't put much weight on this, though, as it's early days with just a handful of results per manufacturer percent -- plus we're fully aware of other evidence suggesting different trends. We'll be keeping an eye on this one, nevertheless. In graph 4, we've applied the same treatment to the popular flavors of propelling cars (regular gasoline engine, premium gas, diesel, hybrid, and plug-in hybrid). For premium, regular, and hybrid, our results closely jibe with the EPA's; the diesels, though, are using 3.4 percent more gallons per 100 miles; the single plug-in example, 8.4 percent less. However, an example of one -- or even eight, in the case of the diesels -- is really only a tantalizing suggestion at this point.
Ever wonder what that "your mileage may vary" sentence at the bottom of the TV ad really means? By far, the biggest variation is your particular mix of city and highway driving, which on average can account for a 26-percent swing in mpg (graph 5). Air-conditioning use plays a much smaller role (graph 6), but, interestingly, more so during city driving than highway, as a slower average speed means more minutes of A/C operation per mile while higher speeds naturally aid in cooling.
Question: Has engine downsizing made smaller engines more sensitive to driving styles and road congestion? What we've found is that, not surprisingly, mileage drops as displacement increases, as shown by the middle curve in graph 7 bracketed by indications of the best and worst extremes. But aggressive driving considerably penalizes cars with smaller-displacement engines (graph 8), though to a lesser degree, the opposite happens in congested traffic (graph 9). Last, we have the impact of both traffic congestion and hot-foot driving regardless of displacement (graph 10). On average, there's an 11-percent swing.
Recently, the Federal Trade Commission began a review of its 39-year-old guidelines for how car companies advertise mileage claims. There's consumer confusion, evidently, over those "40 mpg!" claims in ads -- without explaining that it's the cherry-picked EPA highway number. We're all for the review, and, FTC, if you want advice on real-world mileage claims, give us a call. In the meantime, we'll be methodically testing our way through our second 100 cars, and gaining even more insights along the way. See you at 200!
Definition of Terms
Aggression Penalty This represents how much mpg suffers as the average rate of acceleration increases. Specifically, the Emissions Analytics data model predicts the impact on mileage were your acceleration rate double that prescribed by the test route.
Air-Conditioning Use Throughout the driving cycle, each car’s a/c is repeatedly switched fully on, then off, to sample its influence on mileage. Emissions Analytics sensors can detect the changes in CO and CO2 produced and, consequently, its impact on mileage.
Congestion Penalty The congestion penalty is similar to the aggression penalty but a bit more complex. Here, it predicts a mileage “penalty” based upon two parameters teased from the data: The increased number of stop/start events and a lower average “city” speed.