Every time I take out a test car, I do a mileage check on it. I start with a full tank, drive it for a week, fill it up, and then do the math. And much of the time, I don't achieve the vehicle's official mileage figures.
I always thought that the official figures were achieved by taking out the vehicle on a cloudless day, dropping by the racetrack to find the smallest jockey there, and having him drive it on a glass-smooth road. But it turns out that, compared to the way it's actually done,that would be too realistic.
Instead, according to Natural Resources Canada, here's the story. First of all, the government only compiles the figures (and then prints them in a guide that you can find online at ecoENERGY, or as a hard copy by calling 1-800-387-2000, or dropping by most new-car dealers who will give you one at no charge).
Instead, it's the car manufacturers that do the actual testing, in laboratory conditions (which, in fairness, eliminates outside variables). They use a procedure standardized to government specifications, on a two-wheel chassis dynamometer, with a new vehicle that's got about 6,000 km "run-in" on it. Four-wheel and all-wheel-drive vehicles are tested in two-wheel drive mode, and then the tests are "adjusted to reflect the increased weight and engine load using 4x4 and AWD systems," according to the government's website. On all ratings, "other adjustments are made to reflect the average fuel consumption of vehicle configurations, options and sales mixes sold in Canada," although it doesn't specify exactly what the adjustments are. The automaker submits the figures, and only if they seem truly out of whack does the government get involved in the procedure.
In any case, the automaker has to provide figures for both city and highway use. So on the city course, the test simulates a 12-kilometre, stop-and-go-trip, with an average speed of 32 km/h, and a top speed of 91 km/h. The test runs for 23 minutes and includes 18 stops, with about four minutes of that test time spent idling to represent waiting at traffic lights. The test begins from a cold engine start, similar to starting a vehicle after it has been parked overnight during the summer. When the test is completed, it's done again with a hot engine start, and the first eight minutes of the test are repeated, to simulating restarting a vehicle that has been warmed up, driven, and then stopped for a short time.
Now we're on to the highway, which simulates a 16-kilometre trip, with an average speed of 77 km/h, and a top speed of 97 km/h. The test begins from a hot engine start, and runs 31 minutes, with the speed varying to simulate different kinds of highway and rural roads.
There you have it: slightly over an hour for everything, at a top speed that is not even the posted maximum speed limit on a Canadian limited-access highway.
Remember that, all you readers who write in on a fairly regular basis to complain that my week-long, 200-km-or-so, real-world-driving fuel figure isn't "realistic." I'm already over the official tests long before I even get the car home.