AC on/off Corvette at 55 mph data right?

link to column

Looking at the Corvette at 55 mph, something doesn’t make any sense. With window’s up, the difference of having AC on or off is -0.9 mpg. The mpg difference with the AC off and windows up and down is +3.5 mpg. Together I would expect the AC on, Windows down combo to be the base value (windows up and AC Off) + (delta Windows Down) + (delta AC ON) or 36.3, not 29.6.


Yes, the column points out that the AC off/windows up case is strange. It was tested three times, however, and found to be consistently strange. The reasons for it defy explanation. We were tempted to not even report on it to Cecil, but we had to because the data was what it was, even though there may be an error.

The only possible explanation is that somehow the aerodynamics improved remarkably with the windows down - but then why was the AC on/windows down case poorer than the AC on/windows up case?

We twice tested the mileage meter via driving in mixed driving for extended periods, and tracking gas added at fill-up, and found the meter was only off by less than 1% each time.

Given the small difference and the larger difference in the “AC on” scenarios, I’d say that even “consistently strange” after just 3 tests is still in the range of error (not seeing the error bars here).

The message I’m willing to take home at the moment is ‘AC on’ is consistently worse than ‘AC off’ and that 55 MPH corvette with windows down is “intriguingly uncertain and further funding - involving the purchase of Corvettes for all lab staff - is strongly recommended.”

Pet peeve: those graphs should be bar charts, not line graphs. A line graph makes absolutely no sense for category data.

Better yet would be a 2x2 table showing air conditioning and window position as the two independent variables.

We had some debate about this. My argument was that (a) the expected condition was for the line to descend monotonically from left to right, and thus (b) any exceptions would jump out (which I think they do). Since I was in charge of chart-making, I got to decide. That said, if GM calls up and says they’ll give us a big pile of money to replicate our results, only we need to give them bar charts, we’ll give them bar charts.

Can a Corvette really get 37mpg??? Shocking…

I live in Florida, and therefore I feel I am qualified to announce the proper way to get the best summer gas mileage from any car which I’m familiar:

  1. Use the cruise control.
  2. Drive smoothly.
  3. Windows up.
  4. The slower you go, the better mileage you get. My car gets the best mileage at 55.
  5. “Milk” the A/C. This means turn it on, cool down the interior (and air vents), then turn it off. Keep the blower fan on low while the A/C is off, and the vents closed from outside air (recirculate). Even though the A/C is off, the air vents will continue to blow cool for several minutes, depending on how hot it is outside. When you’re feeling warm, turn the A/C back on, and turn the fan up to a higher power. Repeat as needed. This method works better than you can imagine and really helps to save gas. If your car has a thermostat, experiment to get the same result, but the relief from that fresh blast of cold air can only be achieved by turning the A/C on full power for a few minutes.
  6. If at all possible, windows vented or up and no A/C is best.
  7. Over-inflate your tires to 36-38 psi.

I would personnally blame the computing of the gas consumption here.

I work at an Audi dealer and get to test drive cars every day and I sometimes pay attention to the data displayed on-board and have noticed some strange things. A few examples:

  • Achieving over 40L/100km (I’m Canadian, heh, that equals 5.88 miles/gal) travel gas consumption while the engine shifted gears at no more than 4-4.5k RPM ; it was a 2011 A4 2.0T (ok, it’s turbo, but still, it should never exceed 20-25L/100km)

  • Recording a speed of 3 km/h while on park ; I will admit I’m not too sure what kind of speedometer instrument is installed, and it was a 2007 A4, but it’s certainly not converting the rotation of the camshaft into axle rotations either

  • Displaying the “default” trip average consumption of 8.0L/100km (by default, I mean the calculated combined highway/city fuel consumption that is displayed as soon as you start moving) while excessively revving the engine ; I think the guy didn’t pick the part that he had to switch gears when using the manual mode, otherwise the engine would do it automatically when redlinning (and I did repeat myself a few times throughout the test drive), that was on an automatic 2010 A3 equipped with a 2.0 TFSI (gasoline) engine.

Sure, we could blame Audi, but my guess is if these engineers still haven’t completely figured out the science of fuel consumption calculation, chances are they are not alone…or their measuring instruments are inaccurate.

In any case, I suggest you redo the tests with a manual measurement of fuel spent and distance traveled. Because right now, not knowing the margins of error of your instruments is quite unscientific of a method.

Of course it works like you see it. With the windows down the AC has to work harder to achieve a constant temperature. All the cooled air is lost through the open window and is replaced with warmer ambient external air, which requires further cooling. The AC ‘delta’ has to be more with the window open.

All of this assumes that there is some form of temperature sensor and a target temperature in the system.

Twice during the test the mileage was checked via comparing distance traveled with gasoline added, and backed up by GPS. The Corvette had errors of 0.5% and 0.7%. Which is amazingly accurate.

Interesting that the Corvette MPG calculator was so accurate. However it’s measured on my Prius, it’s at least 10% wrong (on the high side, not doubt to make you think your mileage is even better than it already is). I too check MPG by dividing my miles driven versus fuel consumed. Did Dex do the same checks for the Camry mileage? I’m sure his car must have the same gadget as mine.

I check mileage using the instrumentation and calculation every time I fill up on my 2000 Corvette. They are always less than a mile of one another.

I’m not certain if he did; I think so though.

Does that mean you’re the one who screwed u-- I mean, does that mean you’re the one to politely inform about a typo? The topmost right chart says “25 MPG” when I believe it should be “25 MPH”.

Not that I enjoy picking on Ed (I’m sure that’s Cecils’ job), but it does cause some confusion when first trying to read the graphs.

It is hard to believe the Corvette gets better mileage with windows down, but aerodynamics can be weird. I wouldn’t have thought a pickup gets better mileage with the tailgate on than with it off, but according to Mythbusters it does (and they have some plausible aerodynamic modeling to explain it, as I recall). So, even though the windows-down Corvette is even harder to believe, it just might be real.

That’s what we thought at first - the problem is, windows down with the AC on was worse than windows up with the AC on. The only theory here is that since there was air blowing through the car, the AC compressor had to run constantly to keep temperature, rather than cycling…but that’s just speculation.

It has to work this way (see my post above). It’s not speculation, it’s physics.

If the Corvette has climate control (which it better at that price) then the aircon is trying to maintain a specific temperature. If you continually replace cool air with warm air through the open windows the aircon has to work harder to try to achieve that temperature.


Of course. The point Una was trying to make is that if this were simply a quirky aerodynamic effect, you’d see it under both AC on and AC off conditions. But you don’t. If you have the AC off and you roll down the windows, the Corvette’s gas mileage at 55 MPH goes up. If the AC is on and you roll down the windows, it goes down. The latter is of course the expected result. The former isn’t. We have yet to come up with an explanation for this.

What am I missing? Rolling the windows down evidently improve aerodynamics, improving efficiency. But rolling the windows down with the AC on, according to phb1’s entirely reasonable hypothesis, causes the AC to run at full capacity, thereby decreasing efficiency. We have some idea of the strength of the first, aerodynamic, effect (about 3.5 mpg), but no data on the strength of the second (We have some idea of the drag from AC with windows up, but we’re presuming that the AC drag with the windows down is much more). If the second effect is stronger than the first, then overall efficiency would be lower with windows down and AC on.
One way to partially check phb1’s hypothesis is to note whether the compressor/blower is running more with windows down (not sure if this is at all obvious to a passenger, especially with the noise of open windows, but worth asking if Una noticed anything)

I didn’t make separate calculations of fuel consumed vs miles travelled, I relied on the Camry’s mpg calculation. It does calibrate every minute, and I looked at the pattern to be sure there wasn’t any noticable hiccough.

I’m sure that there would be a moderate margin of error in any calculation, unless I drove (say) a hundred miles at each speed – which would have been impossible for the low speeds or for the windows up/no AC in 93-degree heat.

This is the part we can’t figure out. Everybody assumes, and common sense suggests, that rolling the windows down increases turbulence and drag and reduces efficiency. Car designers subject prototype vehicles to wind tunnel tests; it’s implausible that Corvette designers would consciously design the car to be more efficient with the windows down at highway speeds. Even if it were physically possible, why would you do it? Most people use AC and thus have the windows up - that’s the assumption you design for. That’s what makes our contrary result so hard to understand.