[Geezer]

From Geezer:

It's a lot more than experimental error. That would only apply if all conditions were carefully controlled, but in this case there are so many uncontrolled variables that the results could be very misleading. I would think a simulation with accurate input data would yield much more meaningful results.

Geezer, in our local dialect of the English language "experimental error" means "the uncertainties arising from design, execution, and measurement in an experiment". It is hard to see how the effects that you refer to, which are undoubtedly real ones, are "...much more than..." this..

Ah yes, but if I'm not mistaken, you proposed to run your experiment for a month.

So, it might be a teensy bit tricky to control some things that could have a significant influence on the outcome. A few that spring to mind are: Temperature, precipitation, wind speed, wind direction, humidity, road surface, tire wear, tire pressure, route, load, driver, driver's level of intoxication, mother-in-law in the back seat,.......

I'm sure this is a less than exhaustive list and that you will be able to add quite a few more. My point is, by the time you are done, the summation of these variables could easily swamp any meaningful result. That's why I think a simulation could be a lot more useful, although a carefully controlled "rolling road" environment with a real vehicle would probably be even better.

I wonder if that explains why the automotive industry does it that way?

[CliffordK]

If you could build an accurate fuel consumption gauge, then you could significantly reduce the time involved in the experiment.  And, thus control many other variables, and/or have more trial runs.

I remember a Mythbusters episode where they rigged some kind of a graduated cylinder to the fuel system, so they could run experiments with about 1 liter of fuel at a time.

[Geezer]

Being a glutton for punishment, I took a shot at some actual numbers (I probably mucked it up somewhere too.)

Vehicle weight 1,000 kg
Force on tires 9,810 N
Coeff. rolling friction assumed 0.01

Force to overcome rolling resistance 9,810*0.01 = 98.1 N

Power consumed at 5.55 m/s (20 km/h)  98.1*5.55 = 545 W (0.55 kW)

At 80 km/h (50 mph), power consumed = 2.2 kW (almost 3 HP)

[CliffordK]

At 80 km/h (50 mph), power consumed = 2.2 kW (almost 3 HP)

Whew, not as much as I would have thought.  But...  perhaps representative.
Plus, of course, wind resistance, acceleration/Braking, etc.

With my Ranger EV, I can watch power output. 
Acceleration, I usually hit about 200A x 144V, or about 28KW (and it is pretty poky at that).
Cruising at about 45 MPH, I usually hit about 50A to 100A x 144V, or about 7 to 15 KW.
And, it is a pretty heavy vehicle with all the lead in the back.

Unfortunately I only have an instantaneous reading, not a trip reading, but it is not too far out from Geezer's calculations.

One thing in the Electric Vehicle discussions is that with Lead batteries, at some point with a range of about 40 or 50 miles, adding more batteries (equivalent of fuel) no longer significantly extends the range due to the increase of weight.
--------------------

Apparently any ODB-II equipped car can give you real-time mileage estimates.  Perhaps not as accurate as readings from the pump, but they could be used for comparitive purposes.

http://www.smarthome.com/93001/Linear-Logic-SGE-ScanGaugeE-Fuel-Economy-Gauge/p.aspx

Unfortunately, I don't think I have any ODB-II compatible vehicles,  [xx(]

[Geezer]

Yes, I was a bit surprised the power is so small. Maybe there is something wrong with the calculation, but it is quite straightforward (I think!) Perhaps the coefficient is too optimistic, but 0.01 seems to be around the middle of the range.

[Geezer]

Mind you, here's the other side of the equation: Lets' say our 1000 kg vehicle manages to do 30 mpg (US) while travelling at a steady 50 MPH. That means it consumes 50/30 = 1.66 gallons per hour

Switching to SI units, that's 3.7854*1.66 = 6.31 litres per hour, or 1.75 ml per second.

1 ml of gasoline has a thermal energy of 34 kJ, so thermal energy is produced in the engine at 1.75*34 = 59.5 kJ/s, or 59.5 kW (imagine how quickly that would heat your house!)

If we are optimistic, 20% of that, about 11.9 kW (16 HP), might make it to the wheels.