2016 Odometer Error

Mendel, more than some people seem to think. The statement "to expect averaging to make variables irrelevant is a misunderstanding of statistics." is in itself a misunderstanding of statistics, not to mention the scientific process.

It seems best to view the Prius' MFD numbers as entertainment. As fuzzy1 points out, there is no legal pressure for them to be accurate, whereas gallons pumped are supposedly certified to within 0.5% in many states, an acceptable limit given the difficulty of measuring to greater accuracy. As I said, the fact that the Gen 3 Prius' MFD numbers invariably overstate the MPG versus ODO/gallons pumped by a consistent percentage whereas the Insight 's MFD did not convinces me that something fishy with the Prius' methodology is introducing a systematic error.

Thanks to GPS it's easy to check odometer accuracy and correct that to a reasonable level of precision.

Unless there is input from the moderators, I'll determine MPG as gallons pumped divided by GPS-corrected ODO reading, as Wayne Gerdes does. I'll probably gather GPS data on several runs to see how consistent the correction is. If it is consistent, I'll just use an average correction factor for the ODO reading and recheck it from time to time.

I suppose sloppy numbers can be used "for reference" to compare to other sloppy numbers determined in the same way, but they don't have much value in comparisons to the real world. Ironically, we have a government bureau, the NIST, devoted to accuracy in measurements, and they use "reference standards" as the bedrock against which methods are calibrated.

 

I think the Toyota engineers are succumbing to internal pressures, the numbers are invariably and consistently at the edge of credulity, lol. It's so obvious as to be an embarrassment.

And, I don't really buy into the thesis, that inaccuracies and variables compound: my experience is the longer you keep records the more accurate, and unbudgeable, the results.

With close to 120 fill ups calculated, I'm seeing an unwavering 7.5% positive bias, minor oscillations to either side not withstanding.

Meh, should have held fast on "what do I know".

 

I also seem to remember other comments about the Civic Hybrid being closer, even low, and the Gen2 Prius being closer than the Gen3.

 

Yup, our 06 Civic Hybrid was either spot-on, or low by 0.1, in the liters/100km system. It ain't hard.

Of course, the numbers were more likely be 6 point something, vs 4 point something with the Prius, but at least they were honest, lol.

 

I found even the estimated 5% error on my 2011 Gen 3 to be extremely annoying. If I had good conditions and got, say, 70mpg on the MFD, I'd be really pleased until I realized it was probably more like 66.5mpg. Phooey!

At least subtracting 5% is easy: divide by 10, take half of that, and subtract. None of those gymnastics with the Insight-I.

 

If our display is (for example) 4.5 liters per 100 km, calculated is invariably 4.8 or 4.9. Very easy accommodation, but yeah: frustrating. It's like that hyperbolic acquaintance...

 

I'm not sure, but I would suspect if my fuel economy increased significantly, the .3~.4 difference would become .2~.3. In short: the percent would hold, not that arithmetic difference.

 

Thanks for doing this. It seems that SpeedTracker is accurate enough for our purposes so we don't have to resort to more painstaking methods to measure actual mileage.

I have Toyo tires and so far have done three calibration runs using SpeedTracker, as noted previously:

I have to do a few more runs to pin it down more closely. You are finding a larger error, perhaps car-to-car variation or perhaps a difference in the tires Toyo vs Bridgestone.

Incidentally, SpeedTracker has another screen that displays MPH, which can be used to check the speedo if one cares.

 

I guess it is better everybody's odometer is reading less mileage rather than higher. I think California allows odometer readings to be plus or minus 4% off. Would be more concern with manufacturers being on the plus side to speed up warranty expiration.

 

The plus side has already been hammered away by class action lawsuits over warranty fraud. I know that both Honda and Subaru were hit up, as I received the legal notices for cars I owned (neither of which actually had the purported 'defect'). If other car makers were hit, I missed them due to not having subject cars.

 

Thanks for finding this. The difference in correction factors we see may be variation between different cars?

 

I think Nissan was the other. I am probably like most people and take it for granted that odometer is correct and more concern with the speedometer when switching out wheels and new tires.

Now I am wondering if I am really getting the miles between each fill up or not.

 

You will be getting slightly more miles, from what's been reported above.

 

That's exactly what most car manufacturers used to do, before the class action suits scared them into being honest, or erring appreciably in the opposite direction in an overabundance of caution. I checked odometers on a lot of cars in the old days. Most showed 1-6% high, a few even more. My own Subaru showed 5% high. My Mazda, 1.8% with the original tires. Cheating on the warranty was not their only motivation to exaggerate distance.

 

Let's try again.

Admittedly, "variables" might've been an imperfect choice of words on my part, so I'm sorry if that's the cause of your misunderstanding my point. However, note that I did not say every variable you can imagine is irrelevant. I wrote "those variables," specifically meaning reasons (some examples of which were described in the post I quoted) why the amount of fuel in the tank after fill-ups can vary. Obviously those inevitable filling inconsistencies can cause a sizeable short term error in the ratio of true mpg to displayed mpg.

I hope we can agree that the average number of miles per gallon a car scores over a distance long enough to include several fill-ups should equal the overall distance traveled after an initial fill-up, divided by the sum of fuel volume added at the final fill-up plus all other fuel volume added in the meantime. That means filling inconsistencies in the intermediate fuel additions (or whether they even approximate complete fills) are---yes, I'll say it again!---totally irrelevant. Small inconsistencies between the initial and final fill-ups are not 100% irrelevant, but will diluted to the point of practical insignificance.

If, over any group of several consecutive purchases, I compare actual fuel put in the tank to the the total volume the car claims it consumed (considering that claimed miles divided by claimed mpg=claimed gallons each time I read and reset the trip meter), the ratio of claimed volume to actual volume is remarkably consistent. (Actual mpg to displayed mpg ratio would therefore be the same, ignoring for now the much smaller odometer and pump errors.) That means this car's fudge factor is consistent.

If you've kept good records and similarly analyze the numbers from your car, then do that again for another period, you'll likely also find a consistent fudge factor. It probably won't be the same as mine, even on another 3rd generation, due to manufacturing tolerances, etc. I'm getting a 5% error, plus or minus a tenth or two. Other credible or semi-credible PC members have reported different consistent discrepancies, from maybe 4% up to about 8%. I don't know yet what the corresponding range is for the 4th generation. The fact that all of the 3rd-generation Prii mpg displays, even the best, are exaggerating says to me that a degree of exaggeration is very likely intentionally programmed-in by Toyota. If you'd find that to be acceptable, good for you. I don't like it, but have to accept it, and can correct for it by applying my fudge factor.

 

If you are averaging averages, you are probably getting the wrong answer.
See en.wikipedia.org/wiki/Mean , for measurements of speeds, mpg's, you need to use the harmonic mean.
An arithmetic mean will give you consistently lower numbers.
Or just keep raw data, miles and gallons, add up the totals and divide.