2016 Odometer Error

I know of no legal pressures, so far, that force the car's MPG display to be accurate.

There are legal pressures to make sure the speedometer does not read too low, and the odometer does not read too high. Also, there are legal forces to make sure filling station fuel stations pumps don't deliver less fuel than their displays show, and financial forces to not err significantly in the other direction.

Drivers can easily check speedometers and odometers against other references. Fuel pump accuracy is not so easily checked, but the Gen3 Prius MPG errors we see would happen from fuel pump errors only if the gas sellers were giving us a lot of free fuel out of the goodness of their hearts. If you believe that, I have several bridges to sell you.

Different models and vintages of cars have reputations for different displayed MPG errors. E.g. my 2014 Forester MPG error is less than 1%, much more accurate than my 2010 and 2012 Prii, even despite a much larger odometer error (in the direction that avoids class action lawsuits). If drivers were calculating something wrong, or if the main MPG display error was caused by inaccurate fuel station pumps, then we should not be seeing this difference among car models.

I believe the MPG display errors are built in at the behest of the car maker marketing departments.

 

Let's go over the basic metrics:

1. distance/speed - these come from the brake controller that integrates the four wheel encoders. The primary ones are the front two wheels. So I am running oversized tires on the front wheels of our 2003 Prius and those dictate the reported speed and distance. On our 2010 Prius, the replacement tires are identical and sized to bring the odometer into GPS and mile-marker agreement. So to get accurate odometer, trip meter, and speeds, we first changed the front tires for a set whose 'revs/mile' calibrate the odometer and speedometer. Then we bought the rear pair and moved on.
   
2. MAF (Mass Air Flow) meter - after 45-55 seconds, the Prius maintains a stoichiometric air-to-fuel ratio of 14.7-to-1 which is critical for the three-way catalytic. So it is trivial to calculate the fuel consumed by the mass of air consumed, grams/second. Fortunately, the MAF is temperature compensated so it handle sub-freezing to over 100F air.
   1. Note that the Gen-1 and Gen-2 had to limit exhaust temperature at high power settings to avoid burning out the catalytic converter. For example, the Gen-1 runs a slightly richer mixture above 3,900-4,100 rpm. This is handled by adjusting the fuel trim.
   2. Before the catalytic converter reaches operating temperature, the engine uses built-in and learned trim values. This typically takes 45-55 seconds during which the fuel consumption tends to be higher.
3. Injector timing - there is a linear relationship between the volume of fuel injected and the MAF. However, the density of gasoline varies by feed stock and temperature. Furthermore, there is no metric for fuel density or temperature.

The easy solution is get the current, front tire, revs/mile from the manufacturer web page. Tire Rack also collects these specs. Knowing error between the current front tires and GPS/mile marker, select a pair of replacement tires whose revs/mile eliminates the error. Buy them and verify the error is eliminated and the problem is solved. The rear tires can be replaced whenever.

Bob Wilson

 

Fuzzy1, I think this is backwards. My Gen3 MFD displayed tank MPG values that were consistently 5% HIGHER than the MPG values I got by dividing odometer readings by gallons pumped. Factoring out roughly 1-2% odometer error, that leaves 3% HIGHER. That would mean that the gas pump numbers were HIGHER than the gallons used as reported by the Prius' computers. It doesn't say which number is wrong. But it seems unlikely that the gas stations would be consistently charging me for almost exactly the same percentage more gas than I actually pumped. And as you point out Toyota does have a motive to make the MPG look a little better than it actually is, so perhaps the computer is set to read high, or the method tends to err on the high side?

So we are stuck knowing distance traveled reasonably accurately, but with gas consumption measured by two different methods which disagree by several percent (a systematic error, not variability). And we don't know which method is more accurate.

Here are three sets of odometer data I got with my Model 2 ECO on the Toyo 15" tires at 45psi:
ODO 72.1, SpeedTracker 72.6, correction factor +1.53%
ODO 168.2, SpeedTracker 178.1, correction factor +1.13%
ODO 106.2, SpeedTracker 107.4, correction factor +1.13%
No more data until I return in a couple of weeks.

My only tank so far was 700.9 miles on the ODO, adding 1.13% would give 710.0mi, dividing by 10.477 gal pumped = 67.8 MPG
The MFD displayed 69.0 MPG, a difference of +1.8%

 

Gas stations are allowed to have a maximum volumetric error while pumping fuel. Of course the pumps precision is way higher, so they are programmed to always be at that maximum legal error in favor of the fueling station. That would explain the consistency in mpg overestimate.

I don't know how much this error is where you live. Over here is a 3%, and that's the less you are gonna be deceived when refueling in the best case scenario.

I have measured up to 20% less tank performance once with a Honda Odyssey. This can be done mixing alcohol or ethanol with gas so the energy density is lower an the mpg will suffer instantly without being noticed in the tank gauge or pump display.

Nexus 10 ?

 

Nearly universally in the US, gas pumps are inspected at 5 and 10 gallons. Can you guess where they are the Most accurate ?

 

I have noticed with our 2010 Prius in warmer weather my last fillups were about 2.5 mpg off than computer, 58 mpg indicated 55.5 calculated mpg. In Winter its usually 3-3.5 mpg off.

 

@JohnF: It could be with the 2016 Toyota has reduced the displayed mpg error.


merged.


Here's the last year of calculation for our 3rd gen, displayed error highlighted:

 

Can you calculate statistics for consumption and error? Mean value and standard deviation will tell a lot.

Nexus 10 ?

 

Only laboratory conditions will result in truly accurate measurements, and even in the laboratory there are known and unknown variations and understood variation. No vehicle I know calculates fuel consumption to molecular levels, but relies on volumetric information. The level of accuracy and variation in measurement are subject to the accuracy of the instruments involved and all environmental factors, which can vary widely.

Fuel volume in the tank is affected by temperature. Filling a container to the "brim" on a cold day will result in an overflowing container on a hot day, for example.

Wet or icy roadways induce wheel "spin" that will affect mileage, as well, and this phenomenon extends to different road surface types. Rough roadways cause loss of grip and increase spin.

Even properly inspected gas pumps are allowed a fixed amount of error. See the California standards here. See page 117 for acceptable variance. In Arizona variations of 2.5% from the readout is acceptable.

"And even when pumps are regularly inspected, that's no guarantee.

Anna Werner at KPIX in San Francisco found that in California, 94 percent of pumps pass inspection. But consumers can still be cheated. That's because pumps can pass even when they dispense a little less than what the pump says. It's a margin of error the law allows.

So a high-volume station that routinely sells a little less than a gallon could rake in around $50,000 a year extra - for gas you never get." (web page reference here).
​

Lots of variables to consider that could vary from day to day. In general, any mpg calculation is an estimate only, and is useful only for gross calculations and comparison with other imperfect calculations.

 

I disagree, with pretty much all of that. And the more gas receipts and odometer readings you throw at it, the more accurate the numbers become. Toyota is cooking the books.

I don't in the least miss our previous 06 Honda Civic Hybrid, but one thing it did right: if it said you did (for example) 5.0 liters/100 km, calculated would either agree, or be even slightly better, say 4.9. It's doable.

 

You are confusing random measurement variations with systematic errors.

And, having made measurements to molecular levels, I can assure you that a documented 2-7% systematic difference is hardly "molecular level". In his Gen 3, Mendel observed at least a 7% average difference between MPG displayed on the MFD and that calculated by dividing odometer reading by gallons pumped. I routinely had a 5% difference in my Gen 3, and seem to be getting about a 3% difference in my Gen 4 (of which roughly 1% can be attributed to odometer error).

My 2006 Insight-I's MFD mpg was routinely within 0.5% of the calculated MPG from ODO/gallons pumped. Whether the car had compensating errors built in that offset gas station volumetric errors, I don't know. But over many tanks I never saw differences anywhere near the magnitude that I saw consistently with my Gen 3 Prius. As Mendel says, it appears hopeful that the Gen 4's MFD MPG may be closer to accurate than the Gen 3 was.

For reference, a 3% systematic error in pumping gas corresponds to just under half a cup per gallon, or 5 cups per fillup.

 

I am not trying to be argumentative, but what "books" is Toyota "cooking?" I don't see any claim for the accuracy of the fuel consumption indication system in my owner's manual. Quite the opposite:

And that, my friend, may account for nearly all of the variation/error in Arizona (pump accuracy must be within 2.5% of regulatory measurement).

It seems equally likely that the calculation the Prius is giving is the accurate one, rather than the hand calculation. However, either result is only a reference. Choose your poison: Manual Calculation or Prius Calculation. Just don't mix the two. For me, I go by the in car display, but I do track the mileage using an iPhone, as well.

 

Wow, that Arizona allowance is huge. The California figure in your link, 6 cubic inches in 5 gallons, amounts to 0.5%. The smaller 3 cubic inches for acceptance testing is 0.25%.

I haven't yet found my state's allowed tolerances, but these news items from Texas and DC are more in line with the California rule above than the Arizona rule:
Think the gas pump is wrong? You may be right - Houston Chronicle
Keeping Gas Pumps Honest | NBC4 Washington

 

My state, CT, apparently shares the same 6 cubic inches per 5 gallons rule, or 0.5%:
State inspectors test gas pumps for accuracy - WFSB 3 Connecticut

So the take-home lessons:
(1) If you live in CT, Texas, DC, or CA, 0.5% gas pump thievery cannot explain the much larger discrepancy between MFD and ODO/gas pumped.
(2) Don't buy gas in Arizona!

Hopefully the moderators will post some sort of standard way to calculate MPG's because otherwise comparisons get confusing.

 

A 0.5% error in the pump volume doesn't have much of an impact on calculated mpg. The formula showing is in cell D1:

 

Right. The discrepancy between displayed mpg and odometer miles (or actual miles) divided by volume pumped is much too large to be random variability, manufacturing tolerances, etc. There is admittedly some car-to-car variation in the extent of Toyota's mpg exaggeration, but it's obvious that the exaggeration is intentional.

In the case of my 2011, the ratio of conventionally calculated to displayed mpg remarkably consistently averages 0.950. Correcting for the relatively insignificant odometer error, 0.95×1.003= approx. 0.953. Multiplying displayed mpg by that number should yield honest mpg.

 

I am inclined to agree. I think 50% of the errors are the gas pumps and the rest is instrumentation accuracy. Honda was terrible in their first Civic hybrid. 6MPG off. The Prius dances rings around the Honda.

Also how many pumps meet the standard? I suspect most are pencil whipped or bribed.
They will sell you a gallon close enough that you will never be able to check it in a gas can as being accurate. Also how do you fill your tanks? Does filling fast or slow give you the most gas??????? Doing one top off click??????
Trust but verify! I would trust Toyota over big oil. Even so they are out for a profit like everyone.
Those old Hudson's were great cars BTW.
Hudson Motor Car Company - Wikipedia, the free encyclopedia
Side comment= How is Corolla long term quality these days?

 

Those variables are 99% irrelevant if you average over several fill-ups and do your arithmetic properly. They certainly can't explain the consistent exaggeration of Prius mpg displays.

 

I work as a numbers analyst. Given the variables, I find the numbers on the display to be acceptable when compared with my hand calculations, especially when the manuals explicitly says to use the numbers as a reference. If you want closer numbers you should use different instruments for measurements and carefully control fueling conditions.

And to expect averaging to make variables irrelevant is a misunderstanding of statistics.

 

Yeah, what do we know.