What is the realistic accuracy & precision of typical consumer MPG calculations (tripmeter miles/pump gallons)

Until proven otherwise, I'd say what he's saying there is what you say and what I would expect, that it just uses volume of fuel injected and miles traveled.

No, steve. You are wrong. The amount of fuel you put in is the amount you can squeeze into the empty portion of the tank. The amount you used is the amount that used to be in the tank. You need to fill it to the exact same point each time to get an accurate reading. You may have filled your 72 liter tank to only 71 liters the last time you put in 50 liters to fill the tank. Now, at a different station, with different levels, you may squeeze in 73, or only 68. COSISTANCY is the key - and where most will fall down, because, like you, they just don't REALLY understand.

If the first and last are identical, none of the others matter. The difference of 1, or 10 liters spread across many tanks becomes , more or less, just noise. On the short term, like 1 tank, it can be a pretty large percentage of error.

On 07/20/2017 3:32 PM, trader_4 wrote: ...

That's certainly what the Buick displays indicate--if you reset then the initial instantaneous and average are identical and you can watch the average settle down as each instantaneous reading is added...

In a uni graduate seminar/project class we had access to firmware from GM in order to build competition vehicles. But, being of the age I am, this was long before the "veritable plethora" explosion of microprocessors and the whole system was in one little M68HC11 and none of the display functions currently were yet extant...

That is a combination of snow tires, sloppy roads and cold weather - not just tires

Roger clark forgets ONE major thing. The computer knows how much FUEL is used - the energy density of the fuel doesn't enter into the calculations at all - nor should it. You are measuring how much "fuel" is burned - doesn't matter if it is gasoline or hooch. The computer ASSUMES nothing. It calculates the volume by a combination of the length of the injection pulse and the number of pulses, and the distance travelled by the vss signal (which also provides data to the speedo and odo)

Yep - warming up in the driveway while I scrape the windows .. also I'll run the winter tires just a pound or two soft. John T.

You buy gas by the gallon and mileage is miles per gallon. Summer gas has higher density so you get more gas by weight for your money but it costs more per gallon in the summer. Sounds like you can't win.

If they base their fuel consumption on exhaust gas analysis, rather than actual fuel measurement he is correct. I believe that is how the EPA rest works (and why it is NEVER accurate)

The translation I get from GM is that it will read 25. Makes no sense to consider energy per gallon but it seems they do. I look at it as a rough guide, nothing more. When I turn my car off I get a reading from the last trip. Highway will be in the 25 range, city will be in the 18 range.

Seems GM is not rational. In any case, it is always optimistic in my experience.

and the engine runs richer with low intake air temps too. Amd the bearings and gears are stiffer in the cold, and you are pushing slush aside or crawling over snow - all of which takes more power and fuel.

GM says this:

Roger Clark, senior manager of GM's energy center, explains that the fuel economy gauge makes a calculation by counting the number and duration of pulses made by the fuel injectors as they squirt gasoline into the combustion chambers of the engine. The onboard computer system divides the distance the car travels by this estimated fuel consumption.

I don't see how it could be any clearer that they are simply computing based on volume of gas used and distance traveled.

The latter part, where they talk about the fact that the energy content of the gas makes a difference in MPG doesn't change the fact that it's calculated the way you'd expect.

We have mandated ethanol blended gasoline in the winter months. It's good for a couple of mpgs. That's in early November before warming the car up in below zero weather and running the heater full tilt happens. That causes another drop.

Then they need to hire a decent programmer. Integrating the volume delivered per injector pulse should give you the total volume. There might be a slight correction for the temperature of the fluid.

And if you're like me, the car idling in the driveway until the ice melts off the windows and the interior temperature is somewhere north of zero (F).

Every reading a mom and pop does has inaccuracies that, I posit, are tremendously higher than most people seem to think they are (at least most people who quote mpg figures with decimal places in them).

Most people have a tripmeter reading and a gas pumpmeter reading. Where they fill the tank and reset the tripmeter before driving away.

I can't find any reliable source that says what the accuracy or repeatability of that mom-and-pop tripmeter/pumpmeter calculation, but basic logic dictates that the errors compound such that there is likely (IMHO) no way to get anywhere near decimal-point accuracy, and worse, probably plus or minus 1 mpg is the closest anyone can get in terms of repeatability and precision.

Even the EPA's $360,000 machine only claims plus or minus 2% of the indicared reading. I can't find where I got the notion that a mom and pop can't possibly get closer than about 4% with a tripmeter/pumpmeter mpg calculation - but I'm still seeking those numbers as we speak.

On 07/21/2017 11:51 AM, Mad Roger wrote: ...

...

Why do errors compound in your view?

And, it depends on what you mean in terms of accuracy -- in terms of absolute one needs to know the calibration error of the odometer; most folks are satisfied to just assume it's close enough for the purpose.

If you look at simply a single fillup, it's not unreasonable to expect a few tenths of a gallon difference between the first fillup level and the subsequent; if you try it on shorter distances than a full tank then the fractional error goes up.

OTOH, if one keeps track over longer periods of multiple fillups and take some care to use the same filling pattern and only fills up after using near the full tank capacity, then over time plus/minus targets _will_ tend to cancel out and I have no qualms in believing a relative performance number in the 0.1 mpg can be determined.

As noted, I've done this on long trips a number of times (generally on first trip or so with a new vehicle, either actually new (rare) or (most often) new to me) just to see how it compared with previous and have had quite good comparisons on recent ones with the computer-computed results. These would be over total distances of 1500 to 2000 miles, not just 20 miles test runs.

I measure my gas mileage on every fillup. I get 19 to 20 MPG every fill unless I do a lot of around town driving. Very consistent. I watch it to see if it drops off which would mean something is wrong.