MPG difference between "computer" and "actual"

From what I have seen on the discussion boards, everyone who has a car that computes mpg says the computed mpg is always more than the number you get when you divide miles driven by the gallons it takes to fill up after the trip. Me, too.

I would like to know if there is an actual, scientific explanation for this. What comes to mind is thermal expansion that occurs during the process of filling a tank, or delivering fuel to the engine or something like that.

It is easy to accuse car manufacturers of fibbing to us. It is equally easy to accuse the gas stations of over-reporting the amount of gas you have pumped, thereby charging more for the gas you pumped.

I trust the gas pump and mileage tracking more than a glorified trip computer. I’ve explained it to my wife, that uses the damn thing to figure out how many miles she can go before completely running out of gas. (yes, she runs it almost to complete empty before filling up again. I’ve begged her not to.) I’ve actually seen the ‘miles remaining’ number change just by parking on a slope.

I honesty believe the little computer is a good estimator, but none too accurate.

A truly accurate measure of gasoline mileage is quite difficult. As I remember, Consumer Reports had a 5th wheel that they pulled behind the car and drove a measured distance on a test strip at a specified speed. They fed the gasoline into the fuel system through a container that held a precise amount of gasoline. This was in the old days when are had caburetors. All the cars tested were run the same day under the same weather conditions. Even then, I think that the gasoline mileage was rounded to the nearest 1/2 mpg.

I meassure my mileage over several fill-ups of the tank. My figures are usually within 1 mpg of the “computer” on the dashboard. I monitor the gasoline mileage and if I see a drastic drop, then I check things out. By drastic, it has to be over 3 mpg over a period of time.

I suspect the computers in most cars do a very rough estimate of the fuel consumption by using “known” amounts of metered fuel and comparing that to the mileage. They can do that by knowing the pulse duration of the injector nozzles and the amount of fuel delivered by each nozzle under the designed pressure. But there are some inaccuracies, of course. It is virtually impossible to know exactly what amount of fuel is used at each individual injector. This is due to manufacturing tolerances and more importantly deposits which impede or partially obstruct fuel flow. Presumably, a brand new car estimate should be close because the new injectors can be assumed to be clean. Fuel pressure may vary slightly between cars which can affect the amount of fuel delivered per pulse. Then multiply that by 4-6-8-10 (# of cylinders in your vehicle) and you can get some variations.

Dividing miles traveled by fuel used can be fairly accurate over time. But consider the variables there as well. Fuel delivery at the pump should be reasonably accurate because the pumps are calibrated on schedule. But fuel density varies with temperature, and the fuel density can also depend on the mixture of hydrocarbons and percentage of ethanol (for example, if any).

Fuel tanks don’t normally expand all that much, and would “beer can” if they did, which you would hear when filling the tank. But your shut-off point can vary widely depending on whether you allow the pump to shut off automatically or you manually decide when it’s full. Also consider that the length of the nozzle into the filler neck, the plastic overflow guard and even the manufacturer of the nozzle mechanism can cause variations between stations and between fill ups.

I would suggest you monitor fuel consumption over a period of time (monthly for example) and use that accumulated average to get a better estimate…and hope the weather doesn’t change.

The computer in my Lincoln is extremely accurate; to within a tenth of a mile.
It’s also been verified countless times by comparing it with the tripmeter/tank fill method.

About once a year it will get a spell for a day or so in which it gets a bit stupid with the readings but for the rest of the year it’s near dead-on.

Have you located any technical information that states pulse duration is used is computation of mpg figures? I can see how in theory it could but is this the actual method used?

To what level of accuracy do you think any method of figuring mpg can be achieved?

We have prople posting their mpg with a level of accuracy expressed in hundreths (like 20.55,19.35 etc) I don’t believe it is possible to calculate this close.

Thanks. You have given me some insight as to possible sources of error with the on board measurement of fuel used. It makes me wonder why they don’t use something like the system used by gas stations. If they are operating “legal” on their pumps, the amount delivered should be quite accurate. And, as has been pointed out, over several tanks you can average out the small random errors. What is actually occurring, though, is not random. It is systematic and in my car it is about 5 percent and it sounds like others have about the same amount. One mpg out of 20 is 5 percent. On a gas station pump, that would be unacceptably inaccurate.

I saw something on the TV news last summer, when gas prices were at their peak. They said that the volume changes because the temperature in the underground storage tanks is different from the temperature at the surface where you pump it into the gas tank of your car. I don’t quite see how that logic works.

I remember those CR tests. One of the benefits of doing it their way is that they can calibrate the actual miles driven. Odometer accuracy is taken out of the picture. They also carefully measure the fuel consumed. This should give the best estimate of mpg under actual driving conditions on their test track.

In this case, it is slightly different. The vehicle’s odometer is the same number whether you divide that by gallons at the pump or gallons measured by the on board computer. Perhaps the computer is using a different measure for miles driven? That could make a sensible explanation. For example, your trip odometer does not move while you are backing up, yet miles driven in reality increases. I have already determined that the extra miles driven backwards out of my driveway could not possibly come close to the amount needed to change the calculation of mpg. I’m still stumped on why there is a SYSTEMATIC difference in the two calculations.

I’m beginning to wonder if on board computers do something “stupid” like always rounding numbers up to the next largest value. That kind of error could add up and would not be “random” but “systematic” error.

Different cars, different computers, different info.
06 Ford Escape hybrid. “average fueal economy” is based on the entire grand total miles on the car. Can’t barely change the numbers even after days of in-town high mpg driving.
“miles to empty” is based on very recent driving habits. Pull off of highway driving showing 150 miles to empty and after 10 miles of in-town driving it shows 190 miles to empty ! And vice versa.

They’re just tools to add to what you already know how to do. I don’t blindly drive by them only. I leave mine set on “miles to empty”.

To get accurate results, you need accurate measurements. Measure the fuel flow directly. If the system has a return line, a measure of the return fuel would have to be made, and subtracted from the supply line measure.
A flow-meter gives the most accurate indication of moment to moment fuel use. For a car which gets 20 mpg, that would be a 3 gallon per hour, at 60 mph, rate of usage. This equals, roughly, 0.2 liters per minuets (200 ml per min.). Here is a flow-meter (100 ml to 1000 ml) which would work:

I can think that a computer would use known inputs including injection time which was mentioned which would deliver a calculated amount of fuel knowing the fuel pressure over a known distance from the electric pulses of the speedometer transducer to be able to calculate miles per gallon.

Possible sources of error would be the fuel quantity delivered due to assumed known injection pressure, odometer error and return line fuel was mentioned; don’t know if ours has that. Possibly the energy content quality of gasoline might be a source of error too.

As was inferred, the manually calculated number has it’s own source of errors including gas amount purchased and the odometer. The temperature of purchased gasoline, if higher than the industry standard temperature for your area of the cou try will deliver more volume as read at the pump.

Our 08 at first glance seems to read about 2 mpg high in 30 but when you factor in odometer error, it turns out to read about 1 mpg high. Our 05 also read about 1 mpg high.

From my limited point of view, it would seem that mfrs set up the mpg calculation to err on the high side rather than attempt to land on nominal with some reading less and some vehicles reading more mpg.

The miles remaining number can be calculated from the gas gauge and the present mpg. The system appears to have a fair amount of inertia or lag of sorts so that the miles remaining number does not constantly fluctuate with a constantly changing mpg number.

My on-board computer is not what I would call accurate, but it does tell me that driving 35 mph in 5th gear on a level road uses less gas than in 4th, but when going up a hill, it may be the opposite.  Each time I buy fuel I tell it how much I bought, it knows the rest and then it adjusts it's estimate so the next tank is a little more accurate.  

I will note that fuel pumps are generally very accurate, and the odometers on most newer cars are also accurate assuming OEM size tyres.  The speedometer is usually optimistic about 5-9% intentionally.

Those computers are just toys and have no real value although some of them seem to work.