Real fuel consumption vs. the one reported by computer

In the U.S., the EPA tests cars indoors on rollers. Probably a warm engine, and No hills, rain, winds, etc. It'll improve with tire & engine break-in, plus warmer weather.

 

Welcome to the MisInformation Display. My unit is a relatively 6%-7% optimistic. Add the distance on the tank to the distance to empty then divide by 10 to get a closer approximation. At least the constant optimism is an improvement over the Gen2.

 

How would you estimate the size & volume of a squirt of gas?

With a quantity that small and a count that high, how many digits following the decimal point is reasonable?

Think about sensor measurement and computer resources that are needed. You're bound to be off too, especially with the engine shutting off but not stopping. Estimations get messy. Keeping rounding errors optimistic is the better choice.
.

 

They did quite well on GenII, is GenIII any harder?

 

I have been manually calculating the gas consumption for 6 months on my P3, the actual consumption has always been about 10% less than the computer reports. Am I disappointed? No, I knew the factory figure is never going to be true in the real world. I notice as the weather gets warmer, the improvement is noticeable.

 

I think this is true of a lot of cars, not just the Prius. do not know why. C/D had a recent write-up on that (it is on their website currently, I think). One factor I think is that we as Prius drivers tend to pay more attention to this, so it may seem worse in our cars, but probably is on a par with others. FWIW, I've been averaging about 4-5% difference, but when I gas up and do the math, I almost invariably beat the EPA estimates, so I'm not too unhappy. It would be nice to fix it, though.
~T

 

Yup! I am noticing a similar error. On my latest trip (608 km) the computer says 3.6L/100km, while my calculation says 3.8L/100km.

 

It's 8 - 9 % difference in my case. On my 5 yrs old Renault this difference is less than 1%.

 

I bought my Gen III in July of '09. It has been consistantly about 2.1 mpg off when calculated When it says 60 I'm getting 57.9. It doesn't seem to very by percentage. Don't know if the dealer can correct it, no big deal to me because I know where I'm actually at but I have a calculator in the console and check every tank.
Tony Renier

 

The computer knows how long an injector pulse it applied, and it knows the flow rate of the injectors, so it should be extremely accurate. It's possible that the fuel consumption computation isn't taking account of the time taken to shut off the injector after the pulse finished.

It knows full well when it's spinning the engine without fuel. As I have said repeatedly on the numerous 'unintended acceleration' threads, the engine management computer software is solely responsible for turning each cylinder's injector on at the appropriate point in the cycle, and solely responsible for turning it off again. (In those threads it was to make the point that if the computer halted the program, the car would just stop dead because no fuel would get to the engine and no spark would fire.)

It shouldn't be possible for more fuel to actually be added than the computer dictated, as - once fully warmed up and in closed-loop fuel control - it's monitoring the air/fuel ratio by monitoring the oxygen content of the exhaust, and adjusting the input fuel to keep the gases in the right ratio for the three-way catalytic converter to work correctly.

It's possible that these fuel adjustments aren't accounted for in the fuel consumption calculation. The stoichometric ratio - where the exhaust gases are in the right ratio - for 'pure' gasoline is 14.7:1, but it's different for E10 fuel. I think the car has to run richer - more fuel for the same amount of air. If it's not compensating for this, the results will be wrong.

In addition there may be additional sources of loss, and not necessarily all of the fuel that the pump charges you for ends up in the tank. Some may be lost to vapour recovery in the pump (if fitted) or simply lost as vapour (if not).

 

Adding to Mike's excellent comments is an explanation for the non-engineering types reading this thread:

The proper ratio of fuel to air is very important for a gasoline engine, and is essential for good mileage and low emissions. In the Prius this ratio is controlled by the computer. Likewise, the Prius gets very good mileage and has very low emissions. This fact tells us that the computer in a Prius very accurately measures the amount of fuel squirted into the engine.

If the engine control computer is accurately controlling each squirt of fuel, then why does the displayed mileage not agree with the calculated mileage? The answer has to do with accumulated error. Each squirt of fuel is very, very small. Another computer counts these squirts and records the size of each one, but it likely rounds off the size of each squirt for easy calculation. The size of the round-off is tiny, but the error accumulates because of the large number of squirts required to make a gallon of gasoline.

As an example, let's say we need a gallon of milk, but all we have for measure is a teaspoon. We have to fill the teaspoon 768 times to get our gallon. If our teaspoon is a tiny bit large we add a tiny bit of extra milk each time. It's not much, but we do it 768 times. That can add up to a significant amount of extra milk.

In a similar fashion the same thing can happen to a fuel computer that counts small squirts of fuel.

Generally, in systems like this, some sort of calibration factor is introduced to correct for systemic errors. We can only guess why this wasn't done with the Prius fuel computer. My guess is that someone made a decision to round up, which produces a better marketing figure. Unfortunately this leads to disappointment and frustration when it doesn't match the calculated mileage.

Tom

 

Since the gauge is consistently over, Toyota could have adjusted their calculations to compensate. Obviously, they deliberately chose not to do this - they would rather be on the high side all the time than be 0.0001% low at any time. (they probably guessed - correctly - that most people wouldn't do the actual calculations)

I have been tracking my mpg real/displayed for several months, and the range is from 4%-8% high, or from 1.8-3.4mpg high.

I see less deviation on tanks which were all highway miles. And I saw approximately the same margin of error with snow tires vs the stock tires, interestingly enough...

 

If the numbers are rounded they would be rounded down as well as up. It shouldn't be a factor.