calculated vs. computer MPG - Please post your results

We are seeing about a 5% overstatement of mpg by the computers.

 

My concern is over losing MPG. My 2010 is six months old. When I picked it up the monitor said 575 miles until empty. Since then, it's been continuously decreasing. I'm now registering 403 per fill-up....and that's when I'm down to 1 bar. Every month it drops by at least 20 MPGs. I had a 2007 Camry Hybrid for three years. My MPG until fill-up was always between 465 and 510 MPG for the duratiion of the 3 years that I owned it. That's a fluctuation of only 45 MPG in 3 years. My 2010 Prius has fluctuated nearly 175 MPG in only 6 months. I can't trust the readout.

 

I have approximately 5000 miles on my Gen III since the end of October 09, each tank's hand calculated mileage has consistently been between 2 and 3 MPG less than the computer's estimate.

 

Hi all,
Like many people here, I have been experimenting about 5% discrepancies between the calculated and computer fuel consumption. I have bought my Gen III prius late November, which means I have been using it mostly in cold weather conditions (always below 0°C). I was wondering how the cold weather is affecting the FC metering. In Canada, the fuel is sold volume corrected to 15°C, which means that the fuel is sold by mass rather than by volume. Thus, if you fill-up at -15°C, you will be able to put more fuel mass into your tank than if you fill up at 30°C. Or, if the fuel is at a constant temperature when you fill-up (the tanks are buried in the ground, where the temperature is roughly constant throughout the year), the fuel in your tank will shrink in volume when it cools. How is the fuel consumed calculated in the prius? Is it calculated by volume or by mass? That would make a big difference (the thermal expansion coefficient of fuels are about 995ppm/°C, which means about 3% for a 30°C difference), unless the fuel is brought to a constant temperature before it is injected into the ICE... Also, the retailers volume measurement tolerance is about 0.5% by regulation, which could add-up to the error induced by the prius. Any thoughts?

 

There is no individual fuel flow meter in Prius.
Our consensus is the fuel flow is calculated by sum of fuel injector on time period, so it is based on the volume.
The fuel delivery pipe is located in the engine room, therefore the gasoline temperature becomes warmer than the gasoline in the tank.
I believe we can assume the fuel flow is reasonably consistent (within 1% error) with the injector on time period, and we can forget about the gasoline temperature in the tank.

Ken@Japan

 

Thanks for the explanation. However, do you think the fuel measurement may be biased when the engine is warming up, that is, when the fuel reaches the engine while it is still cold? This would tend to give wrong FC results when many short trips are made under cold conditions...

 

I believe there are a lot of variable about gasoline characteristic here, such as specific gravity (more mass when cold), viscosity (less flow when cold), fuel pressure (maybe higher when cold) and others.

I have made my own fuel consumption meter, called SuperMID M-1, which measures the fuel flow by the fuel injector on time.
When we adjust the fuel parameter (on time vs fuel flow) correctly, we see the fuel flow accuracy within 1% in any ambient temperature.

Therefore, I believe there is no technical barrier to display it within a reasonable accuracy, but there are some sales & marketing strategy...

Ken@Japan

 

Oh... pity, isn't it? Anybody knows if other cars are plagued by such a marketing "feature"?

 

AFAIK,
Honda Insight-II +7%
VW Golf V +10%

Ken@Japan

 

1,700 miles, 43.4 on the car's computer screen, 40.7 according to my calculations from my latest tankful. Disappointing. Not driving hard. Pissing people off behind me, actually, which is probably good, because everybody drives in crisis mode around here, all the time. At least I'm not in the race anymore. That's one thing about this car over the Mercedes I traded for it. The Mercedes had a 3.2l, and it could zip you out of a "situation" if you wanted it to, plus, it had a sport suspension. Encouraged a little civil disobediance which this thing discourages, which is probably good, plus I have a warrantee on this, which expired on the used Mercedes I'd bought. Still, why doesn't Toyota respond to the discrepancies people are having with the computer vs the actual fill-up milage, and why don't they tell you about the drop in milage with cold weather driving before you buy the car? I had a Honda and they were always good at things like that. I drove an Insight before I drove the Prius, and was all set to buy one. The Prius had a bigger, more luxurious feel, so I paid more, and lost the good financing Honda was offering, plus I lost naviagtion, and had to put down $1,200 which Honda didn't require. I wonder how Honda compares to the car's computer, and EPA estimates, cold weather, etc?

Cold. 20's, 30's.

 

Thank you for your data.
It looks your Prius is recent delivery and we are expecting the computer program update to correct the errors.
revised program?
Would you please let us know the manufactured month of your Prius?

Ken@Japan

 

My road trip from Ark. to Big Bend and back through Austin: 2660 miles/59.7 gallons = 44.6 MPG. Computer indicated 47.0 MPG. Computer a bit over 5% high.
Month of manufacture 09/09.

 

Forgive me for not reading all the entries on this thread. But I selected the above as one that best describes what I have experienced ........ and a possible explanation.

My G3 was bought new in July (hot weather) and now it is January (cold). I use a computer program to record/document every gallon, and have found when it was hot, my actual was less that 1 .5 mpg lower that indicated. However, now that it is cold, it is about 3.6 mpg lower. I hope that when warm weather returns, the difference will return to the earlier figure, along with the improved efficiency that I experienced in the warmer weather.

I can live with the difference, but I wish there was consistency in the reading.

 

And my prius 2010 +8%

 

On one of the threads, I read thatbecause the Prius can only average a high of 99.9 mpg, it skews the average . Interesting thought. If true, that would suggest that the more you drive in EV, downhill, hypermiling and etc, the greater the percentage of error will be between actual and indicated.

Any comments?

 

I assumed it can only display 99.9. I would be surprised if the computers limited their math to only that accuracy. I'm pretty sure that is the case since I go down 3000+ feet every day and when I fill up high mine will stay at 99.9 for a long time after I'm down the hill even though I'm actually only averaging about 50 on the flat.

 

This is more than unlikely. This would mean that the computer has to keep in memory all the instantaneous FE data (say, once per second) and average them? It is a lot simpler to divide the trip distance by the total fuel consumed. Only two values to keep in memory...

 

Sorry for bringing an old post up here... However, it explains my main concern about the FE/FC display error. Here it is: if it was an inaccuracy in the fuel measurement system, you would sometimes get a computer MPG that is higher than the calculated MPG, and you would sometimes get a computer MPG that is lower than the calculated MPG. This is what an error would introduce. If the difference between computer and calculated MPG is always positive, then there is a BIAS (systematical or not?) in the measurement system, rather than a systematical error.

One thing I have not read yet on this thread: Does the computer/calculated error changes with the number of start/stop cycles of the car? That is: if you fill-up, run your car for the day on the highway, and burn all the fuel in a single trip, will you get an FE error that differs from the one you would get with a fuel tank used for many small commutes (many start/stop cycles)?

 

The computer must average at least every minute because it can display the result as a bar graph on the dash.
I would have thought the calculation would not have a bias error one way or the other, so it comes down to a fuel consumption measuring error.

 

As I wrote on the #305 post and the #307 post of this thread, it is not difficult to measure a fuel consumption with in 1% error.

Again, I believe there is no technical barrier to display it within a reasonable accuracy, but there are some sales & marketing strategy...

Ken@Japan