1. Attachments are working again! Check out this thread for more details and to report any other bugs.

(Scientific) study on the calculated vs computer fuel economy discrepancy

Discussion in 'Gen 3 Prius Fuel Economy' started by SCote, Oct 1, 2011.

  1. SCote

    SCote Fuel economy scientist :-)

    Joined:
    Apr 13, 2011
    17
    11
    0
    Location:
    Québec
    Vehicle:
    2010 Prius
    Model:
    N/A
    All,

    I have read many posts on the discrepancy between the calculated (e.g. computed from the fuel pump volume and distance covered) and the computed (e.g. indicated by the trip computer) fuel economy data. I thought the problem was interesting, so I decided to study it more deeply.

    I have done a little experiment, for a total duration of one year. Every time I went to the gas station, I wrote the volume, distance, computer FE, etc. Then I put all those on a graph, and started analyzing the problem. It turns what I found is pretty interesting. I have posted everything on my website: lapriusacote.blogspot.com

    Feel free to comment, suggest improvements to the study, etc. I am very open to suggestions.

    I have also another (even larger) study that I will publish on my site sometime this fall. I'll post again when it is ready.
     
    7 people like this.
  2. GrumpyCabbie

    GrumpyCabbie Senior Member

    Joined:
    Dec 14, 2009
    6,722
    2,120
    45
    Location:
    North Yorkshire, UK
    Vehicle:
    2010 Prius
    Model:
    III
    Interesting - keep us all updated.

    Is correct as simple as a software update or does the whole sensor need replacing to correct this for future?

    Is it in Toyota's interest to correct this?
     
    1 person likes this.
  3. Mendel Leisk

    Mendel Leisk Senior Member

    Joined:
    Oct 17, 2010
    54,474
    38,106
    80
    Location:
    Greater Vancouver, British Columbia, Canada
    Vehicle:
    2010 Prius
    Model:
    Touring
    I'd vote yes.
     
  4. SCote

    SCote Fuel economy scientist :-)

    Joined:
    Apr 13, 2011
    17
    11
    0
    Location:
    Québec
    Vehicle:
    2010 Prius
    Model:
    N/A
    I would hope a software update is enough. As you read in my study, there are 2 main issues: calibration and temperature. It looks like they dont consider temperature - but that would probably be easy to fix through software, knowing they already have temperature sensors in the vehicle. Then comes sensor calibration. I did my study only on my Prius. There is no easy way for me to know whether all the fuel sensors are calibrated the same way. Perhaps they are all a bit different, and they don't come with calibration data. So I hypothesise that Toyota uses some sort of "generic" calibration data, which is too optimistic unfortunately. Perhaps changing those generic parameters for more conservative ones would, on the average, make things better for everyone. And I guess that could be done through software (although I dont know their system).

    If any of you is interested in collecting the same data on their Prius during a few months, I would be very interested in seeing the results. It might help clarify some of those questions.
     
  5. F8L

    F8L Protecting Habitat & AG Lands

    Joined:
    Aug 14, 2006
    19,011
    4,080
    50
    Location:
    Grass Valley, CA.
    Vehicle:
    Other Non-Hybrid
    Model:
    N/A
    A lot of other vehicle manufactures are equally optimistic so maybe there is a general method for setting such systems up and calibrating them.
     
  6. cwerdna

    cwerdna Senior Member

    Joined:
    Sep 4, 2005
    12,544
    2,122
    1
    Location:
    SF Bay Area, CA
    Vehicle:
    2006 Prius
    I've only had a chance to glance at your data, but other factors to consider are odometer error and differences introduced as tires wear out.

    See Speedometer Scandal! - Feature - Auto Reviews - Car and Driver (there's also a PDF to download entitled Speedometer Accuracy) and Tire Tech Information - Tire Rolling Resistance Part 3: Changes to Expect When Switching from Worn-Out to New Tires. I posted the C&D article before at http://priuschat.com/forums/gen-ii-...oting/28808-odometer-accuracy.html#post368671.
     
  7. Mendel Leisk

    Mendel Leisk Senior Member

    Joined:
    Oct 17, 2010
    54,474
    38,106
    80
    Location:
    Greater Vancouver, British Columbia, Canada
    Vehicle:
    2010 Prius
    Model:
    Touring
    Honda, at least for the second generation Civic Hybrid, starting in model year 2006, is slightly pessimistic. In my experience in-dash mpg reported turned out to be about 98% of calculated mpg. It can be done. ;)
     
  8. Codyroo

    Codyroo Senior Member

    Joined:
    Oct 24, 2007
    1,826
    514
    6
    Location:
    Pleasanton, Ca
    Vehicle:
    2010 Prius
    Model:
    IV
    I agree with F8L on this one, in general (there can be exceptions) the automakers will side on making the displayed MPG optimistic. Many people wouldn't even know the difference because they don't bother to calculate their MPG. Having the displayed value = truth in their eyes, so why not make the truth a little more "rosy". The displayed value is usually close to 95% accurate, so for most people, this is "good enough".

    However, this board is populated by scientists, engineers, and people who like accurate AND precise data, thus tend to be bothered by a determinate error in the displayed MPG. For my car, the average determinate error is pretty constant over a number of fill ups, so I can adjust the displayed value to the real value with arithmetic.

    My brother's last Mazda also overestimated his MPG as well, about 5 - 7% I think (typically about 1 to 1.5 MPG on a car that got about 24-25 mpg). While 2 points doesn't make a reliable data set, the issue isn't exclusive to the Prius or Toyota. It is more evident in the Prius because a 5% difference on a 50 MPG car seems more significant than a 5% difference on a 20 MPG car.
     
  9. GrumpyCabbie

    GrumpyCabbie Senior Member

    Joined:
    Dec 14, 2009
    6,722
    2,120
    45
    Location:
    North Yorkshire, UK
    Vehicle:
    2010 Prius
    Model:
    III
    Is the error rate similar for all 3 different methods of measurement? ie US mpg, UK mpg and litres/100km? Or are some methods more accurate than others? Is something lost or gained in the conversion?
     
  10. jdcollins5

    jdcollins5 Senior Member

    Joined:
    Aug 30, 2009
    5,131
    1,338
    0
    Location:
    Wilmington, NC
    Vehicle:
    2010 Prius
    Model:
    III
    I have kept records of all of my tank fillups for two years now and my average error is 6.3%, right in line with SCote, and in US mpg.
     
    1 person likes this.
  11. priustexasbob

    priustexasbob Member

    Joined:
    Sep 3, 2010
    356
    86
    0
    Location:
    arrakis
    Vehicle:
    2015 Prius c
    Model:
    Two
    what are y'all basing the difference on, is it the amount you fill your tank up with? Unless there is a way to measure the exact amount you have in the tank there is no way to accurately measure the mpg (mileage/gallons), even a few tenths change the mpg and hence accuracy of the computation. just a thought
     
  12. wesayso

    wesayso Member

    Joined:
    Mar 18, 2011
    385
    35
    0
    Location:
    Hampton, VA
    Vehicle:
    2011 Prius
    Model:
    Two
    There is another conclusion that may strike a chord with the conspiracy theorists among us. All gas pumps are off in favor of big oil. They are charging us for more gallons than we actually pump. Discuss amongst yourselves, LOL. :eek:
     
  13. Mendel Leisk

    Mendel Leisk Senior Member

    Joined:
    Oct 17, 2010
    54,474
    38,106
    80
    Location:
    Greater Vancouver, British Columbia, Canada
    Vehicle:
    2010 Prius
    Model:
    Touring
    The error will be identical, regardless of the method of reporting mileage, units.

    Ours average error is currently at 6.4%. Greatest error was 9.6%, least was 3.8%.

    (There was a couple of anomolies I threw out: my wife tanked up at a station that was only taking cash, didn't fill completely, then I topped it up, long story).

    I'd like to be a fly on the wall when Toyota was setting up the programming:

    "So how much fudging the numbers do you think we can get away with? What's the sweet spot? High enough to push up our high mpg rep, but not so high that the BS starts to trickle out..."
     
  14. djras

    djras New Member

    Joined:
    Sep 1, 2011
    32
    2
    0
    Location:
    Victorville, CA
    Vehicle:
    2009 Prius
    Model:
    II
    This is interesting data and the error seems higher than I would expect. Typically I see reports of 5% error for the MFD higher than calculated . Both of the apparent causes of error COULD be corrected with software. As stated there is an ambient temp sensor, so temperature variations could be included and if the flow rate is wrong, that could also be corrected. The flow correction COULD even be setup for the user to "adjust" after a few fill-ups.

    The method of calculation is irrelevant to the effect MPG, KPL, or L/100K. All of those are know mathematical constants for the conversion and the unit should be based on the user's display and pump combination.

    All US states have a Bureau of Standards which verify the accuracy of the pump. The pumped volume is accurate to 0.5%. Now WHEN the pump stops filling the tank will vary from pump to pump, but any low fill one time will be washed out by a high fill later.

    As stated earlier, most consumers only care about what the display says. They don't bother to (or do not know how to) calculate their MPG at each fill-up. So, it is in the car maker's interest to "fudge" the numbers on the high side and keep the consumer happy (as long as it isn't "too far" off.

    All in all... very good data and analysis!
     
    1 person likes this.
  15. alfon

    alfon Senior Member

    Joined:
    May 17, 2009
    1,370
    270
    0
    Location:
    seaside, oregon
    Vehicle:
    2010 Prius
    Model:
    III
    That may be true with most cars that display MPG's. However
    Prius owners are of a different breed, and most do calculate their mpg's. Just see the response from this forum....

    alfon
     
  16. SCote

    SCote Fuel economy scientist :-)

    Joined:
    Apr 13, 2011
    17
    11
    0
    Location:
    Québec
    Vehicle:
    2010 Prius
    Model:
    N/A
    Good point! Actually I thought of it, but did not include in the calculations. The effect is going to be small, though. But I will do the calculations and make an addendum to my study. I'll post here when ready.

    That is very interesting - thanks for sharing. I guess it does not get cold enough in NC to see seasonal variation on the graph, or does it?

    Agreed. This causes error in the measurements. However, I guess if one pump stops too early, the next one might stop a bit later, and on average, this will have little effect. In my study, I made sure the tank was at least half empty before filling, to minimize that error. I noticed that the 2 times when I filled for only 4 or 5 liters, the discrepancy between gauge and calculated FE was very large - that would probably be explained by the relative importance of fillup error on such small volumes.
     
  17. Codyroo

    Codyroo Senior Member

    Joined:
    Oct 24, 2007
    1,826
    514
    6
    Location:
    Pleasanton, Ca
    Vehicle:
    2010 Prius
    Model:
    IV
    Do you want to collect data from others? I've got an excel spreadsheet set up that has my fuel mileage data from April 2010 to date. It has the miles per tank, fuel added, displayed HSI in mpg, and calculated mpg. I also have a couple of columns that calculates how much gas I would have put in were the HSI correct. I did this so I could see the difference in average calculated MPG vs average HSI reported MPG.

    As a bonus, I just added some columns calculating HSI displayed vs Calc average. The difference does seem to increase in the winter time and decrease in the summer time. Overall difference after 1 year was 5.9%. Current difference (lifetime) is 5.7%. (62.5 mpg HSI displayed vs 59.1 mpg calculated)

    I can email it to you (if given an email address).
     
  18. jdcollins5

    jdcollins5 Senior Member

    Joined:
    Aug 30, 2009
    5,131
    1,338
    0
    Location:
    Wilmington, NC
    Vehicle:
    2010 Prius
    Model:
    III
    Oh yes it does, especially the last two years. We have had record heat of days over 100+F and record cold last winter of days in the 20F range.

    My 6.3% is the average over these periods. It moves up and down from fillup to fillup. This average smooths out all of the tank fillup differences, temperature differences, etc. I keep all of this info in a spreadsheet started when I bought the car in August 2009.
     
  19. SCote

    SCote Fuel economy scientist :-)

    Joined:
    Apr 13, 2011
    17
    11
    0
    Location:
    Québec
    Vehicle:
    2010 Prius
    Model:
    N/A
    It would indeed be interesting to see differences between vehicles. I could try collect data, and produce graphs for each, but I can't promise it will be quick. But I'll do the best I can, I promise.

    If you provide me with data, it means you have no objection in me posting the results on my website, as an addendum to my experiment. I will of course acknowledge the source of the data.

    Here is what I need, as a minimum:

    For each time you filled the tank, you provide:

    • Volume of fuel as measured by pump
    • Distance covered (since last fillup), as indicated by odometer
    • Gauge FE (as indicated by trip computer)
    • Average temperature at the moment you used the fuel (go to the weather sevice website and look at historic data for the period, then put an estimate of the average temperature during that period). It does not have to be exceedingly accurate. Personally, I used the average value of the daily min and max temp.
    I assume each time you filled the tank, you filled it up (until "click").

    Please provide data as a text file (no .xls or other binary format).
     
  20. Codyroo

    Codyroo Senior Member

    Joined:
    Oct 24, 2007
    1,826
    514
    6
    Location:
    Pleasanton, Ca
    Vehicle:
    2010 Prius
    Model:
    IV
    It will take awhile to compile the temperature differences, as I'd have to look up 18 months worth of temperatures, and there will be noise because sometimes I fill up in the morning (on my way to work) or at night (coming home from work).

    As far as filling up, I tend to fill up to one click past the initial click. I do this in case the first filling didn't get a good seal and shut of prematurely (rather than to see if I can fill the damn thing to the neck...gurgle gurgle gurgle). I probably don't need to do this practice, because in the 18 months I've had the car, I haven't run into this issue. Typically, the second click will occur after adding and additional 0.2 and 0.3 gallons of gasoline. That has been consistent.

    If you think this data will have too much noise, let me know and I'll just start collecting data going forward. If you think there is still value, I'll back calculate the avg daily temps.
     
    1 person likes this.