calculated vs. computer MPG - Please post your results

Actually I'd look at as:

• 55.1 actual
• 54.09 corrected indicated MPG
• (54.09-55.1) / 55.1 ~= -1.8% error

I'm pretty sure -1.8% is within the 1-2% error of your previous vehicle. The calibration formula lets you get a number that should be close, not perfect, but much closer to the true MPG.

You are correct that the error gets worse in absolute terms as the MPG numbers increase. In fact, the formula really is only valid within the range of samples that exclude the upper and lower 5%. Now it is possible to substitute another trend line function and reduce the apparent error BUT that may only work for this set of data.

I tried different degree polynomials, the log, power, and other trend line functions but it was obvious that they 'diverted' beyond the sample set in directions that even the outliers do not support. I also agree that I would prefer to have more samples but it is a manual process (hint to PriusChat Gods that be) and time consuming.

I posted the spreadsheet so others would have a chance to review and possibly come up with a better analysis. I just have basic statistical training and there are a host of strange and wonderful formulas and relationships that could be tried. But given the available data, I figured this would be 'good enough'.

Bob Wilson

 

A user " NiHaoMike" in Ecomodder suggested this might be due to ethanol. An interesting hypothesis, the slope seems a little too severe to be explained by E10 ethanol. Also, it intersects the true line between 35 and 40 MPG. I would expect an ethanol effect to share the 0/0 value.

Thoughts?

Bob Wilson

 

Third fill up today:
miles: 549
comp: 56.4
calc: 52.9
temp range: 45-65

I didn't get the average speed for that tank but it is probably around 31 mph.

Getting better FE every tank: 50.9, 51.3 and 52.9 (~1500 miles on car)!

Peter

 

I have heard from a Toyota engineer that Toyota made the FCD to display intentionally 5% better number.

Ken@Japan

 

That is unfortunate since such an error would be so easily detected and then reported. Edmunds has already made such a comment and no doubt the other Prius-skeptics will be quick to jump on it. Regardless, I have another data point:

• 80.5 MPG - indicated
• corrected using 5%: 76.5 MPG
• corrected using statistical formula: 73.7 MPG
• 71.44 MPG - actual
• specifics: 54.3 mi., 0.760 gal., 19 mph average,

Now to find time to repeat the 20 mph target speed test.

For those who may wonder why my 'off the lot' MPG vs mph is different, it could be due to poor technique, failing to drain the battery before the test, but it could also be due to 'new car' performance being significantly better than even 4,000 miles later.

I plan to repeat my benchmarks before taking the car to the dealer for the 5,000 mile oil change that will also include a transaxle oil change. Then from the dealer, I'll repeat the benchmark test to see if there is a measurable change.

Bob Wilson

 

I'm averaging about 35 mph on a commute that's roughly 70% highway and 30% suburbs. The first four fillups (not counting the dealer tank) have been interesting:

Indicated/Actual
47.8/44.7
52.5/53.2
50.4/47.2
49.7/46.6

The second fillup (where the actual was higher than the indicated) was unique in that it was from two road trips rather than regular commuting, so it was 95% highway miles (if not more). The other ones are all consistently a 3.1 or 3.2 mpg difference.

Other things worth noting are that I live in a relatively hilly area so for the suburb part of my commute I probably average in the low 40s at best. Also, this is Southern California in the summer so the AC is always set to auto at 78 degrees.

 

I just assumed something like that must be the case, based on my own results. I think if I were in charge of that subterfuge (assuming I would have accepted the task), I would have put in a "roll-off" with odom reading so the gap continually closes and at least the truth arrives after a little delay! Say, +6% 1st 2000 mi, +4% 2nd 2000 mi, +3% 3rd 2000 MI...... 0% when you hit 10,000 mi. That way, when you buy the car it says, maybe, 53 MPG when you calculate 50 and then as the car wears in and you get better at driving it, the MID stays at 53 and you rise up to meet it in 6-8 months - or whatever.

Funny thing is, my car seems to be doing just that (closing the gap). Wait, they read my mind and put it in my car!!!!!!!

 

I find it hard to believe that Toyota would purposely inflate the number by 5%. Perhaps that technically falls within some government allowance and they figure they'd be at a disadvantage against other manufacturers who might not hesitate, but I don't know. They're the mindshare leader with the Prius, and so basically the game's theirs to lose and a 5% error is asking for toxic publicity.

That said, my latest tank (I know, it's just one tank, but...) was 49.4 MPG by-the-pump and 51.00 indicated. Less than 5% off.

 

-------- FCD -- Pump - Error
Tank 1 : 52.6 - 51.4 - 2.3%
Tank 2 : 48.1 - 45.7 - 5.1%
Tank 3 : 54.1 - 51.2 - 5.4%
Tank 4 : 50.4 - 47.2 - 6.4%
----------------------------
Average: 51.3 - 48.9 - 4.7%

 

I can believe they intentionally inflated the numbers on the 2010.

 

3rd fill-up:
8/29 7.72 gallons - Trip A display: 333.5 miles, 44.60 mpg; calculated: 43.2

I find this a little disappointing, especially since it's going down for each successive tankful. I'm getting about the same as I got on my Gen II for similar drives, but it's a little hard to tell because my Gen II results were more erratic. The overall mpg on my Gen II was 43mpg over 3 years.

 

More data points:


Bob Wilson

 

I'm still not sure what is going on but I now have data on six tanks of gas (where I filled it myself and know it was filled up to approx the same level). As you can see in the attachment - easiest place to see it is the last plot labelled MPG DIFF (MID - Calc) vs Odom.

You can see the MID started with a 7.9 MPG overestimation on tank #1, and this has dropped systematically for every tank until it was down to 1.5 MPG when I filled up today. The curve sure looks like it is headed to zero in a few more tanks. I really don't know if this is something I'm doing, something Toyota did, or maybe there is a simple explanation. Anyway, it sure is interesting!

Anybody else seeing their MPG overestimation dropping off???

 

Unfortunately mine is going in the opposite direction - the discrepancy is getting larger over time. For the first four tanks the overestimate of the FCD has been 2.3%, 5.1%, 5.4% and 6.4%.

 

It appears that most people are getting more random behavior than your rising and my falling MID excess shows. Also, my average excess is 9.4% whereas the average (from bwilson4web's trend line) is 5% for my "lifetime" average calculated mileage of 50.1 MPG. So, I guess we may just be seeing statistical fluctuations and not real effects. If so, you would expect to see our trends stop in a few more tanks. I guess time will tell.

Anyway, I hope that accurate MPG software will be put in everybody's cars when they bring them in for service.

 

Although you've seen what may be a trend after four or five tanks, I think the fact someone else has seen the opposite trend after a similar number of tanks does suggest what you note: Time will tell. More samples are necessary.

I've started to believe that having 14 tanks allows me to establish a real view of the percentage variance as about 5% overstatement. Even so, the numbers do vary considerably from tank-to-tank, and another 5 to 10 tanks may be required before the number may solidify.

 

Guys, this is a no brainer, of course your computer will come up higher than your actual milage. It calculates 100mpg when you move a low distance electrically(ie traffic). So those 100mpgs bring up the average of all the other intervals the computer is more realistially weighing. I could just move my can once every 30 seconds 30 feet for an hour, and that hour(12 intervals) of 100mpg would boost the entire average of the Computers MPG. Other than explaining that, I get just under 60 with the calculated mpg, cause I use the momentum of hills to my advantage well and try to always get over my needed speed before I hit hills so that I don't carry the cars weight but instead perpetuate it. My drive home by intersate gets about 55 whereas by city(5 mins longer) I average around 62. My computers set at about 70mpg(4302 mile) and I've been in very few slow paced 100mpg traffic average boosting situations(500k city, not much traffic until accidents). Became a member just to post this, so hopefully someone reads this and it clicks that the 100mpg slow pace throw the comp way off.

 

Actually in the Gen II the display may have read 99.9 MPG but the internal computer would go as high as 250 MPG. The Gen III should work the same way, but no one knows the highest MPG the computer uses. And you'd have to a lot of gliding to get a large skew from no ICE use.

Wayne

 

Your are confused by the use of MPG. When you read 99.9MPG on your screen, you are in fact getting infinite MPG, not 100MPG. As written elsewhere, using MPG to make averages makes no sense. Think in terms of GPM instead (or L/100km like most of Earth). You are making some strange assumptions... MPG is miles per gallon and has nothing to do with time/duration... It has nothing to do with pace...

A GPM display (or L/100km like most Prius) would avoid the confusion. Running in EV/gliding implies no consumption, so 0L/100km. You can then average something like 5L/100km and 0L/100km; something you can't do between 47MPG and infinite MPG (what you see as 100MPG on the display) (the average would always be infinite!).

 

I just filled up for my 3rd time, unfortunately 3 different stations & pumps, but for what it's worth, here it is (a recap)
Picked up with "full" tank at 117 miles on ODO
1) at 630 miles, filled with 9.449 gals (blinking pip, 23 "miles to empty") 58.4 / 54.3 (7%)
2) at 1098 miles, filled with 9.465 gals (blinking pip, 26 mte) 52.3 / 49.4 calc (5.5%)
3) at 1620 miles, filled with 9.539 gals (blinking pip, 13 mte) 57.4 / 54.8 (4.5%)

Trip 1 fill is off due to the dealer fill and me not getting to reset the trip odometer immediately (I had to RTM) Most of the driving was to/from work with mixed rural (10), town (2), highway (10) and side streets (4).

Trip 2 was a vacation, mostly highway with some low speed steep hills in the middle at the destination.

Trip 3 may be a little suspect as I had to pre-pay and guessed based on the mileage and the previous two tank fillups at that same "miles to empty point". The pump slowed considerably as it approached the pre-pay point so there was no shutoff click. I'm guessing it was close, but not to the rigor of some testers here This was also the mixed driving to/from work.

The current "lifetime" showing in the car is 1620.6 miles 54.2 reported. My Palm app shows 1503.6 miles and calculates 52.8 mpg, not taking into account the 117 initial miles of course. The calculated when I picked it up was a paltry 44.1 mpg :mmph:

I'm also posting on the fueleconomy.gov website.

Robbin