Toyota why is the mpg readout not accurate?????

Knowing that the bias exists is half-the-battle; knowing the magnitude of the bias is the other half-of-the-battle; and, (in true Yogi Berra speak fashion) knowing what to DO with that information is the last half-of-the-battle.

Hence, keeping consistent, accurate, data is important.

 

I am nitpicking for a cause here ...

Imagine the true value to be zero, while the reported values average one. The reported values distribute homogenously around the value of 1 with a range of +/- 1.0 --- from 0.0 - 2.0.

You would say the bias is 1.0, right ?
If we apply that bias as a correction, now the average is 0.0 but the range of error spans -1.0 -- 1.0. Better for some purposes, but not exactly a magic bullet since tolerance remains +/- 1.0

Now imagine meter #2 with the same true value of 0.0, an average of 1.0, a precision of 0.05, and a bias of 1.0. The tolerance is 1.05
After applying the same 1.0 correction factor the meter know shows a cumulative accuracy that is spot on, and individual readings that have a tolerance of 0.05. Excellent!

Mendel's wish for a correction factor is bolstered by the meter's precision: not only will the cumulative average be accurate, but individual readings will be too.

 

This is how I do it:

...which yielded a nice straight line and linear equation:

MPG(tank) ≈ 0.92·MPG(mid) + 6.04; R^2 = 0.995


P.S. -- EPA has always used Harmonic Mean to calculate CAFE numbers.

 

Fuelly takes the total distance traveled and the total amount of fuel used to calculate the average MPG.

 

I like your example but would explain it another way:

Set A <-> B equal to 100 miles:

• A -> B = 100 mi / 50 mpg = 2 gal
• B -> A = 100 mi / 100 mpg = 1 gal
• A -> B -> A = (100 + 100) / ( 2 + 1) = 200 / 3 ~= 66.6 mpg or 66.7 mpg

Having more than one way to achieve the same result often provides a check on both methods. Although I didn't mention it at the time, I admired your energy approach to analysis of the recent Greek trip. I am fond of energy analysis simply because it avoids so many common errors.

Bob Wilson

 

Sagebrush correctly pointed out that I consider our Prius, a 2003 and 2010, to be experimental test subjects so I conduct experiments . . . a lot:

• MPG vs mph
• 12V to 120VAC inverter - what works, fuel consumption, and utility
• transaxle lubricant - service mile change rates and resulting oil tests
• towing
• fuel tank capacity and running out of gas
• driving approaches
  • routes
  • hill climb
  • cruise control
  • warm-up
  • lower bumper blocking
• 4-wheel alignment, toe and camber, caster is a harder problem
• traction battery experiments
  • charge cycles
  • rehydration
  • NHW11 vs NHW20 compare and contrast
• tire diameter testing: stability, handling, and efficiency
• audio-beeping backup LED
• fuel studies of different straight gas brands and comparing straight to E85
• dashcam
• OBD systems
• 3d party GPS
• accelerator encoder rebuild (pioneered by Doug S.)

I share my results not because I'm selling anything but that is part of the experimental methodology, the peer review, looking for errors. It isn't cheap and often finds obscure niches that I often find either difficult to explain (too many steps) or runs counter to 'accepted practice' like tire rotation or popular opinion. But then we drive Prius which immediately puts us in a different class of drivers, the early adopters.

Bob Wilson

 

Information is like electrical power: whether someone chooses to USE it (flip the light switch) or NOT (don't flip the switch) is totally up to THEM.

 

Yep,
A straightforward method that is most practical for more than two sums. I showed an admittedly cumbersome method to try and generalize the idea when to use harmonic means (expressed as a weighted average.)

For mental calcs I find it easiest to use a parametric equation.

 

This is how exactly how Fuelly.com calculates the average MPG, total distance divided by the total amount of fuel used.

 

I've not said anything so far because I did not have anything useful to contribute. Just a comment: this thread is a major reason I have stuck with this forum for as long as I have. Intelligent, reasoned, respectful discussions such as this one are fun to read. As a scientist, to hear people discuss the difference between accuracy and precision is a joy to behold.

If you have any comments on this post, please direct them to me by pm. I don't want to hijack this thread.

 

How I view accuracy vs. precision:

 

Accuracy?
Precision?

I think it's simpler than this.

It's how functional is it in actual use and what are your expectations. That's really it.

Reminds me of an old "Northern Exposure" episode, where Maurice purchase's an expensive rare antique clock. At first he is very proud of it. But it comes to bother him when he discovers that as an aged, mechanical clock it is less than perfectly accurate. It doesn't keep "perfect" time. He ends up getting rid of it, because it can not live up to his expectations.



I don't care if you want to say the Prius mpg read out isn't very accurate, or whether you want to say it isn't very precise. For me? It's useful in it's application on a daily basis, AND it meets my expectations for what it is...that's accurate and precise enough for me.

 

True enough, but some people (me included) expect better-than-mediocre results from their otherwise excellent Prius vehicle(s).

 

I liken the displayed MPG to EPA estimates in that they are just that, estimates. I use the EPA to compare vehicles and the MPG to determine if something might need to be looked at. It does seem rather silly though to put in a display if it's not going to be more accurate than the one in the Prius. Just think of what it would be like if our air pressure and fuel pump gauges were allowed to "estimate".

 

How do you decide if you don't know how close the meter is to correct ?
An engineer could say "very close some of the time, or it depends." Is that all the information you require ?

 

The more likely cause of frustration is a specific type of error: unpredictability: If the owner knew that the clock was always 5 minutes fast he would have adapted. In fact he would have loved the error: not only did it help him get to places on time, the clock had "character."

 

We have such a clock. It has been 3 minutes fast for years. Why don't we reset it? Simply because it plays a tune and my wife likes to hear all the clocks separately, so we only set our Grandfather clock to the correct time.

 

I don't expect an "engineers" degree of precision with an on the dash MPG meter.

For me? The Prius provides ample information regards to fuel usage, mileage and gallons left to go.

Could it be improved? Certainly. If threads like this spur Toyota to make improvements? I'm fine with that.

However, in the meantime, I'm perfectly happy with the information The Prius does provide.

As I said in my very first post...I can subtract.

 

An 'engineer' does not necessarily demand more or less precision that you do. The point is knowing the precision.

By allegory,

Me: Are you satisfied ?
You: YES !
Me: With what ?
You: I don't know, but I sure am happy!

 

Enquiring minds WANT to KNOW...the rest of us are just NOSEY!