calculated vs. computer MPG - Please post your results

I see your point... Interesting, I also wonder.... it must not be zero then, because different engine types burn different amounts of fuel per unit of time during idle... does that make sense?

 

Example:

Lets say I go on a 220 mile trip, that's broken down into 4 segments - 100miles/2 gallons (city) + 0miles/0.1 gallons (stopped) + 120 miles/2 gallons (highway) + 0 miles/0.1 gallons (stopped).

My calculated MPG would be = 220 miles/4.2 gallons of consumed gas = 52.38 MPG


Say the computer disregards averaging the MPG for the short time (small amount of gas) the car is stationary (not saying it does...just what if??), then the computer average would be = 220miles/4 gallons = 55 MPG

...coincidentally, that's very close to the discrepency of 5% reported by most people here.

I'll bet money that some of the discrepency lies in the algorithm, and how the manufacturer handles averaging the MPG during the period when the car is not moving, but still consuming gas.

 

Very good point cody but the MPG does go down when stopped and the ICE is engaged, meaning it's not been disregarded.

What i'm wondering is that a small 1.8 engine does not burn the same amount of gas like a big V6 or V8 engine when stopped for the same amount of time, so how in the world do the trip computers account for that? It cannot be Zero regardless of the car...

 

The obvious answer is in my post #213 in this thread.

Again, the obvious answer is that "time" has absolutely nothing to do with the mpg calculation other than as an auxilliary purpose for displaying some data. The car keeps track of distance traveled and fuel consumed. Period. It uses these values to compute the mpg for any period of time, but time has no part in the calculation.

In order to give you a display that has "time" in it, the incrementing values for distance and fuel can be snapshotted (technical term) at any time. To display the mpg for some time period the values for distance and fuel consumed at the start and end of the period in question are subtracted then run through the simple formula (mpg = d/f * k) where k converts from the collected/incremented units into mpg (i.e. the distance might be in wheel rotations and the fuel in drops. Let's say the car goes 8 ft per rotation and there are 75,000 fuel injection drops per gallon. K is the drops/gal * miles/rotation. In this made up example it have a numerical value of 75,000/660, or ~114.)

It should be obvious from some simple observations and knowledge of what is being calculated that there is no "averaging" going on and no "time" factor involved in the calculations. It is distance traveled and fuel consumed between any two points in time, but not caring about the amount of time. Time is only used as an interval to provide displayed output...not in the math. Time is used for speed (mph) not mpg.

3PriusMike

 

Actually that is not necessarily true.

There is one current hybrid vehicle that is notorious for under-reporting its fuel efficiency. The Civic Hybrid II frequently displays worse FE than what it truly achieves (reports 1-3 MPG worse) and in working this way it also pays a price with the media, first time drivers/testers.

It is any wonder why Honda went back to over-reporting again with the Insight II (1-3 MPG)?

Cheers;

MSantos

 

Not what I read from a site similar to this, but on Honda hybrids.

 

I don't understand the persistent thought that there is a time-sensitive algorithm to determine the fuel economy.

I don't know exactly how it is calculated, I will grant you that. Still, the car produces the miles numerator through the odometer. Whether this is based on wheel revolutions per mile or some other factor, it is calculated independent of gasoline consumption or time.

The gasoline consumption is calculated by its use -- whether an assumed amount per injection or some other calculated amount. In any event, the gasoline consumption is not based on time, either. It is based on gasoline being sent to the combustion chambers in the ICE.

At any given point in time, the calculation may be made: miles traveled divided by gallons consumed. The question is across what period the calculation will be made. The trip odometer will measure the accumulated total since reset. The 1-minute and 5-minute consumption screens will measure across those set intervals.

If you creep through and around a large parking lot for 10 minutes, driving at 5mph and using absolutely no gasoline, then what mileage will you have? If you are looking at the past 10 minutes, only, then you'll have an infinity value (the onscreen reading will be maxed at 99MPG). If you then gun the car and get it up to 60mph in 10 seconds and hold it there for 50 additional seconds, you will change the fuel economy reading. If, for those 60 seconds, your average FE is 20MPG, then what is your average FE for the entire 11 minutes?

Is it (10x99 + 20)/11 ==> 1,010/11 ==> 92MPG? NO, of course not.

You need to know how many miles were traveled during the total time period. I'll assume one-third of a mile for the acceleration and 5/6 mile for the rest of the driving. Therefore, we have 7/6 miles for the period of ICE use and 5/6 mile for the EV jaunt (5mph at 10 minutes), or a total of 2 miles traveled. This is the numerator.

The denominator will be the amount of gasoline used. Here, we don't have the full data, but we can tell we'd be burning serious gasoline if the one-minute average was 20MPG. Because the one minute average was 20MPG, and we traveled 7/6 miles, the gasoline usage would have been 7/6 miles divided by 20MPG or .06 gallons.

Fuel Economy: 2 miles per 0.06 gallons or roughly 33MPG.

If you had sat there and raced the engine with the foot on the brake for 60 seconds and used 0.06 gallons, the FE would have been 0 miles per 0.06 gallons or 0MPG.

If you had sat there as the ICE turned on, and you didn't move, then you'd have used perhaps 0.005 gallons for 0 miles. The FE still would have been 0MPG.

In either case, though, the 1-minute and 5-minute consumption screens would have shown different increments. This is fine. These consumption intervals are not used to calculate the car's fuel economy. They simply are (literally) side bars. They are snap shots in time of FE across those time periods. Without knowing the miles driven or the gasoline used, you cannot take those data and develop any rational basis for determining the car's FE.

 

:focus:

Here is an update on my calculation of how much the MID screen fuel economy over-reports (for my car) compared to the fuel economy determined by dividing the period miles by the fuel purchased as replacement. As the individual tank purchases have variation, the total miles divided by the total gasoline purchased will provide a better view of the FE. This is why I report the individual tank variation and the overall variation.

I still see swings, tank-to-tank, but it really appears as though the variation is settling in at a 5% overstatement on the MID.

Here is a chart showing the past 17 tanks of gas purchased up to my recent measurement at 8,153 miles:



The crowded numbers on the bottom indicate the odometer reading at the time of each fill-up. The y-axis numbers (the percentages) reflect the percentage each MID report exceeded (it ALWAYS exceeded) the fill-up (red line) and cumulative fill-ups (green line) calculations based on gasoline purchases.

 

By the way - I'm not disputing the variation. I'm simply trying to understand the possible reasons why this would be so?

Whether you calculate MPG's, or get it from the computer, the distance traveled should be the same in both cases. So, as far as I'm concerned, distance traveled is the same no matter how MPG's is determined. Therefore, the only factor to account for the discrepency is fuel consumed. That is, assuming you believe that Toyota's algorithm to calculate MPG's is not deliberately fudged upwards, or they have some unique way to calcualte MPG's that we are not considering.

So, assuming Toyota is not trying to inflate its numbers (and I don't believe they would), then the only thing left is how to determine the amount of consumed fuel. As I already said in prior posts, estimating how much fuel one consumed based on the pump number can introduce significant error. The counter to this is - it should average out over many tanks of gas. With modern fuel injection systems, I would assume that the amount of fuel consumed, as determined by the computer/fuel injection system, should be fairly accurate. But maybe not as accurate as I might expect?

 

Not only is distance traveled the same, but the measuring mechanism used in both cases is the same.

I agree that the discrepancy is fuel consumed. But, that is when we are comparing numbers for a single car. The odometer issue is not moot, because the comparison between vehicles will have the odometer measurement issue to deal with (as well as the fuel consumption issue). The biggest difference there likely has to do with the actual tire circumferences of the various tires used.

I think they are intentionally inflating the numbers, though I expect they hoped the inflation would be minimized to the point where at least 90-95% of the cars would fall in the overstatement vs. understatement of fuel economy. My car looks to be at 5%. Other cars are at different values. Just ask Bob Wilson -- there is a pattern, but there are individual differences, too.

 

Only comparing numbers from the same vehicle, so tire pressure, diameter, etc is irrelevant. Distance travelled, as shown on the odometer and determined by the OBC is the same. Same distance value used in both the calculated and computer generated MPG numbers.

From another forum, and different car from the prius -


The difference is:
1. You aren't filling up the tank to the same level each time. Even it you let it quit when the nossle cuts off, its a different fill level. Also different nossles cut off at different times. They aren't all set the same. The -only- way you can know it's filled to the same level is if you fill it -completely-. This is not recommended. It can damage the vapour recovery system.
2. The car calculates the fuel used by the injector firing. This is pretty accurate, but not exact. Your errors indicate a 4-10% accuracy. That's very good!
3. Fuel volume changes with temperature. If the car calculates, say 10,000 injector firings, the actual amount of fuel can be different at different temperatures. The car is unlikely to account for that, as it's not intended to be a lab. quality fuel measurement device. You couldn't afford that!

1 MPG difference at around 50 MPG is 2%. You're fooling yourself expecting that kind of accuracy. Most mileage indicators are +or- about 10%. Add to that the filling inconsistency, and you get the kind of numbers you are seeing.

 

The only real issue I have with the above statements concerns the issue with trying to fill the tank with the same amount of fuel each time. This DOES introduce error into individual, tank-by-tank measurements of MPG. It DOES NOT bring in any significant error when calculating lifetime MPG. The only error across the lifetime will be the difference between the first fill and the last fill.

The other small issue concerns the volume/temp issue. This is somewhat true for the first few minutes (at most) when the car is being run in very cold temps. The temp in the pump will gradually increase to a steady state, and then there will be little variation, if any. Also, the oxygen sensors will adjust the flow if the fuel mix is too rich (colder gasoline), so the error (though real) will be too difficult to measure (my opinion).

 

I think the answer is - There are a number of sources of error, but perhaps (?) the biggest is the error associated in trying to determine the amount of fuel dispensed by the injectors. If I believe what he said, then fuel consumption is based (calculated) on the number of "firings", and is not directly measured by a flow meter. That being the case (??), the amount of fuel injected at each firing can have some variance, depending on atmospheric conditions such as temperature, humidity, altitude, pressure, etc.

But, as this author indicated, it's probably unrealistic to expect accuracies much better than 2%-10%. In a BMW forum, on this same issue, I see where some Bimmers have a way that the owner can calibrate the OBC so that the measured and calculated values match. I'd be surprised if toyota had such an adjustment that it's owners could make?

As long as i'm getting better than the EPA numbers, which I am, then I'm cool.

 

Except for the HCH-II, all other Honda hybrids have mildly optimistic iFCD displays. This is well known trivia in the HCH-II community.

Cheers;

MSantos

 

Yes, it is simple.
I believe Toyota engineers already know true_k value within 1.5% accuracy. However, Toyota did mpg = d/f*(true_k)*0.95, I believe.

I built SuperMID M-1 mileage meter, and it was used at the famous Prius 1400mile Marathon attempt in 2005.
Toyota.com : Hybrid Synergy View : 2005 : Fall : Prius Marathoners Top 100 mpg

The accuracy was minus 1.1%, 109.4mpg displayed vs 110.3mpg pumped.

Ken@Japan

 

The discrepancy between indicated and calculated in my previous CR-V was very low. A few times they matched and but most of the time calculated was around 1% lower. So, I guess Honda does know how to make it accurate, but maybe not so much implemented in the HCH....

 

Here are the stats for my last 4 fill-ups:

Miles Gallons Disp MPG Calc MPG Diff
===== ======= ======== ======== ====
370.....8.26.....46.9....44.79.....4%
333.....7.72.....44.6....43.20.....3%
298.....7.03.....43.7....42.32.....3%
283.....6.42.....45.7....44.08.....4%


The difference seems pretty consistent from tank to tank. My Gen II readings were all over the place.

 

The display is showing about 3mpg strong compared to the calculated. Lifetime (8400 miles) mileage is approx. 53mpg by the computer.