Actual MPG vs. MFD reading Comparison Spreadsheet

that is very nice.

2 observations:

1 - i see you got better over time
2 - i wonder if the variability of the delta will be smaller in the bladerless 2010?

 

#2 - Without a doubt. There is still some variation with gas pumps and filling techniques, but the bladder is the biggest wild card.

Tom

 

Nice job. Thanks on behalf of those who don't already have such a tool.

One small flaw: with the tank-to-tank variability in miles driven, your averages in columns K & L will be off a bit. For example, the 421 mile tank in row 11 contributes significantly more to lifetime results than the 138 mile tank in row 10. You might consider a formula like I have in mine to adjust for the variation. The formula provides a weighted per-tank MPG difference based on miles driven, and then the weighted differences are averaged for the lifetime number. I plugged your numbers into my spreadsheet and it calculated a lifetime 3.3% difference.

I've attached my spreadsheet, which you're welcome to steal from as you see fit. I don't track lifetime fuel economy -- I rely on CleanMPG's database for that -- but I do track fuel economy calculation differences. But the concept of the formula will work for column K as well as column L.

 

I would suspect you are correct. With the fuel bladder causing inconsistent fill volumes, your tabulated economy will likely vary widely from the MFD rating. I saw this on my last tank - it was HOT outside, and I was able to get in more gas than I had before given the state of the fuel gauge. It calculated out to only 43.xxMPG. I'm about 320 miles into the tank and it just hit 1/2. More distance than usual before that 6th bar went dark.

We really have to look at the average over a long period - months or years - to verify the MFD average. I use mymilemarker which shows a nice graph of historic economy.

 

There are other factors at play... and so far, they appear to amplify the guestimation error... comparing my 118,185 mile spreadsheet for the Iconic to my 2010.

Having the engine shut off more often and longer is the biggest influence, I believe.
.

 

Ah, yes, right you are! My apologies. I was focusing mainly on the formula for column L and didn't bother to inspect closely the formula in K. I nice person-u-me-d it was written the same as L. :redface:

 

I always keep my gas receipts, and have never be able to make much sense of the calculated mpg. The MFD mpg I assume is the more accurate, though bladder fluctuations tend to average out over many fillups.

I don't know why the Prius has a bladder in the gas tank--there must be a thread or two on the subject--but the bladder concept was originally used in WWII fighters to reduce flammability of the gas tanks (the collapsing bladder reduced volatile fuel vapors.)

I would assume the bladder serves the same advantage in the Prius--reducing the possibility of gas ignition in a collision. Though it may be there for a different reason?

 

It is there to control evaporated emissions. A good idea whose time has not quite arrived.

 

It's there to eliminate evaporative emissions... since not all aire pollution comes from the tailpipe.

There are other ways to accomplish it, which is what the new Prius employs.
.

 

It's still a good idea, but now they do the same thing with a different tank design that uses a pump system.

Tom

 

I don't like either method. When my fuel log eventually starts (not yet, 323 miles down and 5 pips left), I'll compute an 'MFD expected' gallons, and a running total of this and actual gallons pumped. As the tanks add up, this should produce a reasonable error estimate. But I still need to go through JimboK's formulas to see if they may be equivalent, or nearly so, to mine.

On Kenko's ten tanks, I'm getting a 3% error, but the numbers are still too noisy to be accurate to more than a full percentage point. On JimboK's 41 tanks, there is enough data to get an error of about 2.5%, give or take a few tenths (last line on my altered spreadsheet is 2.66%).

 

You haven't seen anything yet, Grasshopper. Wait until you get a "won't fill properly" episode. I've had one now out of 24 tanks...the first on any vehicle I've ever owned. The result was indicated mpg = 53.8 mpg, and calculated 114.5 mpg, a 60.7 mpg difference.

The bladder is certainly part of the problem, but the vent recovery system is another. Sure would be nice to have a high point manual vent release on it for when the system clearly isn't working properly.

 

I get approximately the same result as you (3.19%).

It's important to calculate the MFD gallons for each tank, then compare the two gallons values. Trying to calculate averages from ratios (mpg) and failing to normalize yields skewed results.

With so few tanks each fill has a large impact on the lifetime average error.

Some of the error (over long time periods) is coming from gasoline that bypasses injectors by entering via the evaporative recovery system (sucked in as unmetered fuel.) Another suspected source of error (at least by me) is that in winter the injector pulse volume will contain higher density fuel and therefore the MFD will undercount total gasoline injected. My current cumulative error stands at 1.1%.

 

This is misguided speculation, as the amount of fuel converted to gas in the recovery system is highly unlikely to significantly change the fuel/air ratio in the combustion chambers. If this were the case, and the mileage difference between calculated and MFD shows a noticeable difference, it means the ICE will also be taking a significant hit on fuel efficiency and power output as it will be burning rich. Burning rich in the combustion chambers means COOLER temperatures for the pistons, which means the ICE will be operating more often to keep temperatures up and MPG will take a hit. Highly doubtful that Toyota has designed the system to be self-defeating.

Another misconception - the fuel system controls air/fuel ratio, not just fuel flow rate; air density also increases with temperature drop and f/a is maintained by the fuel system, adjusting according to power demand. If the fuel system is supplying higher density fuel and not adjusting air intake accordingly, there will again be the situation of fuel-rich combustion taking place and MPG will drop drastically - highly unlikely and not what has been observed by many including myself. Although MPG drops in winter, it's mainly due to temperature regulation in the engine, not due to fuel density variation.

 

Dobey's back to troll me some more...

What is misguided speculation is your application of air/fuel ratio. It is controlled via the O2 sensor feedback in addition to the basic injector/MAF mapping, etc. However, there is not to my knowledge a flow meter on the evaporative emissions canister, nor does it do gas analysis to determine composition. So the control system is going to inject the called for fuel at a given airflow rate and fine tune the injection pulse based on O2, etc. It is my understanding that the MFD calculates fuel consumption based on the injector pulse lengths. If one adds a few tenths of a percent of extra unmetered fuel it will be taken care of by the control system, but one's fuel economy consumption calculations will experience a systematic error. The calcs to my knowledge don't depend on the air input, only on the distance and fuel consumption recorded.

Now it is unclear how significant such an error is. I don't expect it to be very large, but it will introduce a systematic bias. What would be "misguided speculation" is to claim the error didn't exist and therefore cannot be significant.

It is possible that the MFD somehow compensates for this by application of a factor or something even more complex, but it seems unlikely, simply because the magnitude of the error is unlikely to be large.

So it comes down to a question of how much fuel actually bypasses the injectors in a typical tank. Is it 1%? Probably not under typical conditions, but I lack a basis for a mass balance on the evaporative emissions control. 0.1%? 0.01%? Less than that? I don't know, but I'm curious.

Wow, you really don't know what you are talking about. There are feedbacks in the control loops, as well as compensation done. The system is relatively robust. With the system you describe the car's controls would be stymied by changes in ambient temperature or elevation (pressure.) It would be unable to adjust the air/fuel ratio for anything other than its base conditions. It would be unable to compensate for differing fuel compositions. And heaven help it if the calibration of each injector and the MAF were not perfect. It would habitually run rich or lean and be unable to compensate.

Again, you have it backwards. The control system adjusts to target the desired ratio(s). But that doesn't make it psychic about things it doesn't meter or does not have built in/programmed in compensation for.

It is possible that the system is measuring fuel temperature at the injectors (or assuming it based on other variables) and is then doing a density correction. It might also be compensating for changing dimensions in the injectors themselves with temperature. This would add a layer or two of complexity and require some assumptions by the programmer.

MPG loss in winter differs from accountability. The fact that you just tried to equate them reveals how poorly you understand what is being discussed. Whether the mileage is 1 MPG or 100 MPG there can still be discrete accountability losses that are separate. I'm pointing to a systematic permanant accounting issue.

Another example of the difference is the temperature response of the bladder, and its impact on MPG calculations. If the bladder shrinks 10% of the normal fill volume between fills, all else being equal the next fill will indicate 1/9th higher mileage. (e.g. 500 miles on 10 gallons actual = 50 mpg, then refill but only get 9 gallons in due to shrinkage. Calculated MPG for the tank would be 500/9 = 55.56 mpg.) As weather warms the impact is the opposite of the above. This will tend to mask some of the actual mileage loss on the onset of winter, but it will be recouped as spring approaches. There is only a short term accounting impact assuming that a full tank is run at the same base conditions, in which case the overall loss/gain will be zero. On the other hand, unmetered fuel is permanent.

Try again.

 

The control system is going to adjust the fuel/air ratio based on the exhaust gas composition as detected by the O2 sensor. Therefore, any additional fuel injected into the airstream prior to the fuel injectors will be detected by the O2 sensor as additional fuel beyond that demanded by the fuel schedule (which is determined according to load and speed). Therefore the fuel system will adjust the fuel flow to account for this additional fuel from the evaporative gas line by reducing injector pulse duration, and hence f/a ratio is maintained. Q.E.D.

Which is a function of load, speed and f/a ratio as determined by the O2 sensor.

Wrong. You even contradict yourself. If the fuel control system "takes care" of the fuel supply based on f/a ratio, it means the fuel supply is adjusted to take into account the evaporated fuel from the gas tank, i.e fuel supply is reduced because the exhaust gas indicates additional fuel is being burned. Hence the system adjusts for this difference.

So you go from asking the question of how significant this speculative theory is, i.e.

..to making an assertion of fact that it is significant. Why don't you produce the data that shows this? Oh wait...

So you really don't know, you're just speculating. In any case, it's irrelevant, since the f/a ratio is determined and adjusted from the exhaust gases through a mass balance of the chemical composition of the exhaust gases, indicated by the O2 amount. The fuel system adjusts fuel flow to compensate for this.

So now you back off from your assertion of fact and claim curiosity instead. It's not the same as knowledge, and your assertion that evaporative gases from the gas tank significantly impacts MPG indications has no solid basis.

Wow, you really can talk BS. Talking of feedbacks, control loops, compensation without really knowing what they mean.

An erroneous f/a ratio indication (either through faulty instrumentation or intentional misuse) for the engine control system means the engine will run outside of its optimum schedule - there is NO OTHER indication that will tell the engine that its combustion mixture is wrong, and there will be no other thing that will help the engine correct this. Faulty fuel control instrumentation and improperly calibrated meters and sensors will ALWAYS mean the engine will run rich or lean. I thought you were joking here, but evidently not, and since you "know" so much about this, why don't you outline how the engine can correct its fuel supply if the O2 sensor and MAF/VAF meters are faulty? What control loops will the system take into account to correct the fuel/air ratio, and how will the system know that it's at the correct ratio?

Wrong again. The exhaust sensor determines whether the f/a ratio is as per schedule - this includes any fuel introduced in the system prior to the exhaust (and yes, this includes any evaporative gases from the tank). The exhaust sensor doesn't care where the additional fuel comes from, it just knows that it needs to adjust fuel flow to compensate according to fuel schedule. Therefore any additional evaporative fuel from the gas tank will be seen by the exhaust sensor as additional fuel burned, and will reduce fuel supply from the pump accordingly.

Nonsense. You raised the temperature difference in winter as a factor in your theory of variable MPG determination, and wordsmithing isn't getting you out of your hole. What you fail to understand is that fuel density variation is IRRELEVANT in MPG determination by the fuel system, since the exhaust sensor determines f/a ratio by exhaust gas composition, i.e. mass balance, not by initial volumetric f/a mix - fuel and air density variations are accounted for in the mass balance in the exhaust gas composition analysis (i.e. O2 sensing). If there's too much fuel, O2 sensor will indicate this and the fuel supply is reduced until the gas composition is as per correct f/a ratio. It doesn't matter that the fuel density varies with temperature, the system determines the proper ratio through chemical analysis.

Toyota acknowledges that fuel bladder variation can be up to 5L (1.3USgal) at 14F - about 11% difference total for the bladder. This accounts for all the significant MPG calculation differences that people experience. Your talk of being able to detect 1.1% variation in MPG is absolute rubbish: human variation in performing the same tasks repeatedly, with no changes to equipment and environment, can result in up to 70% variation in output. There are many six sigma studies that indicate this, a few I have personally been involved in. So what chance do you think you have of controlling and maintaining your daily environment, car, let alone your personal attributes and disposition to have no significant variations when you conduct your driving similarity tests? Extremely minutely small...

And I do wish you would try harder to gain understanding of your own theories before letting loose... a seemingly authoritative advisory mode doesn't mask the lack of technical basis for your advice, although one must applaud your enthusiasm to help, however misguided.

 

Look, my pet troll has returned. Took him 5 days to come back with a rejoinder. Let's examine it, shall we?

That's pretty much what I said. Unfortunately, you don't seem to appreciate what it actually means (this is a theme for you.) Unless the MFD actually back calculates a fuel quantity (including an estimate of other non-injector sources) based on the adjustment it makes for the O2, it is going to have a systematic error in fuel consumption. Not only that but it has to assume a density as well...something you obviously have not even considered. Are you speculating that it ignores its own injector pulse/length count to account for other fuel sources? Because that is the only way to reconcile what you are saying.

What do you think the metered fuel quantity will be? For example, let's say an extreme amount of fuel, 10% is coming in the form of vapor rather than through the injectors (yes, this would be ridiculously large but is only to make the numbers obvious.) In that case the MFD would see 9 gallons injected...but 10 gallons would actually pass through the combustion chamber. The extra 1 being as a vapor stream rather than liquid injection. You are saying it would ignore the injector pulses that said 9 and report via the MFD that 10 gallons were injected. Fat chance. The fuel density variation and composition difference could exceed that easily.

You can't have it both ways, so which is it? Are you contending that the MFD will ignore the injector pulses that it called for and instead rely on secondary (mass air flow, air/fuel ratio) and tertiary (O2 feedback) to determine the actual quantity of fuel consumed. If so I would love to see your source for such a claim as it appears most dubious speculation. Not impossible, but dubious.

Secondary and tertiary measurements. The air/fuel ratio will differ depending on the actual fuel too... Seems unlikely that a smart engineer would base their calculation on less reliable measurements when the injector pulses are well established and volumetric (like MPG.)

No contradiction. You assume far too much as part of your "misguided speculation." You are making the leap that because the O2 sensor allows it to respond, that it will also adjusts its estimate on the MFD of how much actual fuel was injected. Recognizing that MAF sensors are less reliable than injector pulse lengths, and that fuel composition is variable, I find it highly unlikely that the MFD is making the calculation you assume. In fact, it would likely screw it up since the volume would not necessarily correlate that precisely with combustion composition (think oxygenates or density.)

Bunch of redundant crap skipped...

The only hole is in your head. You can't see the difference between tank-to-tank variability and a cumulative error/departure. Damn, if that's not funny!

Provide a source rather than your idle speculation. I'm open to the possibility of some really complex calcs, but as an engineer I also recognize that what you suggest would tend to introduce several times MORE error than the injector based calculation, making it implausible. And this for a sensor (O2) that is prone to failure! (My Tundra is on its third set as part of recalls.) Plus, the error would be more random than what users here have reported--a bias in one direction. Their results are consistent with my basis, and inconsistent with yours...not that this is anything new to our "discussions."

I'm still comfortable with the notice I've made of a density based departure in winter. Next winter I'll attempt to confirm it.

Smart readers will be guffawing over the above. Tank-to-tank variation is not the same as long term. You just failed Prius 101; this is one of the fundamental things that Prius owners learn. You are in so far past your depth that I'm ROTFLMAO.

I've seen huge tank-to-tank variability, yet there is a trend emerging in the cumulative error...1.1% at present. It moves from tank-to-tank but the range of movement is ever smaller with each tank...just as anyone with an understanding of numbers would anticipate.

So, Sherlock, have you ever considered that if the tank size fluctuation is 1 gallon, that the denominator is changing with each fill? A 1 gallon error is alot in a single tank, but over 100 tanks it becomes rather small. No, they aren't cumulative. All that matters is the initial state of the first tank and the final tank.

Perhaps you are incapable of tracking total fuel consumed, distance covered in each tank, MFD mileage per tank, and therefore calculate MFD indicated gasoline consumption. However, many of us here are quite capable of such calculations. Since the MFD displays in tenths of an MPG (~0.2%) the odometer in unitary miles (~ 0.2-3% per tank, much less over the long haul), and the pumps I use display in hundredths of a gallon (~0.1%), it is not that hard to measure a cumulative difference in the 1% range.

So let me get this straight: you are incapable of recording the total gallons put into the car at each fill? Or incapable of recording the MFD mileage indication and/actual odometer mileage each fill? Or is is that you can't figure out how to backcalculate MFD gasoline consumption using the above? Or perhaps you can do all that but are unaware how to calculate the percent difference in the two?

It's not rocket science, ask for help.

There you go, talking to yourself again. You would do best to follow your own advice.

Try harder next time, you just might learn something.

Better yet, rather than playing the part of the jackass trying to kick down the barn (credit to Sam Rayburn), do something constructive that actually helps the community. Explain the observed behaviour, hotshot. This is something you proved incapable of last time. The departure I see is toward the low end of the range.