Measuring Fuel Economy - Instantaneous and Otherwise

[Mighty Mira]

Originally Posted by The Princess Bride

Vizzini: Inconceivable!
Inigo Montoya: You keep using that word. I do not think it means, what you think it means.

In a similar fashion, I think we may need a rethink about mpg (and l/100km), and instantaneous fuel economy.

As I understand it, the typical instantaneous fuel economy device calculates litres used and distance travelled over a short time, performs a simple division and outputs the result. Am I correct here, and is this the same thing that the Honda Insight does?

From what I can gather judging by reading individual posts here, this is the case.

The thing that we need to remember is that we are attempting to minimize fuel consumption over the average distance the car travels. We should also remember that we live in one place, and any time we move house, the distance between destinations is fairly small in terms of total miles. I.e., for every time we drive up a mountain, we come back down.

Also, we need to understand that it is braking rather than accelerating per se that really costs us in the FE stakes. Every time we accelerate, the extra fuel used increases the car's kinetic energy. Every time we scale a hill, we increase the car's potential energy. Both of these activities bring about future above average FE because the energy from the fuel is still in a form that will add extra km to the car's journey.

If we don't understand this, it may (and most likely, will) lead to optimizing the wrong variable. (For example, I've been driving to minimize fuel for some time now. I took some relatives somewhere in a car that has a trip meter for mpg, and amazed them unintentionally by getting the lowest mpg they had ever seen on that trip. And that surprised them, because I basically drive to minimize braking at all costs. Whereas their idea of FE drives is to accelerate very, very gently but not give a hoot about the braking.

A better function for FE with more inputs would lead to better overall driving. (It might possibly also make testing certain FE mods easier.)

Am I correct in my assumptions here (at least as far as a typical instantaneous FE meter)?

On a similar topic, I notice that we are measuring fuel economy increases due to various modifications as X mpg, where X might be 1-2 or 5-6 for example. This also seems to be a bit primitive (not to say that I can come up with anything better yet). Of course, it is nice to know, just may have limited use for comparing modifications between cars.

To be really pedantic, CdA and Cd figures should be obtained for the aero mods, RRC should be obtained for tires, and (in a fit of wishful thinking), a fuel/kW/rpm for the engine, and some sort of efficiency/speed curve for the driveline. Then you could combine them all in a calculation to estimate what those modifications would do for your own car.

Obviously, going this far is almost certainly overkill unless you are an automobile manufacturer. However, even a partial solution would be better than the one we have.

[SVOboy]

The superMID allows you to track instanteous, last km, trip (total), trip (while engine off, doesn't count EOC stuff as part of the mileage). It also allows you to have a side by side comparison of trips and a lap function which allows you to store and compare up to 20 different pieces of information, so I like it in that feature.

Have you ever seen the post from way back last year about measuring cd?

[Mighty Mira]

SVOBoy: No, I haven't seen that post. But I just found it - excellent!

As to the SuperMID - what does the instantaneous fuel economy do
a) under hard braking
b) under hard acceleration
?

I strongly suspect that under hard acceleration, the SuperMID will give you poor l/100km, while under hard braking it will give good l/100km (instantaneously of course).

The main point of my first post was that humans (and other animals, like rats, hamsters, monkeys, dogs etc.) tend to be excellent at optimizing an input to a given device to get some sort of instantaneous feedback/reward. They soon figure out what input will get what output, and modify their behavior around this.

And that's what we have essentially in a car - a few inputs (accelerator, brake, steering wheel, gear, ignition switch for the adventurous). It took until recently to get a more immediate method of measuring fuel economy than trip odomoter/fuel used/ tank, or a more complex instantaneous method than a vacuum guage.

Animals and humans are much less able to optimize when there is little feedback. And naturally so - as the number of variables exponentially increases, doing so requires either understanding the theoretical basis on which the feedback is given, which is something that people rarely have the inclination or the ability to figure out.

However, if you can combine the two - instantaneous AND accurate feedback, you can maximize FE driving for a given human.

[SVOboy]

Well, when ever I'm braking the engine is off or I'm also engine braking, so it reads 99.9 km/l (0 fuel used), however, if I were in gear I imagine it would read 99.9 until I got below the rpm range of engine braking at which point it would read the same as if I were coasting in neutral (I've noticed that if i 0 throttle it below the point of engine braking the FE is the same as if I were coasting in neutral, or at least very close)

Under hard acceleration it depends on how fast I'm going/gear. If I'm doing a WOT run in first gear it'll be like 5 mpg.

[Mighty Mira]

As an illustration of where optimizing to a first order instantaneous FE reading can get you, try this thought experiment:

Pick a long, flat, straight at a time where you can be assured there will be no cars.

1) Set the odometer:
2) Accelerate hard to 100kph.
3) Stick the car in neutral until you stop, or EOC.
4) Examine total fuel economy of the trip.
5) Now, reset the odometer:
6) Accelerate slowly to 100kph.
7) Slam on the brakes.
8) Now, examine the total fuel economy of the trip.

Now, the first experiment will give you a better FE result than the latter, even though the former gave you atrocious instantaneous fuel economy figures for acceleration part of the trip. (As an added experiment, try accelerating slowly to 100kph then either EOC or neutral driving, see if there is any difference to the jackrabbit start. I bet there is hardly any difference.)

The point of this is that in order to train yourself to be a maximally FE driver, you need to be rewarded for those exact behaviors that give you good FE and punished for those exact behaviors that give you poor FE.

[Mighty Mira]

Originally Posted by SVOboy

Well, when ever I'm braking the engine is off or I'm also engine braking, so it reads 99.9 km/l (0 fuel used).

Exactly. Which tells me that the SuperMID uses the following equation to generate km/l:

Inst. Fuel Economy = Delta Distance/ Delta Fuel.

Let's try another thought experiment:

Say we are travelling at 100kph. Then we slow down: We get the same IFE figure no matter whether we brake hard or EOC.

Question: If the above driver bases his driving habits purely on readouts from the instantaneous fuel economy meter, is he likely to adopt an optimally efficient driving style?

[SVOboy]

I think so, yes, because the point from which we begin to slow down is governed by whether we brake hard or EOC from a distance.

[Mighty Mira]

Originally Posted by SVOboy

I think so, yes, because the point from which we begin to slow down is governed by whether we brake hard or EOC from a distance.

I'm not sure if I understand you correctly. Are you saying that a person who is optimizing on the basis of wanting to increase instantaneous FE will want to keep it at a higher number, for longer? If so, then yes, we will see an increase in EOC and hence an increase in FE (primarily because he has stopped accelerating at an earlier point, hence he has stopped the waste of gas earlier).

OTOH, if the driver is a bit of a moron, he will see identical figures for braking and EOC and thus could say "what's the difference?" and choose braking, as it doesn't require looking ahead and concentrating on the road, leaving more time to eat McDonalds, admire himself in the mirror, or sing along to the radio.

Also, I suggest that such low mpg figures for accelerating hard are severely misleading, since that fuel energy is converted to kinetic energy of the car. I suggest that the difference in FE between a jackrabbit start and very slow acceleration is trivial, provided that the same stopping habits are observed.

[SVOboy]

Well, that is certainly under debate, but let me suggest something to you.

When under WOT conditions the engines run significantly more rich than under lower load acceleration, which leads me to the conclusion that unnecessary amounts of gasoline are burnt to cool the combustion of the engine.

Also, through accelerating slowly I often do not reach top speed before I realize a braking event will occur. So many times I will hit 30 mph and realize the light turned and I should coast rather than jetting to 40 mph and hitting the light and having to brake off 10 extra mph.

[Mighty Mira]

Originally Posted by SVOboy

Well, that is certainly under debate, but let me suggest something to you.

When under WOT conditions the engines run significantly more rich than under lower load acceleration, which leads me to the conclusion that unnecessary amounts of gasoline are burnt to cool the combustion of the engine.

Hmmm. I'm not sure whether that would cancel things out. It might, or it might not. Very interesting that you brought that to my attention.

One thing to bear in mind is that the reason smaller engines are more efficient is that there are less losses in friction than larger engines for a given load, because the smaller engine is delivering a larger power output for the energy wasted as friction.

In fact, the engine needs to expend gas just to idle. And as hybrids are proving with automatic shutoff technology, those losses are significant. If you can produce more power for the same loss, you can become more efficient. And discounting the weight difference, it's possible that if you have a small engine with similar losses to a large engine, if you can run the large engine for a short burst at the same throttle as the small one, then EOC, you should get similar FE number to a small engine and sustained running.

Now, I'm not suggesting that we are going to find the best position for a max FE start at WOT, but perhaps there is a happy medium somewhere between Granny start and WOT.

How could this be tested?

You need to know the steady state 100kph fuel economy of a given car. (Actually, you need to know the speed at which the absolute maximum steady state mpg of the car is at, and accelerate to that speed, not 100kph.)

Then you do several tests. Do various accelerations to the Max FE speed, all the way from Granny through to WOT.

You figure out the distance it takes the Granny start to accelerate to the Max FE speed. Then, you measure the total fuel used over that distance.

For the various other speeds, you can either accelerate to the Max FE speed, hold at that speed until the Granny distance and then measure the fuel used, or you can just measure the fuel used in getting to Max FE speed then calculate the total fuel used.

Whichever acceleration gives you the least fuel used is the most efficient throttle to accelerate at. (With the proviso that it's more than likely that the best throttle to use will vary with both speed and rpm, and not be a constant.)

Also, through accelerating slowly I often do not reach top speed before I realize a braking event will occur. So many times I will hit 30 mph and realize the light turned and I should coast rather than jetting to 40 mph and hitting the light and having to brake off 10 extra mph.

Yes, there is that. And at other times when you slow it down you will see the green light change to red, and if you had accelerated a bit quicker you would have made the light and not had to have incrementally worn your starter motor out that one bit more (or used the gas idling at the lights).

Of course, my above statements are mainly concerned with things like highway on-ramps and those moments where you can time the lights to the point where you know if you don't gun it a little now you won't make the lights. You could also just modify the driving habits to accelerate at optimal FE acceleration, hit the speed that you know the lights will change, then coast. This could both slightly improve FE and piss off the people behind you less.