Driving with Engine Efficiency in Mind

[bikeracer]

Driving with Engine Efficiency in mind

After a long conversation with recently graduated mechanical engineer and Dartmouth alum on the subject of engines, engine efficiency, and hypermiling, I've decided to attempt to articulate a slightly different theory on what I understand to be the traditional approach to hypermiling. The important points of this assertion are that engines are more efficient per unit of fuel when running further open (this has wrinkle in it as Ben Jones/SVOBoy pointed out, namely that engine computer tend to become contrary above 80% open in certain cases) because an engine must always work to overcome its own load (internal friction, compression, and when in gear, drive train drag, etc) and if it is working to overcome that load and working at 10% throttle, and for the purposes of this argument we say it uses 5% of its power running itself, than it's using 1/3 of its power just to run and only putting 2/3 of its power to use on the road. When running more open, the percentage that is turned into useful driving work increases. A mechanical engineer I spoke to suggested numbers somewhere in the range of <20% efficiency at low throttle position and 30% at full-open. This seems to translate to driving technique by imagining that one would do well to drive hard when the engine is running and leave it off at all other times. (Please don't yell at me about these percentages—I'm sure they're wrong but the concept is sound.)

Contrary to popular thought that driving up hills with load and accepting a certain speed loss is the most efficient way to drive, I now think that perhaps the most efficient way would be to lose no speed at all up hills if possible and perhaps use the hills as an opportunity to accelerate to the desired speed so that one would need to spend no time at the top of the hill regaining lost speed before going into a coast. Intuitively, this doesn't seem right. It seems like one could accellerate more efficiently on a flat or downhill section of road--that less time would be spent accelerating under those conditions--but perhaps by using the engine as efficiently as possible on the uphills, it would save something in the long term. When testing this theory today, I found that I was surprsingly, no, convincingly efficient. My trip meter lured me into thinking this is the way to do it. As a counterpoint to the argument that spending less time speeding up is more efficient, I say again that if one is to ask something more of their engine, they should ask everything they need at once and then let it sleep.

On the highway, where maintaining a higher-than-pulse-and-glide speed is necessary for not getting pulled over on a Saturday evening, I translated this technique to using hills to get to 65mph and then driving down the other side with the throttle at a position where there would be a steady but slow loss of speed. I was able to average about 35% better mpg this way than had I driven with a constant throttle position (according to my non-scientific or mathematic analysis of my Scangauge's output).

Also, a story related to me by Ben/SVOBoy makes me suspect that it would work for seasoned hypermilers. The story goes something like "My MPGuino died/temporarily stopped working and i had to drive without it, so i wasn't thinking about shaving MPG's up hills any more, just maintaining speed, and that tank was my best ever.""

hmmm....

[meanjoe75fan]

I've been thinking about this, too.

If we assume the "default" setting (for highway travel, anyway) is the "cruise control" approach, then motorists strive--either by CC or manually--to maintain X mph, regardless of road grade.

Then "alternative setting #1" is DWL: to the greatest extent possible, maintain a constant throttle setting, while allowing speed to fluctuate. This results in LESS vairiation of throttle, and is touted as +FE.

"Alternative setting #2" is EOC: to effectively "pulse+glide" the terrain, by using the uphills as your "pulse," and the ensuing downhill as the "glide." Since this involves 1. zero throttle downhill, and 2. all your throttle on the uphill, this results in MORE variation of throttle, and is touted as +FE.

So, the message seems to be "just don't drive like a cruise control, and you'll get better FE," which seems to stretch the limits of credulity. On the other hand, plenty of accomplished hypermilers use DWL to get better FE, and I don't want to discount their experience. So, I think perhaps the solution is:

Use DWL for short, rolling hills where one can rely on thier momentum to carry them over. In other words, for the short "whoops" in the road, better to DWL and allow a small fluctuation in MPH.

Use EOC for sustained up- and downgrades that are too long and/or steep to count on momentum. In this environment, DWL is insufficient as a stand-alone option, and the terrain seems to be "strongly encouraging" an EOC alternative.

"Anticipate" transitions to up- and downhills in a way that a CC can't. In other words, as you crest a hill and see a long, sustained downgrade, just EOC right there (conditions permitting) instead of accelerating to get back to "cruise setting" before descent. After all, why use gasoline to get up to speed when you can wait 10 secs and get gravity to do it for you?

If I've misunderstood the issue, please let me know, but this seems intuitive righ now.

[bikeracer]

I might not have enough experience in driving techniques to write confidently on DWL efficiency, but it seems that from a purely engine efficiency standpoint, it is the least efficient way to go about driving. The engine is working evenly at a relatively low output throughout the drive.

I think that cruise control fails because it keeps the speed constant on the downhills thus often squandering the period in which the engine can be run very gently—or not run at all—in a speed-losing power zone for maximum efficiency. It seems that the MPGs required to maintain speed is a lot lower than what is required to lose 5mph over, say, 15-30 seconds. By a lot, I mean that maintaining speed I use, for example, 35 mpg, and to lose 5mph in 15 seconds I'm using 60 mpg. Using downhills lets one extend the period of time at which you can spend getting the upper value. By gaining the speed back on the next climb and running the engine in a more efficient throttle zone, one takes full advantage of an opportunity to run the engine.

Why does DWL work?
Does it? Is it simply causing a lower average speed and thusly saving gas?

[user removed]

My neighborhood connects to a 4 lane 55 MPH not divided highway.

In eastern Virginia where I live the elevation ranges from 70 feet above sea level to sea level. Elevation changes are small to non existent.

The first 7 miles of my drive has one traffic light and very few intersections or traffic entering or exiting the road.

I found a site where they were dyno testing a 4 cylinder GM engine. At 20 HP load they used 1 unit of fuel. Increasing the load to 50HP the fuel consumed only rose by .5 units.

Basically this demonstrates that the additional 30 HP only cost half as much fuel as the first 20 HP.

20 HP/1 unit of fuel
50HP/1.5 units of fuel

20 HP per unit
versus
33.33 HP per unit

That's the secret of P&G as I understand it.

A hill that was the perfect grade to allow you to use the 33.33 per unit to increase your altitude, then glide down the down slope with the engine off, would give you a huge increase in mileage over driving on a level grade using the 20 per unit rate of consumption.

It would be a short steeper grade then a much longer slighter grade for the coasting portion of the cycle.

I averaged 70.2 MPG in the Insight using the strategy of trying to keep the instant MPG bar as high as I could going uphill, while letting the grade accelerate my car going downhill.

Speed ranged from 65 to 40 MPH. If I remember right the miles were 655 on 9.6 gallons of gas. Average speed on the road was about 55 MPH.

For the last 17k miles on the Insight my average MPG has been 65. That average started in early December so the yearly average will probably be fairly close to 65, maybe slightly less.



On the right heading east are railroad tracks running parallel to the road. On the left is I64 East.

The grades are very small probably around 1% or even less. Elevation changes of 20 feet take several tenths of a mile.

The down slopes are not enough to maintain speed in most cases.

I find that in this specific circumstances my best mileage is to hit my peak speed at the top of the hills and coast as far as possible while maintaining an average of 45 MPH.

Usually this is a range of 52 to 40 MPH.

That gets me 55 MPG in my Echo and almost 90 in my Insight.

The best measurement of load is MAP or a vacuum gauge. Either will tell you how much of the atmospheric pressure is actually getting into the cylinders. Throttle position is not a good measurement as load changes will result in increased MAP (or reduced manifold vacuum).

As you climb a grade and your MAP increases your engine becomes more efficient.

I found that even on flat stretches in the Insight I can accelerate very slowly, about 3-5 seconds per 1 MPH increase, with very little decrease in fuel mileage. This is demonstrates by the instantaneous bar graph in the Insight. If I accelerate about 7-10 MPH at this very gradual rate, then I can let off the throttle and watch the mileage jump to the 100-140 MPG range for about the same amount of time it took to accelerate to the higher speed.

Using this strategy on that 7 mile stretch of road I have seen close to 90 MPG in the Insight.

regards
Mech

[RobertSmalls]

Mech, I didn't know you were driving an Insight. Gen I or II? You should make a Garage entry for it.

Anyway, I've studied BSFC plots, and yes, you get more power out of each drop of fuel at 80% throttle. Pulse and glide should outperform DWL. However, if you apply heavy throttle on the way up the hill, and leave it in the same gear on the downhill, you'd probably be better off DWL. i.e. DWL works by eliminating the period of being in the wrong gear on the downhill.

My normal method of acceleration is to keep the revs as low as possible, and the throttle around 80% open. This actually results in rather slow acceleration in the Subaru, and if I suddenly decide I want to accelerate quickly, I have to downshift. I run out of the appropriate gearing at 50mph, and my accelerating from 50-74mph is actually brisk.

The Insight's BSFC's minimum is actually 70% throttle, 4000RPM. I'll test the 1-2-5 shift pattern that some Insight enthusiasts use. For a fuel economy run, I might attempt a 1-2-EOC-2-EOC shift pattern.

[user removed]

Robert, my old brain forgot to mention my 02 is a CVT, so no lean burn or gear choices .

regards
Mech

[99LeCouch]

I always thought the proper way to DWL was to very slowly lose speed up the hill down to your minimum speed (60 mph here for a variety of reasons). I've definitely come into some hills doing 70-72 mph, aka the interstate speed limit around me, and very slowly lost the speed climbing up the far side. For my car keeping a steady throttle position for as long as possible works nicely.

What I've noticed works pretty well is "helping" your coast at the top of the downgrade when your RPM's and speed are low from having climbed the hill. This works very well on routes you know. On routes you don't know, you risk speeding by 5 mph or so.

[SVOboy]

Patrick and I are planning to go out and try to do a somewhat controlled test of this theory soon. I was planning on doing it like such:

DWL based on constant LOD using SGII
Hilly P&G using LOD as the pulse target

Does this sound like it makes sense for the issues at hand?

[robchalmers]

I've never really got DWL, to me it seems adverse to let a car slow and load-up on a hill to main tain a throttle % I'm with the OP here. for Hills I use the opposite to the 'dakar' technique of dune-ing. Constant speed running (CSR) up the hill, then coast down the other side. for Highway/Freeway hills that works great. However on shorter sharper hills I tend to go back to true dune-ing , Accelerate on the down - more accleration for Engine useage due to gravity aid. then use that momentum up the next hill.

[RobertSmalls]

Originally Posted by SVOboy

Patrick and I are planning to go out and try to do a somewhat controlled test of this theory soon. I was planning on doing it like such:

DWL based on constant LOD using SGII
Hilly P&G using LOD as the pulse target

Does this sound like it makes sense for the issues at hand?

You should also compare it to cruise control, if possible.