Archives

My last tank spurred me to look into how key starting an engine frequently affects FE. For a little background, this tank was the first composed of my new back road commute and subsequently my first tank with a huge amount of key starting...an average of once every 2 or 3 miles after coasting to a stop sign. I thought I was saving all kinds of fuel per the ScanGauge's reported tank average of 42.0mpg, which is usually a little low. And yet my hand calcs at the pump returned a 40.44mpg tank, which works out to roughly 1/2 gallon more fuel burned than I should have over the 552 miles driven. The question is, where did the fuel go?

I think I know where. My assumption, which is probably at least close to reality, is that the fuel injectors are fully open and spraying gas into the cylinder during a key start. My injectors, which I believe are rated at 270cc/min, should be capable of injecting at a rate of 4.27GPH or 0.0012gal/sec. This means that a typical key start of ~2s consumes 0.0024gal of gasoline.

Compare that to idling. According to the ScanGauge, my engine consumes fuel at a rate of 0.3GPH at idle when fully warmed up. That's 0.000083gal/sec. So how many seconds of idling equals one key start? In theory, it works out to 29 seconds!

Now, how about seeing how much fuel I should have used to start the car over the course of this tank? My estimate is ~220 key starts. 220 starts x 0.0024gal/start = *drum roll* 0.528 gal. A little more fuel than I was "missing" but very, very close!

I can tell you with absolute certainty that most of my coasts followed by a key start do not work out to 29 seconds or more. I can usually count to 10 but seldom more than 20. So this means that it's time to test the theory by adjusting my strategy. Dan thinks, and I think he's right, that this doesn't apply to bump starts. (He responded to a thread at GS...Dan, I hope you don't mind your mention here but I want to give credit for your input.) So one option is to key off when I want but always bump start before coming to a complete stop instead of using the key. The other option is simply to take a NICE-on when it's going to be less than 30s, which would be most of the time for me. If my ScanGauge matches my calculation more closely next time, and I'll try to use that pump again, it's good evidence (though admittedly not hard proof) that it really doesn't pay to cut the engine as frequently as I have been if the key is required to re-start. At least for my car.

Score another point for hybrids, which spin the warm engine past idle and then start injecting fuel.

**EDIT**
Found a possible dumb mistake on my part. See below...

Yes, I'm thinking that when key starting, the ECU will inject extra fuel because it senses the starter curcuit has been activated. When bump starting, the ECU can't sense the event so you get the normal amount of fuel injection. I haven't looked at this very closely to be sure however. I will try to remember to do it today.

I might be an idiot. I assume 100% duty cycle but I also assumed only one injector firing at a time. In a way this makes sense, but I don't really know what the car is doing.

Here's the "Worst case" scenario (everything multiplied by four).

That's 0.0024gal x 4 = .0096gal/start.

0.0096gal/start / .000083 gal/sec at idle = 116s or 1 minute, 56s of idling per key start.

220 starts x 0.0096gal/start = 2.112 gallons used to start the car that many times.

Those numbers don't really pass a "sanity check" when compared to the discrepancy with the ScanGauge. I also can't think of a good reason for an injector to fire at a closed valve. Ok, so I might not be an idiot afterall.

Anybody know how the injectors work on startup? All open or firing in sequence?

sometimes i get greedy and delay myself from reaching the stop light, hoping by the time i get to my goal, i can quickly bump and then only idle for a few seconds before the light turns green. As opposed to stopping completely for a few seconds and then realizing you have to key on anyway because the light is changing.

Only on the intersections where I know the other street gets to hog the green light is when I say "Ok! I hope this thing can crank!"

Still don't have any hard knowledge of what's going on at startup, but I did a little experiment to see if I can find out what the injectors are doing during closed-throttle engine braking. All I did was point the car down-hill at 30mph in 4th gear, take my foot off the gas for a couple of seconds, and toggle the ignition switch. Know what happened? Absolutely nothing. The car didn't shudder, decelerate, or otherwise change behavior when I turned the ignition completely off. Though my butt-dyno is not well calibrated, this would suggest that there is nothing going on in those combustion chambers during that condition.

What this means: While my SG readings may be deceptively high when I FAS constantly, this supports the theory that the SG may register deceptively low FE when one uses a lot of engine braking. (Of course, engine braking means loss of momentum.) Tomorrow morning's planned fill (albeit a bit early by my usual standards) ought to register a calculated number higher than the SG this time, or at least not crazy-low like last time.

I think my next tank will be an attempt to put all of this theory together: I'll go back to using FAS regularly, but only under circumstances that allow for a bump-start to get the engine running again. So rather than approaching stop with FAS-->Brake-->Key Start-->GO, it will be FAS-->Bump-->Engine Brake-->short idle period (generally)-->GO. Of course, that's counting my chickens a bit prematurely. Need to get the info from this fillup first.

ya i just assume if i'm in gear, and foot off gas injection ceases no matter what SG says...

since when rpms get to a point where the engine will stall... well it stalled because it has no gas! plain and simple...

i see it as another tool to add to your hypermiling box, engine braking is still loved since it saves my brakes when i know i'm going to stop sooner or later.

I got antsy and filled rather than waiting until tomorrow.

43.2mpg! ScanGauge said 40.0mpg.

The calculated result is high, as expected, but almost too high for me to trust it. A PR tank is not what I was expecting after the kind of driving I've been doing. Some calculations are needed to see if this passes the sanity test. I did go back to the same station but couldn't use the same pump.

Alright, here's how the math breaks down:

First thing I wanted to check is how the hand calcs for the last two tanks average out vs. the ScanGauge numbers. The average of the first tank and second tank based on hand calculations is 41.6mpg, and the average based on the ScanGauge is 41.1mpg. That's a 1.2% difference, which is in-line with what I generally observe. (The ScanGauge has always reorted slightly lower than hand calculations.) Score one point for the theory that this whole thing is pain, simple pump error.

The next thing I wanted to check is whether or not this fill and the last fill balance out mathematically if I assume that the ScanGauge is reporting correctly. The check: the previous fill resulted used 0.51gal more gas than I expected, so today's fill ought to have required roughly 0.51gal less than expected based on the ScanGauge. In fact, I ended up filling ~0.76gal less than expected. So it's kinda what it should be, but not exactly.

At the end of the day I have no idea what's going on. I can't trust the ScanGauge 100%, I certainly can't trust the pump 100% (although it worked in the past). So I can't draw any kind of conclusion yet on the frequent starting thing. More testing will have to be done.

The real kicker is, now I have this PR tank in the log but there's a good chance it's just dramatic measurement error. Maybe the answer is Wayne's super crazy high top-up fills? I don't necessarily like that idea due to the possibility of fuel loss and needless pollution (his PZEV has a sealed fuel system, my ULEV doesn't.)

Mostly because I wasn't sure if it was right or not, asked around, and you're the first person in a week to say anything to the contrary across two websites. So I ran with the 1 injector x 100% duty cycle as a worst case scenario (equivalent to the four injectors firing in such a sequence that one is always open). If you're right, (and after all this I'm inclined to believe that you are), then it isn't much of an issue. Key starting is worthwhile after virtually any measurable period of FAS. Works for me.

Thanks for the info! I think I'll just get back to driving and taking the results for granted...

Any interest in researching some of the coast-in-gear vs.
coast-in-neutral issues? You theoretically get fuel-cut in the
former but more slowdown drag, and an MPG hit from idling in
the latter ... the question still stands, which is better. At
least for your car, in this case..
.
_H*

Any interest in researching some of the coast-in-gear vs.
coast-in-neutral issues? You theoretically get fuel-cut in the
former but more slowdown drag, and an MPG hit from idling in
the latter ... the question still stands, which is better. At
least for your car, in this case..
.
_H*


I like to

coast in gear
when
I wan't to decelerate or when I can coast in gear between 1000 and 1500 rpms (below 40 mph)

coast in neutral with the engine idling
when
I'm traveling faster than 40mph, don't want to decelerate, and can't coast for at least a minute

coast in neutral with the engine shut off
when
I won't need the engine for at least a minute and I want to maintain kinetic energy, most often when descending long hills.

I'm not claiming this is a best practice but asking for feedback. This is currently in a 2002 Hyundai Elantra, manual transmission of course.

Hi Tim:

___The average car uses ~ 7 - 10 seconds of idle time fuel consumption to light the ICE off and 2 gallons over 200 starts not including the reacceleration phase sounds a bit high.

___Hobbit, a NICE-On vs. FAS scenario through a section of roadway passing by a local gas station into a mall, the FAS allows another 20 - 25% increase in FE over the same 2.2 mile stretch according to my instrumentation.

___Add the two up and the FE goes through the roof :D I agree a clutch start is even more fuel efficient (I have an auto though :() but a key start should not use near the amount of fuel calculated earlier. Our non-diesel Accord 2.3/2.4L ICE’s consume ~ 1.4L/hour at idle. With that, you can back calculate an actual fuel consumption (7 - 10 seconds worth of idle fuel consumption) to bring her up via a key start.

___Good Luck

___Wayne