I have linked a blog called "Daily Fuel Economy Tip in the right column of this blog. I enjoy reading it, and have left quite a few comments. Brian Carr's (the host of the blog) most recent post is a follow up to a previous post regarding turning off the engine at stop lights. I also do this, and I have done a fair amount of googling on the topic.
As anyone reading this will likely be aware, the web is awash in sites discussing fuel saving methods. A lot of these discuss avoidance of idling, and there are large differences in the "break even" times for turning off the engine and avoiding burning fuel while idling versus extra fuel used to start the engine. Brian estimates an increase in m.p.g. of about 5.8% from this technique alone. On his site, I left a comment contemplating whether that was a plausible number, I'll copy the comment here:
"I also do this, however, I wonder if a significant portion of your increased mileage in the second period comes from this. Let’s see if it’s plausible:
If you had driven the 1559.9 miles in the second period with the miles per gallon (31.25) from the first, you would have used 49.9 gallons. Instead you used 47.2 gallons, a savings of 2.7 gallons. Now, I’ve had a Jeep Grand Cherokee Limited with a 4.7 liter engine and now have a Land Rover LR3 with a 4.4 liter engine. The Grand Cherokee used 0.38 gallons per hour at idle, the LR3 about 0.5 gallons per hour. Your car probably uses less than either of these at idle but let’s assume 0.4 gallons per hour. It would have taken you 6.75 hours of idling at stoplights to burn 2.7 gallons. In a 30 day month, that would be about 13.5 minutes of sitting at stoplights each day, every day. Assuming that a stop is an average of 40 seconds (don’t have data, just a guess), that would mean that, every single day, you were stopped at a little over 20 stoplights.
Now, if you have a much smaller engine (I forgot what you drive), that would mean you’d have to be spending even more time at stoplights to have turning the vehicle off at stoplights be primarily responsible for your savings. Do you think that this is the case?
Also note that these calculations assume absolutely NO extra fuel use on startup. I’ve searched the web and can find no figures for extra fuel used on startup, but the assumptions above are clearly the most favorable for fuel savings by engine shutoff."
It would seem that his savings must be due to some other factors as well, yet the numbers above aren't completely out of the question. As mentioned there, I know well how much fuel I'm burning as I sit at a light (or coast to it) but it's been very difficult finding any information on extra fuel used to start a spark ignition internal combustion engine. Various sites claim six seconds, 10 seconds, 30 seconds, and one minute as the break even point. No one cites any data to back up their number, and they vary by a factor of 10. This is not especially helpful.
However, after replying to Brian Carr's post, I tried changing my google search string to "idling versus shutting down engine". This led me to this site. It's the closest I've found to giving me the answer I need to determine whether shutting off the engine at stop lights is a fuel saver. It says the following: "Our research showed that a V6 restart takes about the same fuel as 5 seconds of idling. We expect a V8 to save more and a 4-cylinder less." It doesn't state the nature of the research, whether it was actual measurement by the authors, by associates of the authors, or was located in a literature search. It ain't much but it's all I've found.
As I've mentioned before, I haven't done experiments by varying a single parameter and keeping all others as constant as possible to determine the effect of that parameter. Brian Carr says he has done so and the article from the ASME Florida Section provides some verification. So I suspect that, while Brian's entire 5.8% increase in fuel economy isn't due to this procedure, a significant portion of it is.
I am going to run an experiment to determine this once and for all. I don't have a way to directly measure the fuel consumption on startup, but I'll find a vacant stretch of road forming a closed course a couple of miles long. I'll run it for an hour or so stopping and idling for a measured period every mile and find the total fuel consumed. Then I'll drive the same distance, turning the car off and back on every mile - it won't matter how long the shut off period is. I'll be concentrating very hard on duplicating the acceleration, cruising, and deceleration regime of the first series. I'll start both with a full fuel tank.
At the end of this process, I should have enough data to approximate the break even time and from there, the excess fuel used on startup. No doubt this number won't reflect what's happening before the car warms up, but it should provide reasonably definitive data on the question of turning the car off at lights. I can't take credit for formulating this methodology - I read it on someone's web site. I'd credit the site, but I can't find it now.