Pulse & Glide vs Cruise Control - a study

[Doc Willie]

Here's my first attempt with preliminary data from seven 10-mile runs. Details will be presented at HybridFest.

Speed limit: 35-55 MPH 46.1 MPG 32 MPH Av
P&G: 38-15 MPH 78.9 MPG 23 MPH Av
CC: 23 MPH 67.1 MPG 20 MPH Av
P&G: 40-25 MPH 66.2 MPG 27 MPH Av
CC: 29 MPH 73.3 MPG 25 MPH Av
P&G: 43-5 MPH 62.3 MPG 22 MPH Av
CC: 25 MPH 76.6 MPG 22 MPH Av

Summary: P&G in this range is more efficient only within a certain range (in my case determined by setting the pulse range by the IGN [ingition advance] readings on the ScanGuage. At the fifteen MPH and Max Pulses, cruise control is more efficient.

This data is surprising to me, and I intend to make another set of runs.

[bwilson4web]

Hi,

You may want to check some of the data but this is my best guess at plotting it:

Make sure your data shows: min<average<max -> MPG. For plotting, use the average mph and the Y-axis should be MPG and on the right, the min and max mph.

This data is consistent with what I'm seeing:


Good job!

Bob Wilson

[KandyRedCoi]

great I have been using the CC per recomendations by another member here, I guess I'm going to keep on doing it

P&g works better at slower speeds tho

[efusco]

Any stops? What terrain? What kind of traffic did you deal with? Were all runs the same direction or did you P&G one way and CC the other or is it a round trip? What outside temps? Were all done the same day and approximately the same time? How did you deal with stops (if any)...did you coast to all of them?

Was battery SOC taken into account at all..I'd like to see the exact beginning and ending SOC for each run...the thing is, with such a short run the MPG can be dramatically artificially inflated during the last 1/2 mile of the run by increased use of EV/stealth toward the end. The MPG only takes into account the amount of gas energy, but battery energy could be a factor too.

I think you'd need to repeat each segment 3-5 times and get a better average. (for instance the P&G: 38-15 MPH run should be done 3-5 times and the CC 27mph should be done 3-5 times so you can see how consistant your numbers are). I'm not enough of a statistician to put it together, but I suspect there would be not nearly enough data here to show statistical significance for any of the numbers.

If anything can be taken away here it might be that using CC is not significantly inferior to P&G. My experience here in hilly SW MO with no less than 14 stops during a 14 mile drive and modest traffic is that P&G is dramatically superior...minimum 10-15mpg kind of superior...I've only been doing that run for about 3-4 years for about 40,000 miles of my 72,000 on that particular stretch, but I'm pretty sure about my data!

I'm not slamming your effort, I like it, I just don't want anyone taking anything definitive from it at this point. Someone else experienced with P&G can replicate your data I think we'll be able to see if there is any serious trend.

[bestmapman]

Originally Posted by Doc Willie

Here's my first attempt with preliminary data from seven 10-mile runs. Details will be presented at HybridFest.

Good testing Doc. I need to do something like this myself.

The thing about P&G is it is very skill and technique dependent. We should be wary of claiming a definite results without further testing. Because of the skill and technique variables, I don't think we have a valid result yet. Cruise Contral on the other hand, takes the human element out of the equation. There may be differences car to car, but at least the human variances are negligible.

If we had someone like DiamonLarry run P&G over the same route, we may get quite a different result. He and others are pulling out some astonishing numbers

I know JimboK was thinking of coming up with a set of parametes and having several people replicate the exact detailed set of circumstances to validate the results. I don't know what happened to that idea.

If we get a group of people all trying the same parameters, I would like to participate.

[JimboK]

Originally Posted by bestmapman

I know JimboK was thinking of coming up with a set of parametes and having several people replicate the exact detailed set of circumstances to validate the results. I don't know what happened to that idea.

It's still very much alive. I've just been waiting for a weekend day off with decent weather. It seems like it's been cool or rainy every day off since spring began.

My goal is not only to compare P&G to steady-state, but also to compare two different P&G methods with one another. Finally, I want to compare various P&G acceleration rates with one another.

For the benefit of Willie and other interested parties, here's some background. Bob Wilson and I first started talking about comparison testing last spring in this thread and on GreenHybrid. You'll see in the GH thread that I had a number of practical problems that cropped up during my testing. By the time I finally got into a rhythm, so to speak, I had to unexpectedly replace my tires, pretty much rendering invalid any comparisons of subsequent results to those already obtained. Between that, hot weather (I didn't want to run the AC for the tests); a house hunt, purchase, and move in the fall; and winter, I suspended testing. I'm now ready to begin anew.

The test course and its traffic were part of my problem. (It is linked in either thread if you want to see it.) I think I found a new one a few weeks ago. It's 45 MPH, but traffic seems very light on weekends; it's in the area of some industry stuck in the middle of an otherwise rural area. And it's 4 lanes virtually the whole distance. Net elevation change and the difference between high and low points are probably pretty comparable to the old route, but overall it is flatter, with the changes less frequent and more gradual. From end to end it is 3 miles, so I figure to use about 2.5 of it for actual testing -- about the same as the old course. I will run up and back 3 times for each method tested, recording the MPG after each one-way leg. I will average the 6 numbers for the final result.

As you can see from his test protocol, Bob compared P&G at a range of 25-43 MPH with CC at an equivalent average speed. His P&G technique is different than most of ours (to save space here, I'll just refer you to the links for a description). At a minimum, I want to repeat his test and then compare both results to a more conventional P&G. I want to do the same at other speeds and ranges between 25 & 40 MPH.

As you can imagine, this will be time consuming. So I envision multiple phases of testing spanning multiple days. What I describe above will be phase 1 and will be to determine the best overall method. If P&G proves superior, phase 2 will be conventional P&G in the following ranges:

• 15-25 MPH
• 15-30 MPH
• 15-35 MPH
• 20-30 MPH
• 20-35 MPH

From those results I will pick the top two or three speed ranges, and then during phase 3 will use those to compare various acceleration rates as defined by ICE speed -- say, 1400, 1800, & 2200. Not sure yet which rates I'll use.

We have to be careful to account for confounding variables. Weather in particular can be problematic. First, I want to test on a day with little or wind. It might seem that with an up and back test route, the wind's effect would balance out. But it doesn't. A tailwind coming from your 6 o'clock gives the best push, as you would expect. But wind coming from about 33 degrees off your front is worse than a dead-on headwind. And 90 degree cross winds from either direction hurt fuel economy. We have to control or adjust for temperature changes throughout the day and from one day to the next. Etc., etc. So to make adjustments for variables, I will use Wayne Brown's Prius MPG Simulator. I will have my laptop computer with mobile broadband on board to regularly check temperature and wind conditions throughout the test, and use simulator results to adjust for changes.

Evan mentions state of charge. Bob also mentions its potential effect in his PC post. He didn't control for it in his tests, but he suggests force charging to a defined level before each test to eliminate yet another confounder. I might do this before the steady-state cruise control segments. CC routinely has the car running on battery power alone, and the beginning of one test segment could show a substantially different SOC than the next. I could also do it at the beginning of each overall P&G test, though I don't think it will be necessary for each individual segment -- a properly executed P&G is designed to keep you out of the battery, of course. I can use CAN-View to precisely monitor SOC, and I can either force charge or run a short segment of EV to nudge it up or down to the desired level.

Willie, I agree with Evan about reporting your methodology in as much detail as possible. In my high-speed fuel economy study, I reported methodology and results in a format consistent with what I am accustomed to in medical journals (as you are too, I'm sure). Whether it fits the format of a technical paper like the engineering guys might author, I don't know. It works for me, and it documents the scientific method applied to the topic of interest.

[ericbecky]

Doc Willie has contacted me about present the info at Hybridfest. I think this would be a good topic for the Sunday morning member-only technical sessions on at the hotel. Especially if others have info to present as well.
What do others think?

[bwilson4web]

Originally Posted by bestmapman

. . . Cruise Control [rjw] on the other hand, takes the human element out of the equation. There may be differences car to car, but at least the human variances are negligible.

. . .

If we get a group of people all trying the same parameters, I would like to participate.

When I performed my P&G experiments, the initial ones used cruise control to handle the acceleration and "N" to handle the glide. I did everything I could to minimize human variables.

My only other suggestion is to run the car for at least 30 minutes at speeds above 42 mph. This ensures the engine, transaxle and tires are uniformly warmed up and the battery should be pretty much at 60%. Also, I use randomly chosen runs with a spreadsheet random number generator. Then I run multiple tests recording the results.

I've posted my results before (use search function) but the synopsis:

• P&G (as I've described) requires a significant dV >= 15 mph at speeds under 42 mph
• P&G around the maximum range speed, 18-20 mph, is measurably worse than steady speed of 18-20 mph

There may be other techniques than accelerate under cruise control and glide using "N". But I haven't found any that are documented as independent of the individual driver. Until I have a programmable throttle driven by Graham scanner values, I won't have a way to test these other approaches.

Bob Wilson

[efusco]

Originally Posted by ericbecky

Doc Willie has contacted me about present the info at Hybridfest. I think this would be a good topic for the Sunday morning member-only technical sessions on at the hotel. Especially if others have info to present as well.
What do others think?

Eric, what about a panel discussion format. Sounds like we've got a couple people who are on similar thought processes...have them limit their data presentations to 10 min or so then have a panel discussion of techniques with them and other experts...Wayne Gerdes of course, Jimbo, Dan, Doc, maybe one or two more. Perhaps the sub-goal could be to create an 'ideal' study design that could be reproduced by multiple people at home and then that data compiled...

[Doc Willie]

Originally Posted by efusco

Any stops? What terrain? What kind of traffic did you deal with? Were all runs the same direction or did you P&G one way and CC the other or is it a round trip? What outside temps? Were all done the same day and approximately the same time? How did you deal with stops (if any)...did you coast to all of them?

Was battery SOC taken into account at all..I'd like to see the exact beginning and ending SOC for each run...

One turnaround, one stop sign, two stops to enter turnarounds.

Did round trip P&G, followed by round trip using CC attempting to match same av. MPH.

Temps gradually rising 48-54 deg F over 3 hours.

SOC not taken into account. First run done at speed limit to warm everything up.