Cruise control operated pulse and glide

[bwilson4web]

Hi,

Some of us over in GreenHybrid are planning to do some comparative testing of pulse and glide versus steady-state speed. The first problem is coming up with a reproducible, pulse and glide protocol. The following pulse and glide protocol eliminates the 'foot feather' so we can get consistent, reproducible results:

1. Start at 43 mph on cruise control
2. Pull back cruise control to 'suspend' speed control
3. At 40 mph, put car in "N" for a ballistics glide
4. At 25 mph, put car in "D" and resume cruise control
5. When car reaches 43 mph, GOTO first step

This is a 25-43 mph, pulse and glide cycle, the maximum speed range that can easily be supported by the cruise control. Cruise control 'turns off' at 23 mph so the 25 mph low end avoids having to reset it each time. The acceleration from 25 mph to 43 mph will always follow the the built-in cruise control profile. The "D" mode coasting from 43 mph to 40 mph will trigger the ICE to stop. The ballistics glide from 40 mph to 25 mph will achieve a uniform, minimum drag coast without special effort.

The only drawback of this protocol is going down a grade in "N" violates many state driving codes. However, like traveling 5 mph over the speed limit, this is not a big deal in most jurisdictions and enforcement risks are a different problem.

There are several variations of this protocol:

• 25-43 -- maximum P&G speed range
• 25-38
• 25-33 -- minimum P&G speed range and minimum speed
• 30-43
• 30-38
• 35-43 -- minimum P&G speed range and maximum speed

This will let us develop a performance model that should allow us to compare P&G to an identical, steady-state speed over the same course in the same conditions.

I did some practice runs this afternoon and this is what they look like:



This was on an access road that was not terribly level but enough to get an idea of the equivalent, steady-state mph, 33-34 mph. Last year I was able to plot some steady-state speed runs:


For this test, I'll run a combination of pulse and glide and constant speed runs.

Bob Wilson

[ekpolk]

Bob:

Thanks for posting the info. It will be interesting to see how this plays out. What are you using to collect and preserve your data?

EKP

[onlynark]

good stuff bob, please label your x and y axis on the second graph. Thanks

[efusco]

It's always fun and interesting to try new techniques like this and you should see some good results.

I do have a couple comments:
a)I'm not sure how/why you chose 43mph...I think 40 or 41 would've been more logical in that ICE-off stealth would be, essentially, immediate and you could go into Neutral without having to wait for the ICE to shut off as long. Also, I wonder if using 40mph might 'encourage' the CC to be a little less aggressive when you resume at 25mph...it may be the same. You definately fall well outside the 1700-2300rpm 'sweet spot' with the more aggressive acceleration that the CC gives you.

the main 'flaw' with any kind of test like this is that P&G is a dynamic and variable process. Traffic conditions, type and number of hills, stop signs, and traffic lights all impact the way P&G is used. With any kind of formulatic process you may simplify things, but you take away a lot of what's advantagous about the interactive process of anticipating terrain, stops and traffic conditions to allow you to maximize the efficiency of the process.

[bwilson4web]

<div class='quotetop'>QUOTE(nark @ May 5 2007, 04:12 PM) [snapback]436141[/snapback]</div>

good stuff bob, please label your x and y axis on the second graph. Thanks
[/b]

Opps! Will do but it may not happen right away. Give me a day or so:

Y-axis - Miles Per Gallon
X-axis - Speed in Miles Per Hour

Bob Wilson

[bwilson4web]

Hi,
<div class='quotetop'>QUOTE(efusco @ May 5 2007, 04:21 PM) [snapback]436144[/snapback]</div>

. . .
a)I'm not sure how/why you chose 43mph...I think 40 or 41 would've been more logical in that ICE-off stealth would be, essentially, immediate and you could go into Neutral without having to wait for the ICE to shut off as long. Also, I wonder if using 40mph might 'encourage' the CC to be a little less aggressive when you resume at 25mph...it may be the same. You definately fall well outside the 1700-2300rpm 'sweet spot' with the more aggressive acceleration that the CC gives you.
[/b]

I found that once we've reached the cruise control speed, pulling back to trigger 'ICE off' takes a second or so and I need to move my right hand from the cruise control stalk to the shifter. With a little practice, I can be ready to shift into "N" just as 40 mph shows up on the display. I had tried using 41 mph in one case, the ICE didn't shutdown before the car was in "N". In this mode with the car in "N", the ICE continues to run until put back into "D." For me, it is a question of squeezing in two manual steps and I'm comfortable with a 43 mph to 40 mph interval.

<div class='quotetop'>QUOTE(efusco @ May 5 2007, 04:21 PM) [snapback]436144[/snapback]</div>

the main 'flaw' with any kind of test like this is that P&G is a dynamic and variable process. Traffic conditions, type and number of hills, stop signs, and traffic lights all impact the way P&G is used. With any kind of formulatic process you may simplify things, but you take away a lot of what's advantagous about the interactive process of anticipating terrain, stops and traffic conditions to allow you to maximize the efficiency of the process.
[/b]

In engineering testing, I like to minimize all other variables. I have no doubt that there may be special cases but I need a baseline first. Then we can investigate things like 'terrain driving' and other effects. But if I'm investigating these special cases, expect to see them taken one at a time using a methodology that can be easily (relatively easily) repeated by anyone.

Actually we can compare Friday and Saturday practice sessions. The difference is Friday used an access road with about a +/- 5m change in elevations and the Saturday session used a route with +/- 1m difference:





BTW, I'll be sharing my raw data along with a spreadsheet with formulas and small samples. The source data will consist of time stamps with millisecond resolution:

1. MG1 rpm and MG1 torque in one sample
2. MG2 rpm and torque in one sample
3. battery current
4. MG2 torque and rpm in one sample
5. MG1 torque and rpm in one sample
6. fuel injector time in milliseconds

Each cycle takes a little over 1 second.

Bob Wilson

[JimboK]

Some additional information: I am working with Bob on this. I had been considering doing a test comparing P&G at various speed ranges with one another and with steady speed driving. My goal would be to determine which gives the best overall MPG for a given distance, irrespective of elapsed time. When Bob mentioned on GreenHybrid the desire to compare P&G with steady speeds, with elapsed times consistent, I offered to do such a test along with those I've considered.

My test course is 2.7 miles through an industrial park. It's not a loop, so I will do an out-and-back drive for each method tested. Allowing for time at each end to get up to speed before beginning measurements, I figure the total test to be about 5 miles. It's not completely flat, but it has several flat sections and inclines are modest. It has an approximate 60' elevation change from one end to the other. Here is a link to a map of the course:

http://www.gmap-pedometer.com/?r=918909

About half of the course is four lanes, the other half two. Traffic isn't bad even on weekdays. (My test would be on a weekend.) But I still have to consider the possibility of traffic on the 2-lane segment requiring either repeat tests or limiting the test to the 4-lane segment. The speed limit throughout is 45 MPH, and folks won't tolerate my gliding down to 20 MPH or so without being able to pass.

Agreeing with one of Evan's concerns, I will add tests with my normal acceleration method and rate, using many of these same P&G ranges. This is less reproducible and probably more variable, but it's more realistic for me and others who don't use CC to accelerate. This will also help compare acceleration methods and rates for P&G application. I will use my ScanGauge to maintain RPM within a certain range, probably in the neighborhood of 1800-2200.

At Bob's suggestion, I will aim for at least three runs for each method tested and average the results. This will take some time, as you might figure. Bob, if I find the results within each batch of three to be consistent, I'll probably stop there for the sake of time. If I have an obvious outlier or other significant variation, I'll bump it up to five and discard the highest and lowest before averaging.

Unlike Bob, I don't have a dedicated real time data-logging device. So I'll use an audio recorder to capture data, entered later into a spreadsheet, and they will be limited to what the ScanGauge and the car's OEM instrumentation provide.

[bwilson4web]

Preliminary data using 25-43 PNG and 34 mph CC:
Direction,Temp,Style,MFD MPG,Distance
S,73,CC,73.9,2.6
N,73,CC,84.8,2.6
S,73,PnG,77.8,2.1
N,73,PnG,94.1,2.2
S,73,PnG,84.9,2.1
N,77,PnG,93.0,2.6
S,77,CC,76.5,2.6
N,77,CC,90.2,2.6
S,77,CC,70.0,2.6
N,77,PnG,90.3,2.7

Averaging all five samples:
88.0 - PnG (+11%)
79.1 - CC

Averaging the middle three samples:
87.0 - PnG (+12%)
77.1 - CC

I've recorded the data with a Graham scanner and will do a detailed energy analysis later. However, I'm not happy with the variations in results:


Of the samples, I'll start my detailed analysis with the best and the worst of PnG and CC. This will hopefully give insight to improving the protocol. A cursory look at the data suggests the north bound tests did better than the south bound tests but the relative performance ratios look to be about the same.

Bob Wilson

[thebrattygurl]

Bob -

I think my brain just exploded!

[bwilson4web]

<div class='quotetop'>QUOTE(thebrattygurl @ May 6 2007, 10:57 PM) [snapback]436662[/snapback]</div>

I think my brain just exploded!
[/b]

Scoop them back in because I think I've figure out a fix to the data variability. With Pulse and Glide, I have a great deal of operator input so I can ensure complete cycles. However, with cruise control, there is a cycle of ICE on/off that I have no control. I'm hoping my Graham data provides enough detail that I can adjust for complete ICE on/off cycles. But there is another fix.

I can use the 'forced charge' to put the battery in the same state at the beginning of each run. This will give both Pulse and Glide and steady-state cruise control the same 'boost'. The overall numbers will be unreasonably high but I can document that with a warning. But the main thing is I should achieve groups of results closer together so we can compare the differences.

The risk is my 2.5 mile test course may be too short. Even at 34 mph, a fully charged battery may run for such a large portion that I don't get a full cycle. My only alternative is a stretch of about 15 miles west of Decatur that looks to be fairly flat. It is a divided highway and I can drive it safely with blinkers.

The ultimate answer is a closed, flat track but I'm concerned about fees. If the speeds were slower, I could use some of the large parking lots and I may try one or two just for grins. Most of these lots are posted at 25 mph so 34 mph is a little too obvious but we'll see.

Bob Wilson