Difficulties in Effective A-B-A Testing of Insight 1
 

I would like to discuss something that has been bothering me for a while. A-b-a testing is difficult under any conditions, but it would seem virtually impossible to get accurate testing on an Insight 1.

All of us who have tried a-b-a testing know the problems. There are numerous small pertubations which can lead to inconclusive or outright false results. These pertubations include, but are not limited to, changing weather, winds, inconsistent throttle, warming engines and tires, and changes in tire pressure. With experience, many test operators learn to control many if not all of these variable to acceptable degrees. These tight controls frequently reduce the opportunities to test, but delayed data is preferable to erroneous data.

I am particularly concerned with a-b-a testing on the Insight 1, including my own. In addition to the normal frustrations with stable baselining and testing, there are some extra variables which seem particularly difficult. The three big problems which come immediately to mind are: 1. The lack of cruise control dictates that the operator maintain an almost impossible level of manual throttle control. (As owners know, the car FCD is extremely sensitive to the throttle.) 2. The difficulty of controlling inefficient regeneration causes errors in the fuel consumption measurement. 3. Any deviations of SOC from full will cause background charging which will cause fuel consumption errors.

I personally would like to hear some ideas on controlling these problems. I live in Richmond, VA where we have hills north, south and west. I may be able to find a testable road section by traveling most of the way to Norfolk, but I'm not aware of any such road, and population densities off interstates would argue against any such road.

Perhaps the best testing of aero mods would be achieved by doing Cd measurements by coast down, rather than the traditional a-b-a testing. Any thoughts :confused:

I have to disagree with the logic of your three points. You're not getting ERRORS in measuring fuel consumption, because the car's instruments are pretty darned accurate. Specifically:

1) Any halfway decent driver can get better mpg without using cruise control - in any car, but particularly in the Insight.

2) You can control regen/assist with a MIMA setup.

3) Again, those aren't fuel consumption error, they're actual fuel use.

It does seem less than logical to get a car that's optimized for high mpg, then complain because the optimizations make your testing difficult.

However, I'd suggest a different, and possibly more accurate, method of testing. Find a hill on a non-busy stretch of road, start at the top at a known speed, turn the engine off, and measure your coasting distance.

In general I agree, but when trying to measure small improvements in FE, it is important to stabilize as many variables as possible. A-b-a testing involves the measurements of small changes in FE - a difficult process under the best of conditions. Just read back through the record of some folks here who have tried it. I've done a fair amount in an Echo with aftermarket cruise control so I'm well familiar with the challenge. In the past I've had the best results on level road sections with cruise control.

Buying and installing a MIMA seems like expensive overkill just to do some testing.
:eek::eek:

Agreed, it is actual fuel used. The error for purposes of a-b-a measurement is the potential error in the measurement of fuel used.

Not complaining about the FE. Who with an Insight could not like the high mpg ;)

You are probably correct about the coast down. I mentioned that myself as potentially a better way to go.

BTW, have you driven an Insight?

Jim, I agree, the Insight's fuel economy is particularly twitchy. You can control the assist/regen with the clutch switch mod, and that will save lots of gas, too. But you should test in neutral.

A-B-A test results do not have to be reported in mpg. It would be more relevant and accurate to report the change in your CdA after, before, and after an aeromod, or in your low-speed rolling distance A, B, A tire mods or weight reduction.

If all you have is a long, steady grade, you might be able to tease a CdA number out of it, but the math is much easier when you eliminate the gravity term by testing either on a perfectly flat road (which does not exist), or on a mostly flat road in both directions.

If you are testing powertrain mods, such as a hot air intake or an IMA inhibit switch, obviously you can't test in neutral. I would log your commute mpg, with and without the mod. Large sample size is the key to getting meaningful information out of noisy data. However, if the expected impact is only a few mpg, although the mod may be worth doing, the sample size required to determine the impact of the mod may be prohibitively large.

He has one.... but no garage entry for it which would put his contributions in context (hint, hint!).

If you simply want to know if a particular aero mod is better, worse or indifferent from baseline, I wouldn't worry about the math or finding the "perfect" road. Simply repeat the coastdown over the same stretch (absent other traffic, in consistent weather conditions, with a fully warmed up drivetrain) and note the delta in times. The higher the initial speed and the longer your coast, the better.

It's tasty knowing the actual MPG impact, but "powered" testing is more difficult.

jimepting

I live very near Williamsburg Country Club (mapquest it) off of Route 143 which runs from Williamsburg to Newport News Va.

Route 143 is between I64 and the railroad tracks. The stretch near my house runs 5 miles between traffic lights. Speed limit is 55 MPH. The railraod tracks clearly show any grade changes. There is another stretch that runs close to 3 miles after the single traffic light at Lee Hall. Speed limit on this stretch is 45 MPH.

Not sure about the speeds you want to drive, but I can show you stretches of road in my area where there are elevations changes of less than 5 feet in a mile of road, but those are near the water and speed limit is 35 MPH.

Another option is route 5 running on the north side of the James River from the beltway around Richmond to Williamsburg. Not super flat, but there are some stretches that are pretty level, but it's 2 lanes, while the other ones are 4 lanes.

Route 58 east of Emporia also has some pretty flat sections, divided 4 lane with 55 MPH limit, but watch your speed around Emporia, they love to write tickets, well known speed trap.

I also own an Insight, but it's a CVT. My local run has been 70 MPG the last two days with better weather.

I have driven eastern VA roads for 43 years, so I can probably help you find what you are looking for.

regards
Mech

Neither does Jim. Hint, hint!

Like Robert mentioned above, you can eliminate the background charging of the Insight by installing a Calpod IMA disable switch.

Look on InsightCentral on how to do this. The mod is relatively easy to do, and, like Robert mentioned, save you cash on your fill ups.

It saved me about 3 mpg over the last 3/4 of a year or so.

As for coast downs, I am finding out that they are not so easy to get repeatable results while using my GPS, which stores readings every two seconds.

Metro's method of simply driving and measuring fuel consumption at a steady pace might be better.

Pick a road where you can drive faster, as this will show up the improvements of aero mods quicker than driving slower.

Jim.

Yes, I've owned one for the last 6 years.

As for garage entries, that'd involve me keeping fuel receipts & entering things - work, in other words. Why bother when the Insight records mpg automatically?

Wow! That is tremendously helpful for a new resident of Richmond. I'll certainly be checking out those roads on some sunny Sunday afternoon. Maybe we could meet at Mickey Dees for a cup of coffee.

I'm currently enrolled in CleanMPG winter quarter fuel economy contest. My average in the contest is currently running around 83 mpg for my 2006 MT Insight. (Wife's back surgery has kept her out of the drivers seat;-)
I've really gotten fired up by the work being done on boattail designs in another thread here. Being an retired engineer, I really love the challenge of aero improvements. Been rolling over some designs and design factors, but I see testing as a challenge.

Thanks again - ever so much:thumbup:

Fuel logs are actually optional. All you need to do is name your cars and declare their year, make, model, engine, and transmission.

I don't post records either. I do keep a notebook record in my car so that I can track my improvement over time. I'm also active on two sites, so that would double the entry work.

Biggest problem I have with posted records is that it isn't MY record, it is OUR record, and the wife is horrible at efficient driving. She is binary driver, hard on or hard off:(

I want to find her a good, cheap Echo or Metro so that the Insight becomes MINE!

Jim I am retired as well. I would like to show you my IVT design sometime when it's convenient. I think you might like the concept.

I drove from Williamsburg to Blacksburg, then to Leesburg near Dulles Airport and back to almost Fredericksburg. It was 655 miles. Used Route 5, 288 south of Richmond, 360, 29, and some of route 15, 28, and 17.

Got 70.2 MPG on that trip with a lot of work in hills and elevation changes of over 2500 feet.

Driving to Richmond this weekend for a show.

The old US routes are really pleasant drives, since all the madness seems to be concentrated on the Interstates. The old US routes alos tend to follow rivers which makes the grades less severe, and the scenery can be much more interesting.

Sent you a PM

regards
Mech

Thanks Jim. From another site and consultant I was under impression that this mod only worked on 2000-2004 Insights. I have a 2006. I'll take a look at InsightCentral. I'd certainly install the mod if it worked. I find the difficult to control IMA activation to be something of a frustration, though opinions certainly differ on this feature. My wife obviously uses IMA a lot, from the battery SOC each time she finishes driving:(

My Garage
 

Thanks Robert. I went ahead and updated my profile and created a garage.

Your definitely on the right track.

The Calpod IMA disable only requires some small wire, similar to phone cord, and a two way switch.

The hardest part is attaching the wire to the orange or yellow connector ON TOP of the clutch pedal pivot. You can see it if you lay on your back on the floor board, and look up to the extreme top portion of the clutch mechanism.

If you solder small diameter connector pins to ends of your wires, you can "wedge" the wires right into the back side of the connector at the top of the clutch pedal.

If you try it let me know how you like it.

I use my switch all the time, so it does come in quite handy.

EDIT:
Almost forgot.

There was a comment earlier in this thread that holding the throttle steady on the Insight is hard to do.

The ScanGage TPS readout shows that the Insight can be held steady enough to keep it "frozen" at 17 or so over a five mile stretch on the way home in the summer. Even with slight bumps in the road, it will stay right at 17 for the entire five miles, so it is possible to do. This is a straight stretch out in the boonies, with little traffic either way.

Winter driving, with snow on the ground, makes this almost impossible for me, as the drag on tires keeps changing due to the snow depth.

Jim.

There is one concern I have for getting gas consumption readings, similar to Metro and his tail extension:

1) The car without the under body smoothing panels and tail extension
2) The car with them both attached

The Insight has lean burn, and I'm wondering if the car without the aero aids will actually be able to get into lean burn at 55 to 60 mph on an old country road that is flat enough for this. Once in lean burn, the car would need to able to stay there for at least one mile.

Of course, this would need to be repeatable in both directions and for several iterations.

I live near the Interstate as well, but the A-B-A testing would all have to be done in the wee hours of the night, and it takes about 1.5 hours just to jack up the car high enough to be able to remove all the body panels from underneath the car. And they are held on with lots of fasteners. :D

Now imagine doing this for three test sessions (A-B-A), and getting all this done in one day while there is little traffic on the road, and the atmospheric conditions have not changed much!!

I do however, think that driving at a constant speed will give better results than coast down testing, again similar to Metro. :)

Jim.

Ah Ha. NOW you are beginning to have testing concerns like me ;)

Anyway, a bit of information. My 2006 seems to maintain lean burn well at 55 MPH on level, or nearly level, roads. I don't think you will have any problems with that. The problem I see is getting a stable speed and stable throttle setting as you approach your test segment. Lots of stuff to control and look at. You gotta get past acceleration/deceleration, into lean burn, IMA battery fully charged, stay out of assist - it's a challenge but probably doable with the right road and some practice.

I think that you are correct that the best approach is to test all mods at once. In addition to underbelly and boattail, I'd advocate at least grill block (That big old lower grill opening looks lossy.) and maybe even some mild rear wheel smoothers behind the wheels. There is no point in introducing a lot of "noise" with incremental improvements. Just go all the way to what you are willing to live with. JMHO.

Hi jimepting,

There may be another approach that is somewhat reasonable for comparing aero changes without doing it all in one day.

I still am collecting coast down data on a single road on the way to work each day. This is assuming no snow, and clear road conditions. Most of the time, this road has no traffic at all in either direction.

The idea is that collecting this data will hopefully show the effects of air temperature on the coast down numbers as summer approaches.

With information in hand, one may be able to perform constant speed road testing during calm conditions (which is hard to find in it's own right) and use the temperature guide as a way to correlate different runs on different days, with varying temperatures. The temperature greatly affects the molecular density of the air, and thus the overall drag component. Come to think of it, the temperature also affects the Crr component as well.

I still favor making multiple runs with one setup, on one day, maybe making drastic changes to the car on the weekend, then testing those changes on the way to work.

Maybe not as clean and consistent as doing all the changes in one day, but adding under body panels AND the tail extension to the car would take multiple hours to accomplish just one change.

Practicality speaks!!

Jim.

And those fasteners tend to break on removal after they've been exposed to the elements for a few years, and they are not cheap. Though FYI, you can find them online much cheaper than from the Honda dealer.

Hi James,

I don't have stock under body panels, but I know what you meant.

Mine are mostly held on by 1.5 inch plated plaster board screws.

When things get a little warmer, it will be time to get a look at everything after a long winter's driving with the new panels.

Jim.

I certainly agree about practicality!!! It is a powerful constraint.

I did some experimental driving today on I64 east to Williamsburg. I was interested in the lean burn performance at speed. This route has long gentle hills, pretty much one after another, for most of the distance. I targeted two cruising speeds, 55mph +/-2, and 60 mph +/-2. I was able to pretty much hold his over the entire distances. I think that level of speed control is adequate, since the errors will cancel. I noted that on my car I was able to stay in lean burn at both speeds, with an occasional downshift to 4th, clicking right back to lean burn. On the boattail testing, I think 60 mph gets the nod because we are trying to improve the high cruise performance - if I'm not mistaken.

I think I am going to search for a good interstate section of about 10 miles one way as a test course. (From memory, I believe that I95 south of Richmond is superior to I64 east, but I have to have a look.) For most areas that I have seen the interstate is near optimum for testing at this speed. The occasional truck that approaches from the rear can be warned with a couple of quick flashes, as can any dense traffic approaching from the rear. There seems to never be any forward obstruction at this speed ;)

It seems to me that on something as complicated as a boattail testing, it almost has to be tested and baselined on different days. As you alluded, that will mean that one must observe temperature and barometric pretty closely and duplicate conditions as closely as possible, or be able to make some adjustments as suggested. Of course, tuft testing can be done on most any day unless it is raining. (Any substantial testing in one day usually involves changes in temp anyway, as you noted.)

jimepting,

The question I have, is on the Interstate, how do you pick a time when others will not try to pass you in the other lane??

In Wisconsin, we have I-94 within about 10 miles of the house, but there is so much traffic there, AND there is always someone who is going 70 or 75 that will cause air turbulence for your aero testing.

I agree that driving faster is better to measure the affect of the aero mod, but can't figure out how to accomplish this without doing it late at night on the Interstate.

Jim.

I had not thought about that issue. Do you think it is a serious error inducing problem? Hard to account for I suppose. If one baselines about the same time of day on the same road section, then the net impact on the percentage improvement, if any, would probably be small - maybe:confused:

If not an interstate, then we have to look for a 4 lane section of a blue highway I guess, but the problem doesn't necessarily go away. There is no way to maintain any stable test speed on a two lane.

The only infallible way is a wind tunnel. Maybe someone can find an aeronautical engineering grad student, with wind tunnel access, who needs a thesis subject;) Maybe we can copy the Wright Bros. low speed tunnel where the reynolds number isn't much different.

Try route 58 from Emporia to Suffolk. A few miles east of Emporia it's very flat for something like 40 miles. I once drove 40 miles west on 58 and never got close to more than 15 cars going in the same direction. 55 MPH divided highway.

The week Princess Diana died.

regards
Mech

Hi Mech,

I'll take a look at that. Drove over your way yesterday while doing some experiments with lean burn. Looked at VA 241 as your suggested. It is a nice section of road, but the level uncongested part isn't long enough to be practical for testing. In my case there is an additional problem. That section is so far away from my garage that the "commiting" distance to any tests would be too great. But I fear I'm going to have that problem with anything I use. I want to check out the section of I95 between Richmond and Petersburg, but of course that is truck "alley." At least it is close.

And that's the reason I like this old country road about 15 miles from the house.

It's on the way to work, rarely traveled, but quite bumpy right now, due to the cold weather.

Once I get ready to try some changes to the car, I could drive over there on a Saturday morning, and run laps back and forth. It would be at 60 mph, and maybe not a nice as traveling on the Interstate but certainly less crowded.

Jim.

Sounds like you may have the perfect road. Unfortunately, it's going to be a bit of a trip for me to use that road;) The problem we have around Richmond is a high level of congestion generally. We are on the main north/south coastal route and certainly a part of the east coast megopolis. I don't expect to find a testable 4 lane that doesn't have some truck traffic. I'll need to study the traffic patterns to see when it is at a minimum.

Question: I did some experimental sanding today on a piece of packing type styrofoam. I used 80 grit and I noticed that I had to get the pressure just right to avoid ripping the little balls away whole. Of course that made the surface very rough. Does the construction grade white and pink sand better?

Got under the car this morning and did a bit of looking at the stock Honda bellypan work. It would appear that there is a little additional work that can be done at the rear, in the area of the axel, that would make the boattail perform better, but Honda did a pretty good job otherwise. I think I'm willing to do the slight bellypan work and any boattail I can work out as a single step in the testing and chalk it all to the bellypan.

You don't need to have the road to yourself to do high-speed coastdown testing. In a 65mph zone, you can coast down from 74mph to 60mph with a small break in traffic. In Buffalo, there are expressways where you could do that 20 hours a day.

You'll want to do runs in both directions, to account for gravity. As for Crr, ideally you'd do low-speed testing on the same pavement, so you may have to test for Crr on side roads and just hope the numbers are similar.

Sure, passing cars, gravity, etc will interfere. But you can deal with noisy data by having a very large number of samples.

I answered over here to keep the question with the tail extension discussion... Hope this is OK.

http://ecomodder.com/forum/showthrea...tml#post164865


Jim.