How long should a driving loop be for A-B-A testing?

[Zyzzyx]

Now that I've got a ScanGauge I want to start doing some proper testing for various upcoming aero tweaks.

I get there's a few basic requirements for a stretch of road for testing:

1) Flat
2) Low to no traffic
3) Be able to get up to reasonable speed (50mph?)

But what about number 4? How long should the test section be? Is 1/2 mile long enough? Is 5 miles too long to deal with?


I'm scoping out roads in the area. Been a bit tricky to find some flat stuff of a decent length and have places to turn around at each end. I've got a section not too far from home that's 2.5 miles long and 50mph, but has +35/-13ft elevation gain/loss. Also, it make for a long testing session (did some ABA out there today, took a loooong time).

Got a short section even closer, less than 1 mile long, but its a busier two-lane, and only 40mph. Not so good.


Looking at a section a bit out of town, very deserted two lane. Looks like I could find a 1-2 mile section with less than 30ft elevation change and shoulder room to turn around wherever.



So... how long should I make the test section? I'm leaning towards 1 mile, mostly to make 9 full passes of testing achievable in reasonable time.

[Piwoslaw]

When I did coastdown and cruise control testing, my test track was 2km (1.25 mi), with an extra 1km at each end to turn around and adjust speed. The 2km was enough to coast from 100km/h (62.5 mph) down to about 40km/h (25 mph).

I found out that my test track had a slight slope. Very small, not enough to see, but enough to give a 20% difference between E and W runs. I wonder how much this offset my data, since on a slope the weight plays a larger role.

I know how hard it can be to find a good stretch of road, but if you have 2-3 roads to choose from, pick the one that is closest to straight. On each bend you lose some energy.

[RobertSmalls]

For coastdown testing, you only need enough road to get up to speed, then coast down. Steady state cruise mpg metering (driving in a loop while watching the scangauge) isn't as accurate, because it includes variations in engine and drivetrain efficiency, and inaccuracies in measuring fuel consumption. You can use it to make an mph vs mpg graph, but I wouldn't use it to compare mods.

I've been looking for a good piece of road for coastdown testing. There's a flat, residential side street nearby, but it's only a half mile long, and the locals might be unnerved by a strange car being driven very slowly, back and forth, and back and forth. Poor neighborhoods have poorly paved roads, rich neighborhoods have nervous retirees and housewives.

I could drive over to the university in the evening and pretty much have the place to myself, and people would assume I'm just an engineering student working on a project.

One interesting road near me is actually a three-lane urban expressway. That could be useful for coasting down from 60 mph to 50mph, and turning around for another run is easy. I'd rather have a country road, but there aren't any within fifteen miles.

[usergone]

If you are testing aero mods, the important data is how much fuel you use to maintain speed. Aero doesn't make acceleration more or less efficient. What you are interested in, IMHO, is the FE once you get up to cruising speed.

So, accelerate to a speed where aero will make a difference (preferably with cruise control), reset your trip average, drive for some distance, and observe your numbers. When you switch from A-B and back to A, be sure to have the same speed and use the same lane, start point, and stop point.

That way your acceleration won't skew the data too much (hopefully at all).

Am I mistaken somewhere?

EDIT: I just realized I dug up a sorta old one