Accounting for variable grade
 

Did some preliminary scouting of my coast down test strip and it's just long enough to be useful and is very straight. However, I also put a bullseye level on my dash and it varied a fair bit along the stretch. I didn't have a way to accurately measure it with me, but it was enough to noticeably effect coasting time in the different directions. Compared to all the other roads anywhere near here it's quite flat, just not flat enough. Since I don't want to drive hours every time I want to retest my vehicle, I'm trying to determine a way to account for its variations. If it were a constant grade, it'd be easy, but it's not. Since my current plan is to calibrate my Kiwi MPG's digital speedometer and then record that in a movie on my digital camera, I figure all I need to do is also record an inclinometer at the same time. I'd only need to include enough data from each stretch to account for the variance, so fewer points in constant grade sections, more during the variable parts. The math is straightforward, but the practical concerns with getting the data accurately is concerning. I'd need an inclinometer that is responsive, but doesn't jitter all over the place and is cheap.

Google Earth gives elevations. You can drag the pointer along a stretch of road and see how many feet the elevation changes along the way. You could also use it to help you find a flatter stretch of road.

Bill

I can get distance easily, but the elevation change doesn't show in the measurement tool's results. Some trick I'm missing perhaps?

As the stretch is all inside the same contour ring as it is, using Google Maps to search for a more level area seems problematic, although it could definitely rule places out. The mountains here make both flat and straight tough. Lots of curving in one dimension or another. Seems worth systematically scanning near all the places I drive on an semi-occasional basis (parts of NH and some of southern VT). Haven't searched in Quebec much yet. Since I'm right on the border I should give it a look. I don't want to go out more than about two hours, which is why I was considering contingency plans.

As long as you are using the same stretch of road it shouldn't matter if it isn't flat here and there on the road. A mod that allows you to coast more will allow you to coast more no matter what little road conditions exist. Pick the direction that lets you coast the longest right now and use that one.

If I change out my sloppy wheel bearings for brand new ones packed with new graphite-based bearing grease then it should allow me to coast further regardless of whether or not I'm coasting on a gravel road or a flat, perfect paved road. Granted, the gravel road will introduce new losses that make the mods less measurable, like testing aero mods up a hill, but if it made any notable difference at all you will be able to measure it.

Honestly, if a little bit of road inconsistency makes the benefit of the mod either non-existent or immeasurable it isn't worth doing. Very rarely do I find a perfect road so I can't see myself doing a mod that will only work on those roads.

You make a good point, but my first run is to try and zero in on why the car coasts worse than I expected. Since it's a somewhat subtle effect (well, not glaring) and I therefore am looking to get accuracy in absolute terms. If relative were all I cared about I wouldn't be so concerned. That's why I've been running the math to see just what effect errors in measurement have and also why I want to preserve v^2 effects, rather than lumping them in with general rolling resistance calcs (often approximated as only losses proportional to v). Fortunately I know that the Accent's Cd is supposed to be .31, so that lets me gauge to some extent how my testing procedure is working.

A guy at the auto parts place suggested I talk to the local airport, since they've allowed bikes to do runs up there before. He wasn't sure how long their runways are though. Small private field, but I've never been there. I should measure them in Google Maps.

I checked out an inclinometer at the hardware store and it seems plenty sensitive to change but possibly too jittery. Guess it depends how smooth the road is. Have to give it a go. It was certainly cheap enough at least. I could buy a digital one of course. Thinking about it, a digital one would be easy enough to just make: strain sensor and a weighted arm. Be a great excuse to play with strain sensors! :) So not worth it.

The runway's about 3800 feet, but that doesn't give me any place to get up to speed. My test spot's close to the same length, but I can speed up before the reasonably gradual turn.

Digital inclinometers are all of $25.

I'd suggest a digital accelerometer IC. They're plenty small and pretty cheap, you can find some on eBay, and I believe Sparkfun has them.

I just got a phone with one built in... gunna have to try some things...

That is a fine suggestion. I'll check it out. May be able to use it to do the coast down speed logging as well.

If I play my rationalization cards right, maybe I can justify the purchase of a netbook to log the data. One capable of installing osx86, naturally. :p

Wonder how accurate the accelerometers in the Wii controller are? I already have the libraries to read from it. Been looking for an excuse.

I realized I can just buy a cheap power inverter to run my computer for data logging. So much for a netbook, but maybe I can at least justify a smaller case.

Did a bit of digging and it'd be a cumulative error nightmare. Unless I can find a good excuse to have a general accelerometer lying around I should go for the digital inclinometer, since it'd require no effort. Good idea, though.