coastdown hill
 

This thread inspired by:

1 - psyshack's recent Coast down test , and,

2 - CoyoteX's reports of testing his aero mods by seeking terminal velocity on the same hill (Playing with Lawn Edging)

It's got me thinking: there might be a better hill just a little further away that will have a higher terminal velocity. Stands to reason that the higher that terminal velocity, the more likely aero changes will reveal themselves...

Can you tell if the percent increase or decrease jives with the on the road SG test?

I could try. Most of the aero mods I've tested at a constant speed on the test route have yielded similar sized results on FE - in the range of 2-3%.

So they should have similar end points on the coasting test.

It won't be useful for figuring out actual FE effects, only relative differences between stuff. Which is fine with me. I figure if I can uncover a discernable change on the coastdown, I may then go out and measure it at a higher constant speed on the flat test route.

I'm thinking: monster air dam :)

Air dam = :D
My air dam includes a nice forward belly pan, goes from air dam to approximately the front axle line. For the whole package, I definitely see an improvement since completing it (I had a couple threads here re. the air dam around the last weekend in May).

I have to admit I don't have an apples-to-apples comparison test. The weather is warmer and I boosted my tire pressure a couple pounds. But since then I've been consistently between 30-32 mpg while before it was basically between 27-30.

I see a big improvement in the glides. For any given glide, it goes further now. Many times I can get significantly up the next hill while still gliding where I used to need to go back in gear to get up the hill.

I hope to do rear wheel skirts next. I want to call them pants since my car is named Sven.

Pantless Sven, we want data!

I love that people are trying aero stuff, but it's better when they isolate the mods and gather results. Otherwise how can you be sure you're not fooling yourself?

How hard is it to remove the dam? If it's not difficult, a coastdown ABA wouldn't be too hard to do.

----

On another note, this afternoon I went to the other 4 "big" hills that are nearby and the most I saw was 30 mph max on any of them, so they're not quite as good as the one I went to this AM. (Can you tell I'm getting lots of work done today?)

There's one other hill literally minutes from the house that I'm sure would coax upper 30's mph on the descent, but there's a railway crossing with brutal bumps and a traffic light in the run-out zone.

When you did the hatch test, was that with the hatch just open, but down, or was that with the hatch at the top of it's travel?

Like sooo. . . . . ginormity???

https://aycu23.webshots.com/image/161...8164179_rs.jpg

Um, keep your pants on. Unfortunately removing the air dam would be a fair sized PITA.

I may get a chance to retest it without the air dam. A friend is interested in molding the dam in fiberglass which would require temporary removal of the oem air dam plus my extension as a unit, which is pretty easy. If he can loan me a spare OEM one then I can do a comparison. Don't know if I can/will do a coastdown test but if he provdes an OEM one then at least I'll get to drive it a bit.

MetroMPG: I found where your test was done with the hatch full on open. I looked at the link and searched and I couldn't locate any test that was done with the hatch unlatched, but not raised. Was their a test with the hatch just cracked open, like with a broom handle or something?

That IS ginormous. I was thinking something like that, yes. Only in cardboard & duct tape in its testing form :D

That's too bad. Pants remain on. But if you do get a chance to compare it, go for it.

No, that was Peakster's test, not mine.

Today I duplicated his - the hatch was wide open. I don't believe anyone's done a test with the hatch only unlatched & cracked open a little bit.

How about starting at the same spot on the hill and letting it coast down the hill to a certain spot and see what the speed is? Pick a road sign or tree or something before the tracks and see if you can get a good variance coasting down to that spot. That way you don't have to deal with the crossing and stuff. Also maybe try coming at the downhill run at speed to see if you can find a terminal velocity. If it is a long hill that should not be hard as long as you don't have to hit it at something like double the speed limit or whatever :)

The thing I always look for is a very consistent spot that I can go to one day to the next that always gives me the same speed/distance. So anything that would be more involved than removing a few screws to change I can do over a few days. The only thing I have seen that changes my distances on tests from one day to the next is really big temperature changes so if the temp changes by more than maybe 20 degrees F a day then the numbers wont be quite as good. The best way I have found is to pick a spot or a few spots and keep doing the test anytime you happen to be around there. This helps let you know if that spot has good repeatability for doing tests over the course of several days. Once you have a spot you know exactly how your car does consistently on, testing like this becomes way easier.

I don't think I'll be able to easily find terminal velocity on this hill - it's not long enough (and there isn't much of a "runway" before the top to build up speed either).

I wonder how repeatable your data is. Like if you ran three trials with the same settings, how close would your distances be? Hmm.

That's the key, isn't it?

Next time I try it, I'll be collecting data from multiple runs to answer that question.

Answer to that question:

Well, you could do a "control" run every couple of test runs, and then correct for any apparent improvement across runs. Know what I mean? Not ideal, but pretty defensible.

Well, you know what? This "problem" could be an opportunity to do a test.

I played a little with terminal velocity tests on hills and am thinking that I might get a more accurate test to test result by taking the acceleration factor completely out of it... by seeing what terminal velocity is "reduced" to rather than "increased" to. ie. hit the top of the hill faster than your expected terminal velocity and let aero (and RR) forces decelerate you to terminal velocity...

On paper it shouldn't make a difference but in practice you can hit close to terminal velocity at the top of the hill and get the reading for a much longer duration (than hitting the top at say 50 mph and seeing how high you speed gets by the bottom).

I have not tried it yet though... just seems to make the experiment a little more controllable and eliminates all factors other than aero and RR.

Removing variables is always good.

I need to make a chassis dynamometer to pre-warm my drivetrain. :D

I don't like terminal velocity tests because power increases as the cube of speed. So if you make a really good improvement, worth like 3%, then you terminal velocity only changes by 1%, which might be too small to see.

Excellent point, Bill.

yeah, agreed... terminal velocity margins of error are multiplied.

Force is only a squared factor though, and that is really the measure you are looking for.

Plus the RR, which is a large non-aero factor, weighs in.

What I am seeing in my measurements is a RR force about 30 lbs with a aero force at 60 mph of about 50 lbs... actually one should be able to get the margin of error from this:

let's say +- 1mph on speedo reading which is +2, -almost 2 lbs of aero force.

Cd change of +- 0.1 on my VX yields +-2lbs as well.

So a aero change of +- .02 is the best I could hope to measure.

Also forgetting wind variation... and I am not convinced RR is a constant.

Either way this gives an idea of what is possible.

I am playing with coast down testing a lot, using a video camera on a second clock and the speedo. When I review, I mark the timestamps for each 5 mph tick, the calculate the deltas between in 10mph increments. I average this approach of 5 or so runs. I think I can show that the margin of error is less for a couple of reasons:

the deltas are calculated and relative: ie if I get one reading high or low the next calculation delta will absorb that error.

the averages should take wind and road levelness out of the equation (although 10 runs might be better).

Yeah, I pretty much agree. I guess ideally we would do coast-downs from about 80 mph. :-)

Depending on your video software, you might be able to get times from the software/video rather than the clock, which would let you break it down to ~.04 second intervals.

Anyway, it sounds like you are doing great stuff.

Wow! that makes me wonder, how much shorter it would be in the winter? That may be a big factor in bad winter mpg. I have noticed cars don't seem to coast nearly as good in cold weather, could be a major cause of poor mpg in the cold.

ELF: definitely - for multiple reasons: more viscous lubricants; denser air; stiffer tires (more RR), etc.

...and snow.

Great idea...

Also thinking a DMM hooked up the the speed sensor might give precise readings... if you could read 1/10's of a sec and get speed to 10th as well..
hmmm... wish that damn wind wasn;t there.... or the bumps in the road :-)