Done: A-B-A test on rear wheel skirts - Corolla, Camry Hybrid
 

The rear wheel skirts "test" I ran on my Firefly (Metro) was kind of sloppy: first of all, it wasn't an A-B-A. The skirts were added on top of another significant variable (grille block), so they weren't tested independently. (That made it actually an A-B-BC test.)

Out of curiousity I'm going to re-run the test. I'll be using a different car this time as well: 2007 Toyota Corolla automatic.

Click to zoom...

http://ecomodder.com/forum/attachmen...1&d=1352388336

http://ecomodder.com/forum/attachmen...1&d=1352388287

http://ecomodder.com/forum/attachmen...1&d=1352388287
http://images6.theimagehosting.com/corolla-side.th.jpg

09-06-2007, 11:08 PM

Pretty ideal conditions. Test approach was A-B-C-A, where...

A = no mods in place;
B = rear wheel skirts
C = rear wheel skirts plus "fairing" behind the rear OEM mud flap where the lower aft edge of the skirt was attached

I'll post details and pics tomorrow, but here's the nut shell:

- 3 bi-directional runs (6 one-way runs in each direction) for each of A,B,C,A
- Speed was 85 km/h, cruise control set once, cancelled by brake between runs, (flat, deserted road, little to no wind, no traffic in my lane; infrequent oncoming cars)

A average = 50.24 mpg (US)
B average = 50.53 mpg (US)
C average = 50.62 mpg (US)

This was on a 2007 Corolla automatic. I probably should have run this at a higher speed, like 95 km/h, to eke out more significant results.

So much for not being sloppy:

I should have tested at 95 km/h. I just checked, and that's the speed I used when I did the A-B-BC runs with the Flea and calculated a 2.7% improvement for its rear skirts.

Just based on these averages, the Corolla saw a 0.6% MPG improvement with the skirts alone, and 0.8% improvement with skirts + fairing.

Details:

Temp (°C): 22
Humidity (%): 89
Wind (km/h): SW 4
Pressure (kPa): 101.7

Raw SG readings:

http://ecomodder.com/forum/attachmen...1&d=1196223321

A note on my speed choice for this test: I tend to want to test around 85 km/h because that's a "normal" speed for me. But I should probably test at higher speeds for aero mods because differences, if any, would stand out more clearly from the normal variability of readings. Plus, I'd like people who drive faster to see the potential gains to be had from aero mods instead of dismissing the smaller differences detected at lower speeds.

Also, it's likely that the effect isn't consistent for different vehicles - some will benefit measurably, others may not. Which just makes me want to re-test the Flea's skirts to see if it happens to be one of those vehicles that benefits more than others, or if my original "data" was off.

Better late than never, the Camry results:

http://ecomodder.com/forum/attachmen...1&d=1352387703



Conditions were pretty ideal:

Temp: 25 C / Wind: 2 km/h SSE

Test approach was A-B-A, where...

A = no mods in place;
B = rear wheel skirts plus "fairing" behind OEM mud flap where the lower aft edge of the skirt was attached

http://images6.theimagehosting.com/c...skirts2.th.gif
(note: pic taken the next day after tape had been applied/removed several times already - it was a smoother installation for the test runs.)

- 3 bi-directional runs (6 one-way runs in each direction) for each of A,B,A
- Speed was 95 km/h, cruise control set once, cancelled by brake between runs, (flat, deserted road, little to no wind, no traffic in my lane; infrequent oncoming cars)

A average = 45.9 mpg (US)
B average = 46.3 mpg (US) - 0.7% increase over A

This was on a 2007 Camry hybrid.

But here's the problem: while the average of the six bi-directional A runs is less than the B runs, there was a big difference between the average of the first and second set of A runs:

http://ecomodder.com/forum/attachmen...1&d=1196223598

In fact the average of only the 2nd set of A runs is bigger than the average of the B runs.

If you look at the Corolla test, you'll see something similar: the average of the 2nd set of A runs is higher than the first set (though not higher than the B runs).

All of which suggests to me that:

1) in both cases, the drivetrain was probably still warming up (making the car more efficient) while I was doing the runs. This skewed the results.

I had driven each car for about 25 minutes & 25 km before starting the runs.

2) if #1 is true, then if I had done a 2nd set of B runs after the 2nd set of A runs, the difference between the A & B averages likely would have been greater still.

Unfortunately, the only way to know this for sure would have been to do another set of B runs, and I didn't.

---

In conclusion, my testing bites.

Also, testing for small differences is difficult & time consuming.

And I need to drive much further to ensure a car is fully warmed up next time I try a test.

The standard deviation was pretty darn small for 6 out of the 7 individual configurations of both the Corolla & Camry runs (ie. the A1 run, the B run, the A2 run, etc.):

Corolla: 0.13, 0.75 (the oddball), 0.03, 0.25 mpg
Camry: 0.11, 0.15 & 0.28 mpg

On average: 0.24 mpg for all seven samples. I would say that's pretty good testing resolution!

Yes, more runs per sample would be better. But I still think the biggest problem was that the drivetrain temps hadn't stabilized.

problem with the testing is that I was only recording raw data, so I didn't realize there were differences in the bi-directional averages of the A1 sets vs. the A2 sets until I got home.

I would have needed to figure averages & standard deviation while testing in order to make a call on sample size on the fly.

...Which I suppose are reasonable things to do in the future if I want to have confidence in the testing.

My experience with my car, and as always YMMV, is that the coolant will come up to temperature very quickly, but the oil takes forever to warm up. I have to be driving on the highway for almost 40 minutes, about half of that at 100km /hr on the highway before the oil is most of the way up to temperature (170 F). Even then, the oil is still about 20 degrees below the temperature it'll level out at (195 F). Just the viscosity difference would be significant (but small). I also find with my car that the car runs a hair rich till it's most of the way up to temperature.

I'll be the last person to argue that trying to test outside of using a dyno in a lab isn't fraught with potential problems. It's not easy - it takes a lot of time to try to control as much as possible.