Tested: what's the resolution of coastdown testing? Can it register an open window?

[MetroMPG]

I was curious to know what kind of "resolution" coastdown testing has: in other words, what magnitude of aerodynamic change can be revealed by repeated, timed coasting between two speeds.

We know that large changes will show up, eg, here's Skylark's coastdown results of his Versa's Coroplast boat tail:





But what about the smaller things people are doing? Would they show up in coastdowns?

To try to answer the question, I chose one open window as the aero modification I would try to detect.

The speeds tested were 100 km/h (~60 mph) to 50 km/h (~30 mph)

Test was AA-BBB-AA-B

The results seem to indicate that, for my vehicle, at those speeds, a single open window is just at the edge of detectability - the average difference between the A & B results is just above the standard deviation in the control (A) set.

Any smaller aero "mod" would certainly disappear into the statistical noise / variation of the control runs.

Higher test speeds would presumably result in greater resolution.

[MetroMPG]

So what does an open window represent, in terms of change in Cd and fuel consumption?

I haven't measured it on my car (and can't because my cruise control is still AWOL), but someone else has tested this with a similar car, so we can maybe draw some conclusions:

See: http://ecomodder.com/forum/showthrea...uction-93.html

A similarly shaped car to mine, a VW Golf, was tested at 65 mph with both windows up vs. down.

"We also found that closing windows at freeway speeds improved fuel efficiency by 2-3%."

Let's average that and call it a 2.5% difference.

Based on the aero calculator, to achieve a 2.5% improvement in fuel economy at 65 mph with my car (roughly similar in shape & Cd as the VW GTI tested), Cd would have to improve by 0.01

Considering that VW data is for BOTH windows up/down, that suggests that the resolution of coastdown testing one open window on my car at 100-50 km/h is about half that, or Cd 0.005

Quality of the data set makes that debatable of course.

[Matt Herring]

Nice work Metro...good info to support the windows up theory. Without any numbers to use I can easily tell my Prius is "dragging" with the windows down...even halfway down.

2.5% of mpg with windows down is a huge hit to FE...as much as 1.5 mpg each tank if I did it in my car.

Great work!

[Daox]

Good info Darin. It seems coastdowns aren't quite as good as I guess I would have hoped, but its still not bad. Thanks for testing!

[SVOboy]

Nice stuff, I want to see it graphed though or I wont believe you actually did a test,

[Coyote X]

The easiest good resolution test I have found is from a terminal velocity test. But the biggest problem with that test is it is hard to find a place to do it. For my best hill it is a mile or two at a 5% grade with not a lot of traffic most of the time and the wind always blows the same direction at about the same speed.

Terminal velocity was good enough to tell a difference in the rear wheel covers and things that made a marginally measurable change to mileage. For the stuff that was lost in the noise I had to do a more precise test.

For the absolute best resolution for my testing I would top the hill at 55mph and push the clutch in at a sign post and start coasting. I would hit terminal velocity and keep coasting down the hill at the bottom of the hill there was a sign that I would note my speed. For example if I was doing 60mph crossing the sign. Add something minor like the rear spoiler and I would do 61-62 at the sign at the bottom of the hill.

I used that most days coming home from work to let me know if a tire was a few psi low or if something wasn't quite right with the car. If I was really careful with the test I could see a difference with the back windows just being popped open or not.

The speed at the bottom coasting test works good because it gives the car a long time for the small aero changes to work on the speed of the car. So terminal velocity might only be 0.3mph faster and impossible to measure at 90mph but the slightly better aero made it coast a bit longer on the flat bottom part of the road and that is much easier to measure. It was also very repeatable and variations in runs were pretty much nothing even through several days/weeks for me.

[MetroMPG]

Yeah, a long, steep hill is definitely better - because the average speed of the run will be higher than a coastdown on level ground. So aero changes will have a bigger impact.

I was pretty excited when I thought I had found a decent hill for testing , but I ended up not using it: the speed really was too low (maxed out in the low 30's), plus there's a railway crossing right at the bottom. I did have to jump on the binders for a train once. Crossing the tracks at 30 mph when there was no train coming was pretty rough too.

[MetroMPG]

The good thing about the road I used for these latest coastdown tests is that it runs through a forest where the trees are right up close to the edge of the road.

The road direction is perpendicular to the prevailing wind, so the road itself is really well sheltered if the wind is blowing in its usual direction - it doesn't mess with the testing.

[Southcross]

I wonder how having a single window open would affect FE in regards to cruising "speed", and "vs. running your ventilation fan at full blast"... especially in a car without AC

I would assume that since "drag" increases with speed, below a certain speed the percentage of cD lost (gained? LOL) would have to be substantially less the slower you go. So If your cruising in city at 40mph, you shouldn't get nearly the hit on FE as if your flying down the highway at 75mph with it down... or at least the logic is telling me so LOL

Also, would it be more of a hit to drive with a window rolled down or to have your ventilation fan running full blast (alternator drag). I would think that the opposite of that would be true... bigger hit at slower speeds, less at higher, with the fan on.

Also, I wonder how far down the window rolled down as compared to the hit to FE. cracking both windows, cracking one window... man I wish my car had pop-out windows LOL

[aerohead]

Darin,I'll be of little help on this one.Sorry! I do believe that it may be asking for trouble attempting to apply "counts" from one development vehicle to another.--------- The least squares method from statistics is supposed to provide a good curve from the coastdown scatterplot,and the slope-intercept at any point along the curve would reveal an accurate time/velocity relationship for the deceleration,from which the forces could be derived.--------- The "scale" of the graph should limit the resolution at the standard deviation.If the RR is well established,the curve would "show" any aero changes if they were larger than the signal-to-noise-ratio.------------- Since an open window "may" show up in a windtunnel measurement for a particular vehicle,it follows, that it would also show up in a coastdown ( no test variables other than body change ) for that particular vehicle.If any environmental factors were to vary during the investigation,the data would be trash.----------- Seems like the Scan-gauge or burrette fuel supply ( as used by students at U.C. Davis ) might be easier for this sort of thing.