Do wheel skirts really work?

[trebuchet03]

Originally Posted by landspeed

Another thing not really considered is the effect of wind speed. I know that, on aeroplanes, the faster you go, the flow seperates earlier from the aerofoil. So, maybe this also happens on cars?. It might be that, at 25mph, my car has good aero from the rear hatch, while at 50mph, the Cd increases (in addition to the actual drag increasing), as the air is flowing too fast so it separates at the roof line.

This happens on anything with flow moving around it - the transition point can be found with the Reynolds number equation set to the transition Reynolds number - then solve for the characteristic length. Everything after that point can be considered a turbulent zone and the boundary layer starts getting bigger

No so easy on complex shapes like cars -- so experimentally, you can use tuft testing or smoke testing

[silverinsight]

I'm considering closing up the front wheel arches on my Insight, something like the ice in the first picture.

Attachment 328

After looking at the other two pictures, it looks like my wheel skirts add drag to the car

Attachment 329
Attachment 330

[MetroMPG]

I like Houston Bill's idea of using a video camera to record an analog gauge for coastdown testing. It would permit you to remove most of the human variable of coordinating the stopwatch with a moving target (needle). Tougher is ensuring you're always starting the coastdown from the same speed. This could be accomplished with cruise control & pressing "cancel" at a predetermined point though.

Originally Posted by theclencher

Still, if the effect is that minute, doesn't that say something right there?

Yep: to me it says that aside from a fully tapered boat tail, there's no magic bullet in aerodynamics. Successful aero is the sum effect of a lot of small, but effective, individual elements.

[MetroMPG]

Originally Posted by silverinsight

After looking at the other two pictures, it looks like my wheel skirts add drag to the car

Can we conclude that from water/debris flow patterns?

When I look at the bottom pic, I think we're seeing the effect of water getting channeled upwards and backwards in the skirt's forward seam, and reaching some point where some factor (volume of water?) causes it to spill out of the channel and continue being swept rearwards.

[omgwtfbyobbq]

Originally Posted by theclencher

Intuitively skirts make a lot of sense for lowering Cd. Reality seems a bit different though. I did a skirt/grille block/mirror fold/wiper remove college paper on aero and failed to find A-B differences in coast down tests. Then I did practically the same thing again 23 years later on the Lambo and still have realized no results- although with the coming of warm weather and comparable conditions I'm still hoping to find something.

Also I said "dubious" because some of the most aerodynamic cars in the world have no skirts!!! Why, when they are shooting for ultimate aero and ultimate FE bragging rights, would they omit something so important? I think the answer is, skirts really aren't that important:

They aren't that important compared to under/cooling/rear end flow, but it's still something imo. The trade off of a count or two of drag just isn't worth alienating buyers in the eyes of anyone who designs cars. As for being able to tell from coast down tests, from my back of the envelope *guestimations, a ~mph change in relative speed at ~50mph will result in about the same impact on a coast down test than a reduction of drag from .3 to .29 will. Which implies to me that small items like mirrors, windows up, wheel skirts, etc... will be extremely hard if not impossible to test using a coast down. There's going to some speedometer error, timing error, a likely difference in wind speed. The coast down test would need to be from such a high speed, that it would be needlessly dangerous, if not impossible.

*Going from ~49mph to ~47mph at standard everything is analogous to going from ~22m/s to ~21m/s. In a car with A=~2m^2, and Cd=.3, a drop of 1m/s results in 378N compared to 415N. A reduction in drag of a count, so Cd=.29, results in ~403N of force, which means that a 2mph speed error results in 2.5 the difference that a 1 count drop in drag does. Or that a 1mph speed error at ~50mph will have the same impact of wheels skirts, or a mirror delete, or rolled up windows. For the DIY'er, it's pretty much impossible to work around that level of noise.

[CoyoteX]

The only consistent coast down test I could ever get to work right was to coast down a hill where the car reached terminal velocity. When I experimented with the aero mods I did to my car they all increased my terminal velocity and I could repeat it as many times as I wanted going back and forth with the mods. I could start at the top of the hill within a minimum speed and at the same point on the road I was always at the same speed. I could start at the top of the hill faster than the terminal velocity and the car would still slow down to that speed.

The wheel skirts picked up about 1-2mph on my terminal velocity. Around the same amount as the passenger mirror if I remember it right. So they don't make a huge difference but they do help some. Swapping front bumpers made a much bigger difference, almost 5mph. Rear swift wing on the hatch gave me just a tiny improvement. The needle would sit on the other side of of the mark so maybe .5mph

With all my aero mods I had a hard time reaching terminal velocity at the same point, the car was still accelerating and I was not able to get more speed at the top really due to having to climb the other side of the hill. I need to add nitrous or something to get more speed at the top of the hill

The other big hill I had to go over I could coast down and would always hit 55mph at the same spot on the road. I could change something and the car would vary from that spot where it hit 55mph and I could tell if it was better or worse. But that method was nowhere near as consistent as the first hill where I had a long downhill run.

[omgwtfbyobbq]

Yeeeesssssss. Lemme see. When I'm coasting down, I notice a significant difference depending on passengers, temp, etc... So it seems doable. We'll have something (All in metric or we crash into Mars) like Weight(Crr)+.5(ro)(Speed^2)(CdA)-Weight(gravity)sin(theta)=Force, where theta is the angle corresponding to the grade. A 10.5% grade is a 6 degree angle, so that means the potential energy component of a 14,500N car is about -15,000N. Assuming Crr=.015, the rolling friction coefficient is ~220N, so, for a car with CdA=.7m^2, the fluid friction coefficient must equal ~12,800N. Plug'n'chug, and we get the Speed is ~122m/s. Way fast. But not surprising considering the grade. Since we're sane, we leave it in gear, and end up going much slower since the spinning engine/trans provides drag. Lets say in gear we end up going 38m/s (85mph). This means we have ~1,240N from air drag slowing us down, and the other ~11,560N comes from the engine/trans spinning at whatever rpm. Lets say we drop the CdA from .7, to .68. Now we *still expend (not exactly) ~1,240N for fluid friction, but our speed increases in order to do this. With the .68 CdA we're going ~38.6m/s, which is a ~1.3mph increase in terminal velocity. Granted, there's still the tires slipping a bit, and the *engine/trans drag probably isn't linear, but it's still a nice result imo. I bet a much nicer grade (~2-3%?) with the car in N would yield similar behavior.

So, my BS seems to match up with Coyote X's experience. As long as we find a smooth enough, or steep enough highway hill, it looks like those of us w/o scanguages can figure out what improves CdA the most. Lothar approves!

I should add, that imle it's way easier to notice what the top speed on a decent is because highways tend to have pretty even grades, and the top speed will be held for at least a few seconds. A coast down test otoh, requires the driver to start coasting and a specific point and specific speed. Then accurately give the speed the instant they get to the other location. Or, make note of the location where they see the speed. Either way, there's more room for human error imo. Whereas with terminal velocity, we don't do anything except keep on eye on the speedo for a top speed, which should be present for at least a few seconds.

[CoyoteX]

10% is a bit steep. The hill I was using I hit ~75mph in N with the windows up. On a side note I am now driving a state car to work, 03 or so Taurus and it is more than happy to go past 90 on the same hill

[Bill in Houston]

Originally Posted by omgwtfbyobbq

*Going from ~49mph to ~47mph at standard everything is analogous to going from ~22m/s to ~21m/s. In a car with A=~2m^2, and Cd=.3, a drop of 1m/s results in 378N compared to 415N. A reduction in drag of a count, so Cd=.29, results in ~403N of force, which means that a 2mph speed error results in 2.5 the difference that a 1 count drop in drag does. Or that a 1mph speed error at ~50mph will have the same impact of wheels skirts, or a mirror delete, or rolled up windows. For the DIY'er, it's pretty much impossible to work around that level of noise.

I think that we are talking about two seperate flavors of test:
1- Coast-down: The car is accelerated up to, say, 62 mph. Car is placed in neutral. Time intervals are recorded as the car goes through 60 mph, 55 mph, 50 mph, 45 mph. The times are put into a calculator and a combined Cd and drivetrain/rolling drag are spit out.
2- Terminal velocity: The car is driven down a hill with a consistent grade, and the terminal velocity is measured with the car in neutral. Since the effect is small, repeated trials and strong downhill acceleration (before putting the car in neutral) may be required.

I have not done the math on the coast-down test to see if it has the possibility for better resolution, but I know that one problem with the terminal velocity test is that a 3% reduction in Cd will only change the Vt by 1%, which could be pretty hard to measure, as byobbq was saying. Also, it's just plain hard to know what the Vt is, since you can be within an mph or two of it, and feel like you are topped out. Best to approach from above and below to try to converge on it.

As you guys know, both of these tests are going to be sensitive to temperature and crosswinds, as well as changes in rolling resistance.

Lots of fun.

[Bill in Houston]

Originally Posted by MetroMPG

Tougher is ensuring you're always starting the coastdown from the same speed.

I think that you would start from above, and time as you coast down through the speed range. I think...