Documenting NHW11 Prius rear window flow separation; testing AirTab vortex generators
 

What speed are the photos taken at? 50km/hr like in the article? Are you going to use some AirTabs and see if there is an improvement? Interesting article... Perhaps worth trying something under the front end to see if there is improved stability with no hit to FE, or maybe a slight boost.

Has anybody done any testing with these on a notchback to see the effect?

Bob's Prius is the earlier model that is a notchback.

The next thing I'd be wondering is: does the flow reattach on the end of the deck lid?

Did all the 1st gen Priuses come with rear spoilers? A quick Google image search suggests they did.

No, the FIRST (NHW10) Prius did not have a rear spoiler.

Over in "Prius Technical Stuff" there was a posting that removal of the rear spoiler from the NHW11 (2001-03) model reduces mileage. Furthermore, there is a slot that runs across the bottom with what feels like a 1 cm, rectangular shape. I suspect this slot generates a linear vortex to improve the efficiency of the rear 'wing.'

It is a good question: does the flow reattach? I need to tuff the top of the trunk to see what is happening.

Right now, it looks like a dozen of the Airtabs would keep the flow attached to the rear window. The question is whether or not it might cause something else to 'squish out.' Still it might allow removal of that rear spoiler without an increase in drag.

Most rear spoilers are there to increase down-force for road racing. Somehow, I don't think there has been much discussion about drag reduction.

Bob Wilson

My opinion, based on as-controlled-as-possible A-B-A testing (straight & level road, absent any other traffic, cruise controlled, back to back bi-directional runs) is you'll not likely see any drag reduction using AirTabs. Seen this? http://ecomodder.com/forum/showthrea...olla-2390.html

It's fairly well known that some rear spoilers reduce drag (and rear lift).

They are typically added to most sedan/notchback hybrids (and other high efficiency versions of that body style), where the "regular" edition may have none.

Honda Canada added a small deck lid spoiler to its non-hybrid sedan/coupe as one of a series of measures meant to marginally reduce the car's fuel consumption to make it eligible for federal incentives.

Car and Driver measured drag reduction on both its "Crisis Fighter" ecomodding project cars years ago (See the Pinto project here). Their Datsun 240 project also benefitted (through reduced fuel consumption) from a spoiler addition.

Thanks! A few questions:

1. How much warm-up time and speed before the first "A" test? (Prius warm-up is ~30 min., thermal changes takes about 5-10 minutes.)
2. How much warm-up time before the first "B" test?
3. How much warm-up time before the second "A" test?
4. How many minutes per run?
5. Ambient temperature that day?
6. Start and stop times?

My hypothesis is driving to the test area may have been long enough to warm-up the vehicle. Entering the first "A" protocol would have 'good data.' Then adding the vortex generators would allow the vehicle to 'cool off' somewhat. Then taking off the vortex generators for the second "A" appears to be much closer to the "B". I think I'm seeing the pattern in a plot of the data.

What has peaked my interest in the protocol is the second "A", especially the "West" runs really look different. The "East" runs are more consistent ... given the limited number of runs.

Are the vortex generators still available?

I am thinking about replicating the test using 104 kph (65 mph) that corresponds to the last plateau before MPG drops off.

Thanks,
Bob Wilson

The day of the test, the max temp was 18C - I didn't record the actual early evening temp when I ran the test. It takes me 20-25 minutes to get to the test route (combination of sub/urban and rural driving). The car may also have already been warm prior to setting out (didn't make a note of that).

The airtabs were attached with tape in approximately 2 minutes. The engine was left on during this.

It would have taken less than a minute to remove the tabs, engine left running again.

1.6 km at 90 km/h for the run itself (~64 sec), plus time getting up to speed, plus time coasting to a stop, jotting down the run data & turning the car around. 2-3 minutes per run is probably a good guess.

Yes. I can send them to you if you like. PM me your mailing address.

Some additional data points, I used a spirit level with a scale and found the following:

• 11 degrees - roof at junction with rear window mount/liner
• 21 degrees - top glass near glass mount/liner
• 28 degrees - middle of rear glass
• 31 degrees - bottom of glass near junction with mount/liner

When I did my tuff testing, I only had a middle and lower region tuffted. I have to admit some surprise at the steepness. In fact, now I'm interested in seeing if I can measure the depth of this turbulent area.

I'll copy this data under my original tuff testing and continue the experiment. I have some loaner vortex generators coming and we'll get a chance to see the effect.

I also have a Kovatch "filter separator" pressure gauge that unfortunately has a scale that runs 0-30 psi when I need 0-30" of water. I'll have to modify the spring and calibrate it.

Bob Wilson

One 'wild idea' is to get some small diameter, semi-rigid, plastic tubing and form it to build a boundary layer, suction manifold. The idea is to to suck the boundary layer down and possibly using this approach to keep the boundary layer attached. The challenge is how to build it.

At the hardware store, I can find:

• small diameter copper tubing
• polyethelyne plastic tubing
• PVC tubing - generally no larger than 1/2 inch
• small diameter aluminum tubing

The copper tubing is a little expensive but remains malleable. My concern is that over time, a small gap might close as stress and strain over the 5-6 ft. causes it to flex. To work, the gap needs to remain.

Polyethelyne plastic tubing, the hard, somewhat opaque stuff, is also something that I feel is too flexible. This isn't the vinyl, flex stuff, but still, it is less likely to 'take a hit' and stay closed as copper would.

PVC tubing, as small as possible, looks to be a best answer. Filled with sand and heated, it should 'mold' to fit the rear window curve. Once cooled, it should quickly hold its shape and conform to the upper glass junction. Then using a tapered, balsa fillet and duct tape, flair it right into the glass. A Drimel tool can cut a thin, length-wise gap. The ends can use standard PVC fittings to attach to a vacuum pump. That leaves just the problem of finding a vacuum pump, an air horn pump comes to mind.

I think aluminum tubing might also be used as an alternative to PVC tubing. It should not be as malleable as copper and possibly closer to PVC. Now one 'wild idea' would be to setup a pair of rollers that form an angle and try to reshape the aluminum tube into more of an aerodynamic shape. The advantage is a larger cross-section for the manifold space. Done properly, the ends should still remain circular for standard tubing.

Both vortex generators and vacuum pumps take energy. The vortex generator induces circulation and the vacuum pump needs power. In theory, these energy losses are much less than the improved pressure thrust. But this may be difficult to quantify ... something I'll be thinking about.

Bob Wilson

Hi Bob Wilson,

My guestimate is that the AirTabs are just too big for the boundary layer thickness on cars. They were designed for 18 wheel truck trailers - which have very stagnant flows over the rear third of the trailer. Driving next to 18 wheelers is a bit of a buffeting experience.

Another poster on here found that just a .030 inch thick Dymo tape, cut into sailplane style turbulators was effective in causing the airflow sideways off a windshield to settle down and flow down the side of the car, when placed on the upper third of the A pilar. The idea is that the turbulator converts the flow momentum into turbulence - which is good for causing the sideways flow to turn and flow with the prevailing flow on the side of the car.

In sailplanes, turbulators are used to improve where the seperation point occurs on the wings (further back is better, of course). This seems to be a similar problem to the flow seperation on a car. Except the boundary layer on a car is much greater than on a thin sail plane wing.

The flow off the top of the car has momentum, and cannot bend downwards because of the momentum of the air quickly as the flow needs to conform to the shape of your car. The boundary layer on top of the car is probably about 1/4 to 1/2 inch. So, try some turbulator tapes that are 1/16 to 1/8 inch thick. This might result in reducing the momentum of the flow enough that it will follow the shape of the car. I have fabricated thicker turbulator tapes out of FOAMIES foamed plastic from crafts stores. Some is avilable with a self-adhesive backing. Its available in a wide variety of thicknesses - ideal for experimenting. The zig zag pattern can be cut with Pinking sheers.

I appreciate the suggestion and will try to find some of the references you mentioned. Photos or specific URLs would help as I'm not familiar with these terms or posts.

For now, I want to replicate Julian Edgar's early experiments with an NHW10 and I believe he used the Aerotabs. One concern I have is the part must be robust enough to survive the occasional, small hail storms we have and I have some loaner Aerotabs 'in the mail'. The advantage of a standard part is less 'do-it-yourself' folks can easily follow the same steps. But I have no problem with trying different approaches including a boundary suction manifold and pump.

Thanks,
Bob Wilson

EXECUTIVE SUMMARY

Vortex generators, aerotabs, located on the junction of the roof and rear window have no measurable effect on drag reduction for an NHW11 Prius. Although there is evidence of flow separation without the vortex generators, using these vortex generators show no measurable improvement in aerodynamic dominated, fuel consumption.

METHODOLOGY

The test course is I-565, a six to four lane, divided highway. The end points are given by:

East - 34d 42' 53.47" N, 86d 38' 37.45" W, 196m
West - 34d 38' 27.47" N, 86d 50' 38.37" W, 178m

http://hiwaay.net/~bzwilson/prius/pri_tab_030.jpg

During the test runs, May 9, 7:00-9:00 AM, the wind was from ~290 degrees, a right quartering headwind, increasing from 2-9 mph (3-13 kph.) This was evident in the data as the west-to-east runs had significantly improved MPG in spite of an 18 m increase in altitude. This suggests aerodynamic effects predominated in the data. The temperature ran 73-75F (23-24 C.)
http://hiwaay.net/~bzwilson/prius/pri_tab_020.jpg

The mileage came from the built-in display. The vehicle speed was set to 75 mph (120 kph) with a little under a mile, close to a km, run-up. The cruise control regulated at 75 mph (120 kph) and calibrated by using a Gramin nuvi GPS receiver for the true ground speed (indicated ~73 mph, oversized tires.) As soon as the overpass marking the start point was reached, the MPG was reset and the trip meter. Passing under the ending overpass, the mileage and trip distance were memorized and recorded.

DATA (comma delimited)

miles, MPG(E->W), MPG(W->E), vortex(0=no), comments
12.1, 38.5, 42.5, 1,
24.2, 37.1, 43.0, 0,
36.3, 35.1, 44.1, 1, 1st run required max pedal to avoid traffic
48.4, 36.5, 45.4, 1,
http://hiwaay.net/~bzwilson/prius/pri_tab_010.jpg

ANALYSIS

With the exception of the 3d run headed west when traffic forced maximum acceleration, all other runs are consistent with an increasing wind from the west. There is no evidence of a measurable improvement in MPG using the vortex generators.

I suspect the small, reverse flow behind the rear window probably serves the function of pulling the laminar flow air passing over the roof down. Mythbusters demonstrated this effect with pickup truck tailgate testing.

Mythbusters demonstrated that a stable vortex in the bed of the pickup with the tailgate up had a measurable improvement in mileage as it pulled the laminar flow down. Opening or taking the tailgate off eliminated the stable vortex and resulted in worse mileage.

LESSONS LEARNED

A better time to test would have been between 4:00 AM and 6:00 AM to minimize or eliminate the surface winds.

Testing at the posted speed limit plus 5 mph increased the aerodynamic drag, the primary force of interest, but it also resulted in over taking two slower vehicles running side-by-side and impending loss of lane. The maximum acceleration resulted loss of one run. Select better times than Saturday morning for testing.

Future aerodynamic enhancements need to focus on reducing velocity changes in larger masses of air. Likely areas are:

• bumper air inlet - has excess capacity to handle worst case, maximum power, slow speed climb, a rare condition. Significant mileage improvements are already known in cold weather so a temperature driven, variable air cover may be a better approach.
• tire air blocks - there are small, 2" deep, air blocks in front of each tire. A deeper or curved air block may reduce tire drag effects.
• underbody air cover - a proven technique, this should also reduce road noise although it may complicate maintenance.
• deeper front air dam and side skirts - a proven race track technique, a variable system makes more sense given the number of speed bumps and rough urban streets.

My thanks to "MetroMPG" who loaned me the aerotabs for this testing.

Bob Wilson

Interesting on the vortex generators. More confirmation of what darin found.

:thumbup:

Part of science and engineering is achieving reproducible results and reporting all tests. In this case, we may save folks from spending time trying to reduce rear window drag.

Vortex generators may be useful in other situations or more carefully designed for the specific area. But right now, rear roof mounted, vortex generators have no beneficial effect on our vehicles. Still, some good came out.

I have trustable methodology for aerodynamic effect measurements. I also have an as yet, undocumented bumper air block (a water noodle.) I'm thinking about repeating this test using the bumper block and see if I can quantify the effect. To the best of my knowledge, no one has done this, yet.

Bob Wilson

Sounds like a good idea :thumbup: Looking forward to it.

Thanks for posting your detailed results, Bob.

This is a comment I keep reading, yet the company selling AirTabs markets their product directly to owners of small vehicles as well as large trucks.

As for the thickness of the boundary layer at the trailing edge of the roof, Mitsubishi measured this in their VG development and found it to be approximately 30 mm, which is close to the height of the VGs they used. And which is the same height of the AirTabs style VGs.

It has been speculated that the style of AirTabs vs. the delta-wing shape used by Mitsubishi could produce different results; the delta-wing presents a much smaller projected area than half an AirTab (which produce two vortices, one on each side).

But again the crucial point is, even though optimized for style/placement, Mitsubishi only saw a .006 reduction in Cd in their model with VG's, which is essentially impossible to detect even in as-controlled-as-possible testing that I tried and Bob also undertook.

Note: this thread sparked a discussion of flow separation on the 2010 Prius. See thread here: http://ecomodder.com/forum/showthrea...ues-15082.html