Tail Options: Measuring Real-World Downforce

[Jack Olsen]

With no access to a wind tunnel, and no way I can imagine to measure the actual lift/downforce my 911 is experiencing at speed, it occurred to me that there is still a way I could test different tails and/or wing angles. Regardless of how much lift or downforce the overall shape of the car is generating, the rear decklid/tail assembly's contribution has to be transferred through the deckld to the wheels in a more or less straight path, right? So why couldn't a measuring device be inserted where the decklid latch normally goes in order to measure how much force is being transferred?

In other words, couldn't I temproarily remove the latch mechanism, and put a scale in its place? Whether it's a small electronic scale or even some sort of ratcheting spring device (that would lock, stopped, at the point of greatest compression), it would allow me to measure what the tail is doing once I take the car up to a specific speed on a smooth stretch of road, right?

I'd think this would allow me to see the real-world differences between my ducktail and my 3.8RS tail with wing (and what different wing angles are priducing) without having to rely on my subjective seat-of-the-pants response.

The trick to come up with a cheap, simple device to do the measurements.

Does anyone have any ideas? I'm guessing it would need the ability to measure 0-300 or so pounds. A bathroom scale seems a clumsy solution, but it could maybe be sandwiched between a hinged frame that would compress it like a pair of plyers...

[RickM]

Perhaps an option is to repurpose a barn door or hotwire intake mechanism.

[widebody911]

A strain guage in place of the latching mechs? You'd need to either wire a recorder, or a guage on the dash. The dash guage sounds like a cool idea; you could see real-world changes with respect to winds, other cars, atmospheric conditions, etc.

[ghnat]

Isn't there some metal that changes resistance based on the force applied to it? A simple wire and a calibrated gauge might work...

[djmcmath]

Hmmm ... watching the data real-time seems important. If you're just catching the maximum force, you're liable to get bad data from a gust of wind, or a small bounce in the road, or something. If you have real-time data, a passenger could intelligently ignore "bounces" and other data anomalies, as well as record data for different speeds.

Construction can't be that complicated. The oil level sender could be used as a model -- it's just a variable resistance based on the position of the float. In this case, you'd want something similar, except it'd be a big stiff spring. Since the numbers you produce are only comparative, you don't really need to calibrate the spring -- you just need to know how far down it gets pushed. Ummmm ... so a stiff spring, a variable resistance wiper doodad, 12V power, and wiring up to the passenger seat with a volt-meter to read the results. Too easy.

Next problem?

Dan

[jakermc]

I don't think it would work. Whatever damper you put in there would have to prevent the lid from touching any part of the car as if it did, the downforce would be spread over a wider area making the measurement inaccurate. Keeping the measurement concentrated on a single point, like the hinge area, is also inaccurate as different shaped tails will distribute the downforce differently. Some tails might be 'penalized' by a single point measurement if they produce lots of downforce but spread that force evenly across the engine lid area.

Bottomline, I don't think the downforce is distributed in a straight line. Arguements are welcome. I am not an engineer and I slept in my own bed last night, not a Holiday Inn.

[RickM]

Ok, I'll be cheap and suggest readily available instrumentation.
Use a digital weight scale.....the type used to weigh yourself. It could easily be diassembled and repackaged with the proper connections/wires lengthened so the display is extended into the cockpit.
Several types of hinged or underpinned mechanisms can be configured to provide input to the sclae's "table".

[Chuck Moreland]

I think a useful data point could be gathered this way. But the data would not be complete.

Downforce (lift) isn't just on decklid, but also on the rear window / fenders / roof area. The tail effects airflow over all these areas and consequently their lift too.

Further, the decklid is also supported by the hinges. The amount of force "weighed" at the latch depends not only on the magnitude, but also the location of the center of pressure (COP) on the lid. We don't know the location of the COP. Perhaps added "scales" could be inserted in place of the hinges.

I think something could be made using strain guages or load cells.

[304065]

Jack:

At the end of the day, all the aero measurement in the world is no substitute for segment times. Use your DL-90 to log top speed against RPM and concentrate on being at exactly the same revs or road speed coming on to a long straight, for example. Your top end or revs in fifth will absolutely be a function of aero drag. Go out and do one session with a wickerbill or Gurney flap, then download the results. Do another session with the flap removed, download the results.

The exact same methodology can be used for segment timing corners. Set the aero aids, and be sure you are hitting the exact same reference points in the corners. Use "concrete" points (pun intended") like the beginning and ending of turtlebacks. You can throw out any laps that show up on the track map as being highly deviant.

The beauty of this approach is that you are making measurements under the actual conditions that matter, not in a wind tunnel where you are trying to isolate variables. You already own the equipment. . . it just takes a ton of discipline to get out of the car and walk straight over to the laptop table. . . perhaps Herr Schmidt could assist?

[Jack Olsen]

Originally posted by Chuck Moreland
I think a useful data point could be gathered this way. But the data would not be complete.

You're right. But I'm not talking about replacing Frere's charts. I just want to be able to get a good estimate of differences between setups/tails/etc. It would be nice to see where the stall point (or whatever it's called) is with the wing angle for my particular wing. It would also be nice to see what the difference is between the RS3.8 tail with the factory winglet, no winglet, and the addition of the extensions and the big wing.

Part of what I'm thinking is that I'd like to be able to always run a ducktail, because I like the way it looks. So I would be interested in whether extensions and a wing could be combined with a ducktail to create a track-day combo that looks vintage, but also has the effectiveness of a big wing (in the manner that 996's don't change the whole rear tail configuration when a big wing is added).

Originally posted by john_cramer
At the end of the day, all the aero measurement in the world is no substitute for segment times. Use your DL-90 to log top speed against RPM and concentrate on being at exactly the same revs or road speed coming on to a long straight, for example. Your top end or revs in fifth will absolutely be a function of aero drag. Go out and do one session with a wickerbill or Gurney flap, then download the results. Do another session with the flap removed, download the results.

That's exactly the problem, though. There are so many variables at the track that I think a 'poor man's wind tunnel' approach would actually give me better data as a starting point. I've never seen consistent tire grip lap by lap at the track, much less during different sessions over the course of a day. Driver variations (with my limited skills) are an even more inconsistent factor. Track temperature and surface variations would also reduce the usefulness of my numbers.

A calm stretch of flat highway on a windless night (with tire grip eliminated as a factor) would give me a consistent set of numbers that I could then use with my data logger at the track.

Frere's charts are great. But with my weird combination of RSR body, early car aerodynamics, and home-made wing, it would be nice to get some meaningful data to see where I'm at.

[Bill Verburg]

You want the normal forces at the wheel contact patches, which you are not going to get outside of a wind tunnel.

The best you can do is know the basics of what you start w/ and then the basics of how changes affect things i.e. adding height or angle will increase down force^* and drag, a deeper dam or adding a splitter will increase front downforce w/o affecting drag

things like removing the seams can affect both lift and drag

^*or at least decrease lift

[widebody911]

Originally posted by Bill Verburg
You want the normal forces at the wheel contact patches, which you are not going to get outside of a wind tunnel.

And a rolling wind tunnel, at that.

[Wavey]

Originally posted by ghnat
Isn't there some metal that changes resistance based on the force applied to it? A simple wire and a calibrated gauge might work...

I was thinking the same thing - except the material is the piezo ceramic stuff, like what is used in gas grille igniters. It produces a current when compressed. You could measure the voltage produced by various tails. Not sure how you'd convert that to read in pounds of downforce though. But at least you could get a comparison.

[Cory M]

I've seen the use of pressure sensitive paints in windtunnel testing. As long as the ambient temperature is close to a standard day (25C) I think it could be used for testing outside of the tunnel. However, to really get good information you would have to videotape the test, download the results to your computer, and run it while comparing the results to the manufacturer calibration charts. This is probably very expensive and not practical but I thought I'd throw it out there anyway. We use this type of testing at my work and the analysis is run on a normal PC, but the tests take place in a lab so they are easy to control. Probably more of a science project then you're looking for.

The Swift Engineering wind tunnel is just south of you in San Clemente, you just need to make a scale model of your car

[304065]

Driver variations (with my limited skills) are an even more inconsistent factor.

Then, and I say this with all appropriate respect to you as a car developer, an accomplished driver, and a human being, why bother with aero at all at this point?

I usually come to the same conclusion when thinking about installing vortex generators or building splitters out of S-glass. . .I say, "couldn't I just make one fewer mistake per lap?"

[tangerine911e]

My wife is in the process of getting one of her final year aero students to run a CFD model of a 917. She's also toyed with the idea of running a CFD model of an early 911 (with and without a duck).

She's in the process of building a moving ground for one of her university wind tunnels (and already got a smaller - around 6 inches wide - moving ground in a perspex wind tunnel - plus access to a milling machine! I'm very jealous!!).

If anyone's got a decent 3D model of a 911 I'll see if I can get her to run it (on the PC and wind tunnel). She looked at getting a 3D scan of a decent 1:18 model to run, but the accuracy wasn't there.

Doesn't really help your problem Jack (unless we can magically model your car), but now I'm intrigued...

I might just get her to throw my 1:18 911S into the tunnel. Stay tuned!

[randywebb]

So why couldn't a measuring device....

interesting idea - it could give you useful info. Grady & bill are right - it will be neither complete nor accurate. But all you want is an index of downforce for various options - right? It's do-able. You want a Campbell CR-10 datalogger and a strain gage. You'll want to smooth out the data for bumps. You can hook in a laptop and have it chart the datalogger output for a realtime display. The CR10 used to be about $2k. Other types are available, including plug in cards for PCs. But that is the cheapest, most reliable, easiest to use system I know of. I'm assuming that as a writer, you are also not a systems control engineer...

Yes, the more accurate scales will have strain gages inside them -- a vet scale would be good. Bath scales are all over the map in how they work -- and how well. Save yourself trouble if you go that route and get a scale with a remote output that reads a voltage that is linear with wt. on the scale.

There could be a magazine article in this somewhere too....

[Elombard]

what about submerging a scale in the road (carefully) and run over it at the same speed with diff. tails. Have some one watch the reading. Crude but cheap. How long does a car scale take to get a weight? probably too long.

I wonder if you could bribe a truck scale operater to let you run over the scales at 100 mph a few times :-))

[efhughes3]

OK, who's got the old Car & Driver(?) or Road & Track(?) article where Steve McQueen gave his opinions of the 911. There was a sidebar as I recall that showed how the mag set up a bicycle wheel assmebly off the back of the car showing a very effective result from a simple ducktail. That did show a distinct downforce up and down.

The other thought might be to do a "plastigauge" type of evaluation with clay or something show compression by how much it squeezes.

I think you can get valid results. Ultimately Jack is looking for the best tail downforce, as I recall he's got three tails to choose from, so it really comes down to RELATIVE differences between tails, as you may not get an exact pressure reading of each setup.

Maybe a crude contraption will show an undisputed champion in relative terms?

[H20911]

something like this is what I would consider although i also like the bathroom scale taken apart method too.

http://www.transducertechniques.com/SLB-Load-Cell.cfm

not much weight and easy to set up. the problem that I see with them is that the the lid will not be with out some vibration and the cells might not like that.