911 rake front/rear

[Tyson Schmidt]

Bogieman, can you make one with a flat bottom, but with the rear of the underbelly upturned like on the 964/993? If I can see it, I will believe!

[BogieMan]

Got some CFD results for the effect of rake on a crude 911 cross-section(no tail). 2D results show that the suction under the nose is enhanced by the effect of rake, while the lift on the rear wheels actually increases slightly. Since this is really a 3D problem, the 2D results should be used for trends only.

Rake Total Lift Front Lift Rear Lift
0deg 1,100# +433# +671#
5deg -682# -1,370# +688#

The increase in lift on the tail is due to an increase in underbody pressure at the rear and a slight reduction in separation on the rear deck. The increase in underbody pressure at the rear is consistent with a simple venturi analogy. The air velocity is slowing down underneath the car as it moves aft and car under-body rises higher above the ground.

Please forgive the crude nature of the model. I don't have a 911 or access to one.

I've attached a couple photos, and have more thorough results in .pdf format. If anyone is interested, let me know if you have a site to host the files.


Hope this helps!
-BogieMan

[Bill Verburg]

Very interesting!!

[stlrj]

Looks like Cd goes up significantly too.

[MotoSook]

Nice work Bogie! I would love to get a copy of the pdf file. I can host it if you'll send me the file (<25MB).

[MotoSook]

Without the more detailed report....

Something to think about, most of us would lower our cars further than euro height (just because it's a start), if we were serious about this. Rake could increase, but don't do it without a tail (as others have stated). A diffuser at the rear would help (think Dodge Avenger (yep..look at the first gen avengers)...or as Tyson stated, the later 911's)

Very interesting indeed! Again, nice work Bogie.

[350HP930]

Great work there BogieMan. I'm glad to see my intuitive model on page five is not too far off the mark.

[BogieMan]

Thanks MadRocketSci.

Good job with the sketch - I think that clarified a bunch of stuff, especially when used in conjunction with Bill's tuft photos.

I've mailed the larger file off to Souk who has graciously offered to host it on his site.

-BogieMan

[masraum]

Very cool, thanks for running that!

[MotoSook]

Here are the goods gentlemen. Thanks BogieMan.

BogieMan CFD Doc

I hope you hang out here after the interest for this thread dies down.

[island911]

not too far off the mark. . . of an exceedingly simple CFD model. Don't hurt your elbow there. (patting yourself on the back)

I'm always amazed how much pretty computer gennerated pictures can persuade people. I hate to break it to you guys, but that is a model. . .an approximation . . .not reallity. And, if you don't know what went into the model, you don't WTF you're looking at. SO. . .

Bogieman, interesting; how did you model the ground BC? Mirror?. . .moving plate . . .static, wrt the body?

[MotoSook]

island you are absolutely correct, but we are taking a slice of a very large pie in order to understand some of the phenomena that affect the 911 at speed.

If the engineers could do away with wind tunnel testing and use only CFD data, it would make things much easier, but there is plenty to learn from actual testing, in a wind tunnel and even on a track. I think Bogie, and most of us who have used a CFD program before, understands the limitations of CFD.

Bogie qualified the results.

[masraum]

Guys, I hate to ask, but what would it look like with a tail? Can you run that one?

Just out of curiosity what software is that?

[Randy Webb]

Hey, not only is it just a model -- BUT the "real" equations aren't even solved analytically. The models just simulate particular numerical solutions.

Someday, some 9 year old from India or Russia will solve it tho....

Those were pretty -- thanks for posting them. Now, what assumptions were in the model?

[BogieMan]

Some details of computational analysis as per request.

CFD results obtained with an unstructured Navier-Stokes solver and a yplus gridding < 0.5. Residuals converged to 5 orders of magnitude using multi-grid sequencing and pre-conditioning. Ground plane simulated as an inviscid surface to avoid build of of viscous boundary layer on moving ground plane. This is acceptable for 2D solutions as the principle effect is to dissipate vorticies that are not modeled in 2D. 3 levels of absorbing layers were used in the farfield simulation 150 ft away from the automobile.

This model incorporates 2 of the 3 critical features essential to the 911 rake problem. First it models the venturi effect. Second, it captures viscous lift and separation in ground effect. The third effect, encompassing cross-flows and vorticies are not possible to model in 2D. These are none-trivial effects, yet the results may still be used to demonstrate and quantify the first and second factors.

As stated previously, interpretation of these results should be limited to trends and demonstration of fundamental effects.

-BogieMan

[BogieMan]

Shouldn't be to hard to add a tail. May be able to get to it tonight since I'm waiting for some large grid generation jobs to finish.

The software is Metacomp's CFD++ solver. It's used in the automotive industry by several companies.

-BogieMan

[Randy Webb]

Wonder how "smooth" the bottom of these cars really is? I like seeing the shed vortices off the back -- looks pretty realistic. Maybe we should stick some Kamm tails on our cars...

[Jim Garfield]

I just read through this post and while the "my friends are smarter than your friends" aspect is amusing, it's like watching someone trying to reinvent the wheel and beat a dead horse with it.

I know nothing about aerodynamics other than what I have observed on race cars over the years. But I do have quite a few Porsche related books. A quick scan of the first one that I pulled down, Paul Frere's PORSCHE 911 STORY has a good overview of aerodynamic developments on the 911 from the original Stuttgart University wind tunnel tests of Sept 1968 through the 993. My scanner is out of action but maybe someone could scan & post this?

I'm sure that Porsche's wind tunnel work on the 911 is well documented and published.

Briefly, Frere states without spoilers a 911 at 152 mph has 397 lbs. of total lift, adding a Carrera front and rear spoiler reduces this to 37.5 lbs. of lift. A ducktail instead of a whaletail increases the lift to 76.5 lbs.

"The rear spoiler, which increases the rear wheel grip by no less than 20% at the cars maximum speed, has the additional advantage that, combined with the front dam, it moves the center of air pressure back from a position 10" behind the front wheel axis to a position 16" behind with a resultant noticible decrease in the car's sensivity to side winds. (Which was a huge concern and started them doing wind tunnel testing in the first place) This is further decreased by the fact that the dam and rear spoiler combination also reduces the tendency for the lift to increase in side winds. It is also noteworthy that the aerodynamic aids result in a front/rear lift ratio of 45/55 almost exactly matching the car's weight distribution - an ideal state of affairs.

Extending the front dam downward, which reduced the ground clearance to 7.5" actually created some front down thrust and further reduced drag, but because of the pitching effect over the front axle it INCREASED REAR LIFT."

[BogieMan]

Jim -

Neat results.

No argument here. The increase of rear lift is shown in the CFD results above arising from nose suction and rear underbody pressure. Regarding the levels of lift, results from the Stuttgart tunnel should be much better than 2D computer simulations. The level of lift that you quote is more realistic than the 2D stuff as well.

-BogieMan

[Randy Webb]

So..... what happens if I shave off my drip rails??