Auto Aerodynamics 101 {Continued}
                                by Dan Jones

With the wing, the distances to be concerned with are local (i.e. over and under 
the wing itself and not over the whole car).  This why the wing is raised off the 
body of the car on your '93 Cobra (and my '87 GT).  What the rear wing is attempting 
to do is create downforce (negative lift) on the rear of the car, presumably to 
balance out a rear-biased lift tendency of the car without wing (assuming it's not 
just cosmetic).

Also, the path the air actually travels may be quite different from the contour of 
the vehicle.  For instance, a flat shape with equal distances over and under can 
produce a lot of lift. If you don't believe me, try this experiment at home (just 
don't sue me if you do). Step into the bed of a pick-up truck and lift a 4'x8' sheet 
of plywood over your head.  Be careful to hold the sheet of plywood parallel to the ground, 
while the driver slowly accelerates to 60 mph or so.  Now comes the fun part. Grip 
tightly to the sides of the plywood and quickly tilt the leading edge upward. What
happens? Instant lift (and an impressive, if short lived, Peter Pan imitation). 

What you've just experienced is the influence angle-of-attack has on lift. Take a symmetric 
(top-to-bottom) airfoil shape that does not produce lift when it is aligned parallel to the 
air flow (i.e. is at zero angle of attack) and point it up.  It produces lift. 
Point it down and it produces downforce.

While the physical distance over the top and bottom of the plywood are the same, 
the distance the airflow travels is not. Likewise, you don't need angle of attack 
or even thickness to produce lift/downforce.  A thin curved shape like a Venetian blind slat 
will also produce lift.  This is an extreme example of wing camber.

A little wing theory and several definitions are in order here.  This would
be easier to explain with illustrations, but I'll give it a shot with words.
An airfoil is the 2-dimensional cross-sectional shape obtained by the intersection 
of a wing and a perpendicular plane.  The mean camber line of an airfoil is the 
locus of points halfway between the upper and lower surfaces (measured perpendicular 
to the mean camber line itself). The chord of an airfoil is the straight line 
connecting its leading edge to its trailing edge.  Camber is the maximum distance 
between the mean camber line and the chord line, measured perpendicular to the chord line. 

An airfoil's angle of attack is the angle between the relative wind (the local airflow 
direction) and the airfoil's chord line.  Drag is the component of aerodynamic force 
parallel to the relative wind and lift is the perpendicular component. 

If an airfoil is symmetric (top-to-bottom), it has no camber.  A sheet of plywood has no camber. 
A Venetian blind slat is a shape that has camber but(practically) no thickness. The camber, 
the shape of the mean camber line, and the thickness distribution of an airfoil 
determine its lift and moment characteristics.  Surface roughness also plays a roll but 
is usually treated as a separate design issue. 

Because of camber, wings can have lift at zero degrees angle of attack and because of 
angle of attack, wings (and sheets of plywood) with no camber can still produce lift.  
To separate these effects, aerodynamicists break an airfoil's lift into two components: 
                       Cl = Clo + (Cla * alpha)
  where:

    Cl = coefficient of lift 
    Clo = coefficient of lift at zero angle of attack
    Cla = lift curve slope (the slope of Cl versus alpha)
    alpha = angle of attack

On low speed circuits where downforce is very important, Formula 1 race cars will have multiple, 
highly cambered, wings, oriented at a relatively large negative angle of attack.  All of this 
is done in an attempt to generate downforce.  Since this approach is a relatively 
high drag method of generating lift, you won't see similar set-ups on aircraft 
(they are not limited by wing size rules). 

Front lip spoilers (like those on a Boss 302) produce downforce because they are mounted 
at a relatively large negative angle of attack.  They have all the aerodynamic elegance 
of that piece of plywood, but they work.  A pedestal mounted, cambered, wing would be more 
efficient but probably wouldn't look too good mounted on top of your hood.  Some versions 
of the Lamborghini Countach have a pedestal mounted front wing on the nose. 

By the way, car spoilers really aren't spoilers at all.  On aircraft, spoilers are devices 
which intentionally promote pressure separation.  They are called spoilers because 
they "spoil" lift when they are deployed.  They are generally mounted flush on top 
of a wing and pop-up to reduce lift and increase drag.  You can watch these devices 
at work on airliners when they decelerate in preparation for landing. 

Rear mounted spoilers (like those on Cobra Daytonas), look more like a fixed version of 
an aircraft trailing edge flap.  Trailing edge flaps generate lift/downforce by altering 
the effective length of a wing and its camber line shape. 

Since real wings exist in three dimensions, there are 3-d effects to be concerned with. 
Lateral flow along a wing is called span-wise flow and is usually undesirable. Guides, 
called fences, are often employed on aircraft to reduce span-wise flow.  The circulation 
around the ends of wings is particularly strong and creates large vortices and substantial
drag.  The flow wants to circulate laterally from the high pressure region underneath 
the wing to the low pressure region on top of the wing (in the case of lift). Race cars 
usually employ large vertical end plates on the sides of their wings to reduce this circulation. 

Unfortunately, most of this is probably academic since I have reason to believe the rear wings 
on late model Mustangs don't see all that much flow. A couple of guys at work 
(McDonnell Douglas Aerospace), tufted an '87 LX from the center of the roof to the 
taillights.  They were trying to use vortex generators to increase the flow attachment 
on the rear glass.  Vortex generators are devices which are put in the flow field to 
intentionally induce turbulent flow. They are often used on aircraft to re-attach and
direct flow (especially over control surfaces). Their vortex generators were based on 
aircraft designs and they used a hang glider airspeed indicator on a pole to measure 
the boundary layer thickness across the roof.  They made the vortex generators two 
inches tall to be conservative (the boundary layer was approximately one inch thick 
and a rule of thumb is to make the generators 1.5 times the boundary layer thickness).
They didn't see an improvement in coast down times, but the tufts did appear a little 
better behaved with the vortex generators. They believe the turn at the back of the roof 
may be too sharp to permit attached flow. Some sort of fairing might help there (or maybe a 
switch to a Capri hatch).  They also noted that much of the clean wing flow appeared 
to be coming from around the sides of the car. 

From watching the flow patterns in the rain, one of them concluded that the flow over 
the hatch of his 1986 Camaro was still attached, though it flowed laterally as well as 
longitudinally. 

Taller wings or wings that are mounted farther aft will see cleaner air. Rear wings 
on race cars tend to be mounted high to get them out of separated flow and into clean flow.
The wing on the Dodge Daytona Chargers and Plymouth Superbirds (the ones that made 
them look like shopping carts) are examples of this.  It would be interesting to 
tuft the various rear wings (LX, GT, Cobra, SVO biplane, etc.) to see if any of them are useful.

On a related note, some of the racing Shelby GT-350's had a noticeable gap between the 
fastback window and the roof. Does anyone know if this was for aerodynamic reasons?

I previously mentioned trying to estimate drag from coast down measurements. By taking 
measurements at several speeds, you should be able to separate the effects of rolling 
resistance (roughly proportional to speed) from aerodynamic drag (proportional to the 
square of speed). Data scatter would be a real problem, but you could do a least 
squares curve fit. Also, making runs with and without winds (at the same speeds) 
could be used to isolate the aerodynamic contribution. Supposedly, NASA has done 
some work on coast down drag equations as part of an effort to reduce drag on 
tractor-trailers.  They even put a full size tractor-trailer in the 80' x 120' Ames wind tunnel. 

I dug up some information on Mustang and Thunderbird drag coefficients from a couple of old magazines. 
The January 1984 issue of Sports Car Graphic claimed a Cx of 0.39 for a 1984 SVO Mustang (the model 
without flush headlights) and s 1987 issue of Sports Cars of the World had these numbers: 
                 0.35 for a 1986 base Thunderbird 
                 0.38 for a 1986 Turbo Coupe Thunderbird 
                 0.34 for a 1987 base Thunderbird 
                 0.36 for a 1987 Turbo Coupe Thunderbird 
One of the guys who did the vortex generator experiment, also relayed some information on an 
AIAA presentation made by Corvette engineers about the 'Vette's aerodynamics. They claimed 
the 'Vette has a Cx of 0.30 and said it was a difficult number to achieve with such wide tires.
Note that GM tests without mirrors, so this number may be a bit optimistic.  There was also
a drag hit with the externally mounted 3rd taillight.  GM used to test at several tunnels, 
including one at Lockheed Georgia and one in Canada, but has since built its own tunnel. 
Interestingly, it does not have a rolling mat. 
The engineers admitted this is a compromise but noted they use a boundary layer suction device 
near the front tires. This arrangement apparently yields useful data with less scatter 
than a rolling mat facility. 

The 'Vette engineers also noted that more aerodynamic testing is done for acoustic (noise) 
and cooling reasons than for drag reasons.  At first glance, since it takes energy to make noise, 
you might think a quiet car is a slick car. This is not necessarily so.  Turbulent 
boundary layers are noisier than laminar ones, but they often provide lower drag. 
Of course, you need to trade this off against pressure separation noise. 

P.S. I just saw a set of 351 tunnel port heads in the SVO catalog.  Does 
anyone know if they use faired-in pushrod tubes?  I think the old 427 
tunnel ports just used cylinders to house the pushrods.  Since a cylinder 
is a relatively high drag shape, an airfoil shaped pushrod housing should 
yield a flow increase.