Professional Documents
Culture Documents
1996 161 Facts of Lift
1996 161 Facts of Lift
1996 161 Facts of Lift
P. B. S. Lissaman
Southern California, Univ., Los Angeles, CA
AIAA, Aerospace Sciences Meeting and Exhibit, 34th, Reno, NV, Jan. 15-18, 1996
As an aid to teaching fluid mechanics and understanding dynamic fluid forces, fundamentals of 2D and 3D lift and drag
in an inviscid incompressible fluid are developed, emphasizing simple rigorous derivations, based only on the
momentum equation. Some common beliefs about wings and airfoils are shown to be false or oversimplified. The
standard textbook model, deriving lift from circulation around a circular cylinder, is noted to be unrealistic and
deceptive. Instead, the wing is modeled as a turning vane in a steady flow, and the Joukowsky and Lagally theorems
are derived. An unsteady analysis is developed for lift and drag on finite lifting systems; half the lift occurs in upwash
momentum effects before the flow reaches the airfoil, and both 2D and 3D lift are entirely due to bound vorticity,
according to the same equations. Consequently, the downwash due to the trailing vortex system behind a finite wing
does not contribute to the lift. Effects of flow boundaries, like the wall effect, of multiple bodies, as in multielement
airfoils, and of nonplanar lifting surfaces, like tiplets, are described. It is shown that neither angle of attack, camber,
nor circulation are required for lift. (Author)
Page 1
AIAA-96-0161
LIFT IN THE REAL WORLD OF THREE where the integral extends to infinity in the horizontal
DIMENSIONS (spanwise) and vertical directions.
V= Ujf wtdzdy = 0.
Noting that sin^i - y/li, and/(y/li)dy = lathis As expected we will find an identical term for the
momentum flux through a vertical surface ahead of the
provides, between the limits, Ij, \2 '• wing, so we get that the lift, L, is given by:
V*(z)=U(T/4jtr){li-l2} L= UTb.
which, on inserting the spanwise limits of integration, The important point here is that the lift is created by an
as lj, \2 approach infinity, becomes: upwards induced flow ahead of the wing and a
It is very obvious that the above argument can be Lagally is very useful for complex flows, where it can
extended for a bound vortex of spanwise varying readily be shown that the same results apply, providing
strength, T(y ), so that the general expression for lift that the local speed and direction of flow and force at
will be the integral across the span given by: each singularity is used. This shows, for example, that
for a number of bodies in a uniform steady flow it is
L=UJT(y)dy very easy to get lift (a force normal to the free stream
flow) without circulation. As an illustration, a
symmetrical body of finite thickness at zero inclination
WHAT A DRAG! in the presence of a wall parallel to the flow will
experience a potential force towards the wall. The rule
If we take the 2-D case and apply the same reasoning of is that sometimes you need circulation for lift, and
integrating momentum to mass flux and pressure, p, on sometimes you don't!
the vertical surfaces, we get that the horizontal
contribution, H, due to the rear surface is: Now if there is a trailing system, then the perturbations
due to it are large in the rear plane, usually called the
Trefftz Plane. In fact, the vertical flow induced by the
H = A(U+u) (U+u) + p} dz trailing system is infinite outboard of the wing tip.
Thus, for the 3-D case, we need to add to our pressure
Bernoulli's equation reduces this to: integral the terms due to the trailing vorticity giving the
Trefftz Plane integral:
H = /°{Uu + u2/2 - w2/2} dz.
H = JJ°°{- vt 2/2 - wt 2/2} dzdy
The last two quadratic terms vanish on a distant plane
by virtue of their order of magnitude, and incidentally, These terms do not appear on the forward control
on a finite distanced plane as well, because of the surface, since there are no trailing vortices there.
symmetry. The remaining term is of exactly the same Consequently, there is always a lower pressure on the
nature as the circulation induced term handled in the lift
rear surface due to the high speeds of the circulatory
case. If the body is closed, then u is of order 1/R?, and flow developed by the vortex system, especially near the
we get the famous result that puzzled Jean le Rond wing wake. This gives rise directly to the induced drag.
D'Alembert between 1744 and 1768. According to John This can be evaluated by a direct integration of the
Anderson's marvellous book, Fundamentals of pressure in this vicinity: it dies away rapidly far from
Aerodvnami cs (1984), Jean said: " It seems to me on the wing wake, and, although infinite at places near the
the contrary that this theory, dealt with and studied with wake tips, is integrable in a regular sense.
profound attention gives, at least in most cases,
resistance absolutely zero: a singular paradox which I In the real world, these infinities near the tips do not
leave to geometricians to explain." Vive D'Alembert! exist, being thoroughly suppressed by viscosity, and
He's the guy Who gave US streamlining compressible effects. People who fool around with
doing things at the wingtips to reduce induced drag, and
So much for that D'Alembert's Paradox is valid in two they are an innocent and trusting bunch of enthusiasts,
or thee dimensions, as long as the body doesn't trail any are dealing with this issue. It can be done, and doesn't
vortitity. And don't be shocked, it holds true in defy any laws of nature, but it's a tough way to do
compressible flows too, in the absence of battle with induced drag. It does seem to illustrate one
discontinuities! of Murphy's Laws: that anything done to improve a bad
scene usually makes itworsei
If there is mass generated by the body, say at a rate of
M volume per unit time, then we get the dual of The author confesses to being one of that hapless band
Joukowskt's result, a source thrust, T given by: of induced drag worriers myself. In the Dark Ages,
1968, to be precise, Jerry Lundry and I published a
The approach here, which is the most general attack on which integrates to:
lift, is to change the frame of reference and adopt an
unsteady procedure. Take the frame of reference as E-112 ds
inertially fixed in a basically still fluid with the wing as
moving through it at speed U. This wing makes a lot The rate of increase of the energy in the system per unit
of energy, but no net downwash momentum. Consider time is given by UE, while the rate of doing Work of
the difference in the situation after the wing has moved the wing by moving against the drag, D, in this frame
some distance in unit time. We see (Fig. 6) that an of reference is DU. Consequently the drag of the Wing
extra strip of trailing vorticity has been produced. Now, is given by:
while the momentum is undefined, the impulse is not,
as explained by Horace Lamb in 1879 in his D =1/2 J<I>(y j] *n ds*
monumental tome, Hydrodynamics.
L =U/*(y)]dy.
So this recovers the cqrrect result for lift obtained Fig. 7 Highly Loaded Wing
Again, this is nothing more than Newton's Law, but
applied in a very global fashion. What this says is that Assume the induced flow in the wing plane is normal to
to experience lift in an unbounded flow you have to the inclination, $, of the trailing system at infinity and
generate trailing vorticity. half the magnitude it has at infinity and that all the
bound vorticity is concentrated in the wing plane as a
lifting fine. Then the lift, L, and drag, D, are given by
The spurious method previously detailed gives:
L= {U-sin dy
L = 2Uf*(y)]dy.
If we now insert the results for elliptical loading and its There is a low pressure contribution on the Trefftz
Plane which produces a drag-like force, D*, amounting
associated downwash, namely
to [1/2] O n /*(y )] dy parallel to the far wake
{jib 2/4}<$n = /<E>(y)] dy, we get the maximum lift, vorticity, while the bound vorticity will couple with the
* °f any planar wing to be given by: speed parallel to the far wake to produce upwash and
downwash momentum on the forward and rear plane to
- {^b 2/4} (2/3) 3/2 U2 give a lift like term, L*. parallel to the Trefftz Plane
and of magnitude LI = U cospJ"<I>(y)] dy. Finally
The nominal maximum lift coefficient, CL , based on we must consider the crossflow term, U sin p. This is
free stream dynamic pressure is given in terms of the found to couple with the bound vorticity to provide an
overpressure on the Trefftz Plane and a corresponding
aspect ratio, A, by: suction on the forward control surface, yielding a net
thrust like force, L2, directed normal to the Trefftz
3
CL = {n/2} (2/3) /2 A = 0.86 A. Plane and of magnitude, L-2 = U sin PJ 4>(y )] dy.
Actually, the functional effective lift coefficient on the These three forces can be resolved into a convenient and
wing would need to be two times higher, since the vivid formulation for the exact nonlinear force on a
induced counterflow due to the downswept trailing wing system of any aspect ratio wing, which is given
system will reduce the effective local flow at the wing by the sum of two non-orthogonal forces, L*, and D*,
to (1/2) 1/2 of its free stream speed. expressed by:
The above result, thanks to the stagger theorem, is not L* = U J <fr(y )] dy , normal to the free stream
restricted to high aspect ratio wings, but should also be
correct for slender wings. D*= [1/2] <l>n/*(y)] dy, parallel to the wake vorticity
It now becomes interesting to study the forces from the Of course this recovers the result originally derived by
global point of view of the momentum equation. It is less rigorous arguments, for a rigid, lifting line model.
clear that, having fully defined the internal flow by the
vorticity distribution in the Trefftz Plane, we can chose There is yet another, positively Olympian, approach to
a control volume and exactly compute the force on any the force determination. This is to take a truly Grand
body inside it using only conditions on the boundaries, View of the situation, and to treat it as a wing moving
the only singularity in this case being the trailing at constant speed in a still fluid. The problem is now
vorticity sticking through the Trefftz Plane. Any unsteady, and generation of wake vorticity occurs over a
inertia! control volume fixed with respect to the wing length of Ucosp per unit time. In coordinates oriented
will do. We select two infinite near vertical surfaces with the Trefftz Plane this generates a rate of increase of
8
American Institute of Aeronautics and Astronautics
impulse rate of I * and of kinetic energy E expressed by: at the the non-linear result using some old, familiar and
beloved integrals from the linearized world!
I * = Ucos p/$(y)] dy
E*= Ucos p [l/2]<J>n; <D(y )] dy. THE STAGGER THEOREM AND OTHER
MONKEY BUSINESS
Thus one immediately obtains by using Lamb's results
for the rate of change of impulse the force normal to the The Stagger Theorem, derived by Max Munk in his
vortex wake, L**, and while the rate of work done by terrifying dissertation Isopcrimetrische Probleme aus der
the wing to create E* is given by UD, the force parallel Theorie des Fluges (1918), is much beloved by
to the speed. These two identities give: engineers because its name says exactly what it's about,
it's easy to apply and usually works. The general idea
L**= Ucos P/*(y)] dy is that one may move bound vorticity, or wing
elements, as much as one wants in the streamwise
D= cos p [l/2]4>n/$(y)] dy direction without affecting the lift and drag of the
system, providing one keeps the same circulation
which are the correct results — exact, nonlinear! distribution on an element as it is moved. If one views
events from the Trefftz Plane the theorem becomes self
To develop the above result we have assumed, to all evident, since in that plane one doesn't know how the
intents and purposes, there is some total impulse due to bound vorticity is arranged in a streamwise sense
the whole system, including all the mess around the anyhow. While the induced drag of the system is
wing and at the starting vortex; and after a unit time the unchanged by stagger, the individual drag of
stuff at the front and back remains the same and all one components is profoundly affected.
has added is that little slice, Ucosp thick, of the Trefftz
Plane-like flow having the rate of change of impulse It is interesting to apply this to a pair of lifting systems
calculated above and providing the force of L** normal with collinear bound vorticity. Consider two birds
to the wake. But L** is NOT the entire force on the gliding line abreast. We can readily demonstrate that
system, there is also the component parallel to the wake each experiences the beneficial upwash of the other's
given by D*. Thus the rate of change of impulse as trailing system, and, in fact, helps the other by exactly
calculated by the above procedure does not give the total the same amount, even if the span or lift of the two
force and thus can't be ALL the impulse, which is a systems differ. Thus a swallow flying abreast with an
vector quantity. eagle will experience just the same absolute forward
thrust in Newtons as does the eagle due to the swallow.
We have stumbled upon the Cisotti Paradox (1924), Now if one of the birds falls back in a streamwise
which states that it is a puzzlement that a flat plate direction, then it experiences a stronger upwash due to
subject only to pressure forces can experience a net force the flow outside the trailing system of the other, and
not parallel to the plate. As noted by Garret Birkhoff , thus experiences a larger induced drag saving. The
in his interesting and profound book Hydro-dynamics: a forward system has a correspondingly reduced saving.
study in logic, fact and similitude(1950). this is a
singular point paradox, which is immediately resolved In a long line of birds flying abreast, those in the
for wings of finite leading edge radius, and is called the middle experience favorable upwash from both sides and
nose thrust phenomenon. What is happening in the have an easier time than the tip members. But if the tip
lifting wing case is that the whole linear extent of the members fall back, they will reduce their relative flight
wake is increasing as time passes, and while impulse power requirements and will thus rather naturally select
normal to the wake surface is certainly added, there is a Vee formation which has eqiripartition of induced drag,
also some change in impulse parallel to the wake which and quite substantial savings. Wieselsberger, another
must be related to flow around the leading edge of the Gottingener, wrote a paper on this topic in 1919, and
wing. The remarkable thing is that computing the years later Carl Shollenberger and the author revisited it
change in energy simply by calculating the amount with the help of the IBM 360, which was considered hot
added in the strip by which die wake extends provides stuff in 1966. Thanks to Mr. Watson's admirable
the integral sufficient to calculate the streamwise force calculating engine we were able to do the tiresome sums
on the wing and hence the force parallel to the wake! which Wieselsberger had more sense than to attack by
We should be so Lucky! ! hand. It turns out that for a reasonable tip spacing a
Vee formation can provide more than 56% range
Still-an-all, we might as well quit while we're ahead, increase, which seems sufficiently large that one
and comment that if one does know the wake structure assumes that even a fowl of Limited Brain, like a
in the Trefftz Plane, and has some estimate of its Canada Goose might have noticed, or, at least, that
inclination, one can make an excellent and accurate shot those that didn't are now pate and down jackets! An
10
American Institute of Aeronautics and Astronautics
compute anything unless one has the correct spanwise
Anyhow with all that twisted air in a vortex wake one lift distribution to start with. If that is OK then one
would certainly expect odd things to occur. They has at least a few invariants to give some control on the
definitely do here! And one might well despair of the global features of the rolled-up downstream flow, and
sort of intelligent approximate mathematical analysis one may hope that if the wiggly bits occur at least a
that was good enough for von Karman. This area is a few spans, say six, from the wing itself, that only the
fertile field for computational fluid mechanicers. The first few moments of the vortex distribudon will affect
idea is that the field equations are really quite simple -- the induced flow at the wing, so that one might have a
all you have to do is follow the trajectory of every passable model of the induction of the far wake and add
vortex with infinite precision. It is rather frightening to it to the near wake induction to calculate the lift
attack the free vortex problem numerically. But I distribution. But enough of this conjecture; nobody
suppose some one has to do it! knows what happens here, even with our massive CFD
power. As noted, it is a Bad Scene.
So we must consider a set of two dimensional vortices,
all milling around, responding to the induction of the
others — like, well, like a herd, a most charming THE STAGNATION POINT AT THE REAR
collective noun invented by flan Kroo. If we write END
them in the form appropriate to discrete vortices, each
of strength Fj at location zj, the motion is defined by Mercifully, we have come to the limit of my knowledge
the speed, Vj , of each vortex i. Sure, the speed's of the subject of incompressible lift, actually, we are
singular, but here we can use the principal value. So somewhat beyond that point! The chapter heading seems
this is given by: like a good end; en fin, if we all designed bodies to nave
stagnation points at the back, then D'Alembert would
rest happy, and we would have no drag. Presumably
that would be a Good Thing generally, except for
This herd of vortices does its thing subject to three parachutists!
global constraints:
This paper is some kind of a homage to Leonardo. One
2 Fi = A 2 Fi zi = B wishes we could end with one of bis renderings of the
ZFj Fj In ( Zi . zj ) = C. female head. Unfortunately, our reproduction facilities
can't do his graphics justice, so well conclude with a
The constant A represents conservation of vorticity, B touching quote from that lover of all fluids and all
conservation of lift and C, conservation of energy. things human:
These are physical invariants that we should expect as " Observe, with the approach of evening,
the consequence of any rational wing theory. We may in stormy weather, hou> much grace and,
well ask why the two dimensional unsteady model also sweetness are to be seen on the faces of
respects this. After all it doesn't know it's supposed to
be the backside of a wing! The first two results come men and women in the streets".
directly from the kinematics of two dimensional as reported by Vasari in 1550.
vortices, the third reflects the fact that the vortex is
force free thus no work is done in the course of its REFERENCES
motion so the kinetic energy of the flow is conserved.
Unfortunately the contribution to C is singular at each Anderson, John, D., Fundamentals of Aerodynamics.
vortex. Actually that's not too odd, since a finite vortex McGraw-ffill,N.Y., 1984
does have an unbounded kinetic energy and creates Birkhoff, Garrett, Hydrodynamics Prinr^ton Univ.
infinite induced drag. Press, N.J., 1950
Chaucer, Geoffrey, The Parlement of Foules. 1380 ?
Why do we need to know this for actual wings? Well Giacometti, R., Aerodynamics of Leonardo da Vinci, J.
there are two reasons ~ at the wing itself we need to Roy. Ae. Soc., DC, 1930
know the induced flows to define the proper kinematic Lamb, Horace, A Treatise on the Mathematical Theory
boundary condition to get the lift pressure distribution; of the Morion of Fluids. CUP., 1879
and far downstream from the wing, where we need to Munk, MM., Isoperimetrische Probleme aus der
know what's going on for the sake of some other Theorie des Fluges, Gottingen Dissertation, 1918
innocent airplane flying into this mess of vorticity. Prandtl, L and Tietjens, O.G.. Fundamentals of Hvdro-
and Aerodynamics Dover Publications, 1934
I don't know what anyone has done recently about the Vasari, Giorgio, Lives of the Painters. Sculptors and
roll-up problem as it affects the flow at the wing. This
would seem to be the important place, because one can't Architects. 1550
11
American Institute of Aeronautics and Astronautics