A Variational Theory of Lift

Cody Gonzaleza and Haithem E Tahaa,2

a
University of California, Irvine

This manuscript was compiled on October 19, 2021

In this paper, we revive a special, less-common, variational principle in analytical mechanics (Hertz’ principle of least curvature) to develop a
novel variational analogue of Euler’s equations for the dynamics of an ideal fluid. The new variational formulation is fundamentally different
from those formulations based on Hamilton’s principle of least action. Using this new variational formulation, we generalize the century-old
problem of the flow over a two-dimensional body, to find that lift is a direct consequence of curvature. The developed variational principle
arXiv:2104.13904v3 [physics.flu-dyn] 15 Oct 2021

reduces to the classical Kutta-Zhukovsky condition in the special case of a sharp-edged airfoil, which challenges the accepted wisdom about
the Kutta condition being a manifestation of viscous effects. Rather, we found that it represents conservation of momentum. Moreover, the
developed variational principle provides, for the first time, a theoretical model for lift over smooth shapes without sharp edges where the Kutta
condition is not applicable. We discuss how this fundamental divergence from current theory can explain discrepancies in computational
studies and experiments with superfluids.

Variational Principles | Kutta-Zhukovsky Lift | Ideal Fluids | Hertz’ Principle of Least Curvature

1. A Meager State of Theoretical Modeling of Aerodynamics

T
The problem of the flow over a lifting airfoil is a century-old, classical textbook problem in aerodynamics and fluid mechanics
(1–3). The problem is analytically solvable thanks to three elements. First, the potential flow around a circular cylinder is
readily known since the 1877 seminal paper of Lord Rayleigh (4). Second, the Riemann mapping theorem ensures that any
F
simply connected domain can be (biholomorphically) mapped to the open disc. So, the flow around any two-dimensional
shape can be easily constructed from the cylinder flow via conformal mapping between the cylinder and the shape of interest.
However, this solution is not unique. One can always add a circulation of arbitrary strength at the center of the cylinder, which
RA
does not affect the no-penetration boundary condition at all. Interestingly, this circulation is of paramount importance for lift
production; in fact, it solely dictates the amount of lift generated. Therefore, the potential-flow theory alone cannot predict
the generated lift force; a closure condition must be provided to fix the dynamically-correct amount of circulation. The third
element is the Kutta-Zhukovsky condition, which has traditionally provided such a closure.
The Kutta condition is quite intuitive: if the body has a sharp trailing edge leading to a singularity, then the circulation
must be set to remove such a singularity; it is a singularity removal condition (5). It is quite accurate for a steady flow (at
a high Reynolds number and a small angle of attack); indeed, it is a paradigm for engineering ingenuity where an enabling
D

mathematical condition is inferred from physical observations.

Interestingly, if one wishes to consider a smooth trailing edge, however small the trailing radius (i.e., however close to a
sharp trailing edge), the classical aerodynamic theory collapses; there are no theoretical models that can predict lift on a
two-dimensional smooth body without sharp edges! (only few ad-hoc methods with no theoretical basis). In fact, some authors
even consider the sharp edge as a lifting mechanism; i.e., an airfoil must have a sharp trailing edge to generate lift, see Ref. (6).
For unsteady flows, the status of current aerodynamic theory is even worse. Martin Kutta never claimed that his condition
works for unsteady flows; in fact, for an unsteady case, it is already known that, in the early transient period after an impulsive
start, the flow goes around the trailing edge from the lower surface to the upper surface (see (7), pp. 158-168; (8), pp. 26-36; or
(1), pp. 33-35). That is, since more than a century, it is known that the Kutta condition is not generally applicable to unsteady

Significance Statement
After more than a century since the first successful flight, we managed to discover the general mathematical principle that governs lift
generation over airfoils; we developed a new theory of lift that dispenses with the ubiquitously used empirical condition—aka the
Kutta condition. Interestingly, the developed theory is based on a principle due to Hertz that seems to have gone into oblivion in the
history of mechanics: Hertz’ principle of least curvature. The developed (inviscid) principle reduces to the Kutta condition in the
special case of a sharp-edged airfoil, which challenges the widely accepted wisdom about the viscous nature of the Kutta condition.
Moreover, the developed theory, unlike the classical theory, allows computing lift over smooth shapes (without sharp edges).

H.T. introduced the use of variational principles in the airfoil problem; C.G. conjectured the minimality of curvature; H.T. drew the connection with Gauss’ principle; C.G. reduced it to Hertz’ principle and
conjectured the constraint nature of the pressure force; H.T. proved it; C.G. performed numerical simulations; H.T. wrote the draft of the paper; and both authors finalized the manuscript.

The authors declare no competing interests.

2
To whom correspondence should be addressed. E-mail: hetaha@uci.edu

www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX PNAS | October 19, 2021 | vol. XXX | no. XX | 1–8

flows (even for airfoils with sharp trailing edges at small angles of attack). This is why numerous research reports criticized the
application of the Kutta condition to unsteady flows (5, 9–22). Nevertheless, almost all unsteady models, starting with the
pioneering efforts in 1920’s and 1930’s of Wagner (23), Theodorsen (24), and Von Karman and Sears (25) until the recent work
of Badoo et al. (26), adopted the Kutta condition. There is no legitimate alternative! So, what we have is a meager state of
knowledge and a very confined capability of aerodynamic theoretical modeling: we can only analyze steady flow at a small
angle on a body with a sharp trailing edge! Basically, the aerodynamic theory is encumbered with the Kutta condition; it is
applicable whenever the latter is befitting. It is clear that we lack a general theory for lift computation—equivalently a general
closure principle, alternative to Kutta’s, for potential flow. The above discussion is intimately related to the lifting mechanism.
The accepted wisdom by the fluid mechanics community asserts that the Kutta condition is a manifestation of viscous effects—it
implicitly accounts for viscous effects in a potential-flow formulation. In fact, this view has been challenged by several authors
(6, 27), but without providing a mathematical proof or theoretical basis. It may be the right time to recall Chang’s 2003
New York Times article: What Does Keep Them Up There? (28): ”To those who fear flying, it is probably disconcerting that
physicists and aeronautical engineers still passionately debate the fundamental issue underlying this endeavour: what keeps
planes in the air?".
In this paper, we revive a special variational principle from the history of analytical mechanics: Hertz’ principle of least
curvature. Exploiting such a variational principle, we develop a novel variational formulation of Euler’s equations for the
dynamics of ideal fluids where the Appellian (integral of squared acceleration) is minimized. The new variational formulation is
fundamentally different from the several developments based on Hamilton’s principle of least action (29–38). It even transcends
Euler’s equation in some singular cases where the latter does not provide a unique solution, such as the airfoil problem.
Applying the developed variational formulation to the century-old problem of the flow over an airfoil, we develop a new
variational theory of lift that dispenses with the Kutta condition. The Hertz principle of least curvature is used, as a first
principle, to develop, for the first time, a closure condition alternative to Kutta’s for the potential flow over an airfoil. In
contrast to the Kutta condition, the developed closure condition is derived from first principles. The new variational condition
reduces to the Kutta condition in the special case of a sharp-edged airfoil, which challenges the accepted wisdom about

T
the Kutta condition being a manifestation of viscous effects. Rather, it is found that the Kutta condition is a momentum
conservation mechanism for the inviscid flow over a sharp-edged airfoil. Moreover, the developed theory, unlike the classical
theory, allows treatment of smooth, singularity-free cylinders. F
2. Lack of Dynamical Features in the Current Theory of Aerodynamics: Variational Formulation is the Solution
To solve for the flow field of an incompressible fluid, both the continuity (kinematics) and momentum (dynamics) equations must
RA
be solved simultaneously. However, in potential flow, the governing equation is the Laplacian in the velocity potential (∇2 φ = 0),
which is obtained by combining the continuity equation (a divergence-free constraint: ∇ · u = 0) with an irrotational-flow
assumption (a curl-free constraint: ∇ × u = 0). These are kinematic constraints on the velocity field u. Then, the momentum
equation (Bernoulli’s in this case) is considered next to solve for the pressure field. That is, in potential flow, the flow kinematics
is decoupled from the flow dynamics; the velocity field is determined from purely kinematic analysis without any consideration
for dynamical aspects. Therefore, it is fair to expect that such a pure kinematic analysis is not sufficient to uniquely determine
the flow field ∗ ; the fix must come from a dynamical consideration.
D

Based on the above discussion, a proper closure condition (i.e., a condition that provides circulation dynamics) in potential flow
must come from dynamical considerations. The challenge is: Can we project Euler’s dynamical equations on a one-dimensional
manifold to extract the dynamics of circulation alone?
Dynamical equations of motion can be determined either from a Newtonian mechanics perspective or an analytical mechanics
one. The former stipulates isolating fluid particles and writing the equations of motion for each individual particle even if
the free variables in the system are significantly fewer than the total number of spatial coordinates of all individual particles
due to kinematic or geometric constraints. On the other hand, the analytical (Lagrangian or variational) mechanics approach
allows accepting kinematical constraints, ignoring the unknown forces that maintain them, and hence focusing on the relevant
equations of motion; it provides directly the relevant equations of motion for the free variables by minimizing a certain objective
function.
Projecting this discussion on the potential-flow case, one finds that the kinematical constraints of potential flow allow
construction of the entire flow field u from the circulation free variable Γ only: u = u(x; Γ). That is, while there are infinite
degrees of freedom for the infinite fluid particles, there is only one free variable (the circulation) which, via the potential-flow
kinematical constraints, can be used to recover the motion of these infinite degrees of freedom. Hence, the analytical/variational
mechanics appears to be specially well-suited for this problem; it will provide a single equation for the unknown circulation
without paying attention to the irrelevant degrees of freedom of the fluid particles or the unknown forces that maintain
kinematical constraints. Simply, the first variation of the “objective function" with respect to circulation must vanish—and this
necessary condition should provide a single dynamical equation in the unknown circulation.
Based on the above discussion, two important conclusions are drawn: (i) A true closure/auxiliary condition for potential flow
must come from dynamical considerations; and (ii) Variational principles would be particularly useful to derive such dynamics.
Insofar this story looks appealing, the question has always been: What is a suitable variational principle for the airfoil problem?
What is the special quantity that is being minimized in every flow over a two-dimensional body?
∗
This is not the case for three-dimensional flows because the domain enclosing a 3D surface is simply connected while that around a 2D object is doubly connected: a closed circuit around the 3D surface
is a reducible circuit while that around a 2D object is irreducible; there is no means to determine the circulation of an irreducible circuit.

2 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX et al.
 3. Theoretical Mechanics Approach: Gauss’ Principle of Least Constraint and Hertz’ Principle of Least Curvature
There have been several variational formulations of Euler’s equations; almost all of them are based on Hamilton’s principle
of least action (29–38). However, Hamilton’s principle suffer from some drawbacks that precludes its applicability to the
airfoil problem. In particular, Hamilton’s principle is a time-integral variational principle. That is, it provides the dynamics
over a period of time; hence, it may not be applicable to a steady snapshot of a flow field. Also, Hamilton’s principle is
exactly equivalent to Newton’s equations, so it does not provide more information in the singular cases where Newtonian (and
Lagrangian) mechanics fail to determine a unique solution.
In fact, the search for a suitable variational formulation for the airfoil problem is challenging. For example, minimizing the
kinetic energy over the field yields trivial (zero circulation) at any angle of attack. We found that the deserted principle of least
constraint by Gauss provides a felicitous formulation for the current problem † .
Consider the dynamics of N particles, each of mass mi , which are governed by Newton’s equations

mi ai = F i + Ri ∀i ∈ {1, .., N }, [1]

th
where ai is the inertial acceleration of the i particle, and the right hand side represents the total force acting on the particle,
which is typically decomposed in analytical mechanics into: (i) impressed forces F i , which are the directly applied (driving) forces
(e.g., gravity, elastic, viscous); and (ii) constraint forces Ri whose raison d’etre is to maintain/satisfy kinematical/geometrical
constraints; they are passive or workless forces (40) ‡ . That is, they do not contribute to the motion abiding by the constraint;
their sole role is to preserve the constraint (i.e., prevent any deviation from it).
Inspired by his method of least squares, Gauss asserted that the deviation of the actual motion a from the impressed one F m
is minimum (41). That is, the quantity
N 2
1 Fi
X 
Z= mi − ai [2]
2 mi
i=1

T
is minimum (39, pp. 911-912). Several points are worthy of clarification here. First, Gauss principle is equivalent to (derivable
from) Lagrange’s equations of motion (39, pp. 913-925), so we emphasize that it bears the same truth and status of first
principles (Newton’s equations). Second, in Gauss’ principle, J is actually minimum, not just stationary. Third, unlike the
F
time-integral principle of least action, Gauss’ principle is applied instantaneously (at each point in time). So, it can be applied
to a particular snapshot. Fourth, in some of the singular cases (when the dimension of the tangent space changes), Newtonian
and Lagrangian mechanics (as well as the principle of least action) may fail to determine a unique solution. In contrast, Gauss’
RA
principle is capable of providing a unique solution, see (39), pp. 923; and (42), pp. 71-72.
In the case of no impressed forces
mi ai = Ri ∀i ∈ {1, .., N },
Gauss’ principle reduces to Hertz’ principle of least curvature, which states that the Appellian
N
1
mi a2i
X
S= [3]
2
D

i=1

is minimum. In this case, because kinetic energy is conserved, it can be shown that the system curvature is minimum (39, pp.
930-932). That is, a free (unforced) particle moves along a straight line. If it is a constrained motion, then it will deviate from
a straight line to satisfy the constraint, but the deviation from the straight line path (i.e., curvature) would be minimum.

4. Novel Variational Formulation of the Dynamics of Ideal Fluids

Recall the Euler equations for incompressible flows

ρa = − ∇p, in Ω [4]

subject to continuity

∇ · u =0, in Ω [5]

and the no-penetration boundary condition

u · n =0, on ∂Ω, [6]

where Ω is the spatial domain, ∂Ω is its boundary, n is normal to the boundary, and a = ∂∂tu + u · ∇u is the total acceleration
of a fluid particle.
†
Papastavridis wrote: “In most of the 20th century English literature, GP [Gauss Principle] has been barely tolerated as a clever but essentially useless academic curiosity, when it was mentioned at
all"(39)
‡
For time-varying constraints, only the virtual work of the constraint forces vanishes, but not necessarily their actual work (39, pp. 382-383)

et al. PNAS | October 19, 2021 | vol. XXX | no. XX | 3

 Equation [4] presents Newton’s equations of motion for the fluid parcels. For inviscid flows, neglecting gravity, the only
acting force on a fluid parcel is the pressure force ∇p. In order to apply Gauss’ principle, we must determine whether this
force is an impressed force or a constraint force. Interestingly, for incompressible flows, it is the latter. The sole role of the
pressure force in incompressible flows is to maintain the continuity constraint: the divergence-free kinematic constraint on the
velocity field (∇ · u = 0). It is straightforward to show that if u satisfies Eqs. [5,6], then (43, pp. 261)
Z
(∇p · u)dx = 0, [7]
Ω

which indicates that pressure forces are workless through divergence-free velocity fields that are parallel to the surface; i.e.,
satisfying [6]. That is, if a velocity field satisfies continuity (i.e., divergence-free) and the no-penetration boundary condition
[6], the pressure force would not contribute to the dynamics of such a field. This fact is the main reason behind vanishing the
pressure force in the first step in Chorin’s standard projection method for incompressible flows (44); when the equation of motion
is projected onto divergence-free fields, the pressure term disappears, which is based on the Helmholz-Hodge decomposition (e.g.,
(43, 45)): a vector v ∈ R3 can be decomposed into a divergence-free component u that is parallel to the surface; i.e., satisfying
[6], and a curl-free component ∇f for some scalar function f (i.e., v = u + ∇f ). These two components are orthogonal as
shown in Eq. [7].
From the above discussion, it is clear that the pressure force is a constraint force and the dynamics of ideal fluid parcels are
subject to no impressed forces§ . Hence, Gauss’ principle of least constraint reduces to Hertz’ principle of least curvature in this
case.
Considering the dynamics of an ideal fluid [4], and labeling fluid parcels with their Lagrangian coordinates ξ (initial
positions), we write the Appellian as Z
1
S= ρ0 a2 dξ,
2
where ρ0 = ρ0 (ξ) is the initial density. Realizing that ρJ = ρ0 , where J is the Jacobian of the flow map (30), the Appellian is

T
rewritten in Eulerian coordinates as Z
1 2
S= ρa dx, [8]
FΩ
2
which must be minimum. As such, the dynamics of an ideal fluid can be represented in the Newtonian-mechanics formulation
by Eqs. [4, 5, 6]. We present an equivalent analytical-mechanics (variational) formulation:
RA
Z
1
min S = ρ a2 dx, [9]
2 Ω

subject to continuity

∇ · u =0, in Ω [10]

and the no-penetration boundary condition

D

u · n =0, on δΩ, [11]

5. Dynamical Closure Condition: A Variational Theory of Lift

Consider the standard potential flow over an airfoil (e.g., (1)). It is straight forward to determine a flow field u0 that is (i)
divergence-free, (ii) irrotational, and (iii) satisfies a given (possibly non-homogeneous) no-penetration boundary condition:
u0 · n = g(x), on δΩ, [12]
where g is a given function. This flow field is a solution of Euler’s equation [4]. Let u1 be any velocity field that is (i)
divergence-free, (ii) irrotational, and (iii) satisfies a homogeneous no-penetration boundary condition [11]; e.g., for the cylinder,
u1 can be the potential velocity field due to a unit circulation located at the center of the cylinder. Then, for any arbitrary Γ,
the flow field u(x; Γ) = u0 u(x) + Γu1 u(x) is also a legitimate Euler’s solution for the problem: it satisfies continuity, Euler’s
equation [4] and the given no-penetration boundary condition [12]. That is, Euler’s equation does not possess a unique solution
for this problem (it is a singular case where Newtonian and Lagrangian mechanics fail to determine a unique solution). In
contrast, the developed variational principle is capable of providing a unique solution. In fact, the dynamics of the free-field
Γu1 is exactly amenable to the developed variational principle since it (i) is divergence-free, and (ii) satisfies a homogeneous
no-penetration boundary condition [11]; that is, the pressure force is indeed orthogonal (in the sense of function spaces) to this
field, as shown in Eq. [7], and hence, it evolves freely (under no forces) with a minimum curvature according to Hertz’ principle.
Considering a steady snapshot (i.e., a = u · ∇u), we write the Appellian from [8] as
Z
1
S(Γ) = ρ [u(x; Γ) · ∇u(x; Γ)]2 dx. [13]
2 Ω
§
This will not be the case for a non-homogeneous, normal-flow boundary condition

4 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX et al.
 D=1

0.02 D=0.5

D=0.05
D=0.01

0 D=0

-0.02

T
0.95 1
(a) (b)
F
Normalized Appellian

1.0
RA
K

0.5
K

0.0
D

3 6 9 12 15
0.005 0.05 1.0
Normalized Circulation
(c) (d)

Fig. 1. The first row shows the effect of the shape-control parameter D : D = 0 results in a a Zhukovsky airfoil with a sharp trailing edge; the larger the D , the smoother
the trailing edge; and D = 1 results in a smooth circular cylinder. Unlike the classical theory, the current theory is capable of predicting lift over this spectrum of shapes.
The second row shows the results of the developed variational theory. Panel (c) shows the variation of the normalized Appellian Ŝ = S 4 with the normalized circulation
ρU
180 Γ
Γ̂ = at different angles of attack for the flow over a modified Zhukovsky airfoil with a smooth trailing edge (D = 0.05). For each angle of attack, there is a unique
π 4πU b
∗
value Γ of circulation that minimizes the Appellian: the dynamically-correct circulation according to Hertz’ principle of least curvature. Although Kutta’s circulation ΓK is not
relevant in this case of a smooth trailing edge, it is shown for comparison. Panel (d) shows contour plots of the Appellian versus the shape-control parameter D and the
circulation (normalized by Kutta’s). Each point in the D − Γ plane represents a flow field, not necessarily a physical flow field; Nature selects a unique flow field for each
geometry (i.e., for each D ): the one with the minimum curvature (minimum Appellian). The yellow curve presents the locus of such minimum-curvature choices. For D = 0
(i.e., a sharp-edged airfoil, the minimizing circulation Γ∗ coincides with Kutta’s ΓK ; and for D = 1 (circular cylinder), the minimizing circulation vanishes, implying no inviscid
lifting capability of this purely symmetric shape. That is, the developed theory could capture the spectrum between the two extremes: from a zero lift over a circular cylinder to
the Kutta-Zhukovsky lift over a sharp-edged airfoil. And the fact that the Kutta condition is a special case of the developed inviscid theory challenges the accepted wisdom about
the viscous nature of the Kutta condition. It is not a manifestation of viscous effects, but rather momentum conservation.

et al. PNAS | October 19, 2021 | vol. XXX | no. XX | 5

And the minimization principle [9], derived from Hertz’ principle of least curvature, yields the circulation over the airfoil as
Z
1
Γ∗ = argmin ρ [u(x; Γ) · ∇u(x; Γ)]2 dx. [14]
2 Ω

Equation [14] provides a generalization of the Kutta-Zhukovsky condition that is, unlike the latter, derived from first principles:
Hertz’ principle of least curvature. In contrast to the classical theory, the current one is not confined to sharp-edged airfoils.
Consider a circular cylinder of radius b in the ζ-domain that is mapped to a modified Zhukovsky airfoil of chord length c in
the z-domain through the mapping
1 − D δ2
z=ζ+ ,
1+D ζ
where δ is a constant that depends on the airfoil geometry (maximum thickness and camber) as well as the parameter D, which
controls smoothness of the trailing edge: D = 0 results in the classical Zhukovsky airfoil with a sharp trailing edge, and D = 1
results in a circular cylinder, as shown in Fig. 1(a,b). The airfoil is subject to a stream of an ideal fluid of density ρ with a free
stream velocity U at an angle of attack α.
Figure 1(c) shows the variation of the Appellian as given by Eq. [13] and normalized by ρU 4 versus the normalized circulation
Γ̂ = 180 Γ
π 4πU b
(i.e., the free parameter) at various angles of attack for the flow over a modified Zhukovsky airfoil with a trailing
edge radius of 0.1% chord length (D = 0.05). The figure also shows Kutta’s circulation ΓK = 4πbU sin α (i.e., Γ̂K ' α◦ for
small angles). Note that Kutta’s solution is not really applicable here. The figure shows that at a given angle of attack, the
Appellian possesses a unique minimum at a specific value of the circulation—the dynamically-correct circulation according to
Hertz’ principle.
Figure 1(d) shows contours of the normalized Appellian (in logarithmic scale) in the D-Γ space at α = 5◦ (the picture is
qualitatively similar at other angles of attack). The figure also shows the locus of the minimizing circulation Γ? (i.e., the
variation of Γ? with D). Interestingly, for a sharp-edged airfoil (D = 0), the minimizing circulation coincides with Kutta’s

T
circulation (i.e., Γ? = ΓK ), which implies that the developed minimization principle (14) reduces to the Kutta condition in the
special case of a sharp-edged airfoil. Also, the figure shows that the smoother the trailing edge, the smaller the circulation
(and lift); i.e., an airfoil with a sharp trailing edge generates larger lift than an airfoil with a smooth trailing edge at the same
F
conditions. In fact, the figure presents the other limiting case (a circular cylinder: D = 1) where the classical result of the
non-lifting nature of a circular cylinder in an inviscid fluid is recovered.
RA
6. Discussion
The last section presents a new theory of lift based on Hertz’ variational principle of least curvature. In particular, Eq. [14]
presents the sought general fundamental principle that provides closure in potential flow based on first principles (Hertz’
principle of least curvature), thereby generalizing the century-old theory of Kutta and Zhukovsky. This principle allows, for the
first time, computation of lift over smooth shapes without sharp edges where the Kutta condition fails, which confirms that a
sharp trailing edge is not a necessary condition for lift generation (46, 47). That is, The new variational theory, unlike the
classical one, is capable of capturing the whole spectrum between the two extremes: from zero lift over a circular cylinder to
D

the Kutta-Zhukovsky lift over an airfoil with a sharp trailing edge, as shown in Fig. 1(d). This behavior provides credibility to
the developed theory.
The fact that the minimization principle [14] reduces to the Kutta condition in the special case of a sharp-edged airfoil,
wedded to the fact that this principle is an inviscid principle imply that the classical Kutta-Zhukovky lift over an airfoil with a
sharp edge is both computed and explained from inviscid considerations. This finding challenges the accepted wisdom about the
Kutta condition being a manifestation of viscous effects. Rather, it represents inviscid momentum conservation. That is, the
circulation (and lift) is the one that satisfies momentum conservation—equivalently, it is the one that minimizes the Appellian
in the language of Hertz. This result explains the several inviscid computations that converged to the Kutta-Zhukovky lift
without viscosity (6, 48–50).
The developed theory and obtained results are intimately related to the ability of an ideal fluid (superfluid) to generate lift.
So, it may be prudent to discuss the seeming contradiction between Craig and Pellam’s experimental result (51) and the recent
computations by Musser et al. (50), in the light of the developed theory. Craig and Pellam have experimentally studied the
flow of an ideal fluid (superfluid Helium II) over a flat plate and an ellipse and found a vanishingly small lift force over these
objects at non-zero angles of attack. Hence, they concluded that superfluids are non-lifting and that “the classical viscosity
boundary condition at the trailing edge (the Kutta condition) does not apply" (52)—the last part of this statement about the
viscous nature of the Kutta condition is refuted above.
In contrast to the above hypothesis, Musser et al. (50) recently performed quantum simulations of the Gross-Pitaevskii
equation governing the flow dynamics over an airfoil in a superfluid. Their simulations show a quantized version of the
Kutta-Zhukovsky lift despite the lack of viscosity in their simulations. However, the continuum hypothesis may not be
applicable in their ultra-small-scale simulations.
From the above discussion, there seems to be contradiction between Craig and Pellam’s experiment (51) and the computations
of Musser et al. (50): the former deceitfully implies that superfluids are non-lifting and the Kutta condition is a viscous
condition (i.e., the physics is discontinuous in the limit), while the latter shows the ability of a superfluid to generate the
Kutta-Zhukovsky lift similar to a viscous flow of a vanishingly small viscosity (i.e., the physics is continuous in the limit). In

6 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX et al.
the light of the developed theory, we find no contradiction between the two results. In fact, Craig and Pellam conducted their
experiment on symmetric shapes (a flat plate and an ellipse) for which the current theory predicts no lift, which confirms their
experimental findings (but not their conclusions). Also, the current theory predicts Kutta-Zhukovsky lift over an airfoil with a
sharp trailing edge; and a smaller but non-zero lift over a smooth airfoil (or asymmetric shape), which confirms Musser et al.
computational simulations (50). Therefore, it is concluded that a purely symmetric shape in a superfluid is non-lifting; however,
any asymmetry would grant some lifting capability. The current theory, as well as Musser et al. quantum simulations (50),
invoke an experimental study of the flow of a superfluid (Helium II), similar to Craig and Pellam’s (51), but over a traditional
airfoil shape (i.e., asymmetric shape).
The fact that purely symmetric shapes are non-lifting in an ideal (non-dissipative) fluid is physically plausible, as shown
in Fig. 2(a) presenting the flow of an ideal fluid over a flat plate as predicted by the current theory. These flows must be
reversible to conserve entropy in the dissipation-free environment. Therefore, for reversibility and because of symmetry, one
should not be able to tell whether the flow field, shown in Fig. 2(a), is for a free stream coming from the left at a positive α or
from the right at a negative α. Clearly, this symmetric solution is non-lifting.
For these symmetric shapes, viscosity is important to enable the weak lift over these bodies: the slight change of the effective
body shape due to boundary layer destroys symmetry; the outer inviscid flow over the modified asymmetric body is now lifting.
The flat plate represents an extreme case in this regard: the effect of viscosity is significant due to the singular nature of the
corresponding ideal flow field. Viscosity leads to separation at the leading edge even at small angles of attack, as shown in Fig.
2(b), which presents our Detached Eddy Simulation (DES) of the flow over a flat plate at a Reynolds number of 500,000 and
α = 2◦ , also see Van Dyke’s Album of Fluid Motion (53). A leading-edge separation bubble is clearly seen on the top of the
flat plate in the immediate vicinity of the leading edge. The outer inviscid flow field outside the separation bubble (i.e., over
the modified flat plate: a flat plate with a naturally rounded nose) is lifting; the minimization principle [14], which is equivalent
to the Kutta condition in this case of a sharp-trailing edge, results in a circulation Γ? that is close to Kutta’s.
The developed theory is expected to deepen our understand-
ing of one of the most fundamental concepts in aerodynamics:

T
lift generation over an airfoil.
Finally, it may be prudent to provide a simple explanation
for lift generation in the light of the obtained results and
presented discussion—an explanation that does not invoke
F complex unsteady phenomena to illustrate how/why a par-
ticular steady lift value is attained. It is found that lift is
due to the triplet [4,5,6], equivalently [9,10,11]. In words,
RA
lift is due to (i) momentum conservation, (ii) continuity (the
fluid has to fill the given space), and (iii) the body’s hard
constraint (i.e., presence of the body inside the fluid), which
forces the flow to move around the body without creating
void to maintain (ii). That is, lift is due to conservation
of momentum between the two equilibrium states of the
Fig. 2. Flow velocity, u/U , of a superfluid (using the current theory) and a real fluid freestream and the curved flow around the body. In layman
D

(using Detached Eddy Simulations) over a flat plate at α = 2◦ . terms, the presence of the body inside the fluid forces fluid
particles to go along a curved path; i.e., it provides the
necessary centripetal force for the curved path. And the fluid responds back by an equal amount of force in the opposite
direction according to Newton’s third law—this reaction is the lift force. This explanation is similar to Hoffren’s (27) and
to some extent to Babinsky’s (54); the current paper provides the sought mathematical proof. Inventors and analysts alike
may find it intriguing that a “Dry Water” model (55) provides the main physical mechanism behind one of most fundamental
problems in fluid mechanics.
1. H Schlichting, E Truckenbrodt, Aerodynamics of the Airplane. (McGraw-Hill), (1979).
2. K Karamcheti, Principles of Ideal-Fluid Aerodynamics. (John Wiley & Sons), (1966).
3. LM Milne-Thomson, Theoretical hydrodynamics. (Courier Corporation), (1996).
4. L Rayleigh, On the irregular flight of a tennis ball. Messenger Math. 7, 14–16 (1877).
5. DG Crighton, The Kutta condition in unsteady flow. Annu. Rev. Fluid Mech. 17, 411–445 (1985).
6. J Hoffman, J Jansson, C Johnson, New theory of flight. J. Math. Fluid Mech. 18, 219–241 (2016).
7. OKG Tietjens, L Prandtl, Applied hydro-and aeromechanics: based on lectures of L. Prandtl. (Courier Corporation), (1934).
8. S Goldstein, Modern developments in fluid dynamics: an account of theory and experiment relating to boundary layers, turbulent motion and wakes. (Clarendon Press), (1938).
9. DS Woolston, GE Castile, Some effects of variations in several parameters including fluid density on tbe flutter speed of light uniform cantilever wings, (NACA), Technical Report 2558 (1951).
10. N Rott, MBT George, An approach to the flutter problem in real fluids, (Inst. of Aeronautical Sciences), Technical Report 509 (1955).
11. HN Abramson, HH Chu, A discussion of the flutter of submerged hydrofoils. J. Ship Res. 3 (1959).
12. WH Chu, HN Abramson, An alternative formulation of the problem of flutter in real fluids. J. Aerosp. Sci. 26 (1959).
13. WH Chu, An aerodynamic analysis for flutter in Oseen-type viscous flow. J. Aerosp. Sci. 29, 781–789 (1961).
14. CJ Henry, Hydrofoil flutter phenomenon and airfoil flutter theory, (Davidson Laboratory), Technical Report 856 (1961).
15. SF Shen, P Crimi, The theory for an oscillating thin airfoil as derived from the Oseen equations. J. Fluid Mech. 23, 585–609 (1965).
16. HN Abramson, WH Chu, JT Irick, Hydroelasticity with special reference to hydrofoil craft., (NSRDC Hydromechanics Lab), Technical Report 2557 (1967).
17. SA Orszag, SC Crow, Instability of a vortex sheet leaving a semi-infinite plate. Stud. Appl. Math. 49, 167–181 (1970).
18. BC Basu, GJ Hancock, The unsteady motion of a two-dimensional aerofoil in incompressible inviscid flow. J. Fluid Mech. 87, 159–178 (1978).
19. PG Daniels, On the unsteady kutta condition. The Q. J. Mech. Appl. Math. 31, 49–75 (1978).
20. B Satyanarayana, S Davis, Experimental studies of unsteady trailing-edge conditions. AIAA J. 16, 125–129 (1978).
21. SB Savage, BG Newman, DTM Wong, The role of vortices and unsteady effects during the hovering flight of dragonflies. The J. Exp. Biol. 83, 59–77 (1979).
22. RL Bass, JE Johnson, JF Unruh, Correlation of lift and boundary-layer activity on an oscillating lifting surface. AIAA J. 20, 1051–1056 (1982).

et al. PNAS | October 19, 2021 | vol. XXX | no. XX | 7

23. H Wagner, Über die entstehung des dynamischen auftriebes von tragflügeln. ZAMM - J. Appl. Math. Mech. / Zeitschrift für Angewandte Math. und Mech. 5, 17–35 (1925).
24. T Theodorsen, General theory of aerodynamic instability and the mechanism of flutter, (NACA), Technical Report 496 (1935).
25. T Von Karman, WR Sears, Airfoil theory for non-uniform motion. J. Aeronaut. Sci. 5, 379–390 (1938).
26. PJ Baddoo, R Hajian, JW Jaworski, Unsteady aerodynamics of porous aerofoils. J. Fluid Mech. 913 (2021).
27. J Hoffren, Quest for an improved explanation of lift in 39th Aerospace Sciences Meeting and Exhibit. p. 872 (2000).
28. K Chang, Staying aloft; what does keep them up there? New York Times (2003).
29. R Hargreaves, Xxxvii. a pressure-integral as kinetic potential. The London, Edinburgh, Dublin Philos. Mag. J. Sci. 16, 436–444 (1908).
30. H Bateman, Notes on a differential equation which occurs in the two-dimensional motion of a compressible fluid and the associated variational problems. Proc. R. Soc. Lond. A 125, 598–618 (1929).
31. J Serrin, Mathematical principles of classical fluid mechanics in Fluid Dynamics I/Strömungsmechanik I. (Springer), pp. 125–263 (1959).
32. JW Herivel, The derivation of the equations of motion of an ideal fluid by hamilton’s principle in Mathematical Proceedings of the Cambridge Philosophical Society. (Cambridge University Press),
Vol. 51, pp. 344–349 (1955).
33. P Penfield Jr, Hamilton’s principle for fluids. The Phys. Fluids 9, 1184–1194 (1966).
34. JC Luke, A variational principle for a fluid with a free surface. J. Fluid Mech. 27, 395–397 (1967).
35. RL Seliger, GB Whitham, Variational principles in continuum mechanics in Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences. (The Royal Society),
Vol. 305, pp. 1–25 (1968).
36. FP Bretherton, A note on hamilton’s principle for perfect fluids. J. Fluid Mech. 44, 19–31 (1970).
37. R Salmon, Hamiltonian fluid mechanics. Annu. review fluid mechanics 20, 225–256 (1988).
38. MI Loffredo, Eulerian variational principle for ideal hydrodynamics and two-fluid representation. Phys. Lett. A 135, 294–297 (1989).
39. JG Papastavridis, Analytical mechanics: a comprehensive treatise on the dynamics of constrained systems – Reprint edition. (Word Scientific Publishing Company), (2014).
40. C Lanczos, The variational principles of mechanics. (Courier Corporation), (1970).
41. CF Gauß, Über ein neues allgemeines grundgesetz der mechanik. J. für die reine und angewandte Math. 1829, 232–235 (1829).
42. M Golomb, Lectures on Theoretical Mechanics (mimeographed notes by I. Marx). (Purdue University), (1961).
43. T Kambe, Geometrical theory of fluid flows and dynamical systems. Fluid dynamics research 30, 331–378 (2002).
44. AJ Chorin, Numerical solution of the navier-stokes equations. Math. computation 22, 745–762 (1968).
45. JZ Wu, HY Ma, MD Zhou, Vorticity and vortex dynamics. (Springer Science & Business Media), (2007).
46. J Stack, WF Lindsey, Tests of n-85, n-86 and n-87 airfoil sections in the 11-inch high speed wind tunnel, (NACA), Technical Report 665 (1938).
47. LJ Herrig, JC Emery, JR Erwin, Effect of section thickness and trailing-edge radius on the performance of naca 65-series compressor blades in cascade at low speeds, (NACA), Technical report
(1956).
48. A Rizzi, Damped euler-equation method to compute transonic flow around wing-body combinations. AIAA J. 20, 1321–1328 (1982).
49. H Yoshihara, H Norstrud, JW Boerstoel, G Chiocchia, DJ Jones, Test cases for inviscid flow field methods., (AGARD), Technical Report AR-211 (1985).
50. S Musser, D Proment, M Onorato, WTM Irvine, Starting flow past an airfoil and its acquired lift in a superfluid. Phys. review letters 123, 154502 (2019).
51. PP Craig, JR Pellam, Observation of perfect potential flow in superfluid. Phys. Rev. 108, 1109 (1957).
52. PP Craig, Ph.D. thesis (California Institute of Technology) (1959).

T
53. M Van Dyke, An Album of Fluid Motion. (Parabolic Press), (1982).
54. H Babinsky, How do wings work? Phys. Educ. 38, 497 (2003).
55. RP Feynman, The Feynman lectures on physics: Mainly electromagnetism and matter. (Addison-Wesley) Vol. 2, (1964).
F
RA
D

8 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX et al.