Chapter 2

Theorems of Helmholtz and Kelvin

2.1 Introduction

Helmholtz’ (1858) vortex theorems paved the way for the legendary Ludwig
Prandtl (1921) to invent the lifting line theory, a crowning advancement in theo-
retical aerodynamics. Sommerfeld (1950) remarked that the main contents of
Helmholtz’s theory are conservation laws: It is impossible to produce or destroy
vortices, or, in more general terms, the vortex strength is constant in time.
Sommerfeld noted that Helmholtz’ theorems are correct under the conditions the
fluid is inviscid and incompressible; the external forces possess a single-valued
potential within the space filled by the fluid. Sommerfeld also pointed out that in
Helmholtz’ theorems, apart from the conservation of the vortex strength in time,
there is also a spatial conservation: the vortex strength is constant along each
vortex line or vortex tube, which must either be closed or end at the boundary of the
fluid. In the current literature, Helmholtz’ theorem on the spatial conservation of the
vorticity strength is called Helmholtz’ first vortex theorem, and the theorem on the
time conservation is called his second vortex theorem.
Lighthill (1986) discussed Kelvin’s (1869) theorem on the persistence of cir-
culation and pointed out that Kelvin’s theorem is exact for the Euler model, which
contains the key assumptions that the fluid is inviscid and incompressible. Lighthill
noted: One especially valuable deduction from Kelvin’s theorem is concerned with
“the movement of vortex lines”. This is Helmholtz’s theorem (also exact on the
Euler model), which states that vortex lines move with the fluid.
In vorticity dynamics, the strategy of partitioning the overall flow problem into
its kinematic and kinetic aspects offers advantages in aerodynamic analyses (Wu
1981; Tuncer et al. 1990). In the present study, this strategy is used in a revisit of
the vortex theorems of Helmholtz and Kelvin and in generalizing these theorems to
flows of viscous fluids. It is learned from this revisit that Helmholtz’ first vortex
theorem on spatial conservation is a theorem of vorticity kinematics and, as such, is
valid not only in the fluid region but also in the solid region. This observation leads

© Shanghai Jiao Tong University Press, Shanghai and Springer-Verlag Berlin Heidelberg 2018 17
J.C. Wu, Elements of Vorticity Aerodynamics, Springer Tracts in Mechanical Engineering,
https://doi.org/10.1007/978-3-662-44040-7_2
18 2 Theorems of Helmholtz and Kelvin

to the portrayal of the vorticity distribution as a system of closed vorticity tubes.

With this portrayal, classical aerodynamic theories, including the lifting line theory,
are interpretable on the basis of the vorticity-moment theorem (Wu 1981, Chap. 6).
This portrayal also brings forth opportunities for establishing new approaches for
viscous and unsteady aerodynamic analysis (Wu et al. 2002, Chap. 7).
Helmholtz’ second vortex theorem, or its equivalence Kelvin’s theorem, is a
vorticity-dynamic theorem based on both kinetics and kinematics. The generalized
second vortex theorem states that the vorticity strength in the viscous fluid is not
conserved in time; it diffuses at a predictable rate.

2.2 Kinematic-Kinetic Partition of Flow Problems

As discussed in Chap. 1, the present study deals with the incompressible flow of a
viscous fluid as described by the following set of two differential equations:

rv¼0 ð2:1Þ

ov 1
¼ ðv  rÞv  rp þ mr2 v þ rq ð2:2Þ
ot q

where v is the flow velocity, p, q, and m are respectively the pressure, the density,
and the kinematic viscosity of the fluid, and ∇q represents a conservative body
force.
Equation (2.1), the continuity equation, is a mathematical statement of the law of
conservation of mass for the incompressible flow. Equation (2.2), Navier–Stokes’
momentum equation, expresses Newton’s second law of motion for the viscous
fluid. For the idealized inviscid fluid, the last term in (2.2) is zero and one has
Euler’s momentum equation for the incompressible flow

ov 1
¼ ðv  rÞv  rp þ rq ð2:3Þ
ot q

The overall problem of the incompressible flow of a viscous fluid is partitioned

into two aspects by ascribing (2.1) to kinematics and (2.2), or (2.3), to kinetics.
Kinematics and kinetics, the two branches of dynamics, are typically defined as
follows: Kinematics is the study of motions of themselves, apart from their causes.
Kinetics is the study of changes of motions produced by forces. In ascribing (2.1) to
kinematics, the view is taken that the law of mass conservation imposes a condition
on the incompressible motion of the fluid. Namely, the incompressible velocity field
must be divergence-free. This condition is imposed on the flow at each instant of
time, and is not regarded as a cause of change of the flow with time. This view is
generally accepted in classical studies of the incompressible flow.
2.2 Kinematic-Kinetic Partition of Flow Problems 19

In an elegant treatise, Truesdell (1954) stated: In general the flow of a fluid,

whether perfect or viscous, can be defined by kinematical conditions. In Truesdell’s
view, circulation preservation in time is one of the kinematical conditions on the
flow. He then presented extensive results based on this view. In the present study,
circulation preservation is viewed not as a kinematic condition, but as a kinetic
consequence. Based on this perspective, it is shown that, in the case of the viscous
flow, the kinetics as described by (2.2) causes the circulation to diffuse. However, in
regions of the flow where viscous effects are insignificant, (2.3) brings about cir-
culation preservation with time.
Scholars in fluid dynamics with different research focuses often interpret the
terms kinematics and kinetics differently. Differences in interpretations are, by
themselves, not important to the present study. The recognition that the overall flow
problem contains two aspects, each with its own physical and mathematical char-
acteristics is, however, crucially important. Viewing (2.1) as a constraint and (2.2)
as a cause–effect relationship makes it feasible to treat these two interlaced aspects
individually before uniting the individual results. The advantages of this strategy
are not obvious if (2.1) and (2.2) are both viewed as kinematic constraints. These
advantages become pronounced in studies of the flow of the viscous fluid using
concepts of vorticity dynamics.
The familiar definition of the vorticity field, as stated in Chap. 1, is

rv¼x ð2:4Þ

Equations (2.1), (2.2), and (2.4) constitute a set of vorticity-dynamic equations

describing the flow of the viscous fluid. Of these equations, (2.1) and (2.4) are
kinematic since they deal with motions without reference to forces that cause
motions. The kinetic Eq. (2.2) can be restated in terms of x as

ov
¼ rðh  qÞx  v  m r  x ð2:5Þ
ot

where h = p/q + (v  v)/2 is the total head.

Taking the curl of each term in (2.5) yields the vorticity-transport equation:

ox
¼ r  ðx  vÞ  mr  ðr  xÞ ð2:6Þ
ot

For the viscous fluid, the flow kinetics is represented by either (2.5) or (2.6).
Although (2.6) does not contain a force term explicitly, it is considered a kinetic
equation because it is a forthwith consequence of (2.2). This equation describes the
change of the vorticity field with time. The terms on the right-hand side of (2.6)
represent physical processes causing the change. For the inviscid fluid, the flow
kinetics is described by (2.7) or (2.8)
20 2 Theorems of Helmholtz and Kelvin

ov
¼ rh  x  v ð2:7Þ
ot
ox
¼ r  ðx  vÞ ð2:8Þ
ot

Several useful mathematical identities are derived in §2.3. These identities are
not new. They are included here to facilitate the discussions of the theorems of
Helmholtz and Kelvin. The reader wishing to verify the derivations may wish to
refer to vector identities listed in §3.2.

2.3 Kinematic Preliminaries

In the following discussions, sufficient smoothness of lines (curves) and surfaces in

space and sufficient differentiability of field variables are assumed so that the
divergence theorem, Stokes’ theorem, and other mathematical relations are
meaningful.
A curve in space whose tangent at each point on it is in the direction of a vector
field f is called a vector line of f. A surface comprising all the vector lines of
f passing through a circuit (a closed curve) in space is called a vector tube of
f. Vector lines and vector tubes of the velocity field v are called streamlines and
stream tubes. Vectors lines and vector tubes of the vorticity field x are called
vorticity lines and vorticity tubes.
A vector field f is said to be solenoidal if its divergence, ∇  f, is zero and
lamellar if its curl, ∇  f, is zero. A solenoidal velocity field v is said to be
incompressible and a lamellar velocity field irrotational. In the following discus-
sions, the terms irrotational and lamellar are used interchangeably: f is said to be
irrotational or lamellar if ∇  f = 0.
With Helmholtz’ decomposition, a general vector field f is expressible in the
form

f ¼ rg þ r  h ð2:9Þ

where g is a scalar potential function and h is a vector potential function. By virtual

of the vector identities ∇  (∇g) = 0 and ∇  (∇  h) = 0, ∇g is lamellar and
∇  h is solenoidal. Thus (2.9) decomposes the general vector field f into a
solenoidal field, denoted f* in this study, and a lamellar field denoted f′. A lamellar
field f′ is expressible as ∇g alone and a solenoidal field f* as ∇  h alone.
Consider two circuits C1 and C2 that lie on the same vector tube of f, each circuit
encircles the tube once. Let S1 be a cap of C1 and S2 a cap of C2. Denote the section
of the vector tube between C1 and c by Se. The three surfaces S1, S2, and Se together
2.3 Kinematic Preliminaries 21

form a closed surface So bounding a simply connected region Ro. The divergence
theorem then gives
ZZ ZZ ZZ ZZ
f  n dS þ f  n dS þ f  n dS ¼ r  f dR ð2:10Þ
S1 S2 Se R0

where n is the unit normal vector on the surfaces So directed outward from Ro.
The field f  n is a measure of the strength of the normal component of f on S and
is called the flux of f on S. Since f is tangent to the vector tube, f  n = 0 on Se. For
the solenoidal field f*, the right-hand side of (2.10) is zero. Let n1 and n2 be two
unit normal vectors on S1 and S2 respectively, with their positive directions given
by the axial direction of the vector tube. With this orientation, if n1 = n on S1, then
n2 = − n on S2. On the other hand, if n1 = − n, then n2 = n on S2. In either case,
one obtains from (2.10)
ZZ ZZ
f  n1 dS þ f  n2 dS ð2:11Þ
S1 S2

The two integrals in (2.11) are called respectively the total flux of the field f*
across S1 and S2 or the strength of f* of the vector tube at the S1 and S2. This equation
states that the strength of the vector tube of f* is constant along the path of the tube.
Equation (2.11) expresses a spatial, or kinematic, conservation relationship. If the
solenoidal vector field f* is time-dependent, then the spatial conservation is valid at
each instant of time. Representing f* by ∇  h, one obtains, using Stokes’ theorem
Z ZZ
Þ h  s ds ¼ f   n dS ð2:12Þ
C S

In (2.12), s is a unit tangent vector on the circuit C. The positive senses of n and
s are related by the right-handed screw convention. Namely, as a right-handed
screw turns in the positive s direction, it advances in the positive n direction. The
left-hand side of (2.12) is called the circulation of h around C. Equation (2.12)
states the total flux of a solenoidal field f* across S is equal to the circulation of the
vector potential h of f* around C, of which S is a cap. The three terms total flux
across a surface, strength of vector tube, and circulation are equivalent. These
terms are used interchangeably in the literature, each describing the surface integral,
or its equivalent line integral, in (2.12).
Let C be a time-dependent circuit. Denote the positions of C at the time levels
t1 = t and t2 = t + dt respectively by C1 and C2. Consider the time derivative of the
integration of f  s over C, where f is time-dependent. One writes
22 2 Theorems of Helmholtz and Kelvin

82 3 9
Z >
< Z Z >
=
d 6 7
Þ f ðr; tÞ  s ds  lim 4 Þ f ðr; t þ dtÞ  s ds  Þ f ðr; tÞ  s ds5=dt
dt dt!0>
: >
;
C C2 C1
82 3 9
Z   >
< Z Z >
=
f ðr; t þ dtÞ  f ðr; tÞ 6 7
¼ lim Þ  s ds þ lim 4 Þ f ðr; tÞ  s ds  Þ f ðr; tÞ  s ds5=dt
dt!0 dt dt!0>
: >
;
C2 C2 C1

ð2:13Þ

Let S1 be a cap of C1 and S2 a cap of C2. Using Stokes’ theorem, the last two
integrals in (2.13) can be restated as integrations of (∇  f)  ni, i = 1 or 2, over the
surfaces Si. Let Se be a surface joining C1 and C2 at the time level t. The three
surfaces S1, S2, and Se together form a closed surface So bounding the region Ro. If
s is in a direction such that n1 = n, the outward unit normal vector on S1, then, with
the right-handed screw convention, n2 = − n on S2. Thus
Z Z ZZ ZZ
Þ f ðr; tÞ  s ds  Þ f ðr; tÞ  s ds ¼   ðr  f Þ  n dS þ ðr  f Þ  n dS
C2 C1 S0 Se

ð2:14Þ

The first integral on the right-hand side of (2.14) is an integral over the closed
surface So. This integral can be restated, using the divergence theorem, as an
integration of ∇  (∇  f) over Ro and is therefore zero. On Se, as dt ! 0, one has
ndS ! (vcdt)  (sds), where vc is the velocity of C1. One therefore has
ZZ Z Z
ðr  f Þ  n dS ¼ dt Þ ðr  f Þ  ðvc  sÞdS ¼ dt Þ ½ðr  f Þ  vc   s dS
Se C1 C1

ð2:15Þ

Placing (2.15) into (2.14), and the result into (2.13), one obtains, upon noting
that, as dt ! 0, the integrand in the first term on right-hand side of (2.13) gives (∂f/
∂t)  s,
Z Z   Z
d of
Þ f ðr; tÞ  s ds ¼ Þ  s ds þ Þ ½ðr  f Þ  vc   s ds ð2:16Þ
dt ot
CðtÞ C C

where the superfluous subscripts for the circuit C are dropped. The symbol C(t) is
used, rather than C, in the left-hand side of (2.16) to emphasize the time depen-
dency of the path of integration. The two integrals on the right-hand side of (2.16)
are not differentiated with respect to time. Therefore the time dependency of the
integration path C is not an issue of concern. The line integral of f  s over a moving
(and/or deforming) loop C(t) changes with time because of two contributing factors:
2.3 Kinematic Preliminaries 23

the local change of f on C and the motion of the loop C. The right-hand side
integrals in (2.16) express separately the contributions.
For the lamellar vector field f′ = ∇g, the last term in (2.16) vanishes since
∇  (∇g) = 0. The integral on the left-hand side of (2.16) vanishes since it can be
restated as an integral of (∇  f′)  n over a cap S of C. The first term on the
right-hand side of (2.16) is similarly zero because ∂f′/∂t = ∇(∂g/∂t). Therefore
(2.16) is trivial for the lamellar field. Only the solenoidal part of the general vector
field f, as represented by (2.9), plays a role in (2.16).
Consider the time derivative of the total flux of f(r,t) across S(t)
82 3 9
ZZ < ZZ ZZ =
d
f ðr; tÞ  n1 dS  lim 4 f ðr; t þ dtÞ  n2 dS  f ðr; tÞ  n1 dS5=dt
dt dt!0: ;
ZZ  
S S2 S1
f ðr; t þ dtÞ  f ðr; tÞ
¼ lim  n2 dS
dt!0 dt
8S22 3 9
< ZZ ZZ =
þ lim 4 f ðr; tÞ  n2 dS  f ðr; tÞ  n1 dS5=dt
dt!0: ;
S2 S1
ð2:17Þ

With the surface Se defined earlier, one obtains, using the divergence theorem
ZZ ZZ ZZZ ZZ
f  n2 dS  f  n1 dS ¼  r:f dR þ f  ne dS ð2:18Þ
S2 S1 R0 Se

As dt ! 0, dR ! − (vs1dt)  n1dS and nedS ! (vc  s)dsdt, where vs1 is the

velocity of the cap S1 and vc is the velocity of the circuit C1. One thus obtains
ZZ ZZ ZZ Z
 
f  n2 dS  f  n1 dS ¼ dt ðr  f Þðvs :nÞdS þ dt Þ f: vC  s dS ð2:19Þ
S2 S1 S1 C1

The integrand in the last integral of (2.19) can be restated as (f  vc)  s. Thus
Stokes’ theorem yields
Z ZZ
Þ fðvc  sÞds ¼ r  ðf  vs1 Þ  n dS ð2:20Þ
C1 S1

Placing (2.20) into (2.19) and the results into (2.17), one obtains, upon noting
the integrand in the first integral in the right-hand side of (2.17) is ∂f/∂t and
dropping the subscripts for S:
24 2 Theorems of Helmholtz and Kelvin

ZZ ZZ ZZ ZZ
d of
f ðr; tÞ  n dS ¼  n dS þ ½ðr  f Þvs   n dS þ ½r  ðf  vs Þ  n dS
dt ot
SðtÞ S S S

ð2:21Þ

The total flux of f over S(t) changes with time because the field f is
time-dependent and also because the surface S is time-dependent. With (2.21), the
time rate of change of the flux of f over S(t) is given by the sum of the three
integrals on the right-hand side. The first integral represents the local rate of change
of f on S. The second and third integrals together represent the motion and
deformation of S.
For the solenoidal field f* = ∇  h, (2.21) reduces to
ZZ ZZ ZZ
d  of 
f  ndS ¼  ndS þ r  ðf   vs Þ  ndS ð2:22Þ
dt ot
SðtÞ S S

Consider the time derivative of the integral of f(r, t) over S(t), which is a cap of C(t)
82 3 9
ZZ < ZZ ZZ =
d
f ðr; tÞdS  lim 4 f ðr; t þ dtÞdS  f ðr; tÞdS5=dt
dt dt!0: ;
S S2 S1
82 3 9
ZZ   < ZZ ZZ =
f ðr; t þ dtÞf ðr; tÞ
¼ lim dS þ lim 4 f ðr; tÞdS  f ðr; tÞdS5=dt
dt!0 dt dt!0 : ;
S2 S2 S1

ð2:23Þ

Through analyses similar to those leading to (2.19) and (2.21), one obtains
ZZ ZZ Z
f ðr; tÞdS f ðr; tÞdS ¼ dt Þ f ðvc  nÞds ð2:24Þ
S2 S1 C1

ZZ ZZ Z
d of
f ðr; tÞdS ¼ dS þ Þ f ðvc  nÞds ð2:25aÞ
dt ot
SðtÞ S C1

ZZ  
of
¼ þ ðr  vs Þf þ ðvs  rÞf dS ð2:25bÞ
ot
S

A similar identity for the time derivative of the integral of f(r,t) over the
three-dimensional region R(t) bounded by S(t) is
ZZZ ZZZ ZZ
d of
f ðr; tÞdR ¼ dR þ  f ðvs  nÞdS ð2:26aÞ
dt R ot
RðtÞ S
2.3 Kinematic Preliminaries 25

ZZZ  
of
¼ þ ðr  vv Þf þ ðvv  rÞf dR ð2:26bÞ
ot
R

where vv is the velocity of points in R and vs is the velocity of points on the

boundary S.
Equations (2.25a) and (2.26a) remain valid with the vector field f(r, t) replaced
by the scalar field f(r, t). The mathematical identities presented in this §2.3 are valid
independently of the material medium occupying the space R. Whether a solid
body, a viscous fluid, an inviscid fluid, or a combination of these media is present in
R does not alter these identities. They are kinematic identities in which C needs not
be a material circuit and S needs not be a material surface. vc, vs, and vv represent
the motions of the circuit, the surface, and points in R and need not be the flow
velocity v.

2.4 Helmholtz’ First Vortex Theorem

In (2.11), let f* be the incompressible flow velocity v. One obtains

ZZ ZZ
v  n1 dS ¼ v  n2 dS ð2:27Þ
S1 S2

Equation (2.27) states a well-known fact: The flow rate (total velocity flux)
through a stream tube in an incompressible flow is constant.
Let f* be v and h be the vector potential w of v in (2.12). One has
ZZ Z
v  n dS ¼ Þ w  s ds ð2:28Þ
S C

Equation (2.28) states that the total rate of an incompressible flow across the
area S is identical to the circulation of the vector potential w around the circuit C, of
which S is a cap. Equation (2.27) therefore states that the circulation of the vector
potential w around a stream tube in the incompressible flow is constant along the
path of the stream tube.
Let f* be x in (2.11). One obtains
ZZ ZZ
x  n1 dS ¼ x  n2 dS ð2:29Þ
S1 S2
26 2 Theorems of Helmholtz and Kelvin

Let f* be x and h be v. Equation (2.12) yields

ZZ Z
x  n dS ¼ Þ v  s ds ð2:30Þ
S C

Equation (2.30) states that the total vorticity flux across S is identical to the
circulation (of velocity) around C, of which S is a cap. Therefore (2.29) states: The
circulation of a vorticity tube is constant along each vorticity tube. This statement
paraphrases the first part of Sommerfeld’s (1950) statement about spatial conser-
vation of vorticity: The vortex strength (circulation) is constant along each vortex
line or vortex tube, which must either be closed or end at the boundary of the fluid.
In classical fluid dynamics, the term vortex is typically used in place of the word
vorticity, often in the context of the inviscid fluid idealization. In the present study,
the idea of a vortex always means an approximation of a part of a vorticity field in a
real flow, not a singular element in an inviscid fluid. The terms vorticity tube,
vorticity line, and vorticity flux are therefore preferred in the present study.
Sommerfeld used the terms vortex line and vortex tube interchangeably. In the
present study, except in direct quotation of classical literature, a vortex filament (not
a line) gives the approximate position of a thin vorticity tube in space. The filament
has an infinitesimal (and nonzero) cross-sectional area and is not a line. It represents
(approximates) the vorticity tube and has the circulation of the tube.
If a vortex filament of finite strength C ends in the fluid, then the strength of the
vorticity tube represented by the filament changes abruptly at the ending point from
C to zero. The strength along the vorticity tube is then not a constant. The second
part of Sommerfeld’s statement, that a vortex tube must either be closed or end at
the boundary of the fluid, is therefore a consequence, a corollary, of the first part of
his statement.
Consider a solid body immersed and moving in an infinite fluid. Denote the solid
region by Rs, the fluid region by Rf, the fluid–solid interface by S, and the infinite
unlimited space occupied jointly by the fluid and solids by R∞. The velocity vr of a
rigid body rotating at the angular velocity X is defined by vr = va + X  (r − ra),
where va is the translation velocity of the point r = ra. It is simple to show that
∇  vr = 0 and ∇  vr = 2X. In other words, (2.1) is satisfied in Rs and, as defined
by (2.4), x = 2X in Rs. Equations (2.1) and (2.4) therefore describe the kinematics
of both the fluid and the solid; they are valid equations in the infinite unlimited
region R∞. It is not difficult to generalize the above discussions to flows containing
multiple rigid bodies.
With the no-slip condition, the normal vorticity component is necessarily zero at
the surface of a non-rotating rigid solid body. Lighthill (1963), on the basis of this
observation, concluded: In flows which do not contain rotating bodies, all vorticity
appears in closed loops. Lighthill’s conclusion can also be obtained by observing that
with non-rotating solids, the solid region Rs contains no vorticity. With the vorticity
field present only in Rf, S can only be a part of the surfaces of vorticity tubes. In
consequence, vorticity tubes cannot continue into Rs and must be closed in Rf.
2.4 Helmholtz’ First Vortex Theorem 27

With the no-slip condition, the tangential components of the velocity vector are
continuous on S. Hence the normal component of the vorticity vector is continuous
on S whether or not the solids are rotating. Approaching S from the side of the solid
or the side of the fluid gives the same value of the normal vorticity on S. The
kinetics of the motion in the solid region differs from that in the fluid. The gradients
of the velocity components, hence also the tangential components of vorticity, are
therefore discontinuous on S. The discontinuity occurs whether or not the solid is
rotating. In either case, because the normal component of vorticity is continuous on
S, vorticity flux is continuous on S and Lighthill’s observation can be generalized to
the statement: All vorticity appears in closed loop in the infinite unlimited region
R∞.
Vorticity lines need not be smooth for the idea of vorticity tubes and vorticity
loops to be meaningful. Vorticity tubes may bend at finite angles; they only need to
continue in space. A vorticity line, being a curve in space, has a zero cross-sectional
area and is not associated with a total flux or a circulation. Along the path of a
vorticity line, the value of the vorticity can change. The line may pass through
points of zero vorticity, giving the impression that it ends in space. This occurrence
does not mean that a vorticity tube ends in space.
Vorticity tubes may overlap, or merge. The merged tube appears as a single
vorticity tube with a circulation that is equal to the sum of the circulations of the
component tubes. For the flow containing rotating solids, some of the vorticity
tubes pass through S. On S, a vorticity tube in Rs merges with a vorticity tube in Rf.
In most aerodynamic applications, the circulation of the vorticity tube in Rs (the
cross-sectional area of the tube times 2X) is weak compared to the circulation of the
tube in Rf. The merging of the two tubes near S gives the misleading appearance
that the vorticity tube in Rs does not continue into Rf and vice versa. This
appearance obscures the truth that, because the vorticity flux is continuous on S,
individual vorticity tubes are continued on S; they do not end on S either from the
fluid side or from the solid side. Helmholtz’ first vortex theorem therefore gener-
alizes to the statement: The vorticity strength (circulation) is constant along each
vorticity tube, which must be closed in space. In other words, a vorticity field in R∞
is portrayable in general as a set of closed vorticity tubes (vorticity loops), each with
a constant circulation along its path. This generalized Helmholtz’ first vortex the-
orem is a kinematic theorem stating the spatial conservation of circulation.
Equations (2.1) and (2.4) contain first order spatial derivatives of the velocity
field v. For these equations to be meaningful, v must be not only continuous but
piecewise smooth (once differentiable). The generalized Helmholtz’ first vortex
theorem and many vorticity-kinematic issues involve only integrations of x.
Therefore the vorticity field x needs only be piecewise continuous. Smoothness
requirements for the vorticity in kinetics are more stringent since the vorticity
transport equation (2.6), contains derivatives of x. The kinematic-kinetic partition
of the overall flow problem is in this context important in flow analyses. Without
this partition, it is difficult to resolve many well-known conceptual difficulties and
paradoxes in classical fluid dynamics.
28 2 Theorems of Helmholtz and Kelvin

2.5 Vorticity Loops

As a consequence of the generalized Helmholtz’ first vortex theorem, every vor-

ticity field in the infinite unlimited space R∞ occupied jointly by the fluid and the
solid is portrayable by a set of vorticity loops (closed tubes). The cross-sectional
area S of each loop varies along the path of the loop, but the total flux of vorticity
(circulation) of the loop is constant along the path. Each vorticity loop can be
divided into n thinner vorticity loops simply by partitioning S into smaller areas Si,
i = 1, 2, … , n, with vorticity strengths Ci. These thinner loops are contiguous and
together they occupy the space of the original undivided vorticity loop. The sum of
the vorticity strengths of the thinner loops is the total strength of the original
vorticity loop. A very thin vorticity loop is accurately approximated by a closed
vortex filament (vortex loop), as discussed in §2.4. Thus any vorticity field can be
approximated by a large number of vortex loops in space.
Let the closed path of the vortex loop representing a thin vorticity tube be C. If C
is divided into two open paths C1 and C2 and the two dividing points are connected
by the path C3, then there are two closed paths: a closed path C1* formed by
combing C3 and C1 and a second closed path C2* formed by combining C3 and C2.
Consider two vortex loops: C1* on C1* and C2* on C2*. Let C1* = C2* = C. On
the open path C3, the directions of the loops C1* and C2* are opposite. Thus the
combined circulation of the loops C1* and C2* is zero on C3. The two smaller loops
C1* and C2* together are therefore equivalent, kinematically, to the original loop C.
Thus any vortex loop C is divisible into two smaller vortex loops with the same
vortex strength. Successive divisions give a set of small loops, each with the
circulation C. In aggregate, the small vortex loops are equivalent to the large loop
C. Since the paths of division can be arbitrarily chosen, the set of small loops can be
configured with great flexibility.
Under general circumstances, the vorticity field spreads over a relatively large
flow region. Therefore a large number of vortex loops, each representing a thin
vorticity loop, are needed to produce a reasonably accurate approximation of the
vorticity field. In the limit as the cross-sectional area of each vorticity loop
approaches zero, there is an infinite number of vortex loops and the approximation
becomes precise.
Dividing thin vorticity loops into sets of small vorticity loops produces a large
number of thin and small vorticity loops well suited for flow computations. Dealing
with vorticity loops rather than the vorticity field ensures the satisfaction of the
principle of total vorticity conservation (see Chap. 5), which is necessary for
solution stability in three-dimensional flow computations. The use of a small
number of vorticity loops is convenient, but significant computational inaccuracy is
expected. For streamlined flows, however, the vorticity-loop concept is powerful in
interpreting the lifting-line theory of aerodynamics, as discussed in Chap. 6.
This suggests the possibility of simplified solution procedures for computing
certain types of flows, including unsteady flows. Additional discussions of
vorticity-loop portrayals are given by Wu et al. (2002) in a study of the lift, the
profile drag, and the induced drag on hovering rotors.
2.6 Kelvin’s Theorem 29

2.6 Kelvin’s Theorem

In (2.16), let C(t) be a material circuit Cm moving with the fluid and f = w, the
vector potential of an incompressible velocity field v. The integrand of the last
integral in (2.16) vanishes since (∇  W)  v = v  v = 0. Thus one has
Z Z  
d oW
Þ Wðr; tÞ  s ds ¼ Þ  s ds ð2:31Þ
dt ot
Cm Cm

Equation (2.31) states that the rate of change of the material circulation of the
vector potential W is caused by only the local change of W. The motion of the
material circuit does not contribute to the change.
Let f = v, the velocity field of an incompressible flow. Equation (2.16) gives
Z Z   Z
d ov
Þ v  s ds ¼ Þ  s ds þ Þ ðx  vÞ  s ds ð2:32Þ
dt ot
Cm Cm Cm

where the left-hand side integral gives the material circulation Cm of the velocity
field.
The first integral on the right-hand side of (2.32) represents the contribution of
the local change of v to the change of the material circulation Cm. The last integral
is the circulation of Lamb’s vector x  v around the material circuit Cm and
represents the contribution to the rate of change of Cm by the motion of the material
circuit Cm.
Equation (2.32) is a kinematic equation obtained using only (2.1) and (2.4).
Placing the kinetic equation (2.5) into (2.32) yields
Z Z
dCm
¼  Þ ½rðh  qÞ  s ds  m Þ ðr  xÞ  s ds ð2:33Þ
dt
Cm Cm

The first integral in (2.33) can be restated using Stokes’ theorem as an integral
over Sm, a cap of Cm, of ∇  [∇(h − q)], which is identically zero. One thus has
Z
dCm
¼ m Þ ðr  xÞ  s ds ð2:34Þ
dt
Cm

Equation (2.34) is a vorticity-dynamic relation based on the kinematic equations

(2.1) and (2.4) and the kinetic equation (2.5). Placing (2.7) in (2.32) or letting m = 0
in (2.34) yields
dCm
¼0 ð2:35Þ
dt
30 2 Theorems of Helmholtz and Kelvin

Equation (2.35) is a statement of Kelvin’s theorem on the persistence of circu-

lation: The circulation around a material circuit moving with an inviscid fluid
remains constant.
For the viscous fluid, (2.34) states: The circulation around a material loop
moving with a viscous fluid changes with time as the result of viscous diffusion. It is
known that the viscosities of air and water, the primary flow media of interest in
aerodynamics, are very small. More precisely, the dimensionless Reynolds number
is very large. Therefore, except in flow regions close to solid surfaces, where the
gradient of the vorticity field can be very steep, the right-hand side of (2.34) is small
and the circulation around the material circuit changes very slowly. In many regions
far from solid surfaces, the vorticity gradients are so small that circulation changes
are negligible slow. Such flow regions are inviscid flow regions even though the
fluid is not inviscid.

2.7 Helmholtz’ Second Vortex Theorem

Let f* = x and vs = v. One obtains using (2.22) and (2.6) the following equations:
ZZ ZZ ZZ
d ox
x  n dS ¼  n dS þ r  ðx  vÞ  n dS ð2:36Þ
dt ot
Sm Sm Sm
ZZ ZZ
d
x  n dS ¼ m ½r  ðr  xÞ  n dS ð2:37aÞ
dt
Sm Sm
Z
¼ m Þ ðr  xÞ  s dS ð2:37bÞ
Cm

Letting m = 0 in (2.37), or placing (2.7) into (2.36), one obtains

ZZ
d
x  n dS ¼ 0 ð2:38Þ
dt
Sm

Sommerfeld’s statement (1950) of Helmholtz’ second vortex theorem is: The

vortex strength (total flux of vorticity across a material surface) is constant in time.
In other words, the circulation around a material vorticity tube is independent of
time in an inviscid fluid. Thus Helmholtz’ second vortex theorem and Kelvin’s
theorem on the persistence of circulation are equivalent. In fact, by virtue of (2.12),
(2.38) is equivalent to (2.35) and (2.37) is equivalent to (2.34). The proof provided
for (2.37) in this Section is therefore redundant.
The surface Sm is a cap of the material loop Cm. There exist an infinite number of
caps for each material loop in space. Thus (2.34) is more general than (2.37a). In other
2.7 Helmholtz’ Second Vortex Theorem 31

words, using Stokes’ theorem, (2.34) can be restated as (2.37) with Sm replaced by S.
S can be an arbitrary cross-sectional surface of a material vorticity tube, a cap of a
material circuit, but not necessarily a material cap moving with the fluid.
Equation (2.38) states that the total vorticity flux across the cross-sectional area of
a material vorticity tube in an inviscid flow is independent of time. Letting the
cross-sectional area of a material tube approach zero, one arrives at Lighthill’s (1986)
interpretation of Helmholtz’ second theorem: Vortex lines move with the (inviscid)
fluid. Saffman (1992) reviewed the works of Lamb (1932) and others that culminated
in Lighthill’s statement. Alternative proofs of this statement are presented in the
works of Lamb, Lighthill, Lugt (1996), Saffman, Whitham (1963), and others.
As discussed in §2.6, the circulation of a material vorticity tube in a viscous fluid
changes with time because of viscous diffusion, a process that spreads the vorticity
in the flow. Consider, for simplicity, a Cartesian coordinate system (x,y,z) and a
planar flow with the velocity field v(x,y) and x = x k, where k is the unit vector in
the z-direction. For this planar flow, the vorticity flux x  n is identical to x since
k = n. Thus (2.37) yields
ZZ ZZ
d  
xdS ¼ m r2 x dS ð2:39Þ
dt
Sm Sm

Using (2.25), (2.39) can be restated in the form

ZZ  
Dx
 mr2 x dS ¼ 0 ð2:40Þ
Dt
S

where Dx/Dt = ∂x/∂t + (v  ∇)x is the material (substantial) derivative of x, a

time derivative following the motion of the fluid. Equation (2.39) is given by Lamb
(1932) and a more general equation for a non-planer surface Sm by Wu and Wu
(1998).
Equation (2.40) is obtainable directly by integrating the two-dimensional ver-
sion of the vorticity transport equation (2.6), which is expressible in the form Dx/
Dt = m∇2x. In this form, the vorticity is a scalar field and the vorticity transport
equation is analogous to the familiar diffusion equation. Consider, for example, the
heat conduction (diffusion) equation ∂T/∂t = j∇2T, where T is the temperature field
and j is the heat conductivity (diffusivity). In heat conduction, heat energy, mea-
sured in terms of temperature, is transported in a medium at rest. This process is
irreversible. It equalizes the heat energy, spreading it from higher temperature
regions to lower temperature regions. Observed in a reference frame moving with
the fluid, the viscous diffusion of vorticity is analogous to heat conduction in a
stationary medium. Without viscous diffusion, the total vorticity over any material
surface in the plane of the flow is independent of time. This is analogous to the
temperature associated with each material element of a stationary non-conducting
solid remaining unchanged relative to time. In the flow of the viscous fluid, the total
vorticity in a material surface changes with time as a result of viscous diffusion.
32 2 Theorems of Helmholtz and Kelvin

In three-dimensional flows, the vorticity is a vector field. Equation (2.37) is valid

for an arbitrarily chosen material surface, including an elemental material surface
dSm. Therefore the total vorticity flux over dSm is spread by viscous diffusion when
viewed in a material reference frame moving with the fluid. It is worth under-
scoring that the total flux x  n dSm, not the vector x, is conserved in the inviscid
flow. This total flux is diffused in the viscous flow. In three-dimensions, the surface
element dSm translates and rotates. The total vorticity flux translates and rotates
with dSm.

2.8 Concluding Remarks

The theorems of Helmholtz and Kelvin are most conveniently interpreted using the
idea of the vorticity flux x  n. Helmholtz’ first vortex theorem is recognized as a
theorem of vorticity kinematics stating that the total vorticity flux in a vorticity tube
is constant along the path of the tube. This theorem is valid in the infinite unlimited
region occupied jointly by the solid and the fluid. In consequence, in external
aerodynamics, all vorticity fields are portrayable as sets of vorticity loops, each
with a constant circulation along the path of the loop.
Helmholtz’ second vortex theorem, or its equivalence Kelvin’s theorem, is a
theorem of vorticity dynamics. A generalized statement of this theorem is: The total
vorticity flux in each material vorticity tube changes with time only as a result of
vorticity diffusion across the boundary surface of the tube. It is important to note
that, because this generalized theorem is a vorticity-dynamic theorem, it is appli-
cable in the interior of the fluid region, and not in the solid region. In an inviscid
flow, the total vorticity flux in each vorticity tube remains unchanged with time. In
many flow regions far from solid surfaces, the vorticity gradients are so small that
circulation changes are negligible slow. Such flow regions are effectively inviscid
even though the fluid is viscous. In such flow regions, vorticity lines are effectively
material lines. It is convenient to think of vorticity loops portraying the vorticity
field as moving with the fluid while retaining their strengths (circulation) in such
regions.
In flow regions where viscous effects are important, vorticity lines are not
material lines. They do not move with the fluid. Since the circulation of each
material circuit changes with time, the strength of each vorticity loop is not con-
served in time. It is not sufficient to merely keep track of the movements of the
vorticity loops; the strength and the shape of each loop must be undated continually.
At the present stage of development, potential applications of the generalized
theorems of Helmholtz and Kelvin in aerodynamics are evolving. Based on the
present understanding, new interpretations of classical aerodynamic theories and of
the connection between two- and three-dimensional aerodynamics are possible.
Discussions of these topics are presented by Wu (1981) and in Chap. 6.
References 33

References

Helmholtz H (1858) Uber Integrale der hydrodynamischen Gleichungen, welche den

Wirbelbewegungen entsprechen (On The integrals of the Hydrodynamic Equations
Corresponding to Vortex Motions). Crelles Journal fiir die reine und angewandte Mathematik
55:25
Kelvin, L (Thompson, W.) (1869) OnVortex Motion. Trans R Soc Edinb 25:217
Lamb H (1932) Hydrodynamics. Dover Publications
Lighthill M J (1963) Introduction. Boundary layer theory. In: Rosenhead L (ed) Laminar boundary
layers. DoverPublications
Lighthill J (1986) An informal introduction to theoretical fluid mechanics. Oxford University Press
Lugt HJ (1996) Introduction to vortex theory. Vortex Flow Press
Prandtl L (1921) Applications of modern hydrodynamics to aeronautics. NACA Report No. 116
Saffman PG (1992) Vortex dynamics. Cambridge University Press
Sommerfeld A (1950) Mechanics of deformable bodies. Academic Press
Truesdell C (1954) The kinematics of vorticity. Indiana University Press
Tuncer I, Wu JC, Wang CM (1990) Theoretical and numerical studies of OscillatingAirfoils.
AIAA J 28:1615
Whitham GB. Laminar boundary layers. In: Rosenhead L (ed) The Navier-Stokes equations of
motion. Dove
Wu JC (1981) Theory for aerodynamic force and moment in viscous flows. AIAA J 19:432
Wu JC, Wang CM, McAlister KW (2002) Wake-integral determination of aerodynamic drag, lift
and moment in three-dimensional flows. AIAA Paper p 555
Wu JZ, Wu JM (1998) Boundary vorticity dynamics since Lighthilis 1963 article: review and
development. Theo Comput Fluid Dyn 10:459
http://www.springer.com/978-3-662-44039-1