See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228606964

Growing vortex patches

Article in Physics of Fluids August 2004

DOI: 10.1063/1.1767771

CITATIONS READS

13 23

2 authors, including:

Jonathan S. Marshall
24 PUBLICATIONS 403 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Conformal mapping between multiply connected domains View project

All content following this page was uploaded by Jonathan S. Marshall on 29 April 2015.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
PHYSICS OF FLUIDS VOLUME 16, NUMBER 8 AUGUST 2004

Growing vortex patches

Darren Crowdya) and Jonathan Marshallb)
Department of Mathematics, Imperial College of Science, Technology and Medicine, 180 Queens Gate,
London SW7 2AZ, United Kingdom
Received 10 December 2003; accepted 10 May 2004; published online 6 July 2004
This paper demonstrates that two well-known equilibrium solutions of the Euler equationsthe
corotating point vortex pair and the Rankine vortexare connected by a continuous branch of exact
solutions. The central idea is to grow new vortex patches at two stagnation points that exist in the
frame of reference of the corotating point vortex pair. This is done by generalizing a mathematical
technique for constructing vortex equilibria first presented by Crowdy D. G. Crowdy, A class of
exact multipolar vortices, Phys. Fluids 11, 2556 1999. The solutions exhibit several interesting
features, including the merging of two separate vortex patches via the development of touching
cusps. Numerical contour dynamics methods are used to verify the mathematical solutions and
reveal them to be robust structures. The general issue of how simple vortex equilibria can be
continued continuously to more complicated ones with very different vortical topologies is
discussed. The solutions are examples of exact solutions of the Euler equations involving multiple
interacting vortex patches. 2004 American Institute of Physics.
DOI: 10.1063/1.1767771

I. INTRODUCTION patches. Elcrat and Miller5 have proved the existence of

equilibrium vortex patches close to a stable configuration of
Point-vortex models and vortex-patch models are by far
point vortices. Converting point vortices to patches, or vice
the most popular and well studied in vortex dynamics.1,2
versa, is a common method of producing modified equilibria
Point-vortex models have the advantage of reducing the
from existing ones.
problem to that of tracking a discrete point set while vortex-
Other methods of constructing modified equilibria from
patch models reduce the problem to that of tracking a curve
or set of curves. Such mathematical simplifications prove any given one have recently been proposed. A given steadily
to be of great advantage. rotating equilibrium of point vortices, for example, often ex-
In studying any dynamical system, an important first hibits points of relative rest in a frame of reference corotating
step is to gain a thorough understanding of the possible equi- with the equilibrium. In the context of point-vortex dynam-
librium configurations since they are often attractors in the ics, Aref and Vainchtein6 have proposed the idea of con-
dynamics. For this reason, the subject of vortex statics is structing more complicated point-vortex equilibria by
an important one.3 One of the very simplest nontrivial point- growing new point vortices at any such corotating points.
vortex equilibria is the corotating point vortex pair in which A zero-circulation point vortex is dynamically inactive and
two equal point vortices corotate about the central point of can be placed at any such corotating points without affecting
their line of centers. On the other hand, the simplest example the equilibrium. The idea of Aref and Vainchtein6 is to per-
of a vortex-patch equilibrium is undoubtedly the Rankine form a continuation in the circulation of any such nascent
vortex solution1,2 which describes a circular patch of uniform point vortices in an attempt to create new equilibria involv-
vorticity in solid-body rotation. In this paper, a vortex patch ing more complicated vortical configurations. While there is
is understood to be a region of fluid in which the vorticity is no guarantee that such continued solutions exist, in cases
a uniform constant. where they do, new and even asymmetric equilibria can be
Point vortices and uniform vortex patches are closely constructed.
related. The limit in which the radius of a Rankine vortex Here, we develop this idea in a natural way and consider
vanishes while its vorticity tends to infinity in such a way the possibility of growing new vortex patches as opposed to
that the circulation is fixed is known to yield a point-vortex point vortices at the corotating points of an existing vortical
solution. At the same time, the idea of desingularizing, or equilibrium. This paper illustrates that this is sometimes pos-
regularizing, a point vortex by smearing out the vorticity to a sible and an explicit example is presented in detail using
uniform patch of nonzero area is well known. Dritschel,4 for perhaps the simplest nontrivial equilibriumthe corotating
example, smeared out the vorticity in the rotating point-vortex pair. The mathematical construction is based on
N-polygonal point vortex arrays considered by Thomson1,2 extensions of ideas originally presented in Crowdy7 and de-
to find modified equilibria involving N corotating vortex veloped in Crowdy.8,9 A consequence of the analysis is to
show that the corotating point-vortex pair is, in fact, con-
a
Electronic mail: [email protected] nected by a continuous branch of nontrivial vortex equilibria
b
Electronic mail: [email protected] to the classical Rankine vortex solution. Moreover, the entire

1070-6631/2004/16(8)/3122/9/$22.00 3122 2004 American Institute of Physics

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
Phys. Fluids, Vol. 16, No. 8, August 2004 Growing vortex patches 3123

branch of solutions is describable in mathematical form by

means of analytical formulas although one of the parameters
appearing in these formulas must be determined numeri-
cally.
While this result is of theoretical interest in its own right,
perhaps more important is the more general question which it
provokes concerning how different vortical equilibria, per-
haps with very different vortical topologies, might be con-
nected and how a given equilibrium might be continued,
in a continuous fashion, to a more complicated one. The
explicit example presented here starts with a vorticity distri-
bution consisting of a simple two-point set. This is contin-
ued, continuously, to a distribution consisting of a hybrid
combination of two-point vortices and two-vortex patches.
The latter configuration is then smoothly continued to a dis-
tribution involving four-vortex patches which, finally, coa-
lesce into a single isolated Rankine vortex. Being inherently
nonlinear, the steady Euler equation is renowned for being
difficult to solve. For this reason, a thorough understanding
of how complicated vortical equilibria can be systematically
constructed by continuous deformations of simpler ones is
desirable and is discussed in Sec. VIII.
It should be remarked that the class of solutions involve
vortex patches that are in pure solid-body rotation and are
not the only possible rotating equilibria consisting of two
FIG. 1. Schematic illustrating the vortex configuration of interest. Two-point
patches and two-point vortices. More general solutions in vortices are at (0,1) while two rotationally symmetric uniform vortex
which the fluid in the vortex patches has a nontrivial irrota- patches are located on the positive and negative real axis.
tional component are also possible, but it is likely that this

more general class is not describable in analytical form.

II. THE COROTATING POINT-VORTEX PAIR

4
zz z
S z dz z
S z dz zD
, 3
0 zD
Consider two-point vortices, each of circulation 2,
where D is some fluid region and S(z) is the Schwarz
initially at points (0,1). Introducing the usual complex
function11 of the boundary curve D. For certain special
coordinate zxiy, the associated instantaneous complex
choices of fluid domain D, such a streamfunction can repre-
potential w(z) for this flow is
sent an equilibrium solution of the Euler equation. Moreover,
w z i ln zi i ln zi . 1 in many cases, these special classes of domains can be pa-
It is straightforward to show that such a configuration rotates rametrized using conformal mappings thus effectively lead-
about the origin with angular velocity equal to /2. In a ing to exact solutions of the Euler equation.
frame of reference corotating with this angular velocity, the This constructive method appears to be quite general,
configuration is stationary and the complex velocity field u and we will use a variant of it here. Consider a configuration
i v has the form in which, in addition to the two-point vortices at zi,

two-vortex patches are located symmetrically on the positive
dw 1 1 z and negative real axis. A schematic is shown in Fig. 1. The
ui v i . 2
dz zi zi 2 irrotational fluid region containing the two-point vortices is
denoted D, the two-vortex patches are denoted D 1 and D 2 .
A simple calculation shows that this velocity field has stag-
Let the boundaries of the two patches be D 1 and D 2 . It
nation points at z0,). Equivalently, because these
will be assumed that the vortex patches have the same angu-
points are stagnation points in a corotating frame, one might
lar velocity /2 as the rotating point vortex configuration
equally well refer to these as corotating points following
and, moreover, are in pure solid-body rotation about the ori-
Aref and Vainchtein6 see also Morton10. In the following
gin. The uniform vorticity of each patch is therefore and
section, it is shown how to grow two new vortex patches at
the assumption of solid-body rotation implies that, in the
the corotating points located at z).
corotating frame, the fluid inside the two patches is stagnant.
Now pose that the streamfunction of the flow, in a frame
III. MATHEMATICAL CONSTRUCTION of reference corotating with angular velocity /2, is of the
form 3 where S(z) is now the Schwarz function if it ex-
Crowdy8,9 has demonstrated the theoretical advantages ists of both boundaries of D, i.e., D 1 and D 2 . Recall that
of considering streamfunctions of the form the vorticity is given by 4 zz where subscripts denote par-

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
3124 Phys. Fluids, Vol. 16, No. 8, August 2004 D. Crowdy and J. Marshall

tial differentiation. This means that, in the corotating frame, V. EXACT SOLUTIONS
the vorticity inside D associated with the streamfunction 3
is a uniform constant. Note also that the function S(z), Consider now the conformal mapping given by
which is locally analytic in annular neighborhoods of both P 1 P P
D 1 and D 2 , must satisfy z R , 10
P 1 P e i P e i
S z z 4
where R and are some real parameters and the special
on both D 1 and D 2 . If there are just two-point vortices in function P( ) is defined as
D, it is clear that an additional restriction on the function

S(z) is that it must only have two simple poles in D. P 1

k1
1 2k 1 2k 1 . 11
The associated velocity field is
This is the same special function used by Crowdy9 to con-
i struct exact solutions for annular arrays of vortices. It is re-
ui v 2i z zS z . 5
2 lated to the first Jacobi theta function.12 Note that the map
has a simple pole at the point which therefore maps to
If S(z) satisfies 4 on D 1 and D 2 , it is easy to verify that physical infinity, while it has a zero at which there-
this streamfunction satisfies the kinematic and dynamic fore maps to z0. The conformal map 10 depends on just
boundary conditions on both boundaries D 1 and D 2 of D. three real parameters R, , and .
The kinematic boundary condition is that both D 1 and D 2 It is straightforward to show, by use of its definition 11,
are streamlines, the dynamic boundary condition is that the that P( ) satisfies the functional relations
velocity field must vanish everywhere on these boundaries in
order to be continuous with the stagnant flow inside the P 2 1 P ,
patches. Continuity of velocity on the boundary of a steady
P 1 1 P . 12
vortex patch implies that the hydrodynamic pressures are
continuous.1 These relations 12 are all that are needed to verify directly
that z( ) satisfies
z 1 z , 13
IV. CONFORMAL MAPPING
which implies that for every point on the unit circle map-
7,8
As in Crowdy, the most effective way to construct the ping to z 1 , say, there is a point 1 on the -circle mapping
relevant fluid domains D is to employ conformal mapping to z 1 . This means that the mapping produces two rotation-
techniques. Here, D is an unbounded doubly connected do- ally symmetric vortex patches, as required. Also, 12 can be
main. The Riemann mapping theorem guarantees that any used to verify directly that z( ) satisfies the requirement
such domain can be conformally mapped from some annulus given in 9 which is essentially the condition that the
1 in a parametric plane. The parameter relevant Schwarz functions of the two separate boundaries D 1 and
to any given domain must be determined as part of the prob- D 2 are the same. Finally, noting that
lem. S z z 1
Let the conformal map be z( ) and suppose that the
circle 1 maps to D 1 while maps to D 2 . Note P 1 1 P 1 P 1
that on D 1 , R ,
P 1 1 P 1 e i P 1 e i
S z zz z 1 , 6 14
it is clear that S(z) has just two simple poles in the annulus
where we have used the fact that 1 on 1 and the 1, at the points e i , e i . This means that
conjugate function z( ) is defined by S(z) has just two simple poles in D at the corresponding
conformally mapped points z( e i ) and z( e i ), again
z z . 7 as required. The scaling parameter R is fixed by insisting that
the two-point vortices are at i which is equivalent to the
Note also that on D 2 ,
equation
S z zz z 2 1 , 8 iz e i . 15

where we have used the fact that 2 1 on . In The conformal map 10 therefore satisfies almost all of the
order for 6 and 8 to be consistent it is clear that the requirements for a mapping to an appropriate equilibrium
conformal map must satisfy domain D. The only outstanding condition is to ensure that
the two-point vortices are stationary under the effects of the
z 2 z 9 local non-self-induced velocity field. By symmetry, it is only
necessary to ensure that this condition of stationarity is sat-
for all . isfied at one of the point vortices. Substituting 10 and 14

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
Phys. Fluids, Vol. 16, No. 8, August 2004 Growing vortex patches 3125

into 5 one obtains a formula for the complex velocity field,

in the corotating frame, as a function of and . This is

ui v
iR
2
P 1 P P
P 1 P e i P e i

P 1 1 P 1 P 1
P 1 1 P 1 e i P 1 e i
.

16
On use of the conformal mapping function it can be shown
that, as zi, the velocity field 16 locally has the form
i s
ui v Vo 1 , 17
2 zi
where s is the circulation of the point vortex at zi and V
is the local non-self-induced velocity. The condition that the
two-point vortices are stationary under the non-self-induced
velocity field is equivalent to V0. The Appendix gives ex-
pressions for s and the condition that V0 in terms of the
conformal mapping parameters. By the symmetry, this con-
FIG. 2. A continuous branch of rotating vortex arrays connecting the coro- dition also ensures that the vortex at zi is stationary. As
tating point-vortex pair to the Rankine vortex. The figures illustrate the seen in the Appendix, this equation is independent of R
growing of two-vortex patches at the two stagnation points at ) in the which is simply a scaling parameter but depends only on
corotating frame of the corotating point-vortex pair. The sixth figure is the
limiting state of the solution 10 where crit and where the two patches and . By applying Newtons method to this equation, we-
develop cusps and touch at three distinct points enclosing circular irrota- have found that it can be solved numerically for for given
tional regions centered on the point vortices. The last two figures illustrate in the interval
the desingularization of the point vortices to form a single Rankine vortex.
The circulations s and sp are also shown. 0, crit , 18

FIG. 3. Graph of satellite patch circu-

lations sp and satellite point-vortex
circulations s against . The sum s
sp is found to equal 2 for all val-
ues of . As crit , s , sp .

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
3126 Phys. Fluids, Vol. 16, No. 8, August 2004 D. Crowdy and J. Marshall

. Different values of correspond to different values of the

circulations of both the point vortices and the patches. Let s
denote the point-vortex circulations and let sp denote the
circulation of each of the vortex patches. The expression 16
can be used to derive analytical expressions for s and sp in
terms of the conformal mapping parameters. Figure 2 anno-
tates each configuration with the corresponding values of
these circulations. When 0, sp tends to zero because it
is proportional to the patch area. A superposed graph of s
and sp is shown in Fig. 3 and displays a surprising feature.
To within numerical accuracy in the solution of the station-
arity condition for given a value of , it is found that all
the equilibrium configurations satisfy

s sp 2 . 19

Thus, while the values of s and sp both change with ,

their sum remains constant. We can offer no explanation for
this unexpected result.
Perhaps equally surprising is the nature of the limiting
configuration as crit 0.735. This is shown in the sixth
FIG. 4. Critical configuration for two satellite patches and two satellite point
diagram of Fig. 2. In this limiting state the two-vortex
vortices for crit 0.735 drawn in solid lines shown with the three patches touch at three distinct points, the points of contact
circles zi 1 and z 2 superposed drawn with dashed lines. The taking the form of three cusp singularities in the patch
solid and dashed curves are indistinguishable. In this critical case, the cir- boundaries. As crit , the curvature of the near cusps gets
culation of each satellite patch is , as is the circulation of each satellite
larger as the distance between corresponding near cusps on
point vortex.
the two-vortex patches gets smaller. At crit , within nu-
merical accuracy, it is found that the patches touch and en-
where crit yields a limiting configuration to be discussed in close two exactly circular regions of irrotational fluid with
the following section. The solution to this equation is very the point vortices located at their centers. To test this, Fig. 4
straightforward since it is a Newton iteration on just a single shows the critical configuration superposed with the three
unknown parameter. circles zi 1 and z 2. The boundaries of the limiting
configuration are indistinguishable from these three circles.
It was also noticed that in the limiting configuration, the
VI. CHARACTERIZATION OF THE SOLUTIONS
values of s and sp become equal. By 19, they both tend
The conformal map 10, with the three parameters R, , to . This can be seen clearly in Fig. 3.
and constrained by the two conditions of stationarity of the This feature of the limiting two-patch solutions suggests
point vortices and the normalization 15 yield a continuous that the class of solutions can be continued, in a continuous
one-parameter family of two-patch equilibria. Without loss fashion, even past this limiting state. In view of the fact that
of generality, we set 1. In the limit 0, it is found that the streamlines around the point vortices are exactly circular,
the two-vortex patches become invisibly small and essen- the two-point vortices can be desingularized in the usual way
tially disappear at z) in this limit. This is precisely the and replaced by two Rankine vortices with the same total
location of the corotating points in the pure point-vortex con- circulation as the original point vortices but with gradually
figuration discussed in Sec. II. For this reason, the construc- increasing radius r. If r denotes the vorticity of a Rankine
tive procedure just described essentially effects the job of vortex of radius r and total circulation then
growing two-vortex patches at these corotating points.
As increases, so does the size of each vortex patch. 1
r . 20
Figure 2 shows several configurations for different values of r2

FIG. 5. Contour dynamics simulation of a single revolution of the equilibrium shown in the fourth figure of Fig. 2 with s 4.620, sp 1.663.

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
Phys. Fluids, Vol. 16, No. 8, August 2004 Growing vortex patches 3127

FIG. 6. Superposition of the initial

configuration and final configurations
after 10 turnover times i.e., t20) as
computed by contour dynamics. The
curves sit on top of each other. The
initial configuration is the equilibrium
shown in the fourth diagram of Fig. 2.

The value of r0 is arbitrary provided it is less than, or vortices seems to induce this occurrence. This appears to be
equal, to unity which is where the enclosed Rankine vortices an example of a limiting equilibrium exhibiting touching
meet the circular boundaries of the satellite patches. The sev- cusps. Note that, owing to the presence of these cusps which
enth diagram in Fig. 2 shows the configuration for r0.4. have infinite curvature, it is likely that the limiting states
Indeed when r1, by the previously observed fact that the would be challenging to compute using any numerical
limiting circulation of s is , the uniform vorticity of the scheme which relies on a discretization of the patch bound-
enclosed circular Rankine vortices tends exactly to the value aries. The existence of a closed-form formulas for the solu-
of the uniform vorticity of the satellite vortex patches imply- tions is therefore of great value.
ing that, when r1 then r 1 by 20 and the equilibrium
essentially becomes a single circular Rankine vortex of ra-
VII. CONTOUR DYNAMICS SIMULATION
dius 2, uniform vorticity 1, and total circulation 4.
Overman13 showed that points of nonanalyticity in the As a check on the mathematical solutions, the contour
boundary of a steady vortex patch can either be right-angled surgery code of Dritschel16 for computing the evolution of
corners or cusps. It is known that two corotating vortex vortex patches was modified to include the effect of two-
patches reach a limiting configuration at which they touch at point vortices interacting with the patches. This code was
the same time as a corner develops in the boundary of each initialized using the equilibrium configurations just derived
patch. Saffman and Szeto14 and Kamm15 have investigated in order to check that they simply rotate without change of
such problems. The equilibrium solutions found here exhibit form under the dynamics of the Euler equation. Figure 5
the feature of two equal corotating vortex patches touching shows snapshots of the evolution of the equilibrium in the
in this case, simultaneously at three distinct points as the fourth diagram of Fig. 2 during a single turnover time. Here,
boundaries of the two patches develop cusps, as opposed to time has been rescaled with respect to 2 so, since 1 so
corners. The presence of the straining flow due to the point that the angular velocity is 1/2, then t2 corresponds to a

FIG. 7. Contour dynamics simulation of a single revolution of the critical equilibrium.

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
3128 Phys. Fluids, Vol. 16, No. 8, August 2004 D. Crowdy and J. Marshall

FIG. 8. Contour dynamics simulation of a perturbed equilibrium. The upper point vortex in the equilibrium of the fourth figure of Fig. 2 is displaced upwards
by 0.05. The configuration is robust, but its overall angular velocity of rotation is affected.

single revolution of the array. The snapshots are taken at t replaced by a radius 1/2 Rankine vortex each of vorticity 4.
0.5,1,1.5 and t2 and, indeed, these are clearly found to Similarly, Fig. 10 shows a case in which the point vortices
correspond to quarter revolutions of the configuration. As an are replaced by Rankine vortices of different radii: the upper
additional check, Fig. 6 features a superposition of the initial point vortex is replaced by a Rankine vortex of radius 2/3
condition and the final configuration after 10 turnover times and vorticity 9/4 while the lower point vortex is replaced by
i.e., at t20). Within numerical errors associated with the a Rankine vortex of radius 1/3 and vorticity 9. These simu-
simulation, the initial condition is verifiably an equilibrium lations corroborate the fact that the vortex configurations are
of the equations. Figure 7 shows a simulation of a single indeed equilibria of the Euler equation.
revolution of the critical case in which the two patches touch.
This too appears to be a robust equilibrium of the equations
VIII. DISCUSSION
even though the numerical evolution of the cusp regions of
the interface is a little unsteady. It should be mentioned that, This paper has demonstrated that the corotating point-
owing to the presence of these high curvature regions, a large vortex pair can be continuously deformed, through a series
number of points must be given in the initial conditions for of equilibria describable using exact mathematical formulas,
the contour dynamics simulation in this case. to the classical Rankine vortex. This has been done by the
The preceding calculations also suggest that the equilib- device of growing two new vortex patches at the corotating
ria might well be linearly stable since if unstable, growth of points at z) of the corotating point-vortex pair.
any small numerical inaccuracies might be expected to de- It should be noted that, as in the point-vortex case con-
stabilize the array after sufficiently long times. A detailed sidered by Aref and Vainchtein,6 the success of growing
investigation of the stability properties of this class of solu- patches at corotating points is not guaranteed. Indeed, in the
tions remains to be performed, however, some preliminary present example, one could contemplate adapting the same
investigations using contour dynamics suggests that the equi- methods used here to grow a new central vortex patch at the
libria are robust structures. Figure 8 shows the initial con- corotating point at z0 instead of at z). However, this
figuration given in the fourth diagram of Fig. 2 but perturbed attempt would fail as can be concluded immediately from the
by displacing the upper point vortex vertically upwards by a fact that such a generalized equilibrium would fall within the
distance 0.05. The overall structure remains robust but the class of equilibria consisting of a central vortex patch sur-
angular velocity of rotation is affected. After two revolutions rounded by N satellite point vortices considered recently by
of the unperturbed equilibrium, the configuration has not re- Crowdy.8 In the latter study, it was found that equilibrium
turned to its original orientation but is displaced through solutions of this kind can be found for all N3 but the case
some angle. N2 does not yield solutions.
As a check on the equilibria constructed by desingular- Remarkably, using numerical methods, Cerretelli and
izing the point vortices to Rankine vortices, Fig. 9 shows a Williamson17 have recently found a completely distinct
contour dynamics simulation of two revolutions of the criti- branch of equilibria connecting the corotating point vortex
cal configuration in which the two-point vortices are each pair to the Rankine vortex. Instead of growing new vortex

FIG. 9. Contour dynamics simulation of two revolutions of the continued critical equilibrium in which the two-point vortices are replaced by two Rankine
vortices each of radius 0.5.

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
Phys. Fluids, Vol. 16, No. 8, August 2004 Growing vortex patches 3129

FIG. 10. Contour dynamics simulation of a single revolution of the continued critical equilibrium in which the upper point vortex is replaced by a Rankine
vortex of radius 2/3 and vorticity 9/4 while the lower point vortex is replaced by a Rankine vortex of radius 1/3 and vorticity 9.

patches at the corotating points and then later desingularizing theory of conformal mappings of multiply connected do-
the point vortices as done here, the latter authors approach is mains. In the doubly connected case considered here, the
to start by desingularizing the two corotating point vortices mappings have been constructed by implicit use of the well-
by replacing them with two finite-area patches. The area of developed theory of elliptic functions manifested in the use
the two noncircular patches is gradually increased until the of P( ) which is related to Jacobi theta functions12. For
patches touch. Thereafter, Cerretelli and Williamson17 con- more than two patches, the situation becomes much more
tinue the class of equilibria even beyond this apparently challenging and details remain to be worked out. Finally,
limiting state to construct a class of simply connected patch another interesting generalization is to the case where vortex
equilibria variously dubbed dumb-bells and sausages patches are grown at the corotation points of a corotating
before the truly limiting cats-eye state is reached. En point-vortex pair with different strengths, however, we have
route to this limiting state, the patch becomes a Kirchhoff not studied this case in any detail.
ellipse which, it is well-known,1,2 can be continuously de-
formed through a sequence of equilibrium ellipses of gradu-
ally increasing aspect ratio back to the circular Rankine vor-
tex. By this path of solutions, the corotating point-vortex pair ACKNOWLEDGMENTS
is again continuously connected to the classical Rankine vor- This work is partially supported by a grant from EPSRC
tex. in the United Kingdom. J.M. acknowledges the support of an
All this evidence suggests powerful possibilities for the EPSRC studentship.
future construction of new vortical equilibria based on com-
bined point-vortex and vortex-patch models. By procedures
such as i the desingularization of point vortices to uniform
vortex patches or vice versa, ii the growing of new point APPENDIX: THE STATIONARITY CONDITION
vortices at corotation points of existing equilibria as done
recently by Aref and Vainchtein,6 iii the growing of new The velocity field associated with the exact solutions is
vortex patches at corotation points of existing equilibria as 16 which, near e i can be written in the form
done here, and iv the smooth continuation of touching vor-
tex patches to a merged equilibrium as done here and re-
cently by Cerretelli and Williamson,17 it appears that even
ui v
2
iR A
B , A1

basic equilibria with simple vorticity distributions can be where A( ) is analytic at and where explicit formulas
continuously continued to more complicated ones with more for A( ) and B( ) can be derived from 16.
elaborate vortical topology. Let z 1 denote the position of the point vortex on the
In general, any such continuations must be performed imaginary axis. Then z 1 z( ) and
using numerical methods. However, it appears that there ex-
ist special cases where exact solutions can be identified. Ex- z
zz 1 z 2 A2
act solutions of the steady Euler equation are rare, yet con- 2
sideration of streamfunctions of the form 3 seems to be so that, with some manipulations, we deduce that
unusually successful in producing them as evinced both here
and in previous studies.79 Moreover, the resulting solutions 1 z z
O zz 1 . A3
appear to have a number of surprising characteristics that zz 1 2z
have yet to be explained. This is left for future work as the It follows that the circulation s of the point vortex is
potential of the streamfunction 3 is examined further.
Finally, we mention that the two-patch solutions herein s RA z A4
are examples of exact solutions of the Euler equations in- while the stationarity condition V0 is equivalent to
volving more than one vortex patch. The possibility of ex-
tending the general methods to find equilibria involving z
A A B 0. A5
more than two vortex patches is intriguing but involves the 2z

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
3130 Phys. Fluids, Vol. 16, No. 8, August 2004 D. Crowdy and J. Marshall

1
P. G. Saffman, Vortex Dynamics Cambridge University Press, Cambridge, 11
P. Davis, The Schwarz Function and Its Applications, Carus Mathematical
1992. Monographs The Mathematical Association of America, Washington, DC,
2
P. K. Newton, The N-vortex Problem Springer, New York, 2001. 1974.
3
H. Aref, P. K. Newton, M. A. Stremler, T. Tokieda, and D. L. Vainchtein, 12
M. Abramovitz and I. Stegun, Handbook of Mathematical Functions Do-
Vortex crystals, Adv. Appl. Mech. 39, 1 2002. ver, New York, 1965.
4
D. G. Dritschel, The stability and energetics of corotating uniform vor- 13
E. A. Overman, Steady state solutions of the Euler equation in two di-
tices, J. Fluid Mech. 157, 95 1985. mensions. II. Local analysis of limiting V-states, SIAM Soc. Ind. Appl.
5
A. R. Elcrat and K. G. Miller, Rearrangements in steady multiple vortex Math. J. Appl. Math. 46, 765 1986.
flows, Commun. Partial Differ. Equ. 20, 1481 1995. 14
6 P. G. Saffman and R. Szeto, Equilibrium shapes of a pair of equal uni-
H. Aref and D. L. Vainchtein, Point vortices exhibit asymmetric equilib-
form vortices, Phys. Fluids 23, 2339 1980.
ria, Nature London 392, 769 1998. 15
7 J. R. Kamm, Shape and stability of two-dimensional vortex regions,
D. G. Crowdy, A class of exact multipolar vortices, Phys. Fluids 11,
2556 1999. Ph.D. thesis, Caltech, Pasadena, CA, 1987.
16
8
D. G. Crowdy, Exact solutions for rotating vortex arrays with finite-area D. Dritschel, Contour surgery: a topological reconnection scheme for
cores, J. Fluid Mech. 469, 209 2002. extended interactions using contour dynamics, J. Comput. Phys. 77, 511
9
D. G. Crowdy, On the construction of exact multipolar equilibria of the 1988.
17
two-dimension Euler equations, Phys. Fluids 14, 257 2002. C. Cerretelli and C. H. K. Williamson, A new family of uniform vortices
10 relating to vortex configurations before merging, J. Fluid Mech. 493, 219
W. B. Morton, On some permanent arrangements of parallel vortices and
their points of relative rest, Proc. R. Ir. Acad., Sect. A 41, 94 1933. 2003.

Downloaded 07 Jul 2004 to 155.198.192.80. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp
View publication stats