Vortex Dynamics

Abstract This part introduces the reader to the understanding of fluid motion in
terms of vortex dynamics. First the conceptual background on vortex dynamics is
drawn, developing an intuition why most fluid phenomena involve vortices and why
they are so relevant. Then the reader is accompanied along the life of vortices:
where they come from, why they form as vortices, their interaction with other vor-
tices, up to the final dissipation. During this, vortices reveal to have a major influ-
ence on the wall shear stress along nearby tissues, and the process of vortex formation
is associated to the development of forces on surrounding boundaries. Finally, an
account is worthy of turbulence in terms of vorticity.

2.1 Defnitions

Vortices are fundamental performers in fluid mechanics and they develop in almost
every realization of fluid motion (as extensively shown in the beautiful book by
Lugt 1983). The presence of vortices dominates the corresponding fluid dynamics
and the associated energetic phenomena. Vortices that develop in the large vessels
of the cardiovascular systems play a fundamental role in the normal physiology
and bring about the proper balance between blood motion and stresses on the sur-
rounding tissues.
The fluid velocity is commonly assumed as the principal quantity describing
fluid motion. However, velocity is not able to evidence the underlying dynamical
structure of a flow field, like stresses, mixing, or turbulence, that depend on velocity
gradients. The weakness of a description based on velocity alone is particularly
critical when the fluid motion features the presence of vortex structures. In general,
vorticity is the preferable fundamental quantity for the analysis of incompressible
fluid dynamics. Vorticity, which represents the local rotation rate of fluid particles,
allows emphasizing the structure that hides behind the flow field; it also represents
a complete description of the flow and allows recovering the whole velocity field.

A. Kheradvar, G. Pedrizzetti, Vortex Formation in the Cardiovascular System, 21

DOI 10.1007/978-1-4471-2288-3_2, © Springer-Verlag London Limited 2012
22 2 Vortex Dynamics

Shear layer
Vortex
Vorticity at a point
y

Boundary layer

Fig. 2.1 Vorticity corresponds to the local rotation of a fluid particle. The spatial distribution of
vorticity gives rise to different flow structures. An accumulation of vorticity in a compact region
corresponds to a vortex; an elongated distribution of vorticity corresponds to a shear layer that,
when it is adjacent to the wall, is a boundary layer

In mathematical terms, the vorticity is a vector, indicated with w(t,x) and defined
as the curl is of the velocity field u(t,x), normally expressed as the internal product
between the nabla operator and the velocity field

w (t , x ) = Ñ × u (2.1)

that in Cartesian coordinates components reads

⎡ ∂ uz ∂ u y ⎤
⎢ − ⎥
⎡w ⎤ ⎢ ∂ y ∂z ⎥
⎢ x ⎥ ⎢ ∂u ∂u ⎥
⎢w y ⎥ = ⎢ x − z ⎥
⎢ ⎥ ⎢ ∂z ∂x ⎥ (2.2)
⎢⎣ w z ⎥⎦ ⎢ ∂ u ⎥
⎢ y ∂ux ⎥
⎢ ∂ − ∂ ⎥
⎣ x y ⎦

The interpretation of vorticity is particularly intuitive in a two-dimensional flow

field, when only the x and y components of the velocity field exist. In this case vor-
ticity has only the component z, perpendicular to the plane of motion, w = ∂uy/∂x −
∂ux/∂y, and physically corresponds to (twice) the local angular velocity of a fluid
particle. In fact, a positive vorticity corresponds to a vertical velocity, uy, increasing
horizontally, along x, and a horizontal velocity ux, decreasing vertically. It is easy to
understand, see Fig. 2.1 (leftmost sketch), that this type of velocity differences about
a point represents a rotational motion (Panton 2005, Chap. 3).
2.1 Definitions 23

The relevance of vorticity is not limited to local rotation. The spatial distribution
of vorticity gives rise to different possible fluid structures; that is why vorticity is
commonly considered the skeleton of the flow field and the fundamental quantity to
define the flow structure. A vortex can be loosely described as a fluid structure that
possesses circular or swirling motion; it actually is a region of compact vorticity, a
circulatory motion surrounds a region where vorticity has accumulated. In addition
to vortices, the vorticity map allows recognition of any basic flow structure. A shear
layer, that is an elongated layer of friction between streams with differential motion,
is actually a layer of vorticity, a vortex layer. The boundary layer discussed in the
previous chapter is a vortex layer adjacent to the wall that develops because of the
velocity difference between the outer flow and the fluid attached to the wall for
viscous adherence. The correspondences between velocity and vorticity distribu-
tions are sketched in Fig. 2.1. The intensity of a vortex is normally measured by its
circulation, normally indicated with G, that is the integral of the velocity along a
closed circuit surrounding the vortex, and is equivalent to the sum of all the vorticity
(mathematically, the integral) within the vortex area contained inside the circuit.
The intensity of a vortex layer is measured by the difference of velocity, the velocity
jump Du, between the flow above and below the layer, equivalent, again, to the sum
of the vorticity (mathematically, the line integral) across the layer. Vortices and
vortex-layers are the fundamental vorticity structure in flow fields. Their different
three-dimensional arrangements and combinations give rise to the complexities of
all evolving flows.
The significance of vorticity can be best appreciated by the decomposition, due
to Helmholtz and Stokes (Panton 2005, Chap. 17), of the complete velocity field
into two distinct contributions: a rotational component urot that accounts for the
whole vorticity in the flow field, and one irrotational component uirr that is indepen-
dent from the vorticity content

u = urot + uirr (2.3)

This decomposition allows building an easier path to the intuitive understanding

of the physical phenomena concurring in a complex flow field.
The irrotational component of the velocity field is a particularly simple field, in
incompressible flows. It will be shown below that it follows from the conservation
of mass only (continuity constraint), and does not involves the equation of motion.
The irrotational flow helps to satisfy the instantaneous balance of mass without any
evolutionary mechanism, without fluid dynamics, only kinematic congruence.
A flow without vorticity thus gives rise to an irrotational velocity field only, and does
not depend on the balance of momentum. When required, the equation of motion
can be then employed to derive the pressure distribution from the velocity. In the
case of irrotational flow, this can be performed with the simple Bernoulli equation
for an ideal flow because energy dissipation is absent in an irrotational flow. In fact,
the viscous term of the Navier-Stokes Eq. 1.14, Ñ2u, which can be written for an
incompressible flow as Ñ × w, is identically zero for a flow without vorticity.
24 2 Vortex Dynamics

Fig. 2.2 A vortex is a region where vorticity has accumulated; it is not necessarily a region exhib-
iting circulatory motion. A flow made of a vortex only is made of circular streamlines (left panel).
The streamlines are modified when a uniform vertical flow of moderate (centre) ad high intensity
(right panel) is added. In the three panels the vortex is unchanged, and so is shear in the flow

These considerations on the trivial character of the irrotational component of the

flow stem, mathematically, from the fact that any irrotational vector field can be
expressed as the gradient of a scalar field, the so-called potential j, as uirr = Ñj. In
an incompressible flow, the continuity Eq. 1.5 applied to such a gradient field
becomes the well-known Laplace equation for the flow potential, Ñ2j = 0. This is a
linear equation whose solutions are easily obtained by several methods based on the
values at the boundaries. As a further property, given the linearity of the Laplace
equation, the irrotational velocity component can be expressed as a direct superpo-
sition of several elementary irrotational flows.
The velocity decomposition is the key tool to recognize the presence and the role
of vortices in a flow. A vortex, as said, is a region where vorticity has accumulated;
a vortex is not necessarily a region exhibiting circulatory motion. It may appear as
such or the circulatory pattern may remain hidden behind an irrotational contribu-
tion that covers its rotary features. The velocity field corresponding to an isolated
vortex is purely rotational; its streamlines, shown in Fig. 2.2 (left), rotate about the
vortex and describe a circulatory motion. When an irrotational contribution adds on
top of the same vortex flow, it may modify the apparent vortex signature in terms of
streamlines. To explain this point, let us consider the same vortex of Fig. 2.2 (left)
with an additional uniform flow, a rigid translational motion from top to bottom that
is evidently an irrotational component and does not affect the value of vorticity, of
shear rate anywhere. The resulting flow fields are shown in Fig. 2.2 for increasing
values of the uniform motion (central and rightmost panels). The three fields of
Fig. 2.2 present exactly the same vortex, the same gradients of velocity at all points;
nevertheless from a superficial qualitative view in terms of streamlines the underly-
ing vortex may not be equally recognizable.
Fluid dynamics phenomena related to evolutionary dynamics, friction, dissipa-
tion, forces, boundary layer, vortex formation etc., are dominated by the rotational
part of the velocity field, while the irrotational contribution may have a role in terms
2.2 Dynamics of Vorticity 25

of transport and mass conservation only. Therefore, a flow field can be evaluated
from the dynamics of the vorticity, plus an irrotational contribution to adjust mass
conservation. This is why vorticity, and vortices in which vorticity organizes, are the
fundamental elements of the flow: the skeleton and the sinews of fluid motion
(Moffat et al. 1994).

2.2 Dynamicsof Vorticity

The vorticity is the fundamental quantity that describes a fluid flow. From the
knowledge of the vorticity field only, the entire flow field inside a given geometry
can be reconstructed (technically, by inversion of Eq. 2.1). It is therefore tempting
to analyze the dynamics of a fluid motion following the dynamics of the vorticity
itself. This is often useful because vorticity occupies only a small fraction of the
flow field, and takes standard shapes that allow an immediate characterization of the
whole flow field.
The vorticity field has the further simplifying property that it obeys the same
constraint of the velocity in an incompressible fluid previously expressed in Sect.
1.2 and Eq. 1.4. Mathematically, vorticity is a field with zero divergence (simply
because the divergence of a curl is zero by mathematical definition)

Ñ·w = 0 (2.4)

This means that the vorticity field cannot take arbitrary geometric shapes.
Therefore vorticity typically develops in terms of vortex tubes (whose associated
velocity circulates around the tube) or of vortex layers (associated with a difference
of velocity, a shear rate, across the layer). Moreover, the total vorticity contained
inside a vortex tube is conserved like the discharge in a tube of flow: a vortex tube
cannot terminate abruptly, and must either be a closed ring or terminate by spread-
ing into a vortex layer.
Vorticity is an evolving field that follows deterministic evolutionary laws. Their
mathematical expression can be immediately derived from the conservation of
momentum, the Navier-Stokes Eq. 1.14, rearranged in terms of vorticity. By taking
the curl of the Navier-Stokes equation, the vorticity equation is obtained (Panton
2005, Sect. 13.3)

∂w
+ u ·Ñw = w ·Ñu + νÑ2w (2.5)
∂t

that expresses the law of motion in terms of vorticity. Despite the apparent mathe-
matical complexity, the simple qualitative inspection of this equation permits to
extract some important concepts regarding vortex dynamics. For example, it can be
26 2 Vortex Dynamics

immediately recognized that the vorticity equation does not contain the pressure
(or any conservative force like gravity). In fact, the distribution of pressure has no
direct influence on vortex dynamics; on the contrary, however, pressure strongly
depends on vorticity that rules friction and energy losses.
A first property of the vorticity evolution is that if vorticity is zero at one instant
it remains zero afterward. This is seen by inspection of Eq. 2.5 where all terms are
identically zero when vorticity is zero, and vorticity cannot change in time. Given
that vorticity cannot be created inside the fluid, thus it can only be generated at the
interface between the fluid and the boundary. This apparently simple fact is a funda-
mental element for the study of vortex dynamics: in incompressible flows vorticity
does not appear spontaneously within the fluid, the only place where vorticity can
be created is at the boundary between fluid and tissue. Indeed, the issue of the gen-
eration of vorticity, and vortex formation in particular, is a key one and it will cover
several of the following chapters.
After vorticity is generated, it is subjected to few possible evolutionary phenom-
ena. The primary one is that vorticity is transported with the flow as if it were a
passive tracer (although not effectively passive, because velocity is related to vortic-
ity itself). This is represented by the two terms on the left hand side of (2.5) that
simply describe the variation of vorticity over a particle moving with the flow. The
first term is the time variation of vorticity at the fixed position crossed by the parti-
cle; the second term gives an increase of vorticity when a particle points in a direc-
tion along which vorticity grows (i.e., when velocity is aligned with a positive
gradient of vorticity). They take a form analogous to, for example, the first two
terms in Eqs. 1.11 or 1.14, describing the acceleration on a moving particle.
Therefore, vorticity moves with the local fluid velocity, like a tracer, and can further
change its value in virtue of two additional phenomena.
The first, corresponding to the first term on the right hand side of (2.5), repre-
sents the phenomenon of increase of vorticity by vortex stretching. Consider a small
cylinder of fluid along whose axis the velocity increases, thus velocity is lower at
the base and higher at the cylinder top; as time proceeds, the cylinder elongates, it
is stretched by the velocity gradient (and shrunk in the transversal direction for the
conservation of mass). Well, the vorticity vector behaves in the identical manner as
material fluid, when fluid is stretched the vorticity vector is stretched as well and the
vorticity value increases (Batchelor 1967, Sect. 5.2). This term represents the
stretching and turning of vortex lines (Panton 2005, Sect. 13.5) as if they were lines
of fluid. A further important aspect of this term is that it is exactly zero in a two-
dimensional flow. In a two-dimensional flow, the vorticity is perpendicular to the
plane of motion and there is no velocity gradient out of plane: vorticity stretching is
intrinsically a three-dimensional effect.
Before turning the attention to the last term containing the viscous effects, let us
recapitulate the dynamics of vorticity in the absence of viscous effects. First, an ele-
ment of fluid that contains no vorticity remains without vorticity afterward. This is
the first of the three Helmholtz’s laws for inviscid flow (Panton 2005, Sect. 13.9).
Then, the vorticity is a vector that behaves like a small string element of fluid.
2.3 Boundary Layer Separation 27

It moves with the flow and it is stretched and tilted with it. This is essentially the
second Helmholtz’s law. The third law follows from the fact that vorticity is a field
with zero divergence and the total vorticity contained inside a vortex tube (or a vor-
tex filament, when it is thin enough) is conserved along the filament while it moves
with the flow.
The picture becomes extremely simple and intuitive in a two-dimensional flow,
or in a motion that is locally approximately two-dimensional. In this case, the vor-
ticity vector has a unique nonzero component perpendicular to the plane of motion,
therefore it loses its vector character and can be considered as a scalar. Stretching is
absent and vorticity is simply transported with the flow. The value of vorticity is
stuck onto the individual fluid particles; vorticity simply accumulates into vortex
patches, redistributes into vortex layer, accordingly to the motion of fluid particles.
The last, viscous term in the vorticity Eq. 2.5 introduces the effects of friction
and energy dissipation in terms of vorticity. The action of viscosity on vorticity
introduces a phenomenon analogous to that of heat diffusion or diffusion of a tracer
like ink or smoke (Panton 2005, Sect. 13.4). The distribution of vorticity is smoothed
out by viscosity; a sharp vortex reduces progressively its local strength while it wid-
ens its size in a way that the total vorticity is conserved. In general, the diffusion
process is of a simple interpretation. Like in any diffusive process, the rate of diffu-
sion is higher in presence of sharp vorticity gradients, therefore the magnitude of
viscous dissipation become increasingly relevant where vorticity presents changes
over short distances. This leads to the most important aspect of energy losses in fluid
motion: viscous dissipation is most effective at small scales. The vector property of
viscous diffusion is evidenced when it produces the annihilation of close patches of
opposite sign vorticity. This has a peculiar consequence in three-dimensions. When
two portions of vortex filaments get in contact, the opposite-sign vorticity locally
annihilates; this accompanies the reconnection of the cropped, oppositely pointing
vortex lines (that cannot terminate into the flow). The viscous reconnection phe-
nomenon is the underlying mechanism leading to topological changes, metamor-
phoses of three-dimensional vortex structures, and increased dissipation by
turbulence (see Sects. 2.7 and 2.8).
In summary, the dynamics of vorticity is made by its transport with the fluid ele-
ments, intensification by three-dimensional straining of such fluid elements, and
smoothing by viscous diffusion. A dynamics that sees vorticity arranged into tubu-
lar and sheet-like structures ensuring a continuity of vortex lines.

2.3 BoundaryLayer Separation

As said above, in incompressible flows, vorticity cannot be generated within the

fluid. Vorticity can only develop from the wall in consequence of the viscous adher-
ence between the fluid and the bounding tissue. Vorticity is produced because of the
no-slip condition at the interface between the fluid and the solid surface; it then
28 2 Vortex Dynamics

progressively diffuses away from the wall through the viscous diffusion mechanism
to produce a layer of vorticity at the boundary. The boundary layer thickness cor-
responds to the length at which the viscous diffusion penetrates into the flow, which
is proportional to √nt as taught from Eqs. 1.16 and 1.17. The boundary layer was
introduced in Sect. 1.5 as the region adjacent to the wall where the velocity rises
from the zero value that it takes at the boundary to a finite value away from it.
However, its interpretation as a vorticity layer is more intuitive for addressing vor-
tex formation processes.
The boundary layer has a fundamental importance in fluid mechanics as it repre-
sents the unique source of vorticity in a flow field. It can be easily verified that the
value of vorticity at the wall also corresponds to the wall shear rate and, after mul-
tiplication with viscosity, to the wall shear stress

tw = mw (2.6)

therefore the wall vorticity is often employed as synonymous of wall shear rate
(sometimes, given the constancy of viscosity, also of wall shear stress).
In small vessels, the thickness of the boundary layer is comparable to the diam-
eter and fills the entire flow field. At such small scales, in arterioles and capillaries,
viscous diffusion is the dominant phenomenon; vorticity smoothly diffuses into
the whole flow and vortices, with rare exceptions, are absent. On the contrary, in
large blood vessels or inside the cardiac chambers, the boundary layer often
remains thin and is capable to penetrate for diffusion over a small fraction of the
vessel size. Indeed, until it remains attached to the wall, it has a minor influence to
the flow and only represents a viscous slipping cushion for the outside motion.
However, under many circumstances, it happens that such a thin boundary layer
detaches from the wall and is ejected into the bulk flow. This is the process of
boundary layer separation, when thin layers of intense vorticity enter into the flow
and give rise to local accumulation of vorticity and eventually to the formation of
compact vortex structures.
Boundary layer separation is normally a consequence of the local deceleration
of the flow (Panton 2005, Sect. 20.11; Batchelor 1967, Sect. 5.10). The whole pro-
cess of boundary layer separation is sketched in Fig. 2.3. When flow decelerates,
the boundary layer is subjected to deceleration as well and, because of incompress-
ibility, a local stream-wise deceleration associates with a growth of the thickness at
the same location. This tongue of vorticity is lifted and strained by the outside flow
while the vorticity value at the wall below decreases. As this process progresses,
opposite sign wall vorticity appears and a secondary boundary layer develops
below the separating shear layer. The separation point at the wall, from where the
separation streamline departs, corresponds to the place where vorticity is zero. The
secondary vorticity is itself decelerated in its backward motion and is lifted up.
Eventually, it cuts the connection between the original boundary layer and
the separating vorticity that detaches and enters into the flow. It should be also
reminded that vorticity is not a passive tracer, it is made of velocity gradients and
2.3 Boundary Layer Separation 29

Fig. 2.3 Sketch of the boundary layer separation process. The dark gray indicates layers with
clockwise vorticity, the light gray is counter-clockwise; streamlines and velocity profiles are
drawn. The deceleration of the flow produces a local thickening of the boundary layer due to mass
conservation balance (upper panel). Such emerging vorticity is therefore lifted and transported
downstream by the external flow (see arrows). A shear layer then extends away from the wall and
produces a secondary boundary layer, with oppositely rotating vorticity (mid panel). The separated
clockwise vorticity tends to roll-up while the secondary layer lifts up for the same initial mecha-
nism, because it backward motion is decelerating (see arrows). Eventually, the separating vortex
layer detaches from the boundary layer and becomes an independent vortex structure
30 2 Vortex Dynamics

it represents the underlying structure of the flow itself. Fig. 2.3 shows the qualita-
tive velocity profiles and streamlines that develop in correspondence of the sepa-
rating vorticity field.
Boundary layer separation is thus a consequence of the local deceleration of the
flow. In other terms, separation develops in presence of an adverse pressure gradi-
ent (pressure growing downstream) that pushes from downstream and decelerates
the stream. The most common way to have an adverse pressure gradient is that of a
geometric change: a positive curvature of the wall, like an enlargement in a vessel.
In this case, the velocity decreases, for mass conservation, kinetic energy decreases,
and the value of pressure increases for the Bernoulli balance. Therefore, boundary
layer separation develops behind a stenosis, or at the entrance of an aneurism. An
extreme case of geometric change is that of a sharp edge, this is often found at the
entrance of a side-branching vessel, and certainly on the trailing edge of the leaflets
of the cardiac valves. In the case of sharp edges, the flow deceleration is so local that
the position of boundary layer separation is definitely localizable at the edge. The
vorticity that developed on the upstream side detaches at the sharp edge and leaves
the tissue tangentially.
Geometric changes are not the unique possible sources for the development of a
flow deceleration. Immediately downstream of branch sucking fluid away from a
main vessel, the velocity reduces and an adverse pressure gradient develops.
Similarly, boundary layer separation develops for the so-called splash effect, when
a jet reaches a wall and produces high velocity streamlines that decelerate when
they are deflected along the wall. Finally, the local flow deceleration is often pro-
duced by previously separated vortices. A vortex that gets close to a wall gives rise
to a localized increase (or reduction, depending on its circulation) of the flow veloc-
ity at the wall below, and a corresponding deceleration immediately downstream
(or upstream). The vortex-induced boundary layer separation is a frequent phenom-
enon that may become particularly critical in some applications. In fact, the area of
principal separation is often localizable and properly protected, whereas an unex-
pected separation induced downstream due to a previously separated vortex may
occur at unexpected locations.

2.4 Vortex Formation

The separation of the boundary layer represents the starting phase of the vortex
formation process. The featuring property of any shear layer is the difference of
velocity between its two sides: the farther side of shear layer that detaches from the
wall moves with a speed that is higher than the side closer to the wall. Therefore, the
separating shear layer curves on itself and eventually rolls-up into a tight spiral
shape. Now, during the rolling-up process, the distance between two successive
turns of the vortex layer progressively reduces, with the closest neighboring turns at
the center of the spiral. The viscous diffusion process smears out this tight spiraling
structure into by a compact inner core with a smooth distribution of vorticity (Wu
et al. 2006, Sect. 8.1).
 2.4 Vortex Formation 31

Fig. 2.4 Vortex formation

from a sharp edge obstacle.
The shear layer separates from
the upstream “wetted” wall
and rolls-up into a spiral. The
tight turns in the inner part of
the spiral spread for viscous
diffusion into the inner core of
the formed vortex
[AU1]

The roll-up and the formation of an isolated vortex behind a sharp edge obstacle
are shown in Fig. 2.4. In the case of such a sharp geometric change the boundary
layer separation localizes at the edge, the boundary layer from the upstream “wet-
ted” face of the obstacle leaves the edge tangentially and immediately rolls-up into
a spiral. The vorticity viscous diffusion acts with higher strength in the tighter inner
spiral branch and gives rise to a smooth vortex core.
An isolated forming vortex grows with a self-similar shape until there are no
external disturbances that can influence its formation. Thus until the vortex size is
small enough in comparison to the size of the surrounding geometry. The properties
of the initial self-similar growth can be obtained by simple dimensional arguments.
Assume that the bulk velocity grows proportionally to ta and that separation occurs
from an edge of internal angle b, such that b = 0 is a diaphragm and b = 90 is a cor-
ner. It follows that the flow velocity around the edge is given by Atarl-1 where r is the
distance from the edge, A is a dimensional coefficient, and l = 180/(360 − b) (see
for example Batchelor 1967, Sect. 6.5; Pullin 1978; Saffman 1992, Sect. 8.5). Then,
from dimensional arguments (based on the fact that A and t are the only dimensional
quantities available), the typical length size of the vortex increases in time as tn, with
n = (1 + a)/(2 − l), and the vortex intensity, its circulation, grows like t2n − 1.
These estimates allow evaluating the force acting on the surrounding walls due
to the vortex formation, which turns out to be proportional to t2n − 2 (see Sect. 2.6).
Vice versa, they allow evaluating the flow corresponding to a given force (or pres-
sure difference). For example, in an orifice with zero internal angle, b = 0 (l = 1/2),
a flow time-profile ta associates with a force going like t(4a − 2)/3; this shows that a flow
that increases faster than square root growth (a < 1/2) requires an unrealistic, theo-
retically infinite, effort.
32 2 Vortex Dynamics

[AU1]

Fig. 2.5 Formation of vortices behind a circular cylinder. Oppositely rotating vortices separate
from the two sides of the body in an alternating sequence. The previously separated clockwise
vortex detached from the upper wall translated downstream, a counter-clockwise vortex has been
formed from the lower wall, and a novel clockwise vortex is under formation from the wall above

This typical roll-up phenomenon can be disturbed during its development in par-
ticular when the shear layer is particularly thin. A curved shear layer is, in fact,
subjected to an intrinsic instability that gives rise to the birth of wavy disturbance
and the following roll-up of multiple small double spirals along the shear layer itself
(Pullin 1978; Luchini and Tognaccini 2002; Pedrizzetti 2010). This instability has
the origin in the Kelvin-Helmholtz instability for an infinitely thin vortex sheet
(Batchelor 1967, Sect. 7.1) and it is more effective the thinner the vortex layer is.
The thickness of the shear layer is given by the boundary layer thickness prior to
separation, as given by Eqs. 1.16 or 1.17. In general, a separating shear layer is rela-
tively thin in large vessels, especially when it detaches in a sharp enlargement after
a short converging section, where the boundary layer is kept attached by the spa-
tially accelerating flow. This is the case of the trailing edge of cardiac valves.
The vortex formation from a smooth surface is still described by the picture given
above, where a few additional elements of complexity can be emphasized. First, the
actual position of separation depends on the local flow structure; it cannot be prelimi-
narily identified and may even change during time (Pedrizzetti 1996). Furthermore,
the separation from a smooth surface is inevitably accompanied by a more direct
interaction between the forming vortex and the nearby wall when the viscous dissipa-
tion effects normally support the formation of smoother vortex structures.
One typical example of the external separation from the smooth surface of a bluff
body is shown in Fig. 2.5 featuring the formation of oppositely rotating vortices
from the two sides of a circular cylinder. In such an example such vortices interact
and influence the opposite separation process eventually producing a sequence of
alternating vortices known as the von Karman street that is usually found behind
bluff bodies (Panton 2005, Sect. 14.6). The development of alternating vortices is
quite a common phenomenon when previously separated vortices may influence
vortex formation in nearby regions. It is also present, with some differences, in
internal flows when a vortex formed on one side of a vessel creates a vortex-induced
separation on a facing wall. That, in turn, may induce a weaker further separation in
a sort of wavy pattern extending and decaying downstream.
 2.4 Vortex Formation 33

[AU1]

Fig. 2.6 Formation of vortices at a carotid bifurcation. The accelerating systolic flow (upper
panel, at peak systole) leads to a smooth boundary layer separation at the carotid bulb. After the
peak (lower panel) the vortex just formed at the bulb either interacts with the bulb boundary layer
creating multiple small vortices, and gives rise to a vortex-induced secondary separation in the
oppositely facing wall of the internal carotid artery. The same phenomena in a much weaker ver-
sion are noticeable also on the opposite side at the entrance in the external carotid artery

The internal separation, with the following formation of a vortex inside of a ves-
sel is in general a smoother phenomenon, because the presence of confining walls
does not allow vortices to grow into large structures, keeps vortices more constrained
within smaller scales, and is more influenced by viscous diffusion. Nevertheless, the
presence of a vortex inside a vessel may change the entire flow. It has a blocking
effect that locally deviates the streamlines modifying the wall shear stress distribu-
tion, possibly producing further separations. It changes the unsteady pressure drop,
and in a branching duct it may affect the relative flows division in the daughter ves-
sels. An example is given in Fig. 2.6 that reports the vortex formation in the bulb of
a carotid bifurcation. During the systolic acceleration, the boundary layer separates
tangentially from the common carotid artery and develops a smooth roll-up within
the bulb close to the nearby wall. During deceleration, the formed vortex locally
affects the wall shear stress inside the bulb with multiple opposite sign wall vorticity.
34 2 Vortex Dynamics

It has a blocking effect that deviates the streamlines at the entrance of the internal
carotid artery into a faster jet. It produces secondary vortex-induced separation
inside the internal carotid that eventually (not shown in the picture) gives a second-
ary vortex formation and a further small separation little downstream.
A general analysis of the vortex formation process can be outlined when the flow
enters from a small vessel into a large chamber forming a jet whose head is the
forming vortex. Here, after the very initial roll-up phase, a measure of the length of
such a jet is given by the product of Vt, where V is the velocity at the opening and t
is the time. In this case it is enlightening to define a dimensionless formation time,
FT, introduced by Gharib et al. in 1998, as the ratio of the jet length with respect to
the diameter of the opening D

V ×t
FT = (2.7)
D

The formation time represents a dimensionless number that characterizes the pro-
gression of vortex formation. As such, it allows the unitary description of the vortex
formation processes as they occur under different conditions. In reality, the defini-
tion of formation time has a more profound physical meaning (Dabiri 2009). The
separating shear layer has a strength given by the jump of velocity between its two
sides, given approximately by V, and translates downstream with a velocity that is
again proportional to V, thus it feeds the circulation G of the forming vortex at a rate
dG/dt » V2. The formation time thus also represents the dimensionless measure of
the vortex strength, the circulation G, normalized with VD. The definition (2.7) can
be extended to the case when either V or D vary during time, by integration of the
ratio V/D during the period of vortex formation.
The generality of the formation time concept permits to uncover general proper-
ties of the vortex formation that are common to the different cases. These, which do
not appear in the simple description of vortex formation given above, will be
revealed in the next chapter.

2.5 Three-Dimensional Vortex Formation

The vortex formation process described in the previous chapter is given in terms of
two-dimensional pictures. It allows an immediate and intuitive understanding of the
fundamental phenomenon. Actually, the initial phase of any vortex formation pro-
cess is, with rare exceptions, always two-dimensional and the three-dimensional
influence enters into play at some later stages.
The two-dimensional description implicitly treats the vorticity as a scalar quan-
tity; this is a common practice because vorticity vector has the unique component
that is perpendicular to the two-dimensional plane of motion. On the opposite, in
order to understand the three-dimensional features of vortex formation, it must be
reminded that vorticity is a three-dimensional vector. In particular, that vorticity is a
2.5 Three-Dimensional Vortex Formation 35

solenoidal vector field, with zero divergence as dictated by the condition (2.4).
It means that vorticity behaves like an incompressible flow: vortex lines, lines every-
where tangent to the vorticity vector, must be continuous and cannot originate or
terminate in the flow. In a viscous flow they are always closed lines, although some-
time complicated.
A consequence of the continuity of vortex lines is the concept of vortex tube,
sometime called vortex filament when the tube is thin enough. A vortex tube is a
thick collection of vortex lines, a tube whose lateral surface is made of vortex lines.
The solenoidal nature of vorticity imposes that a vortex tube typically maintains its
individuality during flow evolution. It is transported by the local flow, deformed and
stretched by the velocity gradients, and enlarges because of vorticity diffusion,
while it maintains its individuality. There are, however, situations when a vortex
tube must abandon its individual life. When a tube approaches another tube, the
nearby vortex lines belonging to different tubes begin to wrap one around the other,
with a strong local stretching that is ultimately smoothed out by viscosity. In such
“close encounters” a tube may fuse with another tube, filament, or vortex lines.
Another concept that enters in three-dimensional vortex dynamics is that of self-
induced velocity. Let us remind that a two-dimensional vortex is actually a rectilin-
ear vortex tube that does not vary along the third direction; the corresponding
velocity field is a rotation on and around the tube. A three-dimensional vortex tube,
in general, is not rectilinear and presents a curvature that may change along its
length. Well, because of the relation between velocity and vorticity, a curved vortex
tube corresponds to a velocity field made of a rotation around the tube plus a transla-
tion, or self-induced velocity, of the tube itself.
This is easily visualized: assume that the rotation velocity around the curved
vortex tube is such that the velocity is externally upward and internally downward;
this means that every small portion in the curved tube also pushes downward the
nearby elements (that are in the internal side with respect to the tangent), with an
overall result of a downward translation. The self-induced velocity of a curved vor-
tex filament is (Saffman 1992, Sect. 11.1) proportional to the vortex circulation G,
its curvature 1/R, where R is the radius of curvature, and also weakly influenced by
the ratio R/c where c is transversal size of tube. In formulae, the velocity has an
intensity G/4pR log(R/c) (remind that the logarithm is a very slowly growing func-
tion) and is directed perpendicular to the plane that locally contains the filament.
Therefore the tighter is the curve the higher is the self-induced velocity, with the
additional element that thin filaments are some time faster than fat ones.
With these concepts in mind we can consider and describe vortex formation in
three-dimensional flows.
The simplest case of three-dimensional vortex formation is that from a circular
orifice, in that case the forming vortex tube has the shape of a ring. Vortex rings are
well known objects of fluid dynamics (Shariff and Leonard 1992) that are easily gener-
ated using a piston-cylinder apparatus. A vortex ring is a stable vortex structure, it has
an axial symmetric and vortices with a shape close to a ring also tend to the axisym-
metric shape by an internal homogenization. Because of their stability, vortex rings are
often encountered in nature, including when puffing smoke out of the mouth.
36 2 Vortex Dynamics

Fig. 2.7 Formation of a vortex ring from a circular sharp orifice. The orifice velocity is here con-
stant during time, this snapshot corresponds to a formation time FT = 3. Left panel: distribution of
vorticity on a transversal cross-cut; the vortex core is indicated with a dashed line. Right panel:
three-dimensional view of the vortex ring corresponding to the core of the forming vortex visual-
ized by the l2 method (see text). The shear layer separating from the edge rolls-up into a vortex
ring that corresponds to the jet head. The vortex has a curvature, the self-induced velocity is
directed downstream and adds on top of the background velocity

Figure 2.7 shows one instant, corresponding to a formation time equal to 3, dur-
ing the formation of a vortex ring behind a circular orifice. The vorticity distribution
on a transversal section (left panel) shows the shear layer separating from the orifice
that eventually rolls-up into the jet head; however it must be kept in mind that this
planar picture corresponds to a three-dimensional vortex structure that is more dif-
ficult to represent on paper. The vortex ring corresponding to the vortex core is
shown (right panel) to emphasize the main element of the three-dimensional vortex.
In general, however, there is some ambiguity on the effective delineation of a vortex
boundary. This is not a big issue in two-dimensional systems when the entire vortic-
ity field can be shown in color scale on the picture plane and the different elements
of the vortex structure are immediately recognized, from the separating shear layer,
to the rolling-up spiral, to the vortex core (see the left panel of Fig. 2.7). However,
this case is particularly simple because the vorticity has an axial symmetry and
only the azimuthal component: this flow is conceptually planar. Nevertheless, its
2.5 Three-Dimensional Vortex Formation 37

three-dimensional representation, on the right panel of Fig. 2.7, certainly contains

less complete information, and the choice of the vortex core boundary severely
influences the three-dimensional structure that is eventually visualized.
The definition of a vortex structure is a critical issue in three-dimensional flows,
when vorticity is a vector field arbitrarily extended in three-dimensional space.
A generally accepted definition of a vortex is still lacking. The level of vorticity
cannot be sufficient because the highest magnitudes are typically found in the sepa-
rating vortex layer, or in the boundary layer, structures made by vorticity that have
not yet become a vortex. Loosely speaking, a vortex is a region where vorticity
levels are higher than in the surrounding, where the vorticity distribution presents
some coherence, and also where the relative motion of surrounding fluid elements
follows circular paths or, more precisely, where the fluid is subjected to a centripetal
acceleration. There are several tentative definitions; currently, unless the specific
problem suggests an ad-hoc definition, the most accepted technique to identify a
vortex structure is the so-called l2 method introduced in by Jeong and Hussain
(1995). There the vortex boundary is identified by the constant level of a scalar
quantity, l2, evaluated from the properties of velocity gradient.1 Specific details are
contained in (Jeong and Hussain 1995) where a review of other existing methods is
also given. In brief, physically, this quantity identifies the regions where fluid pres-
sure is minimum with the further care that only pressure gradients imputable to
vorticity (rotational motion) are accounted, leaving aside the influence of the irrota-
tional part of the velocity. Therefore the scalar quantity, l2, takes minimum values
at the center of a vortex where pressure is low because centrifugal forces push the
flow away. This technique has been successful in several applications, ranging from
simple vortex flows to turbulence, and it allows extracting and visualizing the coher-
ent vortex structures and then to build interpretations schemes of their dynamics.
Vortex rings are the simplest vortex structures that exhibit phenomena that are
typical of three-dimensional vortex dynamics. A vortex ring presents a self-induced
velocity proportional to its circulation and its curvature. Such a self-induced velocity
gives rise to a peculiar limiting process of three-dimensional vortex formation that
was first reported by Gharib et al. in 1998. During its formation, the vortex ring is
continuously fed by the rolling-up shear layer separating from the orifice edge, there-
fore its circulation grows and the self-induced vortex translation velocity increases as
well. The self-induced translation velocity of the vortex ring rises until it exceeds the
velocity of the separating shear layer. At this point, the primary vortex detaches from
the layer behind with a phenomenon known as pinch-off. At the same time the newly
separated vorticity cannot reach the escaped vortex and eventually rolls-up in its
wake. In one sentence, vortex ring pinch-off occurs when the velocity of the trailing
jet falls below the celerity of the leading vortex ring (see Dabiri 2009 for a review).

1
The scalar l2 is the intermediate eigenvector of the tensor build by the sum D2 + W2, with D and W
the rate of deformation and rotation tensors, respectively, defined also as the symmetric and asym-
metric parts of the velocity gradient. This sum corresponds to the contribution to pressure curvature
imputable to rotary motion alone, when this contribution to pressure has a minimum two out of
three eigenvalues must be negative, otherwise two of them are positive. Therefore monitoring the
intermediate eigenvalue allows verifying the presence of such vorticity-induced pressure minima.
38 2 Vortex Dynamics

Fig. 2.8 Formation of a vortex ring from a circular sharp orifice. The orifice velocity is here con-
stant during time, this snapshot corresponds to a formation time FT = 5. Left panel: distribution of
vorticity on a transversal cross-cut; vortex cores are indicated with dashed lines. Right panel:
three-dimensional view of the vortex rings corresponding to the core of the forming vortices. The
primary vortex has grown until its self-induced velocity has become larger than that of the shear
layer behind, afterward the principal vortex escaped downstream and the shear layer produces
smaller vortices in its wake

The timing of this limiting process has been found to be well described in terms
of the formation time introduced at the end of the previous chapter. It revealed the
existence of a critical value for the formation number, about FT » 4. Above this limit,
the vortex ring cannot grow as a unique structure and multiple vortices develop in its
wake. Indeed, the self-induced velocity of a curved vortex is proportional to G/D and
the formation number can also be interpreted as the ratio between the velocity of the
vortex ring and that of the shear layer. One example of the vortex ring formation
process for a formation time larger than the critical value is shown in Fig. 2.8.
The case of vortex ring formation after a circular opening represents a prelimi-
nary conceptual basis for the interpretation of the more complex phenomena involved
in the three-dimensional vortex formation from general geometries. Let us move
forward and consider the flow across sharp edge orifices with a slender shape. In this
2.5 Three-Dimensional Vortex Formation 39

case, the opening has a variable curvature and the separating vortex filament shall
initially present a variable curvature along its axis as well. These differences give
rise at least to two potential effects in sequence. First, the self-induced velocity,
which is proportional to the curvature, will be different along the vortex filament and
will progressively deform it. Secondly, the formation time depends on the local cur-
vature and the vortex will reach the limiting critical value at different times along the
different portions of the filament. These phenomena give rise to the deformation and
eventual three-dimensional metamorphoses of the vortex structure separating from a
non-circular orifice. Furthermore, once the vortex loses its stable regular shape, the
three-dimensional interactions give rise to a progressive destruction of the vortex
into smaller elements, which in turn deform into even smaller ones, until they are
dissipated for viscous effects. In general, vortex formation after three-dimensional
geometry may give rise to irregularly shaped structures, these become unstable and
undergo to a rapid energy dissipation. One exemplary case of the three-dimensional
vortex formation from a slender orifice is shown in Fig. 2.9 (Domenichini 2011).
The concept of limiting vortex formation, previously described for vortex rings,
presents a renewed role in the prediction of the transition toward the breaking and
dissipation of three-dimensional forming vortices. In fact, the major disturbance to
the stability of a vortex loop occurs when part of it, the most curved portion, reaches
the limiting formation time and develops small vortices. The limiting formation
time should be computed along the vortex loop with the local value of the orifice
curvature. Its variability gives the differential timing of the limiting process and
represents a measure of the instability of the formed vortex structure. Its smallest
value represents the earliest reach of the limiting process and the beginning of the
transition toward three-dimensional disturbance.
The three-dimensional vortex formation from smooth surfaces, after a constric-
tion like a stenosis or in a vessel enlargement, introduces additional elements of
complexity that do not allow drawing a simple unitary picture of the involved phe-
nomena. The initial instants following boundary layer separation and initial roll-up
are essentially two-dimensional with a moderate influence from the three-dimensional
structure. Afterward, the differential vortex formation leads to developments of
widely different results depending on the differences in the separating geometry, in
the interaction with the nearby walls and with other surrounding elements. Specific
examples of such vortex formation phenomena as they realize in sites of interest for
the cardiovascular circulation are provided through this volume.
An important source of complexity comes from the potential variability in the
localization of the separation. In the two-dimensional description, the separation
departs from the smooth wall at a point that, as said earlier, may change over time.
In three-dimension, when the transversal span is included, the separation point
transforms into a separation line. Until this line has a smooth geometry, it reflects
into the formation of a smooth vortex loop. More often, the separation line is irregu-
lar along its length and deforms in time giving rise to an irregular vortex structure
that immediately develops into a dissipative, chaotic, three-dimensional structure.
It is altogether common that the separation line is not even a closed line but
a segment of finite length. In this case, the separation is localized and remains
 40 2 Vortex Dynamics

[AU1]

[AU2] Fig. 2.9 Three-dimensional vortex formation from a slender orifice, made of two half circles con-
nected by straight segments. The orifice velocity is here constant during time, the four snapshots
correspond to a time sequence where the value of the formation time are FT = 2.5, FT = 3.75,
FT = 5, FT = 6.25. One quarter of the entire space is shown for graphic clarity (allowed by sym-
metry); the vorticity contours are reported on the side planes to help understating the three-dimen-
sional arrangement of the principal vortex filaments. In the initial phase, the formed vortex loop
presents a variable curvature and deforms because of the higher self-induced translation speed in
the more curved parts; such a deformation leads to further changes in the three-dimensional curva-
ture and further deformations. Later on, the vortex reaches the limiting formation phase behind the
circular part of the orifice only, where smaller vortices appear. Afterwards the vortex structure
loses its individuality and becomes a set of entangled three-dimensional elements that rapidly dis-
sipate for viscous stresses
2.6 Energy Loss and Force of Vortex Formation 41

Fig. 2.10 A vessel that presents an obstacle partially obstructing the fluid flow, may give rise to
vortex formation that, in turn, generates a force on the obstacle. The picture reports example of a duct
with a diaphragm (top-left), a vessel with a stenosis (bottom-left), and a trans-valvular flow (right)

incomplete in its three-dimensional formation. For example, a vessel can present, as

it typically does, a constriction on one side only of the wall, and the vortex develops
and rolls-up only locally. Nevertheless, the solenoidal constraint for the vorticity
fields requires that any vortex line cannot terminate into the flow. Therefore the
incompletely formed vortex loop, locally separated from the boundary layer, remains
connected with the smooth vorticity distribution contained in the remaining bound-
ary layer. Such type of vortex structure, with a reverse U-shape, normally called
vortex hairpin, are evidently not stable, their tip is slower than the flow, and rapidly
undergo a three-dimensional dissipative evolution.
In conclusion, three-dimensional vortex formation from circular geometries pro-
duces ring-like vortex loops that are subjected to a limiting process, and gives rise
to multiple rings when this limit is overcome. Formation from irregular geometries
normally leads to unstable vortex structures that are rapidly dissipated. Under rela-
tively simple conditions, the limiting process still allows to evaluate the presence
and the timing of the eventual vortex breakdown. The complexity of the actual phe-
nomena involved should, however, be evaluated in the different specific conditions.

2.6 Energy Loss and Force of Vortex Formation

The previous chapters have evidenced how the vortex formation process dramati-
cally influences the flow motion. In addition to this, formation also reflects in the
generation of dynamical actions on the surrounding tissues and in energetic losses
along the vessel.
Consider an obstacle, like a diaphragm, a valve, a stenosis, that partially obstructs
the otherwise free flowing of the fluid inside a vessel, as shown in Fig. 2.10. In
absence of any vortex formation, pressure would change as dictated by the Bernoulli
balance. However, the development of a vortex provokes an additional pressure
drop (or energy loss) due to the transformation of energy into vortex inertia, and an
additional longitudinal force on the obstacle wall.
42 2 Vortex Dynamics

Fig. 2.11 A schematic portion of a straight duct with an obstacle. The process of vortex formation
gives rise to an irreversible pressure drop Dp, and to a vortex-force on the obstacle. These are typi-
cally estimable from global balances without the need to evaluate local details of the flow field

Let us evaluate the pressure drop in the simple configuration of a rectilinear duct
with a diaphragm inside, as sketched in Fig. 2.11. This example allows evidencing
the pressure drop due to vortex formation only, leaving aside the transformation
between kinetic energy and pressure caused by variation of the duct size. The eval-
uation of pressure losses in a partially obstructed straight vessel can be performed
by the equation of motion (like the Bernoulli theorem, in Eq. 1.9). Using the equa-
tions with particular care (written along the streamlines and summed-up for all
stream-tubes connecting the inlet to the outlet) the pressure drop can eventually be
expressed as

dU r dI
Dp = rL + (2.8)
dt A dt

where any viscous dissipative effects have also been here neglected. Both the two
terms on the right side derive from the inertial part of the Bernoulli theorem inte-
grated over all stream-tubes. The former the inertia of the average motion, this is
a reversible pressure loss that sums to zero in a time-periodic motion when the
acceleration and deceleration and a net sum equal to zero. The latter represents
the additional change of fluid inertia that is imputable to the formation of the
vortex. And it is measured by the variation of the impulse, I(t), of the vortex sys-
tem. The vortex impulse defined in general as a vector quantity (Saffman 1992,
Sect. 3.2)

1
I = ∫ w × xdV (2.9)
2

of which only the longitudinal component enters in (2.8). The apparently compli-
cated definition of impulse (2.9) can be made more explicit when considering the
formation of an individual vortex of growing strength G(t) from a constriction of
area A0. The impulse is then given by the circulation multiplied by the area sur-
rounded by the vortex I(t) » A0G(t), for example the impulse of a vortex ring of
radius R is I = pR2G, and from the pressure drop simplifies to (see also Saffman
1992, Sect. 3.8)
2.6 Energy Loss and Force of Vortex Formation 43

dU A dG
D p = rL +r 0 (2.10)
dt A dt

The pressure loss caused by vortex formation, expressed by the second term of
Eqs. 2.8 or 2.10, is an inertial effect. It is a consequence of the adherence of the fluid
onto the wall because of viscosity, although, it is independent from the actual value of
viscosity. In summary, vortex formation gives an irreversible transformation of energy
into inertia, due to viscous adherence, but independent from the value of viscosity.
The presence of an obstruction in the otherwise rectilinear duct deviates the flow
and provokes the development of a force on the obstacle whose strength increases
during a vortex formation process. This force follows from the difference of pres-
sure between the front and the back faces of the obstacle. The high pressure in the
front face is due to the flow that impacts on it to be deviated (in Bernoulli terms, the
velocity decreased approaching the front face of the obstruction and the pressure
rises). The dramatically lowered pressure on the back face, instead, is very much a
consequence of the vortex generation. It is due to the sharp pressure decrease across
the separating shear layer that connects the wall to the growing vortex. The total
force made by the fluid onto the obstacle can be evaluated by integrating the value
of pressure and shear stresses all over its solid surface. This direct calculation is
often unfeasible because it requires a very accurate knowledge of the fluid properties
very close to the interface with the solid boundary. An alternative approach follows
from the integral balance of momentum (the product of mass and velocity integrated
over the entire volume). It states that the rate of change of momentum within any
arbitrary region of fluid, cleared from the net flow of momentum across the bound-
ing surfaces, can only be imputable to the forces that act on the fluid contained
inside that region (see, for example, Panton 2005, Sect. 5.14). This is again Newton’s
second law, here expressed in integral form for an entire region that reads

dM
+M =G +P (2.11)
dt flux

where the left side contains the rate of change of momentum dM/dt (the acceleration
of the fluid) and the flux of momentum Mflux, (positive when exiting). On the right
hand side, G indicates the volume forces, like gravity, and the symbol P represents
the forces acting on all the surfaces bounding the chosen region. These include both
the contribution of pressure difference at the open ends of the region and the total
stresses acting between the fluid and the obstacle, which is equal and opposite to the
unknown force made by the flow onto the obstacle.
Let us consider again the simple duct with constant cross-section sketched in
Fig. 2.11, with the objective of evaluating the longitudinal force, F, exerted on the
solid obstacle because of the vortex formation process. In this case, the total flux of
momentum is zero because the same amount that enters from the inlet exists at the
outlet, and the volume force G due to gravity is simply the static vertical weight of the
fluid volume weight that can be ignored. The longitudinal balance (2.11) thus
expresses a dynamic equilibrium between the change of momentum, dM/dt, and the
44 2 Vortex Dynamics

surface forces, P. First, the product of the fluid density with the velocity gives the
momentum M over the whole volume. It is easy to understand that, in a duct, the
whole volume can be spanned by a sequence of cross-section slices, and that the inte-
grated velocity is the total fluid discharge, Q = U × A, across that section. When the
lateral walls do not move (or their velocity is negligible) the discharge does not vary
along the length of the duct, and the rate of change of momentum is proportional to
the rate of change of the discharge. Mathematically, it can be expressed as r × L × dQ/dt,
where r is the fluid density, and L is length of the considered portion of the duct. The
surface force, P, presents two different contributions. First, the forces acting on the
open ends: the forward pushing pressure at the inlet section and the backward pushing
pressure at the outlet, which sum up to Dp × A. Second, the total longitudinal force
made by the entire solid boundaries, −F. This includes both the actual force across
the contour of the obstacle and the viscous shear stress on the lateral surfaces that are
normally negligible along short tracks. Summing up, Eq. 2.11 becomes

dQ (2.12)
rL = DpA − F
dt
Insertion of the expression for the pressure drop that was previously evaluated in
Eq. 2.8 shows that the force generated on an obstacle by vortex formation can be
primarily expressed by

dI
F=r (2.13)
dt

neglecting the additional contributions that may come from viscous losses. In gen-
eral, when the tissues involve a more complicated geometry and possibly movable
sidewalls, further terms should be included. These, however, do not relate directly
with the vortex formation and are typically estimable from geometry and average
flow properties.
In the simpler case of a single vortex of circulation G(t) across an area A0, when
the pressure drop is expressed by Eq. 2.10, the force simplifies in

dG
F = rA (2.14)
0 dt

This expression permits to associate a force to those elementary vortex formation

processes when the intensity growth of the vortex is somehow estimable. For exam-
ple, the initial strength of a starting vortex can be evaluated from general vortex
formation concepts, like those introduced in Sect. 2.4. Therefore, when a flow accel-
erates across a sharp orifice, the resulting force during the starting phase turns out
to be proportional to rA0(U2/t)2/3. On the other long-time extreme of a jet, when its
head vortex is far downstream, the separating shear layer, of intensity U, translates
straight downstream with a velocity proportional to U. In this case, see the end of
Sect. 2.4, the emitted circulation rate is dG/dt » U2 and the force about rA0U2.
2.7 Vortex Interactions 45

Recapitulating, the phenomenon of vortex formation generates an additional

force on the obstacle that is proportional to the orifice area and the rate of growth of
the vortex strength. Vortex formation is associated with force generation, an effect
that may be beneficial or detrimental depending on the specific. For example, vortex
formation behind a valvular leaflet supports and regulates its opening/closing
dynamics; vice versa, when a stenosis is subjected to vortex formation it suffers a
streamlined hammering at every heartbeat.

2.7 Vortex Interactions

When two or more vortices come nearby each other, they likely interact in an intense
and irreversible manner. The interaction of vortices involves many different and
very complicated phenomena, the principal are outlined here.
Let us first consider two-dimensional vortices. A vortex is associated with a
rotating flow about it whose velocity is proportional to the vortex circulation, G,
and inversely proportional to the distance, r, from the vortex center: v = G/2pr.
Two vortices that come in close encounter reciprocally induce such a rotation
velocity each other. When such vortices have the same sign they rotate together
one around the other; in addition, the differential velocity within each individual
vortex deforms it and makes the vortices winding up one over the other and even-
tually merge into a single larger one made by the sum of them. This process is
associated with little energy dissipation. On the contrary, two vortices with oppo-
site circulation, a vortex pair, translate together for the self-induced velocity (simi-
larly to what a vortex ring does) along a straight or curved path depending on the
relative strengths. Again, the differential velocity inside each single vortex pro-
duces the winding up of one’s vorticity strip on the other; however, such vorticity
strips are of opposite sign and do not merge rather they annihilate each other and
reduce the individual vortices’ strength.
The close encounter of three-dimensional vortex loops begins with the local
interaction between the closest tubular elements of the two vortices that, initially, is
nearly two-dimensional. Let us first remark that a close encounter between tubular
elements of the same sign is an extremely rare event because the overall self-induced
velocity of the corresponding vortex loops would tend to separate the vortices. Thus,
three-dimensional interaction begins prevalently between two oppositely rotating
portions of a vortex tube. One example of the interaction between two identical
vortex rings is shown in Fig. 2.12. Initially, the local interaction is approximately
the two-dimensional process described above: the nearby oppositely rotating tubu-
lar elements induce the velocity each other and try to translate away. This produces
a local stretching of the three-dimensional vortex tube, a stretching that accelerates
while the tubes become closer and, in a non-symmetric case, would locally wind up
one another. The interacting structures develop increasingly small scales until vis-
cous diffusion becomes a dominant effect, at this point the reconnection of vortex
46 2 Vortex Dynamics

[AU1]

Fig. 2.12 Vortex reconnection, and topological metamorphosis between two impacting vortex
rings of equal circulation; the brightness of the filament indicates the strength of the corresponding
vorticity. When oppositely rotating vortex tubes get close, they produce a local vortex stretching
due to the self-induced velocity (from left to central panels). During stretching, the boundary
between the vortices becomes locally sharper until the filaments fuse one into the other for viscous
effect (from central to right panels). After vortex reconnection a new structure is formed, typically
its geometry is irregular, the vortex is often unstable and short lived

lines occurs: adjacent opposite vorticity is annihilated by dissipation and the vortex
tubes tend to fuse one onto the other (Kida and Takaoka 1994).
The interaction between two identical vortices, like that shown in Fig. 2.12, may
result into a complete vortex reconnection and a relatively simple new vortex tube.
More often, however, one vortex is stronger than the other is, only part of its tubular
structure can reconnect with the other weaker vortex, and the incomplete reconnec-
tion gives rise to new vortices with a complex branched geometry. In general, the
vortex structure resulting from the fusion of previous interacting vortices, typically
presents a very irregular geometry. Differential curvatures, which give sharply vari-
able self-induced velocity and local motion, and differential vorticity strength,
which gives axial flow along the tube, tend to rapidly further deform the vortex,
produce further reconnections, and give rise to smaller vorticity structures. In other
terms, an irregular three-dimensional vortex structure is overall unstable, tends to
destroy itself, and is short lived. The more a vortex is regular, like a vortex ring, the
more it remains coherent and lasts longer.
A special case of vortex interaction, which is particularly relevant in closed sys-
tems like cardiovascular vessels, is the interaction with a nearby wall. The vortex-wall
interaction can be divided into two different phenomena: the irrotational interaction,
that is a consequence of the wall impermeability; and the viscous interaction with the
vorticity in the boundary layer. Let us consider the two effects separately.
First, an isolated vortex induces a rotary motion where streamlines are circular.
When such a vortex approaches an impermeable wall, the streamlines must deform
to avoid crossing the boundary. With reference to Fig. 2.13 (left panel), the modi-
fication of the flow field that satisfies the impermeability condition can be imme-
diately constructed simply by symmetry considerations. It is the irrotational flow
that would be induced by an image vortex of opposite circulation placed symmetri-
cally below the wall. Such an image vortex gives a velocity perpendicular to the
wall that is opposite to that of the real vortex, and thus ensures that the fluid does
2.7 Vortex Interactions 47

Fig. 2.13 The interaction of a vortex with the wall produces two separate effects. First (left panel),
the condition of impermeability is satisfied by a distortion to the vortex-induced flow that is equiv-
alent to having an opposite vortex placed symmetrically below the wall. The presence of such a
“image” vortex increases the tangential velocity next to the wall, and induces a translation velocity
to the otherwise still vortex. The second effect (right panel) is due to viscous adherence, the devel-
opment of a boundary layer and eventually a vortex-induced separation

not penetrate into it. On the contrary, the tangential velocity has the same sign of
that due to the real vortex and therefore the velocity adjacent to the wall increases
(splash effect). In addition, the image vortex also induces a velocity to the real
vortex that accelerates or decelerates (depending on the direction of the circula-
tion) with respect to the background flow because of this image effect. For exam-
ple, a (clockwise) vortex that just formed from a wall underneath is decelerated by
the image below the same wall, while it accelerates when it approaches a wall on
the opposite side.
Second, in addition to the image effect, a vortex near a wall also influences the
development of the boundary layer because of the viscous adherence condition at
such a wall. A vortex creates a local velocity gradient along the wall, acceleration
followed by deceleration (or vice versa depending on the direction of rotation). This
perturbation, as previously discussed in Sect. 2.3, may give rise to a vortex-induced
separation of the boundary layer and to the formation of secondary vortices as it is
sketched in Fig. 2.13 (right panel).
When the vortex-boundary interaction described above applies to a tract of a
three-dimensional vortex tube, it eventually affects the following three-dimensional
dynamics. First, the image effect gives a local stretching and deformation of a vor-
tex filament. Second, when the vortex gets closer, it eventually interacts directly
with the vortex-induced vorticity distribution. This is an interaction between oppo-
sitely circulating vorticity. That gives rise to the local wind-up of the wall vorticity
around the approaching vortex and to reconnection with its vortex lines. Eventually,
the vortex crops by dissipation in the regions closer to the wall; this unbalances the
three-dimensional vortex structures that tends to rapidly further deform and develop
small structures that are eventually dissipated.
48 2 Vortex Dynamics

2.8 AMention to Turbulence

Let us enter smoothly into the realm of fully developed turbulence by deepening a
little further the concept of interaction between three-dimensional vortices intro-
duced in the previous chapter. We have said there that the interaction between two
vortices first deforms the overall, large scale geometry of the vortex loops then,
after sequences of reconnections, breaking of vortices and further deformations, it
eventually transforms the original vorticity into several irregular small structures.
Such small scale elements present sharp velocity gradient, viscous friction, and are
rapidly dissipated.
Now consider that large vortices are continuously generated, formed from the
surrounding boundary. The resulting flow witnesses the simultaneous presence of
these large structures with others of all intermediate sizes from these down to the
smallest vortices dominated by viscosity. A measure of the complexity of such a
flow can be provided by from the amount of such contemporary vortices, measured
by the ratio between the largest scale, say L, and the smallest one, that is indicated
with h. When L is comparable to h, the flow is a regular one. For example, if the
system creates continuously vortex loops of size L, the resulting flow is a sequence
of individual rings that decay as time proceeds because there are no vortices of a
smaller size. On the opposite end, when L is much larger than h, the flow presents
the just generated vortex loop of size L, the previously generated vortices that broke
down into smaller structure of size, say, L/2, and a large number of interacting vor-
tices of progressively small size up to the smallest ones. The order of magnitude of
this complexity can be estimated from the phenomenological theory of turbulence
(due to Kolmogorov in 1941 and reported, for example, in Frisch 1995, Sect. 7.4)

L
≈ Re 3/ 4 (2.15)
h

where Re is the Reynolds number

UL
Re = (2.16)
ν

that was previously introduced in (1.21). When the Reynolds number is large
enough, the flow presents fluctuations on velocity and vorticity over a wide range of
scales, and can be classified as a turbulent flow.
An increased friction between fluid elements and enhanced energy dissipation
with respect to regular fluid motion characterizes turbulence. In fact, the develop-
ment of turbulence is the strategy used by fluids to dissipate the excess energy.
When a fluid motion presents a large density of energy (high velocity), the fluid may
be unable, in a regular motion, to maintain equilibrium between viscous dissipation
and the external energy source; in that case, it increases the particle paths by devel-
oping swirling motions and small scales with higher shear rate to increases viscous
dissipation up to equilibrium. In fact, the Reynolds number also represents the ratio
2.8 A Mention to Turbulence 49

between the kinetic energy introduced in the large scales, proportional to rU2, and
their ability to dissipate with shear stress, grossly estimable as proportional to
rnU/L. When the Reynolds number increases above a certain threshold, smaller
scales develop to enhance dissipation. In other words, regular flow becomes unsta-
ble and turbulence appears. Every realization of flow motion presents a critical
value of the Reynolds number above which the motion develops turbulence. The
value of the critical Reynolds number was mentioned in Sect. 1.6 to be about 2,300
in the case of steady flow in a circular vessel.
Turbulence enhances energy dissipation and therefore it is normally a threat of
excessive energy consumption in the vascular circulation. Another property is the
unpredictability of its chaotic fluctuations that makes turbulent flows difficult to
control, model, and manage. On the other side, turbulence has several positive
implications; first of all, it makes life possible by enhancing mixing and diffusion.
While viscous diffusion is an extremely efficient mechanism to distribute substances
at very small scales, turbulent dispersion dominates mixing at larger scales. For
example, viscous diffusion length, which grows proportionally to √t (see Eq. 1.16),
in water takes a few hundredth of a second to reach 1 mm, a few second for 1 cm,
and over 1 h for 1 m. On the contrary, the accelerated turbulent dispersion domi-
nates the mixing and heat propagation at scales sufficiently larger, typically above a
few millimeters (in general as soon as the Reynolds number is above a few thou-
sands). It is evident how turbulence is ubiquitous in nature and how it ensures the
mixing that is experienced in everyday life.
In general, we may think of turbulence as a system of entangles and interacting
vortex elements of disparate sizes. Ranging from the large size generated by the
boundaries, to the smaller size where the flow is smoothed out by viscous effects.
These vortices are not clear individual structures like those discussed in previous
chapters. These are the turbulent eddies, loosely defined blob of vorticity of some
size and arbitrary shape coming from the breakdown of the large unstable vortices
generated from the surrounding boundaries. Turbulence is thus a sea of eddies that
is stretched and twisted by the velocity field that is induced by vorticity field itself.
Following Davidson (2004, Sect. 2.4), turbulence is a spatially complex distribution
of vorticity that exhibits a wide and continuous distribution of scales and advects
itself in a chaotic manner.
The overall dynamics of turbulence is normally described in terms of energy
cascade (Davidson 2004, Sects. 1.6 and 3.2). An external energy input (slope of a
channel, a pumping pressure) pushes a fluid within its boundaries, across an orifice,
around an obstacle, along an irregular vessel bend. The flow thus generates energetic
vortices whose size is comparable with that of the container, these vortices interact
and produce smaller eddies, which further interact producing turbulent eddies of pro-
gressively smaller size. While eddies become smaller, velocity gradients are larger,
and viscous shear stresses become increasingly capable to dissipate kinetic energy
into heat. At the lower end of this energy cascade, very small eddies are entirely
dissipated and do not generate anything smaller. Thus energy is injected in the turbu-
lent flow at large scales, it cascades toward smaller scales, and it is dissipated in the
smallest scales of the flow by viscous friction. The example in Fig. 2.14 shows the
co-presence of vortices of different scale in a two-dimensional turbulent flow.
50 2 Vortex Dynamics

Fig. 2.14 Vorticity field in a

two-dimensional decaying
turbulent flow. The vorticity
distribution is shown in gray
scale, clockwise vorticity is
white, counter-clockwise is
black. The flow field presents
interacting turbulent eddies,
from vortices to shear layers, [AU1]
with a few order of
magnitude differences
between their sizes

The most common strategy to tackle the problem of turbulence relies of statistical
methods, searching for a description of the average motion (responsible for trans-
port) and of its fluctuations (responsible for dispersion). This is such a common
practice that the study of turbulence is often considered that of statistical fluid
mechanics. However, turbulence is not a random process, it is actually a determin-
istic phenomenon; a turbulent flow is a solution of the mathematical equations gov-
erning fluid motion (the Navier-Stokes Eq. 1.14). The underlying deterministic
nature of turbulence often emerges in the form of coherent structures, a somehow
collectively organized dynamics, developing within a random background. These
coherent structures are often the large scale vortices generated during vortex forma-
tion processes, possibly modified for the presence of turbulence. Or they are vorti-
ces that develop from the instability of a parallel flow, typically vortex layer, like in
boundary layer flows (Davidson 2004, Sect. 4.2.6), in some other cases they emerge
from coalescence of incoherent background vorticity, like the vortices in two-
dimensional flows of Fig. 2.14 or three-dimensional filaments. Coherent vortex
structures typically contain most part of the energy and therefore they represent the
fundamental objects in the analysis of each specific turbulent flow. In other words,
the concepts of vortex formation discussed above find application to turbulent flows
as well, keeping in mind the presence of the turbulent background, and extending
the concept of vortex formation loosely to vortex structures developing for instabil-
ity of smooth flows.
In the cardiovascular system, turbulent flows are rarely encountered. The largest
scales of motion achievable in the arterial network cannot exceed the vessel size, of
a few centimeters at most. The Reynolds number is normally well below 1,000, with
the exception of the very largest vessels. The flow in the ascending aortic and, some-
time, in the left ventricular cavity can reach values of the Reynolds number up to
some thousands, just above the critical threshold during a short interval near the
systolic or diastolic peaks. In any case, when any turbulence develops, it is weak
turbulence with an energetic level that does not influence appreciably the main
References 51

dynamics and vortex formation processes. Such a weak turbulence simply increases
the dissipation level through more intense interaction between larger vortices. It
should be remarked that the highest levels of turbulence, if any, in an unsteady pul-
satile flow are recorded during the deceleration after the peak of the flow. In fact,
although the instantaneous Reynolds number has decreased, the flow has been filled
with the energy during the maximum velocity and has to dissipate such energy dur-
ing deceleration. Deceleration enhances shear layer instability phenomena, and
boundary layer separation, which support turbulence.
Weak turbulence may develop in the diastolic filling of the left ventricle when the
mitral jet impacts onto the walls, as it may occur in a dilated heart with a large car-
diac output. The most frequent appearance of turbulence occurs in the aortic artery,
particularly in the ascending part. Here the tri-leaflet geometry of the aortic valve
provokes a rather complex three-dimensional vortex formation that, associated with
the large Reynolds number (roughly from 3,000–8,000 at peak systole), gives rise to
interactions that produce small scales vorticity and weak turbulence. This is even
enhanced in presence of mechanical valves because of the more irregular interac-
tions between the unnatural geometry of vortices and shear layer behind the artifi-
cial orifice. Specific examples will be treated in the next parts of the book.

Refer
ences

Batchelor GK. An introduction to fluid dynamics. Cambridge: Cambridge University Press;

1967.
Dabiri JO. Optimal vortex formation as a unifying principle in biological propulsion. Annu Rev
Fluid Mech. 2009;41:17–33.
Davidson PA. Turbulence. An introduction for scientists and engineers. Oxford: Oxford University
Press; 2004.
Domenichini F. Impulsive vortex formation from slender orifices. J Fluid Mech. 2011;666:
506–20.
Frisch U. Turbulence. Cambridge: Cambridge University Press; 1995.
Gharib M, Rambod E, Shariff K. A universal time scale for vortex ring formation. J Fluid Mech.
1998;360:121–40.
Jeong J, Hussain F. On the identification of a vortex. J Fluid Mech. 1995;285:69–94.
Kida S, Takaoka M. Vortex reconnection. Annu Rev Fluid Mech. 1994;26:169–77.
Luchini P, Tognaccini R. The start-up vortex issuing from a semi-infinite flat plate. J Fluid Mech.
2002;455:175–93.
Lugt HJ. Vortex flow in nature and technology. New York: Wiley; 1983.
Moffat HK, Kida S, Ohkitani K. Stretched vortices – the sinews of turbulence; large Reynolds-
number asymptotics. J Fluid Mech. 1994;259:241–64.
Panton RL. Incompressible flow. 3rd ed. Hoboken: Wiley; 2005.
Pedrizzetti G. Unsteady tube flow over an expansion. J Fluid Mech. 1996;310:89–111.
Pedrizzetti G. Vortex formation out of two-dimensional orifices. J Fluid Mech. 2010;655:
198–216.
Pullin DI. The large-scale structure of unsteady self-similar rolled-up vortex sheets. J Fluid Mech.
1978;88:401–30.
Saffman PG. Vortex dynamics. Cambridge: Cambridge University Press; 1992.
Shariff K, Leonard A. Vortex rings. Annu Rev Fluid Mech. 1992;24:235–79.
Wu JZ, Ma HY, Zhou MD. Vorticity and vortex dynamics. Berlin/Heidelberg: Springer; 2006.