Chapter 10

The Euler equations

10.1

Acceleration of a fluid parcel

In order to write the equations of motion of a fluid, we must place the observer
in an inertial reference system. Then we have to apply the second principle of
dynamics (Newtons second law) to its parcels. This requires a suitable expression
for the acceleration of the parcels as a function of the velocity field. Bear in mind
that when we speak of the acceleration of a parcel, we mean the time derivative
of its velocity. But, according to the Eulerian approach, we need to express this
quantity by means of a velocity field defined as a function of the time and space
coordinates.
We are in the same situation encountered in section [7.2] for the evaluation of
the time variation of a scalar property following the motion of the parcel. This
relation was extended to an arbitrary vector quantity in (7.2). The only difference
is that now the velocity field that advects the property and the property itself
coincide. Thus, in this case (7.2) becomes
du
u
=
+ (u )u.
dt
t

(10.1)

It should be noted that the time derivative on the left is a total derivative, and
55

56

Franco Mattioli (University of Bologna)

represents the acceleration of the parcel, while the time derivative on the right
is a partial derivative, that is, it represents the rate of change of the velocity in
fixed points of space.
Let us analyze in greater detail the above expression. The acceleration of a
parcel depends on two factors. The former, u/t, also called local acceleration,
represents the acceleration due to the fact that in a given point of space the
velocity of the parcels passing through it can be either increasing or decreasing
in time. The latter, (u )u, also called advective term of the acceleration,
provides the acceleration due to the fact that a parcel can move from a region of
low velocity to one of high velocity, or vice versa.
In a unidirectional flow that is uniform over a horizontal plane and possibly variable
with the vertical coordinate, the advective component of the acceleration vanishes. In
fact, the parcels during their motion do not undergo any change of velocity.
Let us see how this statement can be translated in mathematical terms. Let x be the
direction of the motion. The three components of the advective term in the x-direction
u

u
u
u
+v
+w
x
y
z

all vanish: the first one vanishes because the velocity u does not vary with x, and the
other two because both v and w are zero. Furthermore, the other two components of
the acceleration in the y and z-directions vanish because both v and w are zero.
If the flow is also stationary, the local acceleration is zero as well, so that the total
derivative of the velocity vanishes, in agreement with the fact that the parcels are not
accelerated.

The second term of the acceleration is the source of the greatest mathematical
difficulties encountered in the study of fluid mechanics. It is, in fact, a nonlinear
term, indeed, a strongly nonlinear term. This prevents the use of the many
mathematical tools available for the linear equations of all types, and forces the
adoption and continuous research of new and increasingly advanced mathematical
tools.
Problem 10.1 A rectilinear pipe of variable circular section S = S(x) has a volume rate
of flow Q. Evaluate the advective acceleration of the parcels lying along its central axis,
assuming that their velocity is equal to the average velocity computed along a transversal
section of the pipe.
Problem 10.2 Show that if u = r, then
(u )u = u.
Provide the physical meaning of such an expression.

Principles of Fluid Dynamics (www.fluiddynamics.it)

57

Solution. Let us orientate the coordinate axes in such a way that = k. Developing
the given expression by components, we obtain
!

+v
(yi + xj) = vi + uj = u.
(u )( r) = u
x
y
The last term is nothing but the centripetal acceleration associated to the circular path
of the parcels. In fact, if we decompose r in the sum of the two components rk and r ,
respectively, parallel and perpendicular to the vector , one has
( r) = k [k (rk + r )] = 2 r ,
Comment. On the basis of the examples presented, we see that the advective acceleration
is formed by at least two terms: the tangential acceleration along the straight line tangent
to the trajectory of the parcel, and the centripetal acceleration, related to the curvature
of the trajectory in a direction normal to it. It is possible to show that no other terms
exist in addition to these.
Problem 10.3 If we reverse the flows of the previous examples, does the sign of the
advective component of the acceleration change?

10.2

The Euler equations

Now that we have an expression for the parcel acceleration, we can write the
momentum equation, i.e., the equation which provides an expression of the parcel
acceleration. By applying the second principle of dynamics one has

du
xyz = F xyz.
dt

(10.2)

Here, is the fluid density, xyz the volume of the fluid parcel, m = xyz
its mass and F the force per unit mass exerted on it.
The gravity force is clearly unaltered by the motion. Thus our attention must
be concentrated on the surface stresses.
In section [6.1], we have already studied the behavior of the pressure necessary
for an hydrostatic equilibrium. In order to put a fluid in motion we might change
the values of the normal stresses with respect to the hydrostatic distribution or to
add tangential stresses. For the moment we do not take into account this latter
possibility, because the former appears much simpler to manage. This means
that the equations we are going to derive will not have a general validity.

58

Franco Mattioli (University of Bologna)

In conclusion, we assume that the structure of the pressure forces remains the
same studied in section [6.1] for the static case, with the stresses between parcels
still normal to their surfaces of separation and independent of their orientation.
By inserting in (10.2) the expressions (10.1) and (6.1) found, respectively, for
the acceleration of a parcel and for the force applied to it, we have
u
p
+ g.
+ u u =

(10.3)

These equations are known as the Euler equations, after the name of the author
that derived them for the first time. Written in component form, they become
u
u
u
u
1 p
+u
+v
+w
=
,
x
t
x
y
z
v
v
v
1 p
v
+u
+v
+w
=
,
y
t
x
y
z
w
w
w
w
1 p
+u
+v
+w
=
g.
z
t
x
y
z

(10.4)
(10.5)
(10.6)

Problem 10.4 State the condition for the validity of the hydrostatic balance.
Solution. When dw/dt = 0, i.e., when the vertical acceleration vanishes, (10.6) reduces
to (6.3). In particular, the hydrostatic balance holds for any purely horizontal motion.

A fluid satisfying this property or, more clearly stated, a motion for which
such an approximation is acceptable, is called ideal or perfect. Indeed, this is
not a property of the fluid, but rather of the motion. The same fluid can be
considered either ideal or not, as the case may be.
10.3

Properties of the flows governed by the Euler equations

The Euler equations state that when pressure gradients develop inside a fluid
in addition to those relative to the hydrostatic equilibrium, then the fluid parcels
react by accelerating in the opposite direction, the magnitude of the accelerations
being proportional to the magnitude of the pressure gradients.
Such pressure gradients in the interior of the fluid can be produced by normal
stresses exerted along the external boundaries of the fluid or, in the case of a
liquid, by the displacement of the free surface from its equilibrium position.

Principles of Fluid Dynamics (www.fluiddynamics.it)

59

For example, when we shake a bowl full of water, we give rise to a deformation of the
free surface with respect to the equilibrium condition represented by a horizontal plane.
This generates the pressure gradients that cause the motion of the fluid.

In general these equations apply to motions in which the local or advective

accelerations are very strong and can be balanced only by pressure forces. This
includes strongly accelerated fluids and flows around bodies of complex geometry.
However, their validity often does not extend to the whole domain occupied by
the fluid, because in certain regions other forces come into play to modify this
balance.
From a theoretical point of view, it is meaningful to consider the plane-parallel
flows, in spite of their triviality. They satisfy the boundary conditions over any
horizontal surface, while the distribution of the density and velocity with height
can be arbitrary. The vertical component of the Euler equations (10.6) reduces
to the hydrostatic equation (6.3). The horizontal component (10.4) of the Euler
equations in the direction of the motion (the other equation (10.5) is identically
satisfied term by term) then state that each layer of fluid moves as a rigid body,
where the acceleration of the parcels is proportional to the pressure gradient in
the direction of the motion at the same level, so that the various layers move
independently from each other.
Problem 10.5 Consider a unidirectional flow, that is uniform on the horizontal plane.
Find a relationship between the velocity and the pressure gradient, assuming that both
fields vary sinusoidally in time.
Solution. Let us orientate the flow in the x-direction. The x-component of the Euler
equations yields
1 p
u
=
,
x
t
since in this case the advective term of the acceleration vanishes. Let the velocity field
vary in time as
u = U cos t.
A substitution of this expression in the previous equation leads to
U sin t =
from which

1 p
,
x

p
= U sin t.
x

Thus, the pressure results as being out of phase by 90 with respect to the velocity field.

60

Franco Mattioli (University of Bologna)

Comment. This solution can be easily verified in shallow water waves, and represents
one of the many experimental proofs of the validity (in the appropriate circumstances)
of the Euler equations.

In conclusion, the Euler equations have been proven to be very efficient and
accurate in describing the behavior of a fluid in a large number of circumstances.
Often, however, they hold not for the whole region occupied by the fluid, but
only in the part in which the above hypotheses about the structure of the surface
stresses hold.

10.4

The equations of motion for an incompressible fluid

The Euler equations (10.3) and the continuity equation (8.1) are a set of four
equations in the five variables u, v, w, p and . From a mathematical point of
view, a necessary condition for the solution of a system of equations is that the
number of equations be equal to the number of unknown variables. In this case,
the system is said to be closed. Therefore, in general the equations written so far
are only conditions to be imposed on the motion, but cannot define it uniquely.
If we assume that the density of the fluid is constant everywhere and over
time, then one of the unknowns is automatically eliminated, and the system
formed by the Euler equations and the continuity equation represents a closed
system of four equations in four unknowns.
Moreover, if the fluid is incompressible, then, as we have already seen, the
continuity equation reduces to (8.3). The system formed by the momentum equation, the continuity equation and the incompressibility condition (7.3) becomes
a closed system of five equations in five unknowns.
From a mathematical point of view, the closure of the system is possible in another
case, that is, when the density is a function of the sole pressure. But this assumption is
only apparently mechanical, while, indeed, it implies certain underlying thermodynamic
relationships. For these reasons it will not be considered further here.

Clearly, these equations can also be applied to a compressible fluid when the
motion is purely horizontal and all the fields are uniform in each horizontal plane.
As the pressure and all the other physical quantities do not change during the
motion, the density, too, remains constant so that the fluid behaves as if it were
incompressible.

Principles of Fluid Dynamics (www.fluiddynamics.it)

10.5

61

Initial and boundary conditions for an ideal fluid

To solve a specific problem, however, me must add more information. We

have to define the so-called initial and boundary conditions. The equations, in
fact, describe the evolution of the unknown fields with time, starting from a
given initial state. This state changes from problem to problem, and thus must
be defined each time. Furthermore, the fluid is constrained during the motion by
the solid walls of the container or basin in which it moves and possibly by other
surfaces of separation from the external environment, such as the free surface
of a liquid. Along all these boundary surfaces the various fields are subject to
particular conditions, which, again, must be satisfied in order to uniquely define
a solution.
For an incompressible fluid the initial conditions must give information about
the initial density and velocity distribution. Then, the acceleration and the
density variations can be derived from the equations themselves in such a way
that the new velocity and density fields at an immediately subsequent instant
can be computed.
Indeed, the procedure is not so simple as it might appear because not all the equations
are of an evolutive kind. Here, we only wish to stress that the initial conditions are
necessary input for the solution of any problem.

There are several boundary conditions suitable for a perfect fluid which depend on the particular problem treated. Here, we will consider only the simplest
ones.
In a liquid, it is necessary to provide the pressure field along the free surface.
The condition is
p = pa ,
for z = 0,
where pa is the atmospheric pressure.
Along solid boundaries, we must require that the normal component of the
velocity, named also briefly the normal velocity, be zero. The condition simply
states that the fluid can neither enter a solid wall, nor leave an empty space
between itself and a wall. If the boundary is fixed, the condition reads
u n = un = 0,

(10.7)

where n is the unit vector normal to the wall. If the boundary moves with a
velocity u0 , we must impose
u0 n = u0n ,
where u0n is the normal component of u0 .

62

Franco Mattioli (University of Bologna)

As can be seen, no conditions must be given on the tangential component

of the velocity, which therefore appears as a particular aspect of the solution of
the considered problem. This situation reflects the common experience that a
fluid slides along the solid walls without developing any apparent stress. On the
other hand, we have assumed that in a perfect fluid the only possible stresses are
normal to the surfaces. It is evident that a force normal to the boundary can in
no way change the tangential component of the velocity.
In other circumstances, the boundary conditions can be more involved or
refer only to ideal surfaces separating the fluid governed by the Euler equations
from the rest of the fluid. In all cases, they represent a fundamental piece of
information for the solution of a problem.

10.6

The equations of motion for a compressible fluid

In the most general case, the density of the fluid parcels depends on the
pressure, temperature and possibly on other quantities through the so-called
state equation. This implies that the system formed by the momentum equation,
the continuity equation and the state equation is no longer a closed system.
Hence, it is necessary to introduce as many equations as the number of variables
in addition to the pressure that appear in the state equation. Obviously, suitable
initial and/or boundary conditions must be associated to each of them. Each new
equation represents an additional physical process which often is very different
with respect to the processes represented by the other variables.
Therefore, the number of the equations to be written in order to solve a
problem depends on the kind of physical processes to be taken into consideration
or, in other words, on the schematization that becomes necessary to adequately
describe the behavior of the fluid in those particular circumstances.
Since in the equation of state the temperature is always present, this means
that in a compressible fluid mechanical and thermodynamic processes are always
mixed together. For example, heating a fluid modifies its motion. For an incompressible perfect fluid, conversely, the mechanical and thermodynamic processes
are separated from each other. This means that thermodynamic processes are
completely unrelated to the mechanical ones.
For example, the motion in a homogeneous fluid develops in the same way regardless
of its temperature.

In conclusion, when the fluid can be considered incompressible, that is, when
its density does not depend appreciably on temperature, pressure and possibly

Principles of Fluid Dynamics (www.fluiddynamics.it)

63

other variables, the mechanical processes can be studied on the basis of the
equations found so far without any reference to thermodynamics. A study on the
evolution of its thermodynamic properties runs parallel without interfering with
the study of its motion. Conversely, for a compressible fluid both problematic
areas must be contemporaneously taken into account.
10.7

Historical notes and essential bibliography

The continuity and Euler equations were derived by Leonhard Euler in 1755
[11] both in Lagrangian and in Eulerian form. He had to devise the mathematical
tools suited to deal with a continuous medium, that is, the partial derivatives
and the vector differential operators, by introducing the concept of parcel. This
was a fundamental progress with respect to the laws introduced by Galileo and
Newton, referring to simple point-masses.
A description of the Lagrangian approach is contained in Lamb [18], along
with a significant number of solutions of the Euler equations already found at
the time of the publication of the last edition (the sixth) of the book, in 1932.
Other classical solutions are shown in Milne-Thomson [20].