Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Chapter 4

The total derivative

4.1 Lagrangian and Eulerian approaches

The representation of a fluid through scalar or vector fields means that each
physical quantity under consideration is described as a function of time and
position. But the physics of a system is related to parcels, which move in space.
In particular, we are interested in the change of the properties of a parcel with
time.
In principle we might choose to describe the fluid by means of a set of func-
tions of the various parcels. In other words, we might label each parcel by its
coordinates at a given initial time, and then provide their new coordinates as
time proceeds. Such an approach is indeed possible and is called the Lagrangian
approach, but it is more complicated and less used than the Eulerian approach,
where the various quantities are given as a function of the geometrical points of
space. Hereafter, we will develop only this latter approach, but we must be aware
of the existence of the former because sometimes it is used both in theoretical
studies and in experimental practice.
When a buoy is released in the ocean, the information it furnishes is related to its
position. Since it is swept away by the surrounding water, it will record the properties of
that mass of water, which, to a first approximation, can be thought of as a fluid parcel.

19
20 Franco Mattioli (University of Bologna)

In this case the Lagrangian approach becomes the most logical approach.
The problem arises of how the behavior of the parcels can be related to the
fields defined in geometrical points. In other words, we must evaluate the time
evolution of the properties of the moving parcels as a function of the velocity
field and of the distribution of these properties in space and time.

4.2 The total derivative

For the sake of simplicity, we will first consider the case of a scalar property.
Let
u = u(t, x, y, z)
be the velocity field and
ψ = ψ(t, x, y, z)
a scalar property, such as density, pressure or temperature.
The time variation of a quantity, following the motion of a parcel, will be
denoted by the symbol
d
dt
to distinguish it from the variation of the same quantity in a fixed point of space,
which will be denoted by

.
∂t
The first derivative is called total derivative, and the second, partial derivative or
local derivative.
The symbol
D
Dt
is also very common for the total derivative, which is also called substantial derivative,
material derivative or individual derivative.
Let xp (t), yp(t), zp(t) be the coordinates of a parcel moving in space. Then
the variation of the property ψ of the parcel can be obtained by applying the
rules of the derivative of the function of a function
dψ dψ(t, xp, yp , zp)
= =
dt dt
∂ψ ∂ψ dxp ∂ψ dyp ∂ψ dzp ∂ψ ∂ψ ∂ψ ∂ψ
= + + + = +u +v +w .
∂t ∂x dt ∂y dt ∂z dt ∂t ∂x ∂y ∂z
Principles of Fluid Dynamics (www.fluiddynamics.it) 21

The last step is justified by the fact that dxp/dt is nothing but the velocity u of
the considered parcel. Hence, we can write

dψ ∂ψ
= + (u · ∇)ψ, (4.1)
dt ∂t

or, in a more general and symbolic way,

d ∂
= + u · ∇.
dt ∂t

...
.....
....
..
......
.
.....
.....
2• ψ(t + δt, x + δx)
....
.....
....
.
..
...
δx ....
....
....
......
.
...
...
...
1 • ψ(t, x)
...
...
...
...
..
..
..

Fig. 4.1: While the parcel moves from point 1 to point 2, performing the
displacement δx, the quantity ψ varies from ψ(t, x) to ψ(t + δt, x + δx).

We can also see graphically the meaning of the total derivative (Fig. 4.1). During
the time interval δt, the parcel passes from point 1 to point 2. In the first point the
quantity ψ holds ψ(t, x, y, z), and in the second point ψ(t + δt, x + δx, y + δy, z + δz).
But δx can be evaluated to a first approximation on the basis of the velocity in point 1
and of the time interval δt

δx = u δt, δy = v δt, δz = w δt.

By applying the Taylor series theorem about point 1, we have

ψ(t + δt, x + δx, y + δy, z + δz) = ψ(t + δt, x + uδt, y + vδt, z + wδt) =
∂ψ ∂ψ ∂ψ ∂ψ
= ψ(t, x, y, z) + δt + u δt + v δt + w δt.
∂t ∂x ∂y ∂z

Subtracting ψ(t, x, y, x) from both members of the equation, dividing by δt, and taking
the limit for δt → 0, we obtain the above equation (4.1).
22 Franco Mattioli (University of Bologna)

4.3 The structure of the total derivative

Let us further analyze the structure of the total derivative. The variation in
time of the property ψ of the parcel depends on two factors. The former ∂ψ/∂t
represents the variations due to the fact that at a given point fixed in space
the property can increase or decrease with time. The latter (u · ∇)ψ, called the
advective term of the total derivative, depends on the fact that the parcel, during
its motion, can pass from a region with a given value of ψ to another with a
different, either lower or higher, value of ψ.
The advective term (u · ∇)ψ might also be written as u · ∇ψ, or ∇ψ · u. Both expres-
sions are mathematically correct, but partially hide the underlying physical meaning,
and will be avoided when possible.

In the case of a stationary flux, i.e., a flux independent of time, the term
∂ψ/∂t vanishes everywhere, so that the variation in time of the property is equal
to the advective term. On the other hand, if the flux is uniform in space, then
(u · ∇)ψ is always zero, and the property can vary only if the field simultaneously
varies in all points of space. In this case, the total and partial derivatives coincide.
Obviously, what we have said about a scalar quantity ψ can be extended to
any vector quantity v. The same expression found for a scalar quantity must
be repeated for each component of the vector. In vector notation, the total
derivative of a vector takes the form
dv ∂v
= + (u · ∇)v. (4.2)
dt ∂t

Clearly, if a certain quantity associated to a parcel is conserved in time, its


total derivative is zero. For example, in an incompressible fluid the density ρ of
each parcel is constant in time, so that we have

dρ ∂ρ
= + (u · ∇)ρ = 0. (4.3)
dt ∂t

We should not confuse a homogeneous fluid with an incompressible fluid. In the


former case, the density is always the same for all the parcels. In the latter, the density
can vary passing from one parcel to another, but every parcel maintains the same den-
sity during its motion. A mix of two or more homogeneous fluids is an example of an
incompressible fluid.
Strictly speaking, a fluid is incompressible when its density does not depend on the
pressure. Here and in the following we will adopt instead this more restrictive definition.
Principles of Fluid Dynamics (www.fluiddynamics.it) 23

It should be noted that, in spite of the complexity of its definition in a fixed


reference system, the total derivative is nevertheless a simple time derivative when
referred to the moving parcel. Thus, the usual rules for the ordinary derivatives
hold for the total derivatives as well.

Problem 4.1 Show, by applying the Eulerian expression for the total derivative, that if
ψ and φ are two scalar fields variable in time, then
d(ψφ) dφ dψ
1. =ψ +φ ,
dt dt dt
dψ 1 dψ2
2. ψ = ,
dt 2 dt
1 dψ d log ψ
3. = .
ψ dt dt
Problem 4.2 Show, by applying the Eulerian expression for the total derivative, that
dxp
= u.
dt

Hint. Use the property xp = x.

4.4 The Reynolds transport theorem

The total derivative allows us to follow the properties of an infinitesimal parcel


during its motion. Now, we will extend such an operation to a finite volume of
fluid.
Let us consider a quantity given by a volume integral
Z
Ψ= ψ dV,
V

where ψ is a certain scalar property of the fluid. The time derivative of this
integral in a given volume constant in time is simply given by
dΨ d ∂ψ
Z Z
= ψ dV = dV. (4.4)
dt dt V V ∂t
However, if the volume changes with time, then the results is more compli-
cated. We can write
Z Z
δΨ = Ψ(t + δt) − Ψ(t) = ψ(x, t + δt) dV − ψ(x, t) dV.
V(t+δt) V(t)
24 Franco Mattioli (University of Bologna)

The Taylor expansion theorem in time limited to the linear terms allows us to
write
!
∂ψ ∂ψ
Z Z Z Z
δΨ = ψ + δt dV − ψ dV = δt dV + ψ dV. (4.5)
V(t+δt) ∂t V(t) V ∂t δV

......
..
S(t + δt) ...........
.....
......... .....
........ .. ...... n .....
....
....... .. .
S(t) ...
............. . . .. .......................
....
........... ..............α .....
... . . . ..
. . . . ... . .. .. . . .. ....
...
...... ..
. ..
. ........................... .
.. ..
. .
... .
.. .
.. ..
. ..
. .
.....
......
................................................................. ...........
........... . u ...........
... ......
dS ............
.............................................................................................................................
.

... u δt ....
... ...
... ...
... ...
.. δV ..

Fig. 4.2: The volume δV between the instants t and t + δt can be computed as
the sum of the volumes of the cylinders that we can draw between the surfaces
S(t) and S(t + δt) in the direction of the velocity. Since the volume of the
oblique cylinders is given by the product of the area of their basis dS by their
height u δt cos α, it can be written as dS n · u δt = u · dS δt, where n is the
normal to the surface S at the center of the basis of the cylinder.

If the volume changes according to the velocity of its parcels, that is, if the
volume is always formed by the same matter, the last volume integral can be
transformed into a surface integral using the velocity u at which the surface S
bounding the volume moves. With the help of (Fig. 4.2) we deduce that
Z Z Z
ψ dV = δt ψ u · dS = δt ∇ · (ψu) dV,
δV S V

the last step resulting from an application of the Gauss’ theorem. Then, by using
(4.4), (4.5) becomes
d ∂ψ
Z Z Z
ψ dV = dV + ∇ · (ψu) dV (4.6)
dt V V ∂t V
d
Z Z
= ψ dV + ψu · dS. (4.7)
dt V S

The two terms at the second and third member of the equation correspond,
respectively, to the local and avective terms of the total derivative for infinitesimal
parcels, which now is represented by the first term of the equation. This is known
as the Reynolds transport theorem.
Principles of Fluid Dynamics (www.fluiddynamics.it) 25

A particular application of the theorem is obtained by assuming ψ = 1. In this case


we have
d dV
Z Z
dV = = ∇ · u dV, (4.8)
dt V dt V

where V is the volume of V. Therefore, the volume variation of a mass of fluid depends
on the divergence of its velocity field. By considering an infinitesimal volume δV over
which ∇ · u can be assumed as constant, we can recover the definition of divergence
already seen in (D.2)
1 d
∇·u = δV, (4.9)
δV dt

You might also like