Chapter 3

Conservation laws

Reynolds transport theorem

v dt
Σ(t)
dσ

Σ(t + dt)

Figure 3.1: Evolution of a fluid volume.

To write the conservation equations, it is of interest to determine the variation with time of the
amount of mass, momentum or energy contained in a fluid volume Vf (t) limited externally by a
closed fluid surface Σf (t)

"Z #
d
φ(x̄, t) dV ; φ = ρ, ρv̄, ρ(e + |v̄|2 /2). (3.1)
dt Vf (t)

Since both the integrand φ(x̄, t) and the integration domain Vf (t) vary with time, one can expect
two different contributions associated, respectively, with the unsteadiness of the flow field and
with the motion of the fluid, just as we saw earlier in connection with the material derivative.
To quantify the two contributions, it is convenient to consider the evolution of the fluid volume
between two successive instants of time t and t + dt, as sketched in Fig. 3.1.
 Conservation Laws

Equation (3.1) can be written using the definition of limit

"Z # R R
d Vf (t+dt) φ(x̄, t + dt)dV − Vf (t) φ(x̄, t)dV
φ(x̄, t) dV = lim , (3.2)
dt Vf (t) dt→0 dt
which can also be split into two separate contributions according to
R R
V (t) [φ(x̄, t + dt) − φ(x̄, t)] dV V (t+dt)−Vf (t) φ(x̄, t + dt)dV
lim f + lim f . (3.3)
dt→0 dt dt→0 dt
In evaluating the first term we use the expansion φ(x̄, t + dt) − φ(x̄, t) = (∂φ/∂t)dt + O(dt2 ) to
give in the first approximation
R
V (t) [φ(x̄, t + dt) − φ(x̄, t)] dV
Z
∂φ(x̄, t)
lim f = dV, (3.4)
dt→0 dt Vf (t) ∂t
while the second integral can be evaluated as the convective flux crossing the surface Σf (t) during
the time interval dt to give
R R
Vf (t+dt)−Vf (t) φ(x̄, t + dt)dV Σf (t) φv̄ · n̄dσdt
Z
lim = lim = φv̄ · n̄dσ, (3.5)
dt→0 dt dt→0 dt Σf (t)

finally yielding
"Z # Z Z
d ∂φ(x̄, t)
φ(x̄, t) dV = dV + φ(x̄, t) v̄(x̄, t) · n̄ dσ. (3.6)
dt Vf (t) Vf (t) ∂t Σf (t)

The two terms in (3.6) correspond to the previously anticipated contributions arising from flow
unsteadiness and from fluid motion, respectively.

The solution of a given problem involves in general the variation with time of fluid properties
in a finite region Vc (t) that, depending on the problem, can be fixed or variable in time. That
region is called a control volume, which is limited externally by a control surface Σc (t), a closed
surface whose points move with velocity v̄c (x̄c , t). Since the conservation laws of fluid mechanics
apply to fluid volumes, and not to control volumes, it is of interest to relate the time derivatives
extended to Vc (t) to those extended to a control volume Vf (t) occupying the same space at the
instant considered. For Vc (t), the reasoning that led before to (3.6) allows us to write
"Z # Z Z
d ∂φ(x̄, t)
φ(x̄, t) dV = dV + φ(x̄, t) v̄c (x̄, t) · n̄ dσ. (3.7)
dt Vc (t) Vc (t) ∂t Σc (t)

Bearing in mind that we chose the control volume Vc (t) and its bounding surface Σc (t) to coincide
exactly with Vf (t) and Σf (t) at the instant considered, the integration domains that appear
in (3.6) and (3.7) are identical, and straightforward subtraction of both expressions provides
"Z # "Z # Z
d d
φ(x̄, t) dV = φ(x̄, t) dV + φ(x̄, t) (v̄ − v̄c ) · n̄ dσ. (3.8)
dt Vf (t) dt Vc (t) Σc (t)

This equation, to be used below when deriving the conservation equations for an arbitrary control
volume, indicates that the time variation of the given flow property (mass, momentum or energy)
in a fluid volume Vf (t) equals the time variation in a control volume coinciding with Vf (t) plus
the convective flux across the bounding surface Σc (t) (see (2.39)).

26
 Conservation Laws

Mass conservation
The first conservation law of fluid mechanics states that the mass of a given fluid volume remains
constant, "Z #
d
ρdV = 0, (3.9)
dt Vf (t)
which can be expressed with use made of (3.8) in the form
"Z # Z
d
ρdV + ρ(v̄ − v¯c ) · n̄dσ = 0. (3.10)
dt Vc (t) Σc (t)

The physical meaning underlying the above equation becomes clear by writing the second integral
on the right-hand side of the equation to give
"Z # Z
d
ρdV = − ρ(v̄ − v¯c ) · n̄dσ = 0,
dt Vc (t) Σc (t)

which states that the rate of increment of the mass

R contained in a control volume equals the
mass flux entering across its bounding surface, − Σc (t) ρ(v̄ − v¯c ) · n̄dσ. When the control volume
is fixed in space, the above equation reduces to
Z  Z
d
ρdV + ρv̄ · n̄dσ = 0. (3.11)
dt Vo Σo

Momentum conservation
Volume and surface forces
The forces acting on a fluid (or on a solid) can be classified in two different categories: long-range
forces, also called volume or mass forces, and short-range forces, also called surface forces. The
former, including for instance gravity and inertial forces, decay slowly over distances much larger
than the intermolecular distance d and therefore act over the whole fluid volume (for instance,
gravity decays over distances that are of the order of the Earth radius). Short-range forces, on
the other hand, have a direct molecular origin and decay rapidly with distance, their range of
influence being of the order of the intermolecular distance d.
The mass force acting on a fluid particle of volume dV can be expressed in the form

ρf¯m (x̄, t)dV, (3.12)

where f¯m (x̄, t) is the volume force per unit mass, which equals f¯m (x̄, t) = ḡ when gravity is
the only mass force acting on the system. If a non-inertial reference frame is selected in the
description, then inertial forces must be also considered to give

dΩ̄
f¯m (x̄, t) = ḡ − āo − ∧ x̄ − Ω̄ ∧ (Ω̄ ∧ x̄) − 2Ω̄ ∧ v̄, (3.13)
dt
with āo and Ω̄ representing the acceleration and angular velocity of the non-inertial reference
frame. To write (3.12) we have neglected the variation of f¯m within the fluid particle, an accurate

27
 Conservation Laws

approximation given the long-range nature of these forces. Some mass forces are conservative
and derive from a potential according to f¯m = −∇U , so that, for instance,

ḡ − āo − Ω̄ ∧ (Ω̄ ∧ x̄) = −∇[−ḡ · x̄ + āo · x̄ − (Ω̄ ∧ x̄) · (Ω̄ ∧ x̄)/2]. (3.14)

On the other hand, short-range forces only reach a short distance of the order of d. For a gas,
they are directly related with the flux of momentum emerging due to the random molecular
motion. The molecules crossing a surface separating two adjacent portions of fluid transport
their momentum and energy. This microscopic transport effect produces a surface force at the
macroscopic level , and also gives rise to heat conduction. If the fluid is a liquid, then there exist
additional contributions due to the forces between molecules located on both sides of the surface.
Because of their short-range nature, the force acting on a fluid parcel is proportional to its
surface, so that given a surface element dσ with orientation n̄ separating two fluid elements the
resulting force is given by
f¯n (n̄, x̄, t)dσ, (3.15)
where f¯n is the force per unit surface (stress), a function of x̄, t and, also, of the orientation n̄.
In the convention adopted, f¯n is the stress exerted by the fluid on the side of the surface element
to which n̄ points on the fluid located on the opposite side.

Stress tensor
The dependence of the stress f¯n on n̄ can be shown to reduce to a linear relationship, given below
in (3.19), involving a second-order symmetric tensor τ̄¯ called the stress tensor. To derive this
equation, one applies Newton’s second law to the fluid element shown in Fig. 3.2, whose initial
shape is that of a tetrahedron of sides dx1 , dx2 and dx3 and volume dV = (1/6)dx1 dx2 dx3 . Both

d x3
τ13
τ12
τ23
τ22
τ11
d x2
τ21
τ33
τ32
d x1
τ31

Figure 3.2: Stresses on a tetrahedral fluid element.

the inertial term (mass ρ dV times acceleration Dv̄/Dt) and the resultant mass force ρ dV f¯m are

28
 Conservation Laws

proportional to the volume of the element, while the surface forces acting on the four sides of the
tetrahedron are proportional to their surfaces. Hence, for vanishingly small values of dx ≪ 1,
the momentum equation reduces to a balance between the surface forces. In expressing the
equilibrium balance, we shall use f¯i = (τi1 , τi2 , τi3 ) (i = 1, 2, 3) to denote the stress (force per unit
surface) acting on a surface normal to the unit vector ēi of the reference frame. Consequently,
with the sign convention adopted, the force acting on the tetrahedron-shaped fluid element
through the face contained in a plane normal to ēi is given by −dAi f¯i , whereas that applied
through the oblique face is dAf¯n , with n̄ = (n1 , n2 , n3 ) denoting the unit vector normal to the
oblique face. With this notation, the equilibrium equation reduces to
dAf¯n − dA1 f¯1 − dA2 f¯2 − dA3 f¯3 = 0. (3.16)
Using the geometrical identity dAi = ni dA, where dA denotes the surface area of the oblique
face, allows us to rewrite the previous equation in the form
f¯n = n1 f¯1 + n2 f¯2 + n3 f¯3 = n̄ · τ̄¯, (3.17)

where  
τ11 τ12 τ13
τ̄¯ =  τ21 τ22 τ23  (3.18)
τ31 τ32 τ33
is the so-called stress tensor. Equation (3.17) indicates that the stress acting on a surface of
orientation n̄ can be expressed as a function of the nine different components of the tensor τ̄¯,
which are in principle a function of x̄ and t.

dx

dx τ13
τ12

τ23 τ11
τ22

τ21

τ33
dx τ32

τ31

Figure 3.3: Stresses on a cubic element.

The result is even simpler, in that the stress tensor is actually symmetrical, and therefore contains
only six different components. To show the symmetry, one applies the conservation of angular

29
 Conservation Laws

momentum to the cubic fluid element of side dx shown in Fig. 3.3. In the limit dx → 0, the
surface forces and their associated momenta with respect to any point must be in equilibrium. For
instance, the equation imposing a zero torque in the vertical direction reads dx3 (τ21 − τ12 ) = 0,
indicating that τ12 = τ21 , and similar arguments in the other two directions yield τ23 = τ32 and
τ13 = τ31 , thereby demonstrating the aforementioned symmetry.
The stress f¯n (n̄, x̄, t) can therefore be expressed in the form
 
τ11 τ12 τ13
f¯n = τ̄¯ · n̄ =  τ12 τ22 τ23  · n̄, (3.19)
τ13 τ23 τ33
where the normal stresses τ11 , τ22 and τ33 as well as the shear (tangential) stresses τ12 , τ13 and
τ23 are functions of the position and time. Their dependences on the strain rate and on the local
thermodynamic state of the fluid are to be studied in the next section for the case of newtonian
fluids.
The stress τ̄¯ · n̄ can be split into its normal component (n̄ · τ̄¯ · n̄)n̄ and its tangential component
τ̄¯ · n̄ − (n̄ · τ̄¯ · n̄)n̄, the latter called shear stress. There exist three directions, called the principal
directions of stress, for which the resulting shear stress is identically zero, i.e., the only stress
component is normal, a condition expressed mathematically by the eigenvalue problem

τ̄¯ · n̄ = λn̄. (3.20)

The symmetry of the stress tensor τ̄¯ guarantees the existence of three real roots of the associated
characteristic equation  
τ̄ − λ¯Ī = 0.
¯ 
(3.21)
These roots are the normal stresses in the principal directions, which are of course mutually
orthogonal.
The force acting on a given surface can be obtained by integrating (3.19) to give
Z Z
¯
fn dσ = τ̄¯ · n̄dσ. (3.22)
Σ Σ

For a closed surface, Gauss formula (2.9) enables the previous equation to be written in the form
Z Z
τ̄¯ · n̄dσ = ∇ · τ̄¯dV. (3.23)
Σ V

Evaluating the previous equation for a differential fluid element we conclude that ∇ · τ̄¯ is the
resultant surface force per unit volume, so that (∇ · τ̄¯)dV is the surface force acting on a fluid
particle of volume dV . Noting that the mass of the fluid particle is given by ρdV , straightforward
application of Newton’s second law provides
Dv̄
ρdV = ∇ · τ̄¯dV + ρf¯m dV, (3.24)
Dt
which incorporates also the effect of the mass forces possibly acting on the fluid. By eliminating
dV one arrives at the differential form of the momentum equation
Dv̄
ρ = ∇ · τ̄¯ + ρf¯m , (3.25)
Dt
to be derived in the next chapter.

30
 Conservation Laws

Navier-Poisson equation
If the fluid is at rest in a given reference frame, the surface forces act always normal to the
surface and can be in general written as

f¯n = −pn̄, (3.26)

where p is the pressure, a thermodynamic variable introduced above when writing the equations
of state. Correspondingly, the stress tensor reduces in this case to

τ̄¯ = −p¯Ī, (3.27)

where ¯Ī represents the identity tensor.

Note that if T̄¯ = ∇v̄ − T̄¯ is zero in some reference frame, then it is also zero when a different
d r
reference frame is used, so that the fluid motion reduces to a local translation and rotation as a
rigid body. In an appropriately selected local reference frame the fluid would be at rest, so that
one may conclude that the result τ̄¯ = −p¯Ī applies in general whenever T̄ ¯ = 0.
d
¯
When the rate-of-strain tensor T̄d is non-zero (which occurs independently of the reference frame
employed), besides the pressure forces there exist additional surfaces forces, called viscous forces,
that can be expressed in terms of the viscous stress tensor τ̄¯′ , defined such that

τ̄¯ = −p¯Ī + τ̄¯′ . (3.28)

It must be pointed out that, up to this point, no hypotheses have been made about the nature
of the fluid, other than the existence of local thermodynamic equilibrium. However, in order
to relate the viscous stress tensor to the kinematic state of the flow at a particular point, it
is necessary to make assumptions about the functional form of this relationship. Although it
could be quite complex, for the so-called newtonian fluids, which include most of the fluids of
engineering interest (e.g., air and water), it is observed that there exists a linear proportionality
between the components τij′ of the viscous-stress tensor τ̄¯′ and the components γkl of the rate-
¯ . In principle, since both τ̄¯′ and T̄
of-strain tensor T̄ ¯ are symmetrical, this proportionality
d d
condition leads to the six equations
τij′ = αijkl γkl , (3.29)
involving the 36 proportionality constants αijkl .
Equations (3.29) simplify considerably for isotropic fluids, in that the number of proportionality
constants reduces to only two. The demonstration begins by noting that the principal directions
of strain and the principal directions of stress coincide for an isotropic fluid, and could be used
¯ in diagonal form. In the following, τ ′ and γ
as a local reference frame to express τ̄¯′ and T̄ d i i
represent the stresses and strain rates in the principal directions. Evaluating (3.29) for τ1′ and
bearing in mind the isotropy of the fluid one obtains τ1′ = αγ1 + λγ2 + λγ3 , where the coefficients
α and λ relate the strain rate in a given direction with the viscous stresses that appear in that
direction and in the two perpendicular ones, respectively. Analogous expressions, involving also
α and λ would be obtained for τ2′ and τ3′ . The result can be expressed in the compact form
τi′ = (α − λ)γi + λ∇ · v̄ or, alternatively,
¯ + λ(∇ · v̄) ¯Ī,
τ̄¯′ = 2µT̄ (3.30)
d

where the constants µ = (α − λ)/2 and λ are the first and second viscosity coefficients. This
last expression remains valid when the principal directions are replaced by any other orthogonal

31
 Conservation Laws

reference frame (remember that the expansion rate ∇ · v̄ is also invariant under the transforma-
tion). Following the most commonly used notation, the above expression is written in terms of
the bulk viscosity coefficient µB = λ + 23 µ to give

¯ + (µ − 2 µ)(∇ · v̄) ¯Ī,

τ̄¯′ = 2µT̄ (3.31)
d B
3
which further simplifies for an incompressible fluid, for which the second term vanishes. The
expressions for ∇v̄ and ∇ · v̄ given previously can be used to express τ̄¯′ in terms of the derivatives
of the different velocity components.
As can be seen, the viscous stresses depend on the strain rate, which appears explicitely in (3.31),
and also on the local thermodynamic state, through the viscosity coefficients µ and µB . Besides
these two coefficients, a related quantity, which becomes useful when comparing the molecular
transport of mass, momentum and energy, is the so-called kinematic viscosity ν = µ/ρ. Charac-
teristic values of ν for air and water at ambient pressure are, respectively, νa = 1.48 × 10−5 m2 /s
and νw = 1.14 × 10−6 m2 /s at T = 288 K, and νa = 2.24 × 10−5 m2 /s and νw = 0.31 × 10−6
m2 /s at T = 368 K.
The existence of local thermodynamic equilibrium enables the dependence of µ and µB on the
thermodynamic state of the fluid to be studied in terms of any two independent thermodynamic
variables, for instance p and T . Experiments show that the dependence on the pressure is negli-
gible for liquids and gases, a finding that agrees with the results of the kinetic theory of gases.
As for the temperature dependence, the gas viscosity increases with temperature, because the
faster random molecular motion enhances molecular momentum transport. For liquids, the inter-
molecular forces contribute fundamentally to the viscous stresses. With increasing temperatures
the intermolecular distance also increases, causing the intermolecular force to drop and therefore
producing a decrease in the viscosity (we have all seen how the oil viscosity drastically decreases
as the cooking pan is heated on the stove). For both liquids and gases, so long as the tempera-
ture does not vary significantly, one may correctly assume that the viscosity coefficients remain
constant, an assumption to be employed in the following development. As a final comment, it is
worth mentioning that the bulk viscosity coefficient µB is identically zero for monoatomic gases,
as indicated by the kinetic theory of gases.

Force and torque on a solid submerged in a fluid

The resultant force F̄ of the pressure and viscosity forces acting on a surface Σ of normal vector
n̄ pointing towards the fluid can be computed from (3.22) and (3.28) to yield
Z Z
F̄ = − pn̄dσ + τ̄¯′ · n̄dσ. (3.32)
Σ Σ

Similarly, the corresponding moment M̄ of the pressure and viscosity forces with respect to a
given point x̄o is given by
Z Z
M̄ = − (x̄ − x̄o ) ∧ pn̄dσ + (x̄ − x̄o ) ∧ τ̄¯′ · n̄dσ. (3.33)
Σ Σ

These equations can be used, in particular, to determine the force and torque that a fluid exerts
on an solid body in contact with the fluid, with the surface Σ representing in that case the wetted
surface of the body with the normal unit vector n̄ pointing towards the fluid.

32
 Conservation Laws

The momentum equation

Newton’s second law applied to a fluid volume states that the rate of change of the momentum
equals the resultant force associated with the surface and volume forces acting on the fluid
"Z # Z Z
d
ρv̄dV = τ̄¯ · n̄dσ + ρf¯m dV. (3.34)
dt Vf (t) Σf (t) Vf (t)

Expressing the stress tensor τ̄¯ according to (3.28) and using Reynolds transport theorem (3.8)
to extend the above equation for an arbitrary control volume Vc (t) yields
"Z # Z Z Z Z
d
ρv̄dV + ρv̄(v̄ − v̄c ) · n̄dσ = − pn̄dσ + ′
τ̄¯ · n̄dσ + ρf¯m dV. (3.35)
dt Vc (t) Σc (t) Σc (t) Σc (t) Vc (t)

The above equation states that the rate of change of the momentum contained in a given control
volume equals the sum R of four contributions: the convective influx ofR momentum through its
bounding surface − Σc (t) ρv̄(v̄ − v̄c )· n̄dσ, the resultant pressure force − Σc (t) pn̄dσ, the resultant
R R
viscous force Σc (t) τ̄¯′ ·n̄dσ and the resultant mass force Vc (t) ρf¯m dV . Note that, for a fixed control
volume, momentum conservation reduces to
Z  Z Z Z Z
∂ρv̄
dV + ρv̄v̄ · n̄dσ = − pn̄dσ + ¯ ′
τ̄ · n̄dσ + ρf¯m dV. (3.36)
Vo ∂t Σo Σo Σo Vo

Angular momentum equation

Newton’s second law can be alternatively expressed for the angular momentum in the following
terms: the rate of variation of the angular momentum of a system equals the momentum of the
external forces acting on it. For a fluid particle of mass ρdV located at x̄ the angular momentum
with respect to a given position x̄o is given by (x̄ − x̄o ) ∧ ρv̄dV . Therefore, when the overall fluid
volume is considered, angular momentum conservation provides the integral equation
"Z # Z Z
d
ρ(x̄ − x̄o ) ∧ v̄dV = (x̄ − x̄o ) ∧ (τ̄¯ · n̄)dσ + ρ(x̄ − x̄o ) ∧ f¯m dV, (3.37)
dt Vf (t) Σf (t) Vf (t)

which can be written for an arbitrary control volume Vc (t) in the form
"Z # Z
d
ρ(x̄ − x̄o ) ∧ v̄dV + ρ[(x̄ − x̄o ) ∧ v̄][(v̄ − v̄c ) · n̄]dσ =
dt Vc (t) Σc (t)

(3.38)
Z Z Z
− (x̄ − x̄o ) ∧ (pn̄)dσ + (x̄ − x̄o ) ∧ (τ̄¯′ · n̄)dσ + ρ(x̄ − x̄o ) ∧ f¯m dV.
Σc (t) Σc (t) Vc (t)

Energy conservation
Heat transfer by conduction
We have seen above that the existence of strain rates in the flow field results in the appearance
of viscous stresses between adjacent portions of fluid. Likewise, the existence of temperature

33
 Conservation Laws

gradients gives rise to a heat flux (conduction) whose microscopic origin lies - as that of the
viscous stresses- in the random molecular motion, with the work of intermolecular forces also
contributing non-negligibly in the case of liquids. As expected, the amount of energy transported
is proportional to the surface considered, so that the heat flux (energy per unit time) through a
surface element dσ oriented perpendicular to the unit vector n̄ is given by

qn (n̄, x̄, t)dσ, (3.39)

where qn represents the amount of energy transferred per unit surface per unit time in the
direction n̄. In the convention adopted, qn is positive if the energy is transferred to the fluid
element to which n̄ points from the fluid located on the side which n̄ points away from.
To investigate the dependence of qn on the orientation let us consider once more the tetrahedral
fluid element shown in Fig. 3.2. When its size becomes vanishingly small energy conservation
can be shown to reduce in the first approximation to a balance between the heat fluxes through
the four faces of the tetrahedron

qn dA = q1 dA1 + q2 dA2 + q3 dA3 . (3.40)

In the notation adopted, qn dA is the flux through the oblique face with orientation n̄ and qi
denotes the heat flux per unit surface through a plane perpendicular to ēi , so that −qi dAi is
the energy added per unit time to the tetrahedron due to heat conduction through its face
perpendicular to ēi . Using the geometrical relationship dAi = ni dA and dividing the resulting
equation by dA yields finally
qn = q̄ · n̄, (3.41)
showing that the heat flux in direction n̄ = (n1 , n2 , n3 ) can be expressed in terms of a vector
q̄ = (q1 , q2 , q3 ), termed the heat-flux vector, whose three components are in general functions of
the position and time considered.
Extending (3.39) to the computation of the heat flux through a surface Σ yields
Z
q̄ · n̄dσ. (3.42)
Σ

For a closed surface whose interior is free from localized heat sources Gauss formula (2.9) provides
Z Z
q̄ · n̄dσ = ∇ · q̄dV, (3.43)
Σ V

where V is the volume enclosed by Σ. Consideration of the limit of very small volumes reveals
in this case that ∇ · q̄ is the rate of conductive heat loss per unit volume (energy that abandons
the unit volume per unit time due to heat conduction).

Fourier law
The heat-flux vector can be computed according to Fourier law

q̄ = −k∇T, (3.44)

stating a linear proportionality between the heat-flux vector and the local temperature gradient,
with the minus sign indicating that the energy is transported from regions of higher temperature

34
 Conservation Laws

to regions of lower temperature. The proportionality constant k, called thermal conductivity, is

a thermodynamic property that depends on the temperature but is independent of the pressure.
In gases, where the conductive heat transfer is directly linked with the random motion of the
molecules, a rise in temperature produces an increase in the thermal conductivity, so that, for
instance, for air ka = 0.025 W/(m K) at T = 288 K and ka = 0.03 W/(m K) at T = 368
K. Liquids behave differently, because the work of intermolecular forces play a nonnegligible
role, causing the resulting thermal conductivity to exhibit in most liquids a slight decrease with
increasing temperature. For water, however, the opposite response is observed, so that, for in-
stance, kw = 0.59 W/(m K) at T = 288 K and kw = 0.68 W/(m K) at T = 368 K.

The thermal diffusivity α = k/(ρcp ) - or α = k/(ρc) for a liquid- constitutes a useful measure
of heat transfer by conduction. Its dimensions (squared length per unit time) are those of the
kinematic viscosity ν defined above, so that the ratio P r = ν/α is a nondimensional number,
called the Prandtl number, that serves to compare the molecular transport rates of momentum
and heat for a given fluid. In fluids with P r ≫ 1, such as lubricating oils, the transport of
momentum is much more effective that that of heat. Conversely, fluids with P r ≪ 1 may
transport heat at a rate that is much more rapid than that of momentum transport, that being
for instance the case of liquid metals like mercury, for which the presence of free electrons provides
a very effective means for energy transfer. For gases, the Prandtl number is of order unity (e.g.,
P r = 0.72 for air in a wide range of temperatures), while for water the value, dependent on the
temperature, tends to be somewhat larger than unity (P r = 8.14 at T = 288 K and P r = 1.82
at T = 368 K).

The energy equation

As discussed previously, the energy of a fluid (per unit mass) can be expressed as the sum of
the kinetic energy associated with the fluid velocity , |v̄|2 /2, and
R the internal energy e. The
total energy contained in a fluid volume Vf is therefore given by Vf ρ(e + |v̄|2 /2)dV . According
to the first law of thermodynamics, this quantity may change due to heat addition -by heat
conduction- through the bounding fluid surface Σf (t) or by chemical reaction or radiation inside
the fluid volume. Additionally, it might also change due to the mechanical work exerted by mass
or surface forces. This rate balance can be expressed through the integral equation

"Z # Z Z
d 2
ρ(e + |v̄| /2)dV =− pv̄ · n̄dσ + v̄ · τ̄¯′ · n̄dσ
dt Vf (t) Σf (t) Σf (t)
(3.45)
Z Z Z
+ ρf¯m · v̄dV − q̄ · n̄dσ + (Qc + Qr ) dV,
Vf (t) Σf (t) Vf (t)

where Qc and Qr represent, respectively, the rates of heat release (energy per unit volume per
unit time) due to chemical reaction and radiation. Using Reynolds transport theorem enables

35
 Conservation Laws

the above equation to be written for an arbitrary control volume in the form
"Z # Z Z
d 2 2
ρ(e + |v̄| /2)dV + ρ(e + |v̄| /2)(v̄ − v̄c ) · n̄dσ = − pv̄ · n̄dσ
dt Vc (t) Σc (t) Σc (t)

(3.46)
Z Z Z Z
+ v̄ · τ̄¯′ · n̄dσ + ρf¯m · v̄dV − q̄ · n̄dσ + (Qc + Qr )dV.
Σc (t) Vc (t) Σc (t) Vc (t)

Note that, in computing Qc , one needs to consider separate conservation equations to compute
the fluid composition, with account taken of chemical reactions occurring between the different
chemical species. On the other hand, the computation of Qr = −∇ · q̄r from the radiative heat
flux q̄r requires consideration of separate equations for radiation, accounting for local emission
and absorption, with simplified limits arising in optically thin and optically thick environments.
All these complexities are clearly beyond the scope of the present introductory course and are
no longer considered here.

36