The Navier-Stokes Equations
The Navier-Stokes Equations
The Navier-Stokes Equations
In many engineering problems, approximate solutions concerning the overall properties of a fluid
system can be obtained by application of the conservation equations of mass, momentum and en-
ergy written in integral form, given above in (3.10), (3.35) and (3.46), for a conveniently selected
control volume. This approach necessitates in general introduction of simplifying assumptions,
regarding in particular the spatial distributions of the different variables (e.g., uniform velocity
at the open boundaries) and the neglect of terms that are anticipated to give a relatively small
contribution to the overall balances. This integral approach is however not suitable when one is
interested in computing local properties of the flow (e.g., distributions of velocity v̄(x̄, t), density
ρ(x̄, t), pressure p(x̄, t), etc). For that purpose, the conservation equations in integral form
"Z #
d
ρdV = 0, (4.1)
dt Vf (t)
"Z # Z Z Z
d
ρv̄dV =− pn̄dσ + τ̄¯′ · n̄dσ + ρf¯m dV, (4.2)
dt Vf (t) Σf (t) Σf (t) Vf (t)
"Z # Z Z
d 2
ρ(e + |v̄| /2)dV =− pv̄ · n̄dσ + v̄ · τ̄¯′ · n̄dσ
dt Vf (t) Σf (t) Σf (t)
(4.3)
Z Z Z
+ ρf¯m · v̄dV − q̄ · n̄dσ + (Qc + Qr )dV,
Vf (t) Σf (t) Vf (t)
need to be transformed with use made of the Gauss formula to express the rate balances occuring
locally, providing a set of partial differential equations to be integrated with boundary and initial
conditions.
This equation is to be satisfied regardless of the choice of Vf if and only if the identity
∂ρ
+ ∇ · (ρv̄) = 0 (4.5)
∂t
holds at each point of space. This is the continuity (or mass conservation) equation, stating that
the sum of the rate of local density variation and the rate of mass loss by convective outflow
equals zero. An alternative expression is
1 Dρ
= −∇ · v̄, (4.6)
ρ Dt
indicating that the variation of density following the fluid particle is exclusively due to the rate
of volume variation. Note that for an incompressible (constant-density) fluid, the continuity
equation reduces to
∇ · v̄ = 0, (4.7)
a result anticipated previously, whereas for steady gas flow one obtains
∇ · (ρv̄) = 0, (4.8)
that is, the convective rate of mass loss per unit volume is zero.
∂
(ρv̄) + ∇ · (ρv̄v̄) = −∇p + ∇ · τ̄¯′ + ρf¯m . (4.9)
∂t
Using now (4.5) and (2.26) provides
Dv̄ ∂v̄
ρ =ρ + ∇(|v̄| /2) − v̄ ∧ (∇ ∧ v̄) = −∇p + ∇ · τ̄¯′ + ρf¯m ,
2
(4.10)
Dt ∂t
which is simply Newton’s second law expressed per unit volume of fluid, a result anticipated
in (3.25).
For an incompressible fluid with constant viscosity ∇ · τ̄¯′ = −µ∇ ∧ (∇ ∧ v̄), so that (4.10) reduces
to
Dv̄ p
= −∇ − ν∇ ∧ (∇ ∧ v̄) + f¯m . (4.11)
Dt ρ
The viscous force per unit volume can be alternatively expressed as1 −µ∇ ∧ (∇ ∧ v̄) = µ∇2 v̄,
yielding
Dv̄ p
= −∇ + ν∇2 v̄ + f¯m , (4.12)
Dt ρ
1
It is to be noted that in cartesian coordinates each component of the vector ∇2 v̄ is given simply by the
Laplacian of the corresponding velocity component, whereas additional terms appear in cylindrical and spherical
coordinates, thereby complicating the computation.
38
The Navier-Stokes equations
Note that (4.7) and (4.11) suffice to determine the velocity and pressure fields for an incompress-
ible flow with constant viscosity. For such flows, which include those involving water, these two
equations are therefore decoupled from the energy equation, which could be used a posteriori to
determine the temperature field. Also of interest is that, when the mass force is conservative,
introduction of f¯m = −∇U enables (4.11) to be written in the form
Dv̄ p + ρU
= −∇ − ν∇ ∧ (∇ ∧ v̄). (4.13)
Dt ρ
This last expression suggests that, for the motion of perfect liquids, introduction of the reduced
pressure P = p + ρU (e.g., P = p + ρgz when f¯m = ḡ) may simplify the description significantly.
For instance, when the pressure does not enter in the boundary conditions, the solution described
with use made of P becomes entirely independent of the mass forces.
A well-known result arises for steady incompressible flows with conservative mass forces when the
flow conditions are such that the effect of viscosity is negligible, so that the momentum equation
reduces to
p |v̄|2
∇ +U + − v̄ ∧ (∇ ∧ v̄) = 0 (4.14)
ρ 2
as can be seen by using (4.13) with ν = 0 together with
Dv̄
= ∇(|v̄|2 /2) − v̄ ∧ (∇ ∧ v̄), (4.15)
Dt
the corresponding steady form of (2.26). The vector equation (4.14) can be projected along
streamlines by multiplying at each point in space by the unit vector v̄/|v̄|. Since the contribution
of the last term is identically zero, the projection reduces to
∂ p |v̄|2
+U + = 0, (4.16)
∂l ρ 2
where ∂/∂l = |v̄|−1 (v̄ · ∇) denotes the derivative along a stream line. It then follows that for
frictionless steady flow of an incompressible fluid the quantity
p + ρU + ρ|v̄|2 /2 = Cl (4.17)
remains constant along any given stream line, with a value Cl that in general is different for
different stream lines. Equation (4.17) is the so-called Bernoulli’s theorem, that provides a
useful relationship between the variations of kinetic energy, potential energy and pressure along
streamlines. Note that, if the flow is irrotational, the constant Cl is the same for all streamlines,
because ∇(p/ρ + U + |v̄|2 /2) = 0, as follows from (4.14).
39
The Navier-Stokes equations
D
ρ (e + |v̄|2 /2) = −∇ · (pv̄) + ∇ · (v̄ · τ̄¯′ ) + ρf¯m · v̄ − ∇ · q̄ + Qc + Qr . (4.19)
Dt
If the mass force derives from a steady potential such that f¯m = −∇U with ∂U/∂t = 0, then
the above equation admits the alternative writing
D
ρ (e + |v̄|2 /2 + U ) = −∇ · (pv̄) + ∇ · (v̄ · τ̄¯′ ) − ∇ · q̄ + Qc + Qr . (4.20)
Dt
A separate equation for the mechanical (kinetic) energy |v̄|2 /2 can be derived by taking the dot
product of (4.10) and v̄. Since v̄ · [v̄ ∧ (∇ ∧ v̄)] = 0, the product provides the scalar equation
D
ρ (|v̄|2 /2) = −v̄ · ∇p + v̄ · (∇ · τ̄¯′ ) + ρv̄ · f¯m , (4.21)
Dt
which is a purely mechanical energy law. As can be seen, the rate of variation of the kinetic
energy following the fluid particle equals the work rate of the mass force ρv̄ · f¯m plus the work
rate of the surface forces associated with the translational motion of the fluid particle v̄ · (∇ · τ̄¯) =
v̄ · (−∇p + τ̄¯′ ). Subtracting now (4.21) from (4.19) yields
De
ρ = −p∇ · v̄ + τ̄¯′ : ∇v̄ − ∇ · q̄ + Qc + Qr , (4.22)
Dt
clearly indicating that the rate of variation of the internal energy is associated with heat addition
and with the deformation work rate of the surface forces. The term −p∇ · v̄ corresponds to the
compression work, which is a reversible contribution, in that when the fluid volume is reduced
(∇ · v̄ < 0) the compression work produces an increase in the internal energy; conversely, the
internal energy decreases for positive expansion rates ∇ · v̄ > 0. The term φv = τ̄¯′ : ∇v̄ is the
deformation work due to viscous forces. It is non-negative (φv ≥ 0)2 , and corresponds to the
rate of mechanical energy dissipation per unit volume and unit time. Note that both −p∇ · v̄
and τ̄¯′ : ∇v̄ are independent of the orientation, position and motion of the reference frame.
¯ + (µB − 2 µ)(∇ · v̄)2 = 2µ [(γ11 − γ22 )2 + (γ11 − γ33 )2 + (γ22 − γ33 )2 + 6(γ 2 + γ 2 + γ 2 )] + µB (∇ · v̄)2 ,
¯ : T̄
φv = 2µT̄d d 12 13 23
3 3
thereby demonstrating that φv ≥ 0 provided µ > 0 and µB > 0.
40
The Navier-Stokes equations
(4.24)
"Z Z #
− −p∇ · v̄dV + φv dV .
Vf (t) Vf (t)
The way we have written (4.24) and (4.25) is intended to give a clear picture of the balances of
work and energy in the flow field. Thus, the work done by mass forces in the interior of the fluid
volume as well as that done by surface forces on its bounding surface (the right-hand-side terms
in the first line of (4.24)) are devoted to increase the net amount of mechanical energy contained
in the fluid volume. On the other hand, the heat added by conduction, radiation and chemical
reaction (the right-hand-side terms in the first line of (4.25)) is employed directly to increase
the total internal energy. Viscous dissipation as well as compression work, written for clarity
in separate lines in (4.24) and (4.25), are the mechanisms transforming mechanical energy into
internal energy, with only the latter being reversible.
41
The Navier-Stokes equations
indicating that, when heat addition, temporal pressure variations, and viscous forces are all
simultaneously negligible, the combination h + |v̄|2 /2 + U remains constant in the evolution of
each fluid particle.
(4.31)
Z Z
− (q̄/T ) · n̄dσ + (Qc + Qr )/T dV,
Σf (t) Vf (t)
for the variation of the total entropy contained in a fluid volume. The term involving (Qc +
Qr )/T represents a volumetric source of entropy associated with chemical reaction and radiation.
On the Rother hand, q̄ · n̄/T is the entropy flux across the surface due to heat conduction, so
that − Σf (t) (q̄/T ) · n̄dσ is the corresponding rate of entropy gain. Viscous dissipation gives
R R
a nonnegative contribution Vf (t) (φv /T )dV . Finally, the term − Vf (t) [(q̄ · ∇T )/T 2 ]dV is an
additional source of entropy that appears due to conduction inside the control volume. This
contribution is also nonnegative, as can be seen with use of (3.44).
42
The Navier-Stokes equations
of the hypotheses introduced can be expected to fail under extreme conditions (rarefied flows,
nanofluidics applications, etc), the anticipated range of validity of the mathematical formulation
derived covers most applications of interest in engineering.
ρ = ρo and e = eo + cT (4.38)
∇ · v̄ = 0 (4.40)
Dv̄
ρ = −∇p − µ∇ ∧ (∇ ∧ v̄) + ρf¯m (4.41)
Dt
DT ¯ : T̄
¯ + k∇2 T + Q + Q ,
ρc = 2µT̄ d d c r (4.42)
Dt
which are written out below for cartesian, cylindrical and spherical coordinates.
43
The Navier-Stokes equations
Equations (4.32), (4.33) and (4.34) describe all fluid motions (within the range of validity of the
various assumptions introduced in their derivation). The difference between two flow problems
lies in their initial and boundary conditions, which need to be specified to enable the integration
of the Navier-Stokes equations to be performed.
For unsteady fluid motion, at the initial instant we must define the velocity field v̄ = v̄o (x̄)
together with the thermodynamic state, given for instance in terms of p = po (x̄) and T = To (x̄)
(or any other pair of independent thermodynamic variables). For an incompressible fluid, the
initial velocity field must necessarily satisfy ∇ · v̄o = 0. Because of the presence of the term
∂ρ/∂t in (4.32), in the case of gas flow there is no such restriction. Clearly, no initial conditions
are required for the study of steady or periodic fluid flows.
The Navier-Stokes equations determine the flow evolution within the flow field, and must be
therefore complemented with appropriate boundary conditions on the boundaries of the flow
field. In typical configurations, the fluid may be confined by solid walls, as occurs in wall-
bounded flows. The fluid velocity in contact with the wall will be assumed to be equal to the
wall velocity, i.e., at x̄ = x̄p we impose v̄ = v̄p , which reduces to v̄ = 0 when the wall is at rest
relative to the reference frame used in the description. This so-called non-slip condition comes
from assuming that the interaction of the molecules near the wall with the wall is equivalent
to that occurring between neighboring particles. For the velocity field to be continuous, no
velocity discontinuity is permitted between the wall and the adjacent fluid. There exists ample
experimental evidence confirming the validity of the non-slip condition, at least for flows in which
the mean free path λ is much shorter than the macroscopic length of the flow field L. Note that,
when this is not the case, the non-slip condition must be relaxed, along with the condition of
local thermodynamic equilibrium, thereby complicating the associated analysis.
We shall also assume that there exists local thermodynamic equilibrium at the wall, so that the
fluid temperature in contact with the wall is that of the wall (T = Tp at x̄ = x̄p ). There must also
exist an equilibrium of the heat fluxes at the wall surface, providing an additional equation that
reduces to q̄ · n̄ = ∂T /∂n = 0 at x̄ = x̄p when the wall is thermally insulated (adiabatic wall). In
principle, in integrating the conservation equations for the fluid, one could use alternatively the
temperature distribution Tp or the heat flux ∂T /∂n as boundary condition for the temperature
at the wall x̄ = x̄p . In many problems, however, neither quantity is known a priori, and they
have to be determined as part of the solution, which involves the integration of the Navier-Stokes
equations for the fluid together with the heat conduction equation in the solid, coupled with the
conditions of thermodynamic equilibrium and heat flux balance at the wall surface x̄ = x̄p .
The boundary conditions needed when the fluid extends to infinity (i.e., to distances much larger
than the characteristic macroscopic size of the flow field L), as occurs in external aerodynamic
problems, include specification of the boundary velocity field and its accompanying thermody-
namic state. These external boundary conditions can be written in the form v̄ = v̄∞ (x̄, t),
p = p∞ (x̄, t), and T = T∞ (x̄, t), for |x̄| → ∞. For instance, to study the motion of a uniform
stream of velocity U∞ over a solid body of characteristic size L, we shall impose v̄ = U∞ ēx ,
p = p∞ , and T = T∞ at |x̄|/L ≫ 1.
As an example, consider the motion a liquid stream that moves with velocity U∞ (t)ēx over a
spherical body of radius R. If the density and transport properties of the fluid are constant
and the effect of radiation and chemical reaction are negligible, the problem reduces to that of
44
The Navier-Stokes equations
integrating
∇ · v̄ = 0 (4.43)
∂v̄
ρ + ρv̄ · ∇v̄ = −∇p + µ∇2 v̄ − ρgēz (4.44)
∂t
∂T ¯ : T̄
¯ + k∇2 T.
ρc + ρcv̄ · ∇T = 2µT̄ d d (4.45)
∂t
Boundary conditions must be imposed on the surface of the body
|x̄| = R : v̄ = T − Tw = 0, (4.46)
where Tw is the temperature on the surface of the body, different from the value T∞ of the free
stream. Far from the body the velocity should approach that of the free stream whereas the
pressure equilibrates the gravity and inertial forces according to
dU∞
|x̄| ≫ R : v̄ = U∞ (t)ēx , T = T∞ , p + ρgz + ρ x = constant. (4.47)
dt
If the fluid is initially at rest (i.e., U = 0 for t ≤ 0), appropriate initial conditions are
Tw − T∞
t = 0 : v̄ = 0, T = T∞ + , (4.48)
|x̄|/R
where the temperature field is that obtained for a fluid at rest through solution of the corre-
sponding reduced energy equation ∇2 T = 0 with account taken of the spherical symmetry.
U z
8
y
x
T Tw
8
45
The Navier-Stokes equations
′ ∂vx 2 ′ ∂vx ∂vy
τxx = 2µ + (µB − µ)(∇ · v̄) τxy =µ +
∂x 3 ∂y ∂x
′ ∂vy 2 ′ ∂vy ∂vz
τyy = 2µ + (µB − µ)(∇ · v̄) τyz =µ +
∂y 3 ∂z ∂y
′ ∂vz 2 ′ ∂vx ∂vz
τzz = 2µ + (µB − µ)(∇ · v̄) τzx =µ +
∂z 3 ∂z ∂x
′ ∂vr 2 ′ ∂ vθ 1 ∂vr
τrr = 2µ + (µB − µ)(∇ · v̄) τrθ =µ r +
∂r 3 ∂r r r ∂θ
′ 1 ∂vθ vr 2 ′ ∂vθ 1 ∂vz
τθθ = 2µ + + (µB − µ)(∇ · v̄) τθz =µ +
r ∂θ r 3 ∂z r ∂θ
′ ∂vz 2 ′ ∂vr ∂vz
τzz = 2µ + (µB − µ)(∇ · v̄) τzr =µ +
∂z 3 ∂z ∂r
∂vφ
Spherical coordinates (∇ · v̄ = 1 ∂ 2
r 2 ∂r (r vr ) + 1 ∂
r sin θ ∂θ (vθ sin θ) + 1
r sin θ ∂φ )
′ ∂vr 2 ′ ∂ vθ 1 ∂vr
τrr = 2µ + (µB − µ)(∇ · v̄) τrθ =µ r +
∂r 3 ∂r r r ∂θ
v
′ 1 ∂vθ vr 2 ′ sin θ ∂ φ 1 ∂vθ
τθθ = 2µ + + (µB − µ)(∇ · v̄) τθφ =µ +
r ∂θ r 3 r ∂θ sin θ r sin θ ∂φ
v
′ 1 ∂vφ vr vθ cotθ 2 ′ 1 ∂vr ∂ φ
τφφ = 2µ + + + (µB − µ)(∇ · v̄) τφr =µ +r
r sin θ ∂φ r r 3 r sin θ ∂φ ∂r r
46
The Navier-Stokes equations
Continuity equation
∂vx ∂vy ∂vz
+ + =0
∂x ∂y ∂z
Momentum equation
2
∂vx ∂vx ∂vx ∂vx ∂p ∂ vx ∂ 2 vx ∂ 2 vx
ρ + vx + vy + vz =− +µ + + + ρfmx
∂t ∂x ∂y ∂z ∂x ∂x2 ∂y 2 ∂z 2
2
∂vy ∂vy ∂vy ∂vy ∂p ∂ vy ∂ 2 vy ∂ 2 vy
ρ + vx + vy + vz =− +µ + + + ρfmy
∂t ∂x ∂y ∂z ∂y ∂x2 ∂y 2 ∂z 2
∂vz ∂vz ∂vz ∂vz ∂p ∂ 2 vz ∂ 2 vz ∂ 2 vz
ρ + vx + vy + vz =− +µ + + + ρfmz
∂t ∂x ∂y ∂z ∂z ∂x2 ∂y 2 ∂z 2
Energy equation
2
∂T ∂T ∂T ∂T ∂ T ∂2T ∂2T
ρc + vx + vy + vz = φv + k + + + Qc + Qr
∂t ∂x ∂y ∂z ∂x2 ∂y 2 ∂z 2
where
" #
∂vx 2 ∂vy 2 ∂vz 2 ∂vx ∂vy 2 ∂vx ∂vz 2 ∂vy ∂vz 2
φv = µ 2 +2 +2 + + + + + +
∂x ∂y ∂z ∂y ∂x ∂z ∂x ∂z ∂y
47
The Navier-Stokes equations
Continuity equation
1 ∂rvr 1 ∂vθ ∂vz
+ + =0
r ∂r r ∂θ ∂z
Momentum equation
∂vr ∂vr vθ ∂vr v2 ∂vr
ρ + vr + − θ + vz =
∂t ∂r r ∂θ r ∂z
∂p ∂ 1 ∂ 1 ∂ 2 vr ∂ 2 vr 2 ∂vθ
− +µ (rvr ) + 2 + − 2 + ρfmr
∂r ∂r r ∂r r ∂θ 2 ∂z 2 r ∂θ
∂vθ ∂vθ vθ ∂vθ vr vθ ∂vθ
ρ + vr + + + vz =
∂t ∂r r ∂θ r ∂z
1 ∂p ∂ 1 ∂ 1 ∂ 2 vθ ∂ 2 vθ 2 ∂vr
− +µ (rvθ ) + 2 + + 2 + ρfmθ
r ∂θ ∂r r ∂r r ∂θ 2 ∂z 2 r ∂θ
∂vz ∂vz vθ ∂vz ∂vz ∂p 1 ∂ ∂vz 1 ∂ 2 vz ∂ 2 vz
ρ + vr + + vz =− +µ r + 2 + + ρfmz
∂t ∂r r ∂θ ∂z ∂z r ∂r ∂r r ∂θ 2 ∂z 2
Energy equation
∂T ∂T vθ ∂T ∂T 1 ∂ ∂T 1 ∂2T ∂2T
ρc + vr + + vz = φv + k r + 2 2 + + Qc + Qr ,
∂t ∂r r ∂θ ∂z r ∂r ∂r r ∂θ ∂z 2
where
( " #
∂vr 2 1 ∂vθ vr 2 ∂vz 2
φv = µ 2 + + + +
∂r r ∂θ r ∂z
)
1 ∂vz ∂vθ 2 ∂ vθ 1 ∂vr 2 ∂vr ∂vz 2
+ + r + + +
r ∂θ ∂z ∂r r r ∂θ ∂z ∂r
48
The Navier-Stokes equations
Continuity equation
1 ∂ 2 1 ∂ 1 ∂vφ
2
(r vr ) + (vθ sin θ) + =0
r ∂r r sin θ ∂θ r sin θ ∂φ
Momentum equation
!
vθ2 + vφ2
∂vr ∂vr vθ ∂vr vφ ∂vr ∂p ∂ 1 ∂ 2
ρ + vr + + − =− +µ (r vr ) +
∂t ∂r r ∂θ r sin θ ∂φ r ∂r ∂r r 2 ∂r
1 ∂ ∂vr 1 ∂ 2 vr 2 ∂ 2 ∂vφ
sin θ + 2 2 − 2 (vθ sin θ) − 2 + ρfmr
r 2 sin θ ∂θ ∂θ r sin θ ∂φ2 r sin θ ∂θ r sin θ ∂φ
!
∂vθ ∂vθ vθ ∂vθ vφ ∂vθ vr vθ vφ2 cotθ 1 ∂p 1 ∂ 2 ∂vθ
ρ + vr + + + − =− +µ 2 r +
∂t ∂r r ∂θ r sin θ ∂φ r r r ∂θ r ∂r ∂r
1 ∂ 1 ∂ 1 ∂ 2 vθ 2 ∂vr 2 cotθ ∂vφ
(vθ sin θ) + 2 2 + 2 − 2 + ρfmθ
r 2 ∂θ sin θ ∂θ r sin θ ∂φ2 r ∂θ r sin θ ∂φ
∂vφ ∂vφ vθ ∂vφ vφ ∂vφ vr vφ vθ vφ 1 ∂p 1 ∂ 2 ∂vφ
ρ + vr + + + + cotθ = − +µ 2 r +
∂t ∂r r ∂θ r sin θ ∂φ r r r sin θ ∂φ r ∂r ∂r
1 ∂ 1 ∂ 1 ∂ 2 vφ 2 ∂vr 2 cotθ ∂vθ
(vφ sin θ) + 2 2 + 2 + 2 + ρfmφ
r 2 ∂θ sin θ ∂θ r sin θ ∂φ2 r sin θ ∂φ r sin θ ∂φ
Energy equation
∂T ∂T vθ ∂T vφ ∂T
ρc + vr + + =
∂t ∂r r ∂θ r sin θ ∂φ
1 ∂ 2 ∂T 1 ∂ ∂T 1 ∂2T
φv + k 2 r + 2 sin θ + 2 2 + Qc + Qr ,
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2
where
( " #
∂vr 2 1 ∂vθ vr 2 1 ∂vφ vr vθ cotθ 2
φv = µ 2 + + + + + +
∂r r ∂θ r r sin θ ∂φ r r
)
sin θ ∂ vφ 1 ∂vθ 2 ∂ vθ 1 ∂vr 2 1 ∂vr ∂ vφ 2
+ + r + + +r
r ∂θ sin θ r sin θ ∂φ ∂r r r ∂θ r sin θ ∂φ ∂r r
49
The Navier-Stokes equations
50