Sunteți pe pagina 1din 9

Conservation equations in uid dynamics

Bassi Andrea
1 ottobre 2018

Indice

1 Basics 1
1.1 Lagrangian and Eulerian approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Lagrangian approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Eulerian approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The material derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Reynolds Transport Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Conservation Equations 3
2.1 Mass conservation (continuity) equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Momentum conservation equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Kinetic and Mechanical energy equations . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Energy conservation equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

i
1 Basics
1.1 Lagrangian and Eulerian approaches
1.1.1 Lagrangian approach
The Lagrangian approach, which is the most common when dealing with solid mechanics, is based on the
concept of material volume: a portion of matter that
ˆ Always contains the same particles (material points)

ˆ Can change and move with time with a certain, sometimes unknown, trajectory

In this approach we follow the material volume in its history inside the system we are studying. To do
so and to provide a proper study of the physics of the system, we need to know exactly the trajectory
for every single material point within the material volume.
This is a clear limitation and introduces an important complexity for uid dynamics studies, since the
knowledge of such trajectories isn't obviously possible.
However, conservation equations always deal with the same material points, therefore a Lagrangian
approach is suitable for this kind of operation.
For the mentioned disadvantages with respect to uid dynamics we need to change our point of view, but
it will also be important to nd a bridge between the two approaches so that we can express conservation
equations valid for the Lagrangian approach in the newly adopted reference.

1.1.2 Eulerian approach


In uid dynamics the most used and widespread point of view is the Eulerian approach, which is based
on the denition of a control volume ΩCV : a portion of space that
ˆ Doesn't always contain the same particles (material points).

ˆ Doesn't usually change nor move with time, or changes and moves with known trajectory and/or
dynamics
In this case physical quantities lose their reference to a single particle/material volume and become eld
quantities. For example, while in the Lagrangian approach we may be interested in changes in pressure
for a certain particle p(t), in an Eulerian reference we want to know the pressure eld p(x, y, z, t).
In the Eulerian approach we may say that we watch particles as they ow by, remaining anchored to a
certain portion of space and without following the particle in its whole trajectory.

1.2 The material derivative


To express the derivative in time of a generic quantity referred to a material point φ in an Eulerian
approach, we can consider rstly the equivalent eld quantity φ(r, t) and write
dφ(r, t) dφ(x, y, z, t)
= (1.1)
dt dt
Now as we know, for composed functions we can apply the well known chain rule for derivation. Therefore
starting from the previous results we can write
dφ ∂φ ∂φ dx ∂φ dy ∂φ dz
= + · + · + · (1.2)
dt ∂t ∂x dt ∂y dt ∂z dt
which we may write in a more concise form
dφ ∂φ
= + v · ∇φ (1.3)
dt ∂t

1
The rst term is the local derivative and the second term is known as the advective derivative.
We can now isolate and identify this new operator called substantial or material derivative, which
will be indicated with D/Dt and is equivalent to:
D ∂
:= +v·∇ (1.4)
Dt ∂t

This derivative follows a particle of xed identity (i.e. follows a Lagrangian approach) considering the
quantity as a eld quantity and the above mentioned facts make it particularly convenient for expressing
laws of conservation and mechanics in an Eulerian context.

1.3 Reynolds Transport Theorem


The Reynolds Transport Theorem, sometimes indicated as RTT, is a fundamental extension of the Leib-
niz theorem that together with the concept of material derivative lls the gap between Lagrangian and
Eulerian approach.
In its most general form it is an expression that allows to evaluate the derivative of a global quantity
as the local derivative integrated over a certain volume plus the contribute given by the motion of the
mentioned volume.
Z Z
dΦ ∂φ 0
= dΩ + φ(v · n)dA (1.5)
dt Ω(t) ∂t ∂Ω(t)

Where v is the velocity evaluated at the boundary of the volume. The rst term is an expression of
the local change of φ integrated over time, while the second term is the net ux across the volume
boundary surface.
Applying this theorem to the property referred to a material volume we can write
Z Z
dΦm ∂φ
= dΩ + φ(v · n)dA (1.6)
dt Ωm (t) ∂t ∂Ωm (t)

Where v is the velocity of particle points evaluated at the boundary of the volume.
Switching to an Eulerian approach we can still write that
Z Z
dΦCV ∂φ
= dΩ + φ(vb · n)dA (1.7)
dt ΩCV (t) ∂t ∂ΩCV (t)

Where now vb is referred to the points belonging to the geometric surface of the selected control volume.
Now, we can surely assume that at a certain time t0
(
Ωm (t0 ) = ΩCV (t0 )
(1.8)
∂Ωm (t0 ) = ∂ΩCV (t0 )

While at this time it is certainly true that Φm = ΦCV it is clear that


dΦm dΦCV
6= (1.9)
dt dt
Indeed, we can take equations (1.6) and (1.7) we can write
Z
dΦm dΦCV
(1.10)

= + φ((v − vb ) · n)dA
dt t0 dt
t0 ∂ΩCV (t0 )

It is now clear that the dierence of the velocity of a point particle on the surface of the volume and the
one of the boundary surface is equal to the net ux velocity.
We can now insert equation (1.7) in this equation to get to the nal result:
Z Z
dΦm ∂φ
= dΩ + φ(v · n)dA (1.11)
dt t0 ΩCV (t) ∂t ∂ΩCV (t0 )

2
This equation basically allows us to exploit a control volume as a reference for calculating the rate of
change of referred to quantity a material volume which is coincident with the above-mentioned control
volume. The rst term is the local rate of change of the volumetric quantity integrated over the control
volume, while the second one is the ux of the volumetric quantity over the border of the material volume
coincident with the control volume at the initial time.
This implies a constraint over the choice of one of the two volumes, since they must coincide. However,
this is not a problem because we decide to choose freely the control volume and to evaluate properties
of the material volume coincident to the rst, since in uid dynamics we do not have special preferences
about the material volume to choose.

2 Conservation Equations
Now that we have collected the basic tools to deal with conservation equations in an Eulerian approach,
let's study them both from an analytical and from a physical perspective.

2.1 Mass conservation (continuity) equation


The continuity equation expresses the concept of mass conservation, which is based on the following
equation, with respect to a material volume :
dM
=0 (2.1)
dt
Using the RTT as expressed in the form presented in equation (??) we can rewrite equation (2.1) with
respect to a generic control volume, which as previously said, happens to be more suitable for uid
dynamics study: Z Z
dM d
= ρdΩ + ρv · ndA = 0 (2.2)
dt dt ΩCV ∂ΩCV

If we consider the control volume to be xed we can rewrite the previous equation bringing the derivative
inside the integral Z Z
dM ∂ρ
= dΩ + ρv · ndA = 0 (2.3)
dt ΩCV ∂t ∂ΩCV

A rst consideration to make is the one regarding the physical meaning of each term. While the rst
one takes into account changes of the density inside the control volume, the second one is responsible for
mass uxes across the control volume border. In fact, with the hypothesis of steady state, the previous
equation becomes Z
dM
= ρv · ndA = 0 (2.4)
dt ∂ΩCV

which basically means that at steady state, the mass ux entering the control volume must be compen-
sated by the outgoing mass ux.
An alternative form of equation (2.4) can be obtained by applying the Gauss theorem:
Z
dM
= ∇ · (ρv)dΩ = 0 (2.5)
dt ΩCV

And imposing the equation to the integrand in order to maintain the arbitrarily of the choice of the
control volume, we can say, that at steady state
∇ · (ρv) = 0 (2.6)
If we apply Gauss theorem to equation (2.3) we can get to a form in which we have only integrals on the
control volume Z  
dM ∂ρ
= + ∇ · (ρv) dΩ = 0 (2.7)
dt ΩCV ∂t

3
Since the control volume is chosen arbitrarily, we can impose that the integral in equation (2.7) is null
not due to an opportune integration domain, but thanks to the fact that the integrand function is null.
∂ρ
+ ∇ · (ρv) = 0 (2.8)
∂t
∂ρ
+ v · ∇ρ + ρ∇ · v = 0 (2.9)
∂t
By looking at equation (1.4) we can recognize the material derivative that appears in equation (2.9) and
write it more explicitly:

+ ρ∇ · v = 0 (2.10)
Dt
For a uid which has the property of being incompressible, we can say that the material1 derivative of
its density is null, and therefore for those uids the continuity equation becomes
∇·v =0 (2.11)
We can now link this result with incompressible ows, since one implies the other and viceversa:

∇·v =0⇔ = 0 ⇔ Incompressible ow (2.12)
Dt
It is then sucient to determine that a ow is incompressible that the velocity eld has null divergence.

2.2 Momentum conservation equation


To express the momentum conservation consequences, indicating the momentum with P = ρv we have
to initially write it in the most general form for a material volume:
N
dP X
= Fi (2.13)
dt i=1

We can imagine the forces that appear in equation (2.13) to be composed surface forces Fs and volume
forces Fv . So we can write the following equation
Ns Nv
dP X
(2.14)
X
= Fsi + Fvi
dt i=1 i=1

Switching to the Eulerian approach we need to consider the control volume expression for the momentum:
Z Z
dP d(ρv) d
= = ρvdΩ + ρv(v · n)dA (2.15)
dt dt dt ΩCV ∂ΩCV

We can also write an integral expression for the two types of forces we introduced before.
ˆ Volume forces can be expressed as the integral over the CV of a general volumetric force f . In this
way we can write
Nv Z
(2.16)
X
Fvi = f dΩ
i=1 ΩCV

The most common case is the one in which the only volume force acting on the system is the
gravitational force. Maintaining the form introduced in the previous equation and substituting the
general force f we can then nally write
Nv Z
(2.17)
X
Fvi = ρgdΩ
i=1 ΩCV
1 Please pay attention to the fact that an incompressible uid implies that the derivative of the density of the particle
is constant in time. On the other hand, the partial time derivative of the density of a point inside the CV is not necessarily
null: we could have dierent particles owing with dierent density (but constant for them) and this would mean that
our xed point would experience a non-null derivative of the density over time. Clearly this eect must be justied and
weighted by the remaining term in the material derivative.

4
ˆ Surface forces can be expressed as the integral over the CV surface of the supercial forces acting
on the CV border. If we consider an innitesimal surface dA and its outgoing normal vector n we
can indicate as σn the vector that represents the force per surface unit acting on that innitesimal
surface.
According to Cauchy's stress theorem, merely by knowing the stress vectors on three mutually
perpendicular planes, the stress vector on any other plane passing through that point can be found.
So we will use this theorem to indicate a general stress acting on a surface with a stress tensor τ .
In some cases which are quite frequent in uid dynamics, this tensor is decomposable in a diagonal
tensor −pI which represents the hydrostatic pressure forces and a viscous stress tensor τ v
τ = −pI + τ v (2.18)
Therefore, we can write the surface forces using the following form:
Ns Z
(2.19)
X
Fsi = τ · ndA
i=1 ∂ΩCV

Which can be written as


Ns Z Z Z Z
(2.20)
X
Fsi = −pI · ndA τ dev · ndA = −pndA τ dev · ndA
i=1 ∂ΩCV ∂ΩCV ∂ΩCV ∂ΩCV

Thanks to these relations we can nally write this form as a consequence of the momentum conservation,
considering equations (2.17), (2.19) and (2.15).
Z Z Z Z
dP d
= ρvdΩ + ρv(v · n)dA = ρgdΩ + τ · ndA (2.21)
dt dt ΩCV ∂ΩCV ΩCV ∂ΩCV

By applying the Gauss theorem and considering the CV as a xed one, we can update the previous
equation to the following form:
Z Z Z Z
dP ∂(ρv)
= dΩ + ∇ · (ρvv)dΩ = ρgdΩ + ∇ · τ dΩ (2.22)
dt ΩCV ∂t ΩCV ΩCV ΩCV

Which is equivalent to Z  
∂(ρv)
+ ∇ · (ρvv) − ρg − ∇ · τ dΩ = 0 (2.23)
ΩCV ∂t
As we did for the continuity equation and for the very same reasons mentioned before, we ask for the
solution that equals the integrand to zero:
∂(ρv)
+ ∇ · (ρvv) − ρg − ∇ · τ = 0 (2.24)
∂t
Which is equivalent to
∂(ρv)
+ ∇ · (ρvv) = ρg + ∇ · τ (2.25)
∂t
We can develop the right term to try to simplify it
∂(ρv) ∂ρ ∂v
+ ∇ · (ρvv) = v + v∇ · (ρv) + ρ + ρv · ∇v (2.26)
∂t ∂t ∂t
This equation, considering the denition of substantial derivative given in equation (1.4) can be rewritten
as    
∂(ρv) ∂ρ ∂v
+ ∇ · (ρvv) = v + ∇ · (ρv) + ρ + v · ∇v (2.27)
∂t ∂t ∂t
We can now express more explicitly the previous equation and simplify it thanks to the conservation
equation in the form of equation (2.8). The rst of the two addends at the right side is in fact equal to

5
zero, according to the above mentioned conservation equation.
Therefore equation (2.26) becomes2
 
∂(ρv) ∂v Dv
+ ∇ · (ρvv) = ρ + v · ∇v = ρ (2.30)
∂t ∂t Dt

We can apply this important result to simplify the momentum conservation equation in the previous
form, given by equation (2.25):
∂(ρv) Dv
+ ∇ · (ρvv) = ρ = ρg + ∇ · τ (2.31)
∂t Dt
Now we can write a more explicit form by expressing the material derivative and taking the density to
the second member of the equation
∂v 1
+ v · ∇v = g + ∇ · τ (2.32)
∂t ρ

Or, inserting the viscous stress tensor


∂v 1 1
+ v · ∇v = g − ∇p + ∇ · τ v (2.33)
∂t ρ ρ

These are a partial dierential equations (PDE) known as Navier-Stokes equations that introduce as
unknowns the velocity eld and the stress tensor. Therefore, the problem will need the use of constitutive-
phenomenological equations.

2.2.1 Kinetic and Mechanical energy equations


Applying the scalar product to the momentum conservation equation with respect to the velocity vector
v we can write the kinetic energy equation: let's do it starting from equation (2.33).
First of all we explicit the material derivative
∂v Dv
+ v · ∇v = (2.34)
∂t Dt
Then we apply the above mentioned scalar product
Dv 1 1
· v = g · v − v · ∇p + v · (∇ · τ v ) (2.35)
Dt ρ ρ
The reason why this equation is called kinetic energy equation is now clear: if we apply basic dierential
rules, equation (2.35) becomes
 
1 2
D v
Dv 2 Dek 1 1
·v = = = g · v − v · ∇p + v · (∇ · τ v ) (2.36)
Dt Dt Dt ρ ρ
Now every term is clear: a contribution to a change in kinetic energy inside a control volume can be
given by the interaction between gravitational (volume) forces and the motion eld, by an appropriate
pressure gradient with respect to the motion eld and by the right interaction between the latter and
2 This result is always true for every expression of the form
∂(ρφ)
+ ∇ · (ρvφ) (2.28)
∂t
Which always respects equation
∂(ρφ) Dφ
+ ∇ · (ρvφ) = ρ (2.29)
∂t Dt

6
viscous stresses.
The nal form of the kinetic energy equation can now be written

D(ek ) 1 1
= g · v − v · ∇p + v · (∇ · τ v ) (2.37)
Dt ρ ρ

To add meaning to this equation we can consider the following equations which are meant to develop
some terms in order to have a better look at the role each element is playing:
v · ∇p = ∇ · (vp) − p∇ · v (2.38)
v · (∇ · τ v ) = ∇ · (τ v · v) − τ v : ∇v (2.39)
In addition to this we can consider the gravitational term to be part of the gravitational potential energy,
so that3
Dep
g·v =− (2.40)
Dt
If we bring this energy form to the left member, multiply each element for the density ρ and apply the two
equations (2.38) and (2.39) introduced before we can write the so-called mechanical energy equation,
which evaluates the contributes of various elements to the system mechanical energy:
Dem
ρ = −∇ · (vp) + p∇ · v + ∇ · (τ v · v) − τ v : ∇v (2.41)
Dt

Now we can explain the meaning of every term:


ˆ ∇ · (vp) accounts for the mechanical power exchanged by the system with the ambient thanks to
pressure forces, which is a negative contribution if the work is made by the system.
ˆ p∇ · v accounts for the mechanical energy converted reversibly in internal energy. Pay attention to
the fact that this contribution is negative if the uid is expanding and negative if it's contracting.
ˆ ∇ · (τ v · v) indicates the forces exchanged with the outside ambient thanks to viscous eects.

ˆ τ v : ∇v takes into account the irreversible conversion of mechanical energy in internal energy due
to viscous eects. It is always non-negative, and therefore it is always a loss of mechanical energy
if we consider the negative signum it appears with.

2.3 Energy conservation equation


We can write the general form of energy conservation equation as the following one, with respect to a
general material volume
dE
= L̇← + Q̇← (2.42)
dt
Where the direction of the arrow indicates that the two quantities (mechanical power and thermal power)
are positive if directed inside the material volume.
Now we can express the energy derivative using the RTT:
Z Z
dE d
= ρedΩ + ρev · ndA (2.43)
dt dt ΩCV ∂ΩCV

Now we can express the mechanical power and the thermal power exchanged with the ambient in an
integral form
3 Pay attention to the fact that the negative signum is due to the fact that when gravitational force and velocity scalar
product is positive, the element we are considering is losing gravitational energy.

7
ˆ Mechanical power L̇← can be written as the sum of two contribution, one given by gravitational
forces and the other one given by the eect of stresses on the system. In quantitative terms we can
rewrite it as Z Z
L̇← = ρg · vdΩ + (τ · v) · ndA (2.44)
ΩCV ΩCV

Using Gauss theorem we can rewrite the last term and update the previous equation to this form:
Z Z
L̇← = ρg · vdΩ + ∇ · (τ · v)dΩ (2.45)
ΩCV ΩCV

ˆ Thermal power Q̇← can be written as


Z
Q̇← = − q · vdA (2.46)
∂ΩCV

Where the negative signum is necessary to compensate the fact that the normal vector is outgoing
and the thermal power is positive only if it's directed inside the CV.
We can now express the whole energy conservation equation considering a xed control volume and
applying the Gauss theorem when needed:
Z   Z  
dE ∂(ρe)
= + ∇ · (ρev) dΩ = ρg · v + ∇ · (τ · v) − ∇ · q dΩ (2.47)
dt ΩCV ∂t ΩCV

Asking as usual for the solution that considers only the integrands, we come up with the following form
∂(ρe)
+ ∇ · (ρev) = ρg · v + ∇ · (τ · v) − ∇ · q (2.48)
∂t
From equation (2.29) we can rewrite the rst term in a simple version:
De
ρ = ρg · v + ∇ · (τ · v) − ∇ · q (2.49)
Dt
We can consider that the total energy of a system is given by its kinetic energy (if we consider the
gravitational eects as forces and not as potential energy) and its internal energy. Thus, the previous
equation becomes:
Dek Du
ρ +ρ = ρg · v + ∇ · (τ · v) − ∇ · q (2.50)
Dt Dt
From equation (2.37) we can substitute the kinetic energy:
Du
ρg · v − v · ∇p + v · (∇ · τ v ) + ρ = ρg · v + ∇ · (τ · v) − ∇ · q (2.51)
Dt
Now we can simplify equal terms and write
Du
ρ = τ : ∇(v) − ∇ · q (2.52)
Dt
Expressing the stress tensor
Du
ρ = −p∇ · v + τ v : ∇(v) − ∇ · q (2.53)
Dt
Where we can nd the following terms:
ˆ p∇ · v is the contribution given by the reversible trasformation of mechanical energy in internal
energy
ˆ τ v : ∇(v) is the contribution given by the irreversible trasformation of mechanical energy in internal
energy
ˆ ∇ · q is the contribution of thermal uxes coming from other systems

S-ar putea să vă placă și