Lecture 3 - Conservation Equations

Applied Computational Fluid Dynamics

Instructor: André Bakker

http://www.bakker.org
© André Bakker (2002-2006)
1
Governing equations
• The governing equations include the following conservation laws
of physics:
– Conservation of mass.
– Newton’s second law: the change of momentum equals the sum of
forces on a fluid particle.
– First law of thermodynamics (conservation of energy): rate of change
of energy equals the sum of rate of heat addition to and work done
on fluid particle.
• The fluid is treated as a continuum. For length scales of, say,
1µm and larger, the molecular structure and motions may be
ignored.

2
Lagrangian vs. Eulerian description
A fluid flow field can be thought
of as being comprised of a large
number of finite sized fluid
particles which have mass,
momentum, internal energy, and
other properties. Mathematical
laws can then be written for each
fluid particle. This is the
Lagrangian description of fluid
motion.
Another view of fluid motion is
the Eulerian description. In the
Eulerian description of fluid
motion, we consider how flow
properties change at a fluid
element that is fixed in space
and time (x,y,z,t), rather than
following individual fluid
particles.

Governing equations can be derived using each

method and converted to the other form. 3
Fluid element and properties Fluid element for
conservation laws
• The behavior of the fluid is described in
terms of macroscopic properties:
– Velocity u.
– Pressure p. (x,y,z) δz
– Density ρ.
– Temperature T. δy
– Energy E. δx
z
• Typically ignore (x,y,z,t) in the notation. y
x
• Properties are averages of a sufficiently
large number of molecules.
Faces are labeled
• A fluid element can be thought of as the North, East, West,
South, Top and Bottom
smallest volume for which the continuum
assumption is valid.
Properties at faces are expressed as first
two terms of a Taylor series expansion,
∂p 1 ∂p 1
e.g. for p : pW = p − δ x and pE = p + δx 4
∂x 2 ∂x 2
Mass balance
• Rate of increase of mass in fluid element equals the net rate of
flow of mass into element.
∂ρ δxδyδz
• Rate of increase is: ∂ ( ρδxδyδz ) =
∂t ∂t
• The inflows (positive) and outflows (negative) are shown here:

∂ ( ρ w) 1
ρw + . δz δ xδ y
∂z 2
∂ ( ρ v) 1
ρv + . δ y δ xδ z
∂y 2

∂( ρu) 1
∂ ( ρu ) 1 ρu + . δ x δ yδ z
ρu − . δx δyδz ∂x 2
∂x 2

∂ ( ρv ) 1
ρv − . δy δxδz
z ∂y 2
y
x ∂ ( ρw) 1
ρw − . δz δxδy 5
∂z 2
Continuity equation
• Summing all terms in the previous slide and dividing by the
volume δxδyδz results in:

∂ρ + ∂( ρu) + ∂( ρv) + ∂( ρw) = 0

∂t ∂x ∂y ∂z

• In vector notation: ∂ρ + div ( ρ u) = 0

∂t
Net flow of mass across boundaries
Change in density
Convective term

• For incompressible fluids ∂ρ /∂ t = 0, and the equation becomes:

div u = 0.
∂u ∂ v ∂w ∂ui
• Alternative ways to write this: + + = 0 and =0
∂x ∂y ∂z ∂xi

6
Different forms of the continuity equation

Finite control volume Finite control volume fixed

fixed in space mass moving with flow

∂ D
ρ dV + ρ U ⋅ dS = 0 ρ dV = 0
∂t V S Dt V
Integral form Integral form
Conservation form Non − conservation form
U

Infinitesimally small Infinitesimally small fluid element of fixed

element fixed in space mass (“fluid particle”) moving with the flow

∂ρ Dρ
+ ∇ ⋅ ( ρU ) = 0 + ρ∇ ⋅ U = 0
∂t Dt
Differential form Differential form
Conservation form Non − conservation form 7
Rate of change for a fluid particle
• Terminology: fluid element is a volume stationary in space, and a
fluid particle is a volume of fluid moving with the flow.
• A moving fluid particle experiences two rates of changes:
– Change due to changes in the fluid as a function of time.
– Change due to the fact that it moves to a different location in the fluid
with different conditions.
• The sum of these two rates of changes for a property per unit
mass φ is called the total or substantive derivative Dφ /Dt:
Dφ ∂φ ∂φ dx ∂φ dy ∂φ dz
= + + +
Dt ∂t ∂x dt ∂y dt ∂z dt

• With dx/dt=u, dy/dt=v, dz/dt=w, this results in:

Dφ ∂φ
= + u.grad φ
Dt ∂t 8
Rate of change for a stationary fluid element
• In most cases we are interested in the changes of a flow property
for a fluid element, or fluid volume, that is stationary in space.
• However, some equations are easier derived for fluid particles.
For a moving fluid particle, the total derivative per unit volume of
this property φ is given by:
Dφ ∂φ
(for moving fluid particle) ρ =ρ + u.grad φ (for given location in space)
Dt ∂t

• For a fluid element, for an arbitrary conserved property φ :

∂ρ + div ( ρ u) = 0 ∂ ( ρφ )
+ div ( ρφ u) = 0
∂t ∂t
Continuity equation Arbitrary property

9
Fluid particle and fluid element
• We can derive the relationship between the equations for a fluid
particle (Lagrangian) and a fluid element (Eulerian) as follows:

∂ ( ρφ ) ∂φ ∂ρ Dφ
+ div ( ρφ u) = ρ + u.gradφ + φ + div( ρu) = ρ
∂t ∂t ∂t Dt

zero because of continuity

∂ ( ρφ ) Dφ
+ div ( ρφ u) = ρ
∂t Dt

Rate of increase of Net rate of flow of Rate of increase of

=
φ of fluid element φ out of fluid element φ for a fluid particle

10
To remember so far
• We need to derive conservation equations that we can solve to
calculate fluid velocities and other properties.
• These equations can be derived either for a fluid particle that is
moving with the flow (Lagrangian) or for a fluid element that is
stationary in space (Eulerian).
• For CFD purposes we need them in Eulerian form, but (according
to the book) they are somewhat easier to derive in Lagrangian
form.
• Luckily, when we derive equations for a property φ in one form,
we can convert them to the other form using the relationship
shown on the bottom in the previous slide.

11
Relevant entries for

x-momentum u Du ∂ ( ρu )
ρ + div( ρuu)
Dt ∂t

y-momentum v Dv ∂ ( ρv )
ρ + div( ρvu)
Dt ∂t

z-momentum w Dw ∂ ( ρw)
ρ + div( ρwu)
Dt ∂t

Energy E DE ∂ ( ρE )
ρ + div( ρEu)
Dt ∂t

12
Momentum equation in three dimensions
• We will first derive conservation equations for momentum and
energy for fluid particles. Next we will use the above relationships
to transform those to an Eulerian frame (for fluid elements).
• We start with deriving the momentum equations.
• Newton’s second law: rate of change of momentum equals sum
of forces.
• Rate of increase of x-, y-, and z-momentum:
Du Dv Dw
ρ ρ ρ
Dt Dt Dt
• Forces on fluid particles are:
– Surface forces such as pressure and viscous forces.
– Body forces, which act on a volume, such as gravity, centrifugal,
Coriolis, and electromagnetic forces.

13
Viscous stresses
• Stresses are forces per area.
Unit is N/m2 or Pa.
• Viscous stresses denoted by τ.
• Suffix notation τij is used to
indicate direction.
• Nine stress components.
– τxx, τyy, τzz are normal stresses.
E.g. τzz is the stress in the z-
direction on a z-plane.
– Other stresses are shear
stresses. E.g. τzy is the stress in
the y-direction on a z-plane.
• Forces aligned with the direction
of a coordinate axis are positive.
Opposite direction is negative.

14
Forces in the x-direction
∂τ zx 1
(τ zx + . δz )δyδz ∂τ yx 1
∂τ yx 1 ∂z 2 − (τ yx − . δy )δxδz
(τ yx + . δy )δxδz ∂y 2
∂y 2

∂p 1 ∂p 1
( p − . δx )δyδz − ( p + . δx)δyδz
∂x 2 ∂x 2
∂τ xx 1 ∂τ xx 1
− (τ xx − . δx)δyδz (τ xx + . δx)δyδz
∂x 2 ∂x 2

z
∂τ zx 1
y
− (τ zx − . δz )δxδy
x ∂z 2

Net force in the x-direction is the sum of all the force components in that direction.
15
Momentum equation
• Set the rate of change of x-momentum for a fluid particle Du/Dt
equal to:
– the sum of the forces due to surface stresses shown in the previous
slide, plus
– the body forces. These are usually lumped together into a source
term SM:
Du ∂ (− p + τ xx ) ∂τ yx ∂τ zx
ρ = + + + S Mx
Dt ∂x ∂y ∂z
– p is a compressive stress and τxx is a tensile stress.
• Similarly for y- and z-momentum:

Dv ∂τ xy ∂ (− p + τ yy ) ∂τ zy
ρ = + + + S My
Dt ∂x ∂y ∂z
Dw ∂τ xz ∂τ yz ∂ (− p + τ zz )
ρ = + + + S Mz
Dt ∂x ∂y ∂z 16
Energy equation
• First law of thermodynamics: rate of change of energy of a fluid
particle is equal to the rate of heat addition plus the rate of work
done.
• Rate of increase of energy is ρDE/Dt.
• Energy E = i + ½ (u 2+v2+w2).
• Here, i is the internal (thermal energy).
• ½ (u 2+v2+w2) is the kinetic energy.
• Potential energy (gravitation) is usually treated separately and
included as a source term.
• We will derive the energy equation by setting the total derivative
equal to the change in energy as a result of work done by viscous
stresses and the net heat conduction.
• Next we will subtract the kinetic energy equation to arrive at a
conservation equation for the internal energy.

17
 Work done by surface stresses in x-direction
∂ (uτ zx ) 1
(uτ zx + . δz )δyδz
∂ (uτ yx ) 1 ∂z 2 ∂ (uτ yx ) 1
(uτ yx + . δy )δxδz − (uτ yx − . δy )δxδz
∂y 2 ∂y 2

∂ (up ) 1
(up − . δx)δyδz ∂ (up ) 1
∂x 2 − (up + . δx)δyδz
∂x 2
∂ (uτ xx ) 1
− (uτ xx − . δx)δyδz ∂ (uτ xx ) 1
∂x 2 (uτ xx + . δx)δyδz
∂x 2

z
y
∂ (uτ zx ) 1
x − (uτ zx − . δz )δxδy
∂z 2
Work done is force times velocity.
18
Work done by surface stresses
• The total rate of work done by surface stresses is calculated as
follows:
– For work done by x-components of stresses add all terms in the
previous slide.
– Do the same for the y- and z-components.
• Add all and divide by δxδyδz to get the work done per unit volume
by the surface stresses:

∂ (uτ xx ) ∂ (uτ yx ) ∂ (uτ zx ) ∂ (vτ xy )

− div ( pu) + + + +
∂x ∂y ∂z ∂x
∂ (vτ yy ) ∂ (vτ zy ) ∂ ( wτ xz ) ∂ ( wτ yz ) ∂ (uτ zz )
+ + + + +
∂y ∂z ∂x ∂y ∂z

19
Energy flux due to heat conduction
∂q z 1
∂q y 1 (q z + . δz )δxδy
(q y + . δy )δxδz ∂z 2
∂y 2

∂q x 1 ∂q x 1
(q x − . δx )δyδz (q x + . δx )δyδz
∂x 2 ∂x 2

z
y
x

∂q z 1 ∂q y 1
(q z − . δz )δxδy (q y − . δy )δxδz
∂z 2 ∂y 2

The heat flux vector q has three components, qx, qy, and qz. 20
Energy flux due to heat conduction
• Summing all terms and dividing by δxδyδz gives the net rate of heat
transfer to the fluid particle per unit volume:

∂q x ∂q y ∂q z
− − − = − div q
∂x ∂y ∂z

• Fourier’s law of heat conduction relates the heat flux to the local
temperature gradient:
∂T ∂T ∂T
qx = − k qy = − k qz = − k
∂x ∂y ∂z

• In vector form: q = − k grad T

• Thus, energy flux due to conduction: − div q = div (k grad T )
• This is the final form used in the energy equation.
21
Energy equation
• Setting the total derivative for the energy in a fluid particle equal
to the previously derived work and energy flux terms, results in
the following energy equation:

DE ∂ (uτ xx ) ∂ (uτ yx ) ∂ (uτ zx ) ∂ (vτ xy )

ρ = − div( pu) + + + +
Dt ∂x ∂y ∂z ∂x
∂ (vτ yy ) ∂ (vτ zy ) ∂ ( wτ xz ) ∂ ( wτ yz ) ∂ (uτ zz )
+ + + + +
∂y ∂z ∂x ∂y ∂z
+ div (k grad T ) + S E
• Note that we also added a source term SE that includes sources
(potential energy, sources due to heat production from chemical
reactions, etc.).

22
Kinetic energy equation
• Separately, we can derive a conservation equation for the kinetic
energy of the fluid.
• In order to do this, we multiply the u-momentum equation by u,
the v-momentum equation by v, and the w-momentum equation
by w. We then add the results together.
• This results in the following equation for the kinetic energy:

D[ 12 (u 2 + v 2 + w 2 )] ∂τ xx ∂τ yx ∂τ zx
ρ = − u.grad p + u + +
Dt ∂x ∂y ∂z
∂τ xy ∂τ yy ∂τ zy ∂τ xz ∂τ yz ∂τ zz
+v + + +w + + + u. S M
∂x ∂y ∂z ∂x ∂y ∂z

23
Internal energy equation
• Subtract the kinetic energy equation from the energy equation.
• Define a new source term for the internal energy as
Si = SE - u.SM. This results in:

Di ∂u ∂u ∂u ∂v
ρ = − p div u + τ xx + τ yx + τ zx + τ xy
Dt ∂x ∂y ∂z ∂x
∂v ∂v ∂w ∂w ∂u
+ τ yy + τ zy + τ xz + τ yz + τ zz
∂y ∂z ∂x ∂y ∂z
+ div(k grad T ) + S i

24
Enthalpy equation
• An often used alternative form of the energy equation is the total
enthalpy equation.
– Specific enthalpy h = i + p/ρ.
– Total enthalpy h0 = h + ½ (u2+v2+w2) = E + p/ρ.

∂ ( ρ h0 )
+ div( ρ h0u) = div(k grad T )
∂t
∂ (uτ xx ) ∂ (uτ yx ) ∂ (uτ zx ) ∂ (vτ xy )
+ + + +
∂x ∂y ∂z ∂x
∂ (vτ yy ) ∂ (vτ zy ) ∂ ( wτ xz ) ∂ ( wτ yz ) ∂ (uτ zz )
+ + + + +
∂y ∂z ∂x ∂y ∂z
+ Sh

25
Equations of state
• Fluid motion is described by five partial differential equations for
mass, momentum, and energy.
• Amongst the unknowns are four thermodynamic variables: ρ, p, i,
and T.
• We will assume thermodynamic equilibrium, i.e. that the time it
takes for a fluid particle to adjust to new conditions is short
relative to the timescale of the flow.
• We add two equations of state using the two state variables ρ and
T: p=p(ρ,T) and i=i(ρ,T).
• For a perfect gas, these become: p=ρ RT and i=CvT.
• At low speeds (e.g. Ma < 0.2), the fluids can be considered
incompressible. There is no linkage between the energy equation,
and the mass and momentum equation. We then only need to
solve for energy if the problem involves heat transfer.

26
Viscous stresses
• A model for the viscous stresses τij is required.
• We will express the viscous stresses as functions of the local
deformation rate (strain rate) tensor.
• There are two types of deformation:
– Linear deformation rates due to velocity gradients.
• Elongating stress components (stretching).
• Shearing stress components.
– Volumetric deformation rates due to expansion or compression.
• All gases and most fluids are isotropic: viscosity is a scalar.
• Some fluids have anisotropic viscous stress properties, such as
certain polymers and dough. We will not discuss those here.

27
Viscous stress tensor
• Using an isotropic (first) dynamic viscosity µ for the linear
deformations and a second viscosity λ=-2/3 µ for the volumetric
deformations results in:

τ xx τ xy τ xz
= τ yx τ yy τ yz
τ zx τ zy τ zz
∂u 2 ∂u ∂v ∂u ∂w
2 µ − µ div u µ + µ +
∂x 3 ∂y ∂x ∂z ∂x
∂u ∂v ∂v 2 ∂v ∂w
= µ + 2 µ − µ div u µ +
∂y ∂x ∂y 3 ∂z ∂y
∂u ∂w ∂v ∂w ∂w 2
µ + µ + 2µ − µ div u
∂z ∂x ∂z ∂y ∂z 3

Note: div u = 0 for incompressible fluids. 28

Navier-Stokes equations
• Including the viscous stress terms in the momentum balance and
rearranging, results in the Navier-Stokes equations:

∂ ( ρu ) ∂p
x − momentum : + div( ρuu) = − + div( µ grad u ) + S Mx
∂t ∂x
∂ ( ρv) ∂p
y − momentum : + div( ρvu) = − + div( µ grad v) + S My
∂t ∂y
∂ ( ρw) ∂p
z − momentum : + div( ρwu) = − + div( µ grad w) + S Mz
∂t ∂z

29
Viscous dissipation
• Similarly, substituting the stresses in the internal energy equation
and rearranging results in:

∂ ( ρi )
Internal energy : + div ( ρiu) = − p div u + div (k grad T ) + Φ + S i
∂t

• Here Φ is the viscous dissipation term. This term is always

positive and describes the conversion of mechanical energy to
heat.
2 2 2 2
∂u ∂v ∂w ∂u ∂v
Φ=µ 2 + + + +
∂x ∂y ∂z ∂y ∂x
2 2
∂u ∂w ∂v ∂w 2
+ + + + − µ (div u) 2
∂z ∂x ∂z ∂y 3
30
Summary of equations in conservation form
∂ρ
Mass : + div ( ρ u) = 0
∂t
∂ ( ρu ) ∂p
x − momentum : + div( ρuu) = − + div( µ grad u ) + S Mx
∂t ∂x
∂ ( ρv) ∂p
y − momentum : + div( ρvu) = − + div( µ grad v) + S My
∂t ∂y
∂ ( ρw) ∂p
z − momentum : + div( ρwu) = − + div( µ grad w) + S Mz
∂t ∂z
∂ ( ρi )
Internal energy : + div ( ρiu) = − p div u + div (k grad T ) + Φ + S i
∂t
Equations of state : p = p ( ρ , T ) and i = i ( ρ , T )
e.g. for perfect gas : p = ρ RT and i = CvT
31
General transport equations
• The system of equations is now closed, with seven equations for seven
variables: pressure, three velocity components, enthalpy, temperature,
and density.
• There are significant commonalities between the various equations.
Using a general variable φ, the conservative form of all fluid flow
equations can usefully be written in the following form:

∂ ( ρφ )
+ div( ρφ u ) = div(Γ grad φ ) + Sφ
∂t

• Or, in words:
Net rate of flow
Rate of increase Rate of increase Rate of increase
of φ out of
of φ of fluid + = of φ due to + of φ due to
fluid element
element diffusion sources
(convection)

32
Integral form
• The key step of the finite volume method is to integrate the differential
equation shown in the previous slide, and then to apply Gauss’
divergence theorem, which for a vector a states:
div a dV = n ⋅ a dA
CV A

• This then leads to the following general conservation equation in integral

form:
∂
ρφ dV + n ⋅ ( ρφ u) dA = n ⋅ (Γ grad φ ) dA + Sφ dV
∂t CV A A CV

Net rate of Net rate of

Rate of Net rate of
decrease of φ due increase of φ due
increase + to convection = + creation
to diffusion
of φ of φ
across boundaries across boundaries

• This is the actual form of the conservation equations solved by finite

volume based CFD programs to calculate the flow pattern and
associated scalar fields. 33