1.

INTRODUCTION TO CFD

SPRING 2016

1.1 What is computational fluid dynamics?

1.2 Basic principles of CFD
1.3 Stages in a CFD simulation
1.4 Fluid-flow equations
1.5 The main discretisation methods
Appendices
Examples

1.1 What is Computational Fluid Dynamics?

Computational fluid dynamics (CFD) is the use of computers and numerical methods to solve
problems involving fluid flow.
CFD has been successfully applied in many areas of fluid mechanics. These include
aerodynamics of cars and aircraft, hydrodynamics of ships, flow through pumps and turbines,
combustion and heat transfer, chemical engineering. Applications in civil engineering include
wind loading, vibration of structures, wind and wave energy, ventilation, fire, explosion
hazards, dispersion of pollution, wave loading on coastal and offshore structures, hydraulic
structures such as weirs and spillways, sediment transport. More specialist CFD applications
include ocean currents, weather forecasting, plasma physics, blood flow and heat transfer
around electronic circuitry.
This range of applications is very broad and involves many different fluid phenomena. In
particular, the CFD techniques used for high-speed aerodynamics (where compressibility is
significant but viscous and turbulent effects are often unimportant) are very different from
those used to solve the incompressible, turbulent flows typical of mechanical and civil
engineering.
Although many elements of this course are widely applicable, the focus will be on simulating
viscous, incompressible flow by the finite-volume method.

CFD

11

David Apsley

1.2 Basic Principles of CFD

The approximation of a continuously-varying quantity in terms of values at a finite number of
points is called discretisation.
The following are common to any CFD simulation.

(1)

The flow field is discretised; i.e. field

variables (, u, v, w, p, ) are
approximated by their values at a finite
number of nodes.

(2)

The equations of motion are discretised:

continuous
curve

discrete
approximation

derivatives algebraic approximations

(continuous)
(discrete)

df f
f f

2 1
dx x x2 x1

x
x

(3)

The resulting system of algebraic equations is solved to give values at the nodes.

1.3 Stages in a CFD Simulation

The main stages in a CFD simulation are:
Pre-processing:
formulation of the problem (governing equations and boundary conditions);
construction of a computational mesh (set of control volumes).
Solving:
discretisation of the governing equations;
solution of the resulting algebraic equations.
Post-processing:
analysis of results (calculation of derived quantities: forces, flow rates, ... );
visualisation (graphs and plots of the solution).

CFD

12

David Apsley

1.4 Fluid-Flow Equations

The equations of fluid flow are based on fundamental physical conservation principles:

mass:
change of mass = 0

momentum: change of momentum = force time

energy:
change of energy = work + heat
In fluid flow these are usually expressed as rate equations; i.e. rate of change =
Additional equations may apply for non-homogeneous fluids (e.g. multiple phases, or
containing dissolved chemicals or suspended particles).
When applied to a fluid continuum these conservation principles may be expressed
mathematically as either:

integral (i.e. control-volume) equations;

differential equations.

1.4.1 Integral (Control-Volume) Approach

This considers how the total amount of some physical
quantity (mass, momentum, energy, ) is changed within a
finite region of space (control volume).

For an arbitrary control volume the balance of any physical

quantity over an interval of time is
Change = Amount In Amount Out + Amount Created
In fluid mechanics this is usually expressed in rate form by dividing by the time interval (and
transferring the net amount passing through the boundary to the LHS of the equation):
NET FLUX
RATE OF CHANGE

SOURCE
(1)

throughboundaryof V inside V
inside
V

The flux, or rate of transport through a surface, is further subdivided into:

advection movement with the fluid flow;
diffusion net transport by random molecular or turbulent motion.

RATE OF CHANGE ADVECTION DIFFUSION SOURCE

throughboundaryof V
inside V
inside V

(2)

The important point is that this is a single, generic equation, irrespective of whether the
physical quantity concerned is mass, momentum, chemical content, etc. Thus, instead of
dealing with lots of different equations we can consider (Section 4) the numerical solution of
a generic scalar-transport equation.
The finite-volume method, which is the subject of this course, is based on approximating
these control-volume equations.

CFD

13

David Apsley

1.4.2 Differential Equations

In regions without shocks, interfaces or other discontinuities, the fluid-flow equations can
also be written in equivalent differential forms. These describe what is going on at a point
rather than over a whole control volume. Mathematically, they can be derived by making the
control volumes infinitesimally small. This will be demonstrated in Section 2, where it will
also be shown that there are several different ways of writing these differential equations.
Finite-difference methods are based on the direct approximation of a differential form of the
governing equations.

1.5 The Main Discretisation Methods

i,j+1

(i) Finite-Difference Method

Discretise the governing differential equations; e.g.
ui 1, j ui 1, j vi , j 1 vi , j 1
u v
0

x y
2x
2y

i-1,j

i,j

i+1,j

i,j-1

(ii) Finite-Volume Method

vn

Discretise the governing integral or control-volume equations; e.g.

net mass outflow (uA) e (uA) w (vA) n (vA) s 0

uw

ue
vs

(iii) Finite-Element Method

Express the solution as a weighted sum of shape functions S(x); e.g. for velocity:
u (x) u S (x)
Substitute into some form of the governing equations and solve for the coefficients (aka
degrees of freedom or weights) u.
This course will focus on the finite-volume method.
The finite-element method is popular in solid mechanics (geotechnics, structures) because:

it has considerable geometric flexibility;

general-purpose software can be used for a wide variety of physical problems.

The finite-volume method is popular in fluid mechanics because:

it rigorously enforces conservation;

it is flexible in terms of both geometry and the variety of fluid phenomena;

it is directly relatable to physical quantities (mass flux, etc.).

CFD

14

David Apsley

In the finite-volume method ...

(1) A flow geometry is defined.

(2) The flow domain is decomposed into a set of control volumes

or cells called a computational mesh or grid.

(3) The control-volume equations are discretised i.e.

approximated in terms of values at nodes to form a set of
algebraic equations.

(4) The discretised equations are solved numerically.

CFD

15

David Apsley

APPENDICES
A1. Notation
Position/time:
x (x, y, z) or (x1, x2, x3) position; (z is usually vertical when gravity is important)
t
time
Field variables:
u (u, v, w) or (u1, u2, u3) velocity
p
pressure
(p patm is the gauge pressure; p* = p + gz is the piezometric pressure.)
T
temperature

concentration (amount per unit mass or per unit volume)

Fluid properties:

density

dynamic viscosity
( / is the kinematic viscosity)

diffusivity

A2. Statics
At rest, pressure forces balance weight. This can be written mathematically as
dp
or
(3)
g
p gz
dz
The same equation also holds in a moving fluid if there is no vertical acceleration, or, as an
approximation, if vertical acceleration is much smaller than g.
If density is constant, (3) can be written
(4)
p* p gz constant
p* is called the piezometric pressure, combining the effects of pressure and weight. For a
constant-density flow without a free surface, gravitational forces can be eliminated entirely
from the equations by working with the piezometric pressure.
In compressible flow, pressure, density and temperature are connected by an equation of
state; the most common is the ideal gas law:
p RT ,
R R0 /m
(5)
where R0 is the universal gas constant, m is the molar mass and T is the absolute temperature.
For ideal gases, temperature is related to internal energy e or enthalpy h (per unit mass) by
e cv T
h c pT
where cv and cp are the specific heat capacities at constant volume and constant pressure
respectively.

CFD

16

David Apsley

Examples
The following simple examples develop the control-volume notation to be used in the rest of
the course.

D=10 cm

u=8 m/s

10 cm

Q1.
Water (density 1000 kg m3) flows at
2 m s1 through a circular pipe of
diameter 10 cm. What is the mass flux C
across the surfaces S1 and S2?

2 m/s

45

S1

S2

Q2.
A water jet strikes normal to a fixed plate as shown.
Compute the force F required to hold the plate fixed.

Q3.
An explosion releases 2 kg of a toxic gas into a room of dimensions 30 m 8 m 5 m.
Assuming the room air to be well-mixed and to be vented at a speed of 0.5 m s1 through an
aperture of 6 m2, calculate: (a) the initial concentration of gas in ppm by mass; (b) the time
taken to reach a safe concentration of 1 ppm.
(Take the density of air as 1.2 kg m3.)

Q4.
A burst pipe at a factory causes a chemical to seep into a river at a rate of 2.5 kg hr1. The
river is 5 m wide, 2 m deep and flows at 0.3 m s1. What is the average concentration of the
chemical (in kg m3) downstream of the spill?

CFD

17

David Apsley

2. FLUID-FLOW EQUATIONS

SPRING 2016

2.1 Introduction
2.2 Conservative differential equations
2.3 Non-conservative differential equations
2.4 Non-dimensionalisation
Summary
Examples

2.1 Introduction
Fluid dynamics is governed by conservation equations for:

mass;

momentum;

energy;

(for a non-homogenous fluid) other constituents.

Equations for these can be expressed mathematically in many ways, notably as:

integral (control-volume) equations;

differential equations.
This course will focus on the control-volume approach because it is easier to relate to the real
world, is naturally conservative and forms the basis of the finite-volume method. However,
the equivalent differential equations are often easier to write down, manipulate and, in a few
cases, solve analytically.
Although there are different fluid-flow variables, most satisfy a single
generic equation: the scalar-transport or advection-diffusion equation.
(You have met a 1-d version in your Water Engineering course.)
NET FLUX
RATE OF CHANGE
SOURCE

inside V
inside V

through boundary of V

(1)

The flux through the bounding surface can be divided into:

advection1: transport with the flow;

diffusion: net transport by molecular or turbulent fluctuations.

RATE OF CHANGE ADVECTION DIFFUSION SOURCE

throughboundaryof V
inside V
inside V

(2)

The finite-volume method is a natural discretisation of this.

Some authors but not this one prefer the term convection to advection. This author prefers convection to be
reserved for the transport of heat.

CFD

21

David Apsley

2.2 Conservative Differential Equations

2.2.1 Mass (Continuity)
Mass conservation: mass is neither created nor destroyed.
Rate of change of mass in cell = net inward mass flux
or, with the more conventional flux direction:
rate of change of mass in cell + net outward mass flux = 0
d
(mass)
dt

(mass flux )

(3)

faces

u
For an arbitrary control volume or cell:
mass of fluid in the cell:
V
mass flux through one face: C u n A u A

un
A

A conservative differential equation for mass conservation can be

derived by considering a small Cartesian control volume as shown.

z
n

If density and velocity are averages over cell volume or cell face as
x
appropriate, then
d(V )
(uA) e (uA) w (vA) n (vA) s (wA) t (wA) b 0

dt

net outward mass flux

rate of change of mass

Writing V xyz and Aw = Ae = yz etc,

d(xyz )
[(u ) e (u ) w ]yz [(v) n (v) s ]zx [(w) t (w) b ]xy 0
dt
Dividing by the volume, xyz:
d (u ) e (u ) w (v) n (v) s (w) t (w) b

0
dt
x
y
z
Taking the limit as x, y, z 0:

(u ) (v) (w)

0
t
x
y
z

(4)

This analysis is analogous to the finite-volume procedure, except that in the latter the control
volume does not shrink to zero; i.e. it is a finite-volume, not infinitesimal-volume, approach.

CFD

22

David Apsley

(*** Advanced ***)

For an arbitrary volume V with closed surface V:
d
u dA 0
dV

dt V

V
mass in cell

(5)

net mass flux

For a fixed control volume, take d/dt under the integral sign and apply the divergence
theorem to turn the surface integral into a volume integral:

t (u) dV 0
V

Since V is arbitrary, the integrand must be identically zero. Hence,

(u) 0
t

(6)

Incompressible Flow
For incompressible flow, volume as well as mass is conserved, so that:
(uA) e (uA) w (vA) n (vA) s ( wA) t ( wA) b 0

net outwardVOLUME flux

Substituting for face areas, dividing by volume and proceeding to the limit as above produces
u v w
(7)

0
x y z
This is usually taken as the continuity equation in incompressible flow.

CFD

23

David Apsley

2.2.2 Momentum
Newtons Second Law: rate of change of momentum = force
rate of change of momentum in cell + net outward momentum flux = force
d
(mass u)
dt

(mass flux u) F

(8)

faces

For a cell with volume V and a typical face with area A:

momentum of fluid in the cell
= mass u
(V )u
momentum flux through one face = mass flux u (u A)u

u
V

un
A

Momentum and force are vectors, giving (in principle) 3 equations.

Fluid Forces
There are two main types:

surface forces (proportional to area; act on control-volume faces)

body forces (proportional to volume)

(i) Surface forces are usually expressed in terms of stress:

force
or
stress
force stress area
area
The main surface forces are:

pressure p: always acts normal to a surface;

viscous stresses : frictional forces arising from relative motion.

For a simple shear flow there is only one non-zero stress component:
u
12
y
but, in general, ij is a symmetric tensor with a more complex
expression for its components. In incompressible flow2,
u v
12 ( )
y x
or, more generally,
u u j
11
ij ( i
)
x j xi
21

U
22
12
21

y
x

12

22

There is a slightly extended expression in compressible flow; see the recommended textbooks.

CFD

24

David Apsley

11

(ii) Body forces are often expressed as forces per unit volume, or force densities.
z

The main body forces are:

gravity: the force per unit volume is

g ge z

(For constant-density fluids, pressure and weight can be combined as a piezometric

pressure p* = p + gz; gravity then no longer appears explicitly in the flow equations.)

centrifugal and Coriolis forces (apparent forces in a rotating reference frame):

axis

R
2

centrifugal force:
( r ) 2R

Coriolis force:
2 u

In inertial frame

In rotating frame

(Because the centrifugal force can be written as the gradient of some quantity in this
case 12 2 R 2 it can also be absorbed into a modified pressure and hence removed
from the momentum equation; see the Examples).
t

Differential Equation For Momentum

Consider a fixed Cartesian control volume with sides x, y, z.

x
For the x-component of momentum:
d
(Vu )
(uA) e u e (uA) w u w (vA) n u n (vA) s u s (wA) t u t (wA) b u b

d
t

net outward momentum flux

rate of change of momentum

( p w Aw pe Ae ) viscous and other fo rces

pressure force in x direction

Substituting cell dimensions:

d
(xyz u ) [(u ) e u e (u ) w u w ]yz [(v) n u n (v) s u s ]zx [(w) t u t (w) b u b ]xy
dt
( p w pe )yz viscous and other fo rces

CFD

25

David Apsley

Dividing by volume xyz (and changing the order of pe and pw):

( p pw )
d(u ) (uu ) e (uu ) w (vu) n (vu) s (wu ) t (wu ) b

e
dt
x
y
z
x
viscous and other fo rces
In the limit as x, y, z 0:

(u ) (uu ) (vu) (wu )

p

2 u other forces
t
x
y
z
x

(9)

Notes.
(1)
The viscous term is given without proof (but you can read the notes below).
2
2
2
2
is the Laplacian operator 2 2 2 .
x
y
z
(2)
The pressure force per unit volume in the x direction is given by (minus) the pressure
gradient in that direction.
(3)
The y and z-momentum equations can be obtained by inspection / pattern-matching.

(*** Advanced ***)

With surface forces determined by stress tensor ij and body forces determined by force
density fi, the control-volume equation for the i component of momentum may be written
d

(10)
u i dV u i u j dA j ij dA j f i dV
dt V

V
V
V

momentumin cell

net momentum flux

surface forces

body forces

For fixed V, take d/dt inside integrals and apply the divergence theorem to surface integrals:

(u i ) (u i u j ) ij

f i dV 0

t
x j
x j

As V is arbitrary, the integrand vanishes identically. Hence, for arbitrary forces:

(u i ) (u i u j ) ij

fi
(11)
t
x j
x j
The stress tensor has pressure and viscous parts:
ij p ij ij ,

(12)

(u i ) (u i u j )
p ij

fi
(13)
t
x j
xi x j
For a Newtonian fluid, the viscous stress tensor (including compressible part) is given by
u u j 2 u k
ij ( i
ij
)
x j xi 3 xk
If the fluid is incompressible and viscosity is uniform then the viscous term simplifies to give
(u i ) (u i u j )
p

2 ui f i
t
x j
xi

CFD

26

David Apsley

2.2.3 General Scalar

A similar equation may be derived for any physical quantity that is advected and diffused in a
fluid flow. Examples include salt, sediment and chemical pollutants. For each such quantity
an equation is solved for the concentration (i.e. amount per unit mass of fluid) .
Diffusion occurs when concentration varies with position and leads to net transport from
regions of high concentration to regions of low concentration. For many scalars this rate of
transport is proportional to area and concentration gradient and may be quantified by Ficks
diffusion law:
rate of di ffusion diffusivit y gradient area

A
n
This is often referred to as gradient diffusion. A common example is heat conduction.

For an arbitrary control volume:

amount in cell: V
advective flux:
diffusive flux:
source:

(u A

A
n
S = sV

(mass concentration)
(mass flux concentration)
(diffusivity gradient area)

u
V

un
A

(source density volume)

Balancing the rate of change, the net flux through the boundary and rate of production yields
the scalar-transport or (advection-diffusion) equation:
rate of ch ange net outward flux source
d
(mass )
dt

(mass flux
faces

A) S
n

(Conservative) differential equation:

()

(u ) (v ) (w ) s
t
x
x y
y z
z

(14)

(15)

(*** Advanced ***)

The integral equation may be expressed more mathematically as:
d

dV (u dA s dV
dt V
V
V

(16)

For a fixed control volume, taking the time derivative under the integral sign and using
Gausss divergence theorem as before gives a corresponding differential equation:
()
(u ) s
(17)
t

CFD

27

David Apsley

2.2.4 Momentum Components as Transported Scalars

In the momentum equation, if viscous force A (u / n) A is transferred to the LHS it
looks like a diffusive flux; e.g. for the x-component:
d
u
(mass u ) (mass flux u
A) other forces
dt
n
faces
Compare this with the generic scalar-transport equation:
d

(mass ) (mass flux A) S

dt
n
faces
Each component of momentum satisfies its own scalar-transport equation, with
concentration,
velocity component (u, v or w)
diffusivity,
viscosity
source, S
other forces
Consequently, only one generic scalar-transport equation need be considered.
In Section 5 we shall see, however, that the momentum components differ from passive
scalars (those not affecting the flow), because:

equations are nonlinear (mass flux involves the velocity component being solved for);

equations are coupled (mass flux involves the other velocity components as well);

the velocity field must also be mass-consistent.

2.2.5 Non-Gradient Diffusion

The analysis above assumes that all non-advective flux is simple gradient diffusion:

A
n
Actually, the real situation is a little more complex. For example, in the u-momentum
equation the full expression for the 1-component of viscous stress through the 2-face is
u v
12
y x
The u/y part is gradient diffusion of u, but the v/x term is not. In general, non-advective
fluxes F that cant be represented by gradient diffusion are discretised conservatively (i.e.
worked out for particular cell faces), but are transferred to the RHS as a source term:
d

(mass ) [mass flux

A F ] S
dt
n
faces

2.2.6 Moving Control Volumes

Control-volume equations are also applicable to moving control volumes, provided the
normal velocity component in the mass flux is that relative to the mesh; i.e.
u n (u umesh ) n
The finite-volume method can thus be used for calculating flows with moving boundaries3.
3

See, for example: Apsley, D.D. and Hu, W., 2003, CFD Simulation of two- and three-dimensional free-surface
flow, International Journal for Numerical Methods in fluids, 42, 465-491.

CFD

28

David Apsley

2.3 Non-Conservative Differential Equations

The conservative differential equations are so-called because they can be integrated directly
to give an equivalent integral form involving the net change in a flux, with the flux leaving
one cell equal to that entering an adjacent cell. To do so, all terms involving derivatives of
dependent variables must have differential operators on the outside. e.g. in one dimension:
x2

f ( x 2 ) f ( x1 ) g ( x) dx
x1

df
g ( x)
dx
(differenti al form)

(integral form)
i.e.

flux out flux in source

(*** Advanced ***)

The three-dimensional version uses partial derivatives and the divergence theorem to change
the differentials to surface flux integrals.
As an example of how the same equation can appear in conservative and non-conservative
forms, consider a simple 1-d example:
d 2
(conservative form can be integrated directly)
( y ) g ( x)
dx
dy
(non-conservative form, obtained by applying the chain rule)
2y
g ( x)
dx

Material Derivatives
The time rate of change of some property in a fluid element moving with the flow is called
the material (or substantive) derivative. It is denoted by D/Dt and defined below.
Every field variable is a function of both time and position; i.e.
(t , x, y, z)
As one follows a path through space, changes with time because:

it changes with time t at each point; and

it changes with position (x, y, z) as it moves with the flow.

(x(t), y(t), z(t))

Thus, the total time derivative following an arbitrary path (x(t), y(t), z(t)) is
d dx dy dz

dt t x dt y dt z dt
The material derivative is the time derivative along the particular path following the flow
(dx/dt = u, etc.):
D

(18)

u v w
Dt t
x
y
z
In particular, the material derivative of velocity (Du/Dt) is the acceleration.
One can now write a non-conservative but more compact form of the governing equations.
CFD

29

David Apsley

For a general scalar the sum of time-dependent and advective terms (i.e. total rate of
change) is
() (u) (v) (w)

t
x
y
z
(u )
(v)
(w)



u
v
w
t x
x y
y z
z
t
(by the product rule)
(u ) (v) (w)

u
v w

t
x
y
z
t
x
y
z

0 by continuity

D / Dt by definition

(by collecting similar terms)

Dt

(19)

Thus, using the material derivative, the time-dependent and advection terms in a scalartransport equation can be combined in a much more compact (but non-conservative) form. In
particular, the momentum equation becomes
Du
p

2 u other forces
(20)
D
t
x

massacceleration

This form is simpler to write and is used both for convenience and to derive theoretical
results in special cases (see the Examples). However, in the finite-volume method it is the
conservative form which is discretised directly.
(*** Advanced ***)
The derivation of (19) above is greatly simplified by use of suffix notation and the
summation convention:
() (u i
(u i )

u i
t
x
t xi
x
t
(u i )



ui

t
x
t
xi

0 by continuity

D / Dt by definition

D
Dt

If the differential transport equations are derived directly from corresponding integral
equations by shrinking to a point then the conservative forms arise from using fixed control
volumes (Eulerian approach) and the non-conservative forms by using control volumes
moving with the flow (Lagrangian approach).

CFD

2 10

David Apsley

2.4 Non-Dimensionalisation
Although it is possible to work entirely in dimensional quantities, there are good theoretical
reasons for working in non-dimensional variables. These include the following.

All dynamically-similar problems (same Re, Fr etc.) can be solved with a single
computation.
The number of relevant parameters (and hence the number of graphs needed to
convey results) is reduced.
It indicates the relative size of different terms in the governing equations and, in
particular, which might conveniently be neglected.
Computational variables are of a similar order of magnitude (ideally, of order unity),
yielding better numerical accuracy.

2.4.1 Non-Dimensionalising the Governing Equations

For incompressible flow the governing equations are:
u v w
continuity:

0
x y z
Du
p
momentum:
2 u (and similar equations in y, z directions)
Dt
x

(21)
(22)

Adopting reference scales U0, L0 and 0 for velocity, length and density, respectively, and
derived scales L0/U0 for time and 0U 02 for pressure, each fluid quantity can be written as a
product of a dimensional scale and a non-dimensional variable (indicated by an asterisk *):
L
x L0 x * ,
t 0 t*,
u U 0u* ,
0 * ,
p p ref ( 0U 02 ) p * ,
etc.
U0
Substituting into mass and momentum equations (21) and (22) yields, after simplification:

u *
x *

v *
y *

Du *
Dt *

w*
z *

p *
x *

1 *2 *
u
Re

(23)

where

Re

0U 0 L0

(24)

From this, it is seen that the key dimensionless group is the Reynolds number Re. If Re is
large then viscous forces would be expected to be negligible in much of the flow.
Having derived the non-dimensional equations it is usual to drop the asterisks and simply
declare that you are working in non-dimensional variables.
Note.

The objective is that non-dimensional quantities (e.g. p*) should be order of

magnitude unity, so the scale for a quantity should reflect its range of values, not
necessarily its absolute value. In incompressible flow it is differences in pressure that
are important, not absolute values. Since flow-induced pressures are usually much

CFD

2 11

David Apsley

smaller than the absolute pressure one usually works in terms of the departure from a
constant reference pressure pref.

Similarly, in Section 3 when we look at small variations in density due to temperature

or salinity that give rise to buoyancy forces we shall use an alternative nondimensionalisation:
0 *
with * typically varying between 0 and 1.

2.4.2 Common Dimensionless Groups

If other types of fluid force are included then each introduces another non-dimensional group.
For example, gravitational forces lead to a Froude number (Fr) and Coriolis forces to a
Rossby number (Ro). Some of the most important dimensionless groups are given below.
If U and L are representative velocity and length scales, respectively, then:

Re

UL UL

Fr

Ma

U
c

Mach number (compressible flow; c = speed of sound)

Ro

U
L

Rossby number (rotating flows; = angular velocity of frame)

gL

U 2 L
We

Reynolds number (viscous flow; = dynamic viscosity)

Froude number (open-channel flow; g = gravity)

Weber number (interfacial flows; = surface tension)

Note.
For flows with buoyancy forces caused by a change in density, rather than open-channel
flows, we sometimes use a densimetric Froude number instead; this is defined by
U
Fr
( / ) gL
Here, g is replaced in the usual formula for Froude number by ( / ) g , sometimes called
the reduced gravity g: see Section 3.

CFD

2 12

David Apsley

Summary

Fluid dynamics is governed by conservation equations for mass, momentum, energy

(and, for a non-homogeneous fluid, the amount of individual constituents).

The governing equations can be written in equivalent integral (control-volume) or

differential forms.

The finite-volume method is a direct discretisation of the control-volume equations.

Differential forms of the flow equations may be conservative (i.e. can be integrated
directly to something of the form fluxout fluxin = source) or non-conservative.

A particular control-volume equation takes the form:

rate of change + net outward flux = source

There are really just two canonical equations to discretise and solve:
mass conservation (continuity):
d
(mass) (mass flux ) 0
dt
faces
scalar-transport (or advection-diffusion) equation:
d

(mass ) ( mass flux A )

S
dt
n
faces
rate of change
advection
diffusion
source

Each Cartesian velocity component (u, v, w) satisfies its own scalar-transport

equation. However, these equations differ from those for a passive scalar because they
are non-linear, are coupled through the advective fluxes and pressure forces and their
solutions are also required to be mass-consistent.

Non-dimensionalising the governing equations, allows dynamically-similar flows

(those with the same values of Reynolds number, etc.) to be solved with a single
calculation, reduces the overall number of parameters, indicates whether certain terms
in the governing equations are significant or negligible and ensures that the main
computational variables are of similar magnitude.

CFD

2 13

David Apsley

Examples
Q1.
In 2-d flow, the continuity and x-momentum equations can be written in conservative
differential form as

(u ) (v) 0
t x
y

p
(u ) (uu ) (vu) 2 u
t
x
y
x
respectively.
(a)

Show that these can be written in the equivalent non-conservative forms:

D
u v
( ) 0
Dt
x y
Du
p

2 u
Dt
x
D

where the material derivative is given (in 2 dimensions) by

u v .
Dt t
x
y

(b)

Define carefully what is meant by the statement that a flow is incompressible. To

what does the continuity equation reduce in incompressible flow?

(c)

Write down conservative forms of the 3-d equations for mass and x-momentum.

(d)

Write down the z-momentum equation, including the gravitational force.

(e)

Show that, for constant-density flows, pressure and gravity forces can be combined in
the momentum equations via the piezometric pressure p + gz.

(f)

In a rotating reference frame there are additional apparent forces (per unit volume):
2R
centrifugal force:
or
( r )
Coriolis force:
2 u
axis
where is the angular velocity of the reference frame, u is the
fluid velocity in that frame, r is the position vector (relative to a
R
point on the axis of rotation) and R is its projection perpendicular
to the axis of rotation. By writing the centrifugal force as the

gradient of some quantity show that it can be subsumed into a

r
modified pressure. Also, find the components of the Coriolis
force if rotation is about the z axis.

(*** Advanced ***)

(g)
Write the conservative mass and momentum equations in vector notation.
(h)

CFD

Write the conservative mass and momentum equations in suffix notation using the
summation convention.

2 14

David Apsley

R
2

Q2. (Some exact solutions of the Navier-Stokes equation)

The x-component of the momentum equation is given by
Du
p

2 u
Dt
x
Using this equation derive the velocity profile in fully-developed, laminar flow for:
(a)
pressure-driven flow between stationary parallel planes (Plane Poiseuille flow);
(b)
constant-pressure flow between stationary and moving planes (Couette flow).
Ideally, you should devise your own coordinate system to describe
and analyse this problem. However, to be specific, assume flow in
the x direction, with bounding planes y = 0 and y = h. The velocity
is then (u(y),0,0). In part (a) both walls are stationary. In part (b)
the upper wall slides parallel to the lower wall with velocity Uw.

u(y)
x

(c) (*** Advanced ***) In cylindrical polars (x,r,) the Laplacian 2 is more complicated. If
axisymmetric, with fully-developed velocity (u(r ),0,0) then
1 u
2u
(r )
r r r
Derive the velocity profile in a circular pipe with stationary wall at r = R (Poiseuille flow).

Q3. (*** Advanced ***)

By applying Gausss divergence theorem, deduce the conservative and non-conservative
differential equations corresponding to the general integral scalar-transport equation
d

dV (u ) dA s dV
dt V
V
V

Q4.
In each of the following cases, state which of (i), (ii), (iii) is a valid dimensionless number.
Carry out research to find the name and physical significance of these numbers.
(L = length; u = velocity; z = height; p = pressure; = density; = dynamic viscosity;
= kinematic viscosity; g = gravitational acceleration; = angular velocity).
(a)
(b)

(c)

(d)

CFD

(i)

p p ref

U
UL
(i)
;

g d

dz

(i)
du
dz
U
(i)
;
L

(ii)

p p ref

U
UL
(ii)
;

(iii) U 2 ( p pref )

(iii) UL

1/ 2

(ii)

U
;
gL

(iii)

(ii)

gL
;
U

(iii)

2 15

p p ref
g
U
L

David Apsley

Q5. (Exam 2008; part (h) depends on later sections of this course)
The momentum equation for a viscous fluid in a rotating reference frame is

Du
p 2u 2 u
Dt

(*)

where is density, u = (u,v,w) is velocity, p is pressure, is dynamic viscosity and is the

angular-velocity vector of the reference frame. The symbol denotes a vector product.
(a)

If (0,0, ) write down the x and y components of the Coriolis force ( 2 u ).

(b)

Hence write down the x- and y-components of equation (*).

(c)

Show how Equation (*) can be written in non-dimensional form in terms of a

Reynolds number Re and Rossby number Ro (both of which should be defined).

(d)

Define the terms conservative and non-conservative when applied to the differential
equations describing fluid flow.

(e)

Define (mathematically) the material derivative operator D/Dt. Then, noting that the
continuity equation can be written
(u ) (v) (w)

0,
t
x
y
z
show that the x-momentum equation can be written in an equivalent conservative
form.

(f)

If the x-momentum equation were to be regarded as a special case of the general

scalar-transport (or advection-diffusion) equation, identify the quantities representing:
(i)
concentration;
(ii)
diffusivity;
(iii)
source.

(g)

Explain why the three equations for the components of momentum cannot be treated
as independent scalar equations.

(h)

Explain (briefly) how pressure is determined in a CFD simulation of:

(i)
high-speed, compressible gas flow;
(ii)
incompressible flow.

CFD

2 16

David Apsley

Q6.
(a)

(b)

In a rotating reference frame (with angular velocity vector ) the non-viscous forces
on a fluid are, per unit volume,
p g 2R 2 u
( I)
(II)
(III)
(IV)
where p is pressure, g = (0,0,g) is the gravity vector and R is the vector from the
closest point on the axis of rotation to a point. Show that, in a constant-density fluid,
force densities (I), (II) and (III) can be combined in terms of a modified pressure.
Consider a closed cylindrical can of radius 5 cm and depth 15 cm. The can is
completely filled with fluid of density 1100 kg m3 and is rotating steadily about its
axis (which is vertical) at 600 rpm. Where do the maximum and minimum pressures
in the can occur, and what is the difference in pressure between them?

Q7. (Exam 2011 part)

The figure below depicts a 2-d cell in a finite-volume CFD calculation. Vertices are given in
the figure, and velocity in the adjacent table. At this instant = 1.0 everywhere.
(a)

Calculate the volume flux out of each face. (Assume unit depth.)

(b)

Show that the flow is not incompressible and find the time derivative of density.

(-1,2)
y

(6,2)

w
x

CFD

(0,0)

e
n
w
s

e
s

Face

(4,0)

2 17

Velocity (u,v)
u
v
4
10
1
8
2
2
1
4

David Apsley

3. APPROXIMATIONS AND SIMPLIFIED EQUATIONS

SPRING 2016

3.1 Steady-state vs time-dependent flow

3.2 Two-dimensional vs three-dimensional flow
3.3 Incompressible vs compressible flow
3.4 Inviscid vs viscous flow
3.5 Hydrostatic vs non-hydrostatic flow
3.6 Boussinesq approximation for density
3.7 Depth-averaged (shallow-water) equations
3.8 Reynolds-averaged equations (turbulent flow)
Examples

Fluid dynamics is governed by equations for mass, momentum and energy. The momentum
equation for a viscous fluid is called the Navier-Stokes equation; it is based upon:

continuum mechanics;

the momentum principle;

shear stress proportional to velocity gradient.

A fluid for which the last is true is called a Newtonian fluid; this is the case for most fluids in
engineering. However, there are important non-Newtonian fluids; e.g. mud, cement, blood,
paint, polymer solutions. CFD can still usefully be applied for these.
The full equations are time-dependent, 3-dimensional, viscous, compressible, non-linear and
highly coupled. However, in most cases it is possible to simplify analysis by adopting a
reduced equation set. Some common approximations are listed below.
Reduction of dimension:

steady-state;

two-dimensional.
Neglect of some fluid property:

incompressible;

inviscid.
Simplified forces:

hydrostatic;

Boussinesq approximation for density.

Approximations based upon averaging:

depth-averaging (shallow-water equations);

Reynolds-averaging (turbulent flows).

CFD

31

David Apsley

3.1 Steady-State vs Time-Dependent Flow

Many flows are naturally time-dependent. Examples include waves, tides, reciprocating
pumps and internal combustion engines.
Other flows have stationary boundaries but become time-dependent because of an instability.
An important example is vortex shedding around cylindrical objects. Depending on the
Reynolds number the instability may or may not progress to fully-developed turbulence.
Some computational solution procedures rely on a time-stepping method to march to steady
state; examples are transonic flow and open-channel flow (where the mathematical nature of
the governing equations is different for Mach or Froude numbers less than or greater than 1).
Thus, there are three major reasons for using the full time-dependent equations:

time-dependent problem;

time-dependent instability;

time-marching to steady state.

Time-dependent methods will be addressed in Section 6.

3.2 Two- Dimensional vs Three-Dimensional Flow

Geometry and boundary conditions may dictate that the flow is two-dimensional (at least in
the mean). Two-dimensional calculations require considerably less computer resources.
Two-dimensional may include axisymmetric. This is actually easier to achieve in the
laboratory than Cartesian 2-dimensionality because of side-wall boundary layers in the latter.

3.3 Incompressible vs Compressible Flow

A flow (not a fluid, note) is said to be incompressible if flow-induced pressure or temperature
changes do not cause significant density changes. Compressibility is important in high-speed
flow or where there is significant heat input.
Liquid flows are usually treated as incompressible (water hammer being an important
exception), but gas flows can also be regarded as incompressible at speeds much less than the
speed of sound (a common rule of thumb being Mach number < 0.3).
Density variations within fluids can occur for other reasons, notably from salinity variations
(oceans) and temperature variations (atmosphere). These lead to buoyancy forces. Because
the density variations are not flow-induced these flows can still be treated as incompressible;
i.e. incompressible does not necessarily mean uniform density.
The key differences in CFD between compressible and incompressible flow concern:
(1)
whether there is a need to solve a separate energy equation;
(2)
how pressure is determined.

CFD

32

David Apsley

Compressible Flow
First Law of Thermodynamics:
change of energy heat input work done on fluid
A transport equation has to be solved for an energy-related variable (e.g. internal energy e or
enthalpy h e p / ) in order to obtain the absolute temperature T. For an ideal gas,
or
h c pT
e cv T
cv and cp are specific heat capacities at constant volume and constant pressure respectively.
Mass conservation provides a transport equation for , whilst pressure is derived from an
equation of state; e.g. the ideal-gas law:
p RT
For compressible flow it is necessary to solve an energy equation.
In density-based methods for compressible CFD:

mass equation ;

energy equation T;

equation of state p.

Incompressible Flow
In incompressible flow, pressure changes (by definition) cause negligible changes to density.
Temperature is not involved and so a separate energy equation is not necessary. The
Mechanical Energy Principle:
change of kinetic energy = work done
is equivalent to, and readily derived from, the momentum equation. In the inviscid case it is
often expressed as Bernoullis equation (see the Examples).
Incompressibility implies that density is constant along a streamline:
D
0
Dt
but may vary between streamlines (e.g. due to salinity differences). Conservation of mass is
then replaced by conservation of volume:
u v w
or

0
(volume flux ) 0

x y z
faces
Pressure is not derived from a thermodynamic relation but from the requirement that
solutions of the momentum equation be mass-consistent (Section 5).
In incompressible flow it is not necessary to solve a separate energy equation.
In pressure-based methods for incompressible CFD:

incompressibility volume is conserved as well as mass;

requiring solutions of momentum equation to be mass-consistent equation for p.

CFD

33

David Apsley

3.4 Inviscid vs Viscous Flow

If viscosity is neglected, the Navier-Stokes equations become the Euler equations.
Consider streamwise momentum in a developing 2-d boundary layer:
u
u
p
2u
(u
v ) 2
x
y
x
y
mass acceleration = pressure force + viscous force
Dropping the viscous term reduces the order of the highest derivative from 2 to 1 and hence
one less boundary condition is required.
y

Viscous (real) flows require a no-slip (zero-relative-velocity)

condition at rigid walls the dynamic boundary condition.

viscous

Inviscid (ideal) flows require only the velocity component

normal to the wall to be zero the kinematic boundary condition.
The wall shear stress is zero.

U
inviscid

Although its magnitude is small, and consequently its direct influence via the shear stress is
tiny, viscosity can have a global influence out of all proportion to its size. The most important
effect is flow separation, where the viscous boundary layer required to satisfy the non-slip
condition is first slowed and then reversed by an adverse pressure gradient. Boundary-layer
separation has two important consequences:

major disturbance to the flow;

a large increase in pressure drag.

separation

rec
irc
u
flo latin
w
g

CFD

34

David Apsley

3.4.1 Inviscid Approximation: Potential Flow

Velocity Potential,
In inviscid flow it may be shown1 that the velocity components can be written as the gradient
of a single scalar variable, the velocity potential :

,
, w
(concisely written: u )
u
v
x
z
y
Substituting these into the continuity equation for incompressible flow:
u v w
(concisely written: u )

0
x y z
gives
2 2 2

0
x 2 y 2 z 2
which is often written
(Laplaces equation)
(1)
2

Stream Function,
In 2-d incompressible flow there exists a function called the stream function such that

,
v
u
x
y
If the flow is also inviscid then it may be shown that the fluid is irrotational and
u v

0
y x
Substituting the expressions for u and v gives an equation for :
(Laplaces equation)
2

(2)

In both cases above the entire flow is completely determined by a single scalar field ( or )
satisfying Laplaces equation. Moreover, since Laplaces equation occurs in many branches
of physics (electrostatics, heat conduction, gravitation, optics, ...) many good solvers exist.
Velocity components u, v and w are obtained by differentiating or . Pressure is then
recoverable from Bernoullis equation:
p 12 U 2 constant (along a streamline)
where U is the magnitude of velocity.
The potential-flow assumption often gives an adequate description of the flow and pressure
fields for real fluids, except very close to solid surfaces where viscous forces are significant.
It is useful, for example, in calculating the lift force on aerofoils and in wave theory
(Hydraulics 3). However, in ignoring viscosity it implies that there are neither tangential
stresses on boundaries nor flow separation, which leads to the erroneous conclusion
(DAlemberts Paradox) that an object moving through a fluid experiences no drag.
1

Since pressure acts perpendicularly to a surface and cannot impart rotation, an inviscid fluid can be regarded as
irrotational ( u 0 ), and so the velocity field can be written as the gradient of a scalar function.

CFD

35

David Apsley

3.5 Hydrostatic vs Non-Hydrostatic Flow

The equation for the vertical component of momentum can be written:
Dw
p

g viscous forces
Dt
z
For large horizontal scales the vertical acceleration Dw/Dt is much less than g and the viscous
forces are small. The balance of terms is then the same as in a stationary fluid:
p
i.e. pressure forces balance weight.
g
z
With this approximation, in constant-density flows with a free surface the pressure is
determined everywhere by the position of the free surface:
p patm g (h z) ,
where h h( x, y)
This results in a huge saving in computational time because the position of the surface
automatically determines the pressure field without the need for a separate pressure equation.
z
patm
h-z
h(x)

p p atm g (h z )

The hydrostatic approximation is widely used in conjunction with the depth-averaged

shallow-water equations (see Section 3.7 below and Dr Rogers part of the course). However,
it is not used in general-purpose flow solvers because it limits the type of flow that can be
computed.

CFD

36

David Apsley

3.6 Boussinesq Approximation for Density2

Density variations may arise at low speeds because of changes in temperature or humidity
(atmosphere), or salinity (water) which give rise to buoyancy forces. The effect of these
density changes can be significant even if the fractional change in density is small.
The Boussinesq approximation retains density variations in the gravitational term (giving
buoyancy forces) but disregards them in the inertial (mass acceleration) term; i.e. in the
vertical momentum equation:
Dw
p
p

g 0 g ( 0 ) g
Dt
z
z
the Boussinesq approximation is simply to replace on the LHS by the constant density 0.
The approximation is valid if relative density variations are not too large; i.e. / 0 1 .
This condition is usually satisfied in the atmosphere and oceans.
On the RHS of the momentum equation, the part of the weight resulting from the constant
reference density 0 is usually subsumed into a modified pressure p* p 0 gz , so that
0

( p 0 gz )
Dw

( 0 ) g ...
Dt
z

The relative change in density is typically proportional to the change in some scalar (e.g.
temperature or salinity):
0
or
( 0 )
0 0 ( 0 )
0
where is the coefficient of expansion. (The sign adopted here is that for temperature, where
an increase in temperature leads to a reduction in density the opposite would be true for
salinity-driven density changes.) The vertical momentum equation can then be written
Dw
p *
where
p* p 0 gz
0

0 ( 0 ) g

Dt
z
buoyancy force

Temperature variations in the atmosphere, brought about by surface (or cloud-top) heating or
cooling, are responsible for significant changes in airflow and turbulence.

On a cold night the atmosphere is stable. Cool, dense

air collects near the surface and vertical motions are
suppressed; the boundary-layer depth is 100 m or less.

stable boundary layer

u
mixing depth

On a warm day the atmosphere is unstable. Surface

heating produces warm; light air near the ground and
convection occurs; the boundary layer may be 2 km
deep.

convective boundary layer

Vertical dispersion of pollution is very different in the two

cases.

Note that several other very-different approximations are also referred to as the Boussinesq approximation in
different contexts e.g. shallow-water equations or eddy-viscosity turbulence models.

CFD

37

David Apsley

3.7 Depth-Averaged (Shallow-Water) Equations

This approximation is used for flow of a constant-density fluid with a free surface, where the
depth of fluid is small compared with typical horizontal scales.
In this hydraulic approximation, the fluid can be regarded
as quasi-2d with:

horizontal velocity components u, v;

depth of water, h.
Note that the depth h may vary due to changes in the levels of
the free-surface, the bed, or both.

z
h(x,t)
u

By applying mass and momentum principles to a vertical column of constant-density fluid of

variable depth h, the depth-integrated equations governing the motion can be written (for the
one-dimensional case and in conservative form) as:
h (uh)

0
t
x
( 12 gh 2 ) 1
(uh) (u 2 h)

( surface bed )
t
x
x

2
1
The 2 gh term comes from (1/ times) the hydrostatic pressure force per unit width on a
water column of height h; i.e. average hydrostatic pressure ( 12 gh ) area (h 1). The final
term is the net effect of surface stress (due to wind) and the bed shear stress (due to friction).
These equations are derived in the Examples and in Dr Rogers part of this course.
The resulting shallow-water (or Saint-Venant) equations are mathematically similar to those
for a compressible gas. There are direct analogies between

discontinuities: hydraulic jumps (shallow flow) and shocks (compressible flow);

critical flow over a weir (shallow) or gas flow through a throat (compressible).
In both cases there is a characteristic wave speed ( c gh in the hydraulic case; c p/
in compressible flow). Whether this is greater or smaller than the flow velocity determines
whether disturbances can propagate upstream and hence the nature of the flow. The ratio of
flow speed to wave speed is known as:
u
Froude number:
in shallow flow
Fr
gh
u
Mach number:
in compressible flow
Ma
c

CFD

38

David Apsley

3.8 Reynolds-Averaged Equations (Turbulent Flow)

The majority of flows encountered in engineering are turbulent. Most, however, can be
regarded as time-dependent, three-dimensional fluctuations superimposed on a much simpler
(and often steady) mean flow. Usually, we are only interested in mean quantities, rather than
details of the time-varying flow.
The process of Reynolds-averaging (named after Osborne Reynolds3, first Professor of
Engineering at Owens College, later to become the Victoria University of Manchester) starts
by decomposing each flow variable into mean and turbulent parts:
u u
u
mean

fluctuatio n
mean

fluctuation

The mean may be a time average (the usual case in the laboratory) or an ensemble average
(a probabilistic mean over a large number of identical experiments).
When the Navier-Stokes equation is averaged, the result is (see Section 7):

an equivalent equation for the mean flow,

except for

turbulent fluxes, u v etc. (called the Reynolds stresses) which provide a net
transport of momentum.
For example, the viscous shear stress
u
visc
y
is supplemented by an additional turbulent stress:
turb u v
In order to solve the mean-flow equations, a turbulence model is required to supply these
turbulent stresses. Popular models exploit an analogy between viscous and turbulent transport
and employ an eddy viscosity t to supplement the molecular viscosity. Thus,
u
u

u v ( t )
y
y
This is readily incorporated into the mean momentum equation via a (position-dependent)
effective viscosity. However, actually specifying t is by no means trivial see the lectures on
turbulence modelling (Section 8).

Osborne Reynolds experimental apparatus including that used in his famous pipe-flow experiments is on
display in the basement of the George Begg building at the University of Manchester. A modern replica is in the
George Begg foyer.

CFD

39

David Apsley

Examples
Q1.
Discuss the circumstances under which a fluid flow can be approximated as:
(a)
incompressible;
(b)
inviscid.

s
Q2.
By resolving forces along a streamline, the steady-state momentum
U
equation for an inviscid fluid can be written

U
p
U
g sin
s
s
where U is the velocity magnitude, s is the distance along a streamline and is the angle
between local velocity and the horizontal. Assuming incompressible flow, derive Bernoullis
equation. (This question demonstrates that, for incompressible flow, the mechanical energy
principle can be derived directly from the momentum equation.)

Q3.
A velocity field is given by the velocity potential x 2 y 2 .
(a)
Calculate the velocity components u and v.
(b)
Calculate the acceleration.
(c)
Calculate the corresponding streamfunction, .
(d)
Sketch the streamlines and suggest a geometry in which one might expect this flow.

Q4.
For incompressible flow in a rotating reference frame the force per unit volume, f, is the sum
of pressure, gravitational, Coriolis and viscous forces:
f p ge z 2 u 2u
where ez is a unit vector in the z direction and is the angular velocity of the rotating frame.
(a)

If the density is uniform, show that pressure and gravitational forces can be combined
in a piezometric pressure (which should be defined).

(b)

If the density varies, describe the Boussinesq approximation in this context and give
an application in which it is used.

(c)

Show how the momentum equation (with Boussinesq approximation for density) can
be non-dimensionalised in terms of densimetric Froude number, Rossby number and
Reynolds number:
U0
U
U L
Fr
,
Ro 0 ,
Re 0 0 0
L0

( / 0 ) gL0
where 0, L0, U0 are characteristic density, length and velocity scales, respectively,
and is a typical magnitude of density variation.

CFD

3 10

David Apsley

Q5. (Exam 2015 part)

The vector momentum equation for a viscous fluid of variable density is
Du
(*)

p ge z 2u
Dt
where t is time, is density, u = (u,v,w) is velocity, p is pressure, is dynamic viscosity, g is
the acceleration due to gravity and ez is a unit vector in the z direction.
(a)

Define the operator D/Dt mathematically and explain its physical significance.

(b)

Show that, for a constant density 0, the pressure and gravitational terms can be
combined as a single gradient term involving the piezometric pressure.

(c)

If density variations occur as a result of temperature changes, then

0
( 0 )
0
where is temperature and is the coefficient of thermal expansion. Describe the
Boussinesq approximation in this context, apply it in Equation (*), and state the
conditions under which it is valid.

(d)

Show that, with the Boussinesq approximation, Equation (*) can be nondimensionalised as
Du
1
1 2
p 2 e z
u
Dt
Re
Fr
where all variables are now non-dimensional, and Re and Fr are, respectively, the
Reynolds number and densimetric Froude number (both to be defined).

CFD

3 11

David Apsley

Q6.
(a)

(b)

In flow with a free surface, by taking a control volume as a column of (time-varying)

height h(x,y,t) and horizontal cross-section x y, assuming that density is constant,
13 and 23 are the only significant stress components, the horizontal velocity field
may be replaced by the depth-averaged velocity (u, v) and the pressure is hydrostatic,
derive the shallow-water equations for continuity and x-momentum in the form
h (hu ) (hv)

0
t
x
y
z
( surface) 13 (bed )
(hu ) (hu 2 ) (hvu)

gh s 13
t
x
y
x

Provide an alternative derivation by integrating the continuity and horizontal

momentum equations for incompressible flow:
u v w

0
x y z
(u ) (u 2 ) (vu) (wu )
p

13
t
x
y
z
x z
over a depth h = zs zb.
For part (b) you will need the boundary condition that the top and bottom surfaces
z z s ( x, y) and z z b ( x, y) are material surfaces:
z
z
z
D
or
(z zs ) 0
w s u s v s 0 on z = zs
Dt
t
x
y
and similarly for zb, together with Leibniz Theorem for differentiating an integral:
b( x)
b( x)
d
db
da
f

dx f (b) f (a)
f ( x ) dx
dx a ( x )
dx
dx
a ( x ) x

Note: this is easily extended to consider additional forces such as Coriolis forces and other
stress terms (horizontal diffusion). Dr Rogers will cover this in the second part of the
course.

CFD

3 12

David Apsley