Lectures for ES30A/D, FUNDAMENTAL FLUID MECHANICS

Term 2, Weeks 1-5, 2005.

R.M. Kerr, University of Warwick [email protected]
Tel: 74718, Messages: 22014
January 24, 2005
Preliminaries
0.1 Class times and places 0.2 Office hours
• Mondays 17:05-18:05 L4 • Tuesdays 16:00-17:00 F338
• Fridays 16:00-17:00 F338
• Tuesdays 17:05-18:05 L4 • Other times:
– Mon+Weds afternoons Maths C2.29
• Fridays 17:05-18:05 L5 – Most mornings: F338

0.3 Web sites

• Reading notes: http://www.eng.warwick.ac.uk/staff/rmk/rmk files/ES30AD 0405.pdf

• Notes to insert:
http://www2.warwick.ac.uk/fac/sci/eng/euo/students/?url=students/notes/es30a/index.htm

0.4 Main TEXTS

The book that probably covers the greatest portion of this course, especially for those carrying on to take ES30D is:

• E.L. Houghton & P.W. Carpenter (2003) Aerodynamics for Engineering Students, Ed. 5., Oxford: Elsevier Science.
Some of the topics are also covered in these standard textbooks:

• B. Massey (1998) Mechanics of Fluids 7th Ed., Stanley Thornes (Pub.) Ltd.

• F.M. White (1999) Fluid Mechanics 4th Ed. McGraw-Hill.

• F.M. White (1991) Viscous Fluid Flow 2nd. Ed., McGraw-Hill.

Other books that are referenced are:

• Schlichting, H. Boundary Layer Theory, 7th Ed., Chap. 5.

• D.J. Tritton Physical Fluid Dynamics 2nd Ed. Clarendon.

• J.D. Anderson (1991) Fundamentals of Aerodynamics 2nd Ed., McGraw-Hill.

• T. Cebeci & P. Bradshaw Momentum Transfer in Boundary Layers Hemisphere.

• F.S. Sherman Viscous Flow McGraw-Hill.

• M. van Dyke An Album of Fluid Motion Parabolic Press.

0.5 Assessment and Approach

ES30A is a 5 week introduction to fluid dynamics for both Mechanical and Civl Engineers. Civil Engineering students then follow ES30B
while the remaining students continue on to ES30D.

0.6 Assessment
• Three-hour written examination.

• 3 questions each for ES30A, ES30B and ES30D.

• Full marks can be obtained by correct answers to 4 questions, two from ES30A and two from either ES30B or ES30D.

• The questions will be based either on the lecture notes, the set exercises or the case studies.

• 2002 Examination Results: 27% of students got 1st class marks (i.e. scored better than 70%). Average was around 60%.

In order to do well on the course it is essential that:

1. You create a good set of notes – the handouts are only a guide and are not intended to be a complete set of notes.

2. You keep up with the course and seek help from me if you do not understand something; and

3. You do the set exercises – models answer will be made available in the SRC and on the inserted notes web site above.

You will not be able to cope with the course by only attending the lectures without any self-study. Past examination question can be
found on the inserted notes website. Examination questions prior to 1999 are not relevant.
1 Introduction
The physical laws that govern fluid flow are deceptively simple. Paramount amoung them is

1. Newton’s second law of motion which states that:

MASS × ACCELERATION = APPLIED FORCE

In fluid mechanics we prefer to use the equivalent form of

RATE OF CHANGE OF MOMENTUM = APPLIED FORCE

2. Principles of conservation of mass and, where appropriate, conservation of energy.

The remaining physical laws required relate solely to determining the forces involved. For a wide range of engineering applications the only
forces involved are the

body forces due to the action of gravity (which, of course, requires the use of Newton’s theory of gravity; but only in a very simple
way);

pressure forces (these are found by applying Newton’s laws of motion and require no further physical laws or principles); and

viscous forces. To determine the viscous forces we need to supplement Newton’s laws of motion with a constitutive law. For pure
homogeneous fluids (such as air and water) this constitutive law is provided by the Newtonian fluid model, which as the name suggests
also originated with Newton. In simple terms the constitutive law for a Newtonian fluid states that:

VISCOUS STRESS ∝ RATE OF STRAIN

1.1 Continuum model

These simple physical laws are, of course, merely theoretical models. The principal theoretical assumption is that the fluid consists of
continuous matter – the so-called continuum model.
At the microscopic level, a fluid is not a continuum, but consists of myriads of individual molecules. In most engineering applications
even a tiny volume of fluid (measuring, say, 1 µm3 ) contains a large number of molecules.
However, a typical molecule travels on average a very short distance (known as the mean free path) before colliding with another.
In typical engineering aerodynamics application the m.f.p. is less than 100 nm, which is very much smaller than any relevant scale
characterising quantities of engineering significance. Owing to this disparity between the m.f.p. and relevant length scales, we may
expect the equations of fluid motion, we can apply averaging. This is the continuum model and is obeyed to great precision by
the fluid flows found in almost all engineering applications.
This expectation is supported by experience. When pure homogeneous fluids are involved the only exceptions are very small-scale
motions (e.g., nanotechnology), where the relevant scales are comparable to the m.f.p.; and rarefied gas dynamics (e.g., re-entry vehicles),
where there are so few molecules present that the m.f.p. becomes comparable to the dimensions of the vehicle.
 In engineering and for some other scientific applications we often wish to understand flows of mixtures of pure fluids and particulate
matter, e.g., blood, toothpaste, slurries. These are usually modelled as fluids that obey more complex constitutive laws termed
collectively non-Newtonian fluids. We will not consider non-Newtonian fluids in any detail on this course.
The physical laws governing fluid flows may be simple but, even for the simplest geometries, such as the flow past a flat plate
or around a sphere, the resulting fluid flows can be exceeding complex especially at high Reynolds numbers. This complexity still
presents a major challenge to the engineer.
Why do such simple physical laws lead to such complex flows? There is no simple answer. However, part of the reason is the

• effect of viscosity at high Reynolds number. Since the Reynolds number is a measure of the ratio of inertial forces to viscous
forces, we might expect viscous effects to become progressively weaker as the Reynolds number is increased. What actually happens,
instead, is that the viscous effects become concentrated in a thin layers or vortices.

• Near boundaries, the layer which forms at the surface becomes progressively thinner as the Reynolds number is increased.

• Or the vorticity is created by viscous effects acting at the surface and is concentrated in the thin boundary layer.
The formation of thin vortical regions would not matter greatly if they remained attached to the surface of the body. But, owing to
the effects of the combination of pressure and viscous forces:

• the boundary layer inevitably separates from the surface and the thin vortical layers move into the flow field where they have a
tendency to form complex structures.

The most extreme manifestation of this is turbulent flow which is a highly complex, unsteady, three-dimensional flow. Turbulent
flow inevitably occurs at high Reynolds numbers. In fact, when the Reynolds number becomes sufficiently large even the attached vortical
flow in the boundary layer becomes turbulent. Turbulent flow is the subject of a 4th Year MEng Module.

1.2 Basic Objectives

The basic objectives of the course are:

(i) To show you how the underlying physical principles governing fluid flow can be used to derive the governing equations of motion for
fluid flow.

(ii) To give you an understanding of the physics behind the mathematical equations, so that you can understand when it is appropriate to
make the various assumptions and approximations used by engineers. This is essential if you are to avoid blindly using the powerful
commercial software now available for fluid dynamics. Engineers must have a physical understanding of the computational calculations
and results that they use. otherwise it is easy to be mesmerized by the advanced graphics used to present the computational output
and to make catastrophic mistakes.

(iii) We will also consider a series of different approaches used by engineers to obtain useful approximate results for particular engineering
applications. Some of these engineering applications will be used as illustrative examples in the various case studies presented during
the course. All the visual and other material used for each case study will be available on the course web sites.
1.3 Outline
The course is set out according to the following plan:

ES30A Fundamental Fluid Dynamics I

• Chap. 1 – Introduction and revision of mathematics

• Chap. 2 – Derivation and properties of the governing equations of motion for fluid flow – the Navier-Stokes equations. Dimensional
Analysis.

• Chap. 3 – Some useful exact solutions of the N-S equations; flow between parallel plates; stagnation-point flow.

• Chap. 4 – Inviscid Flow. Since this was considered in some detail in the Yr. 2 course it will only very briefly considered here.

• Chap. 5 – Flow in porous media. Inertial effects can be neglected for these flows, allowing the governing equations to be simplified.

• Chap. 6 – Vortices and vorticity. Here we will briefly consider some of the organized vortical structures typically found at high
Reynolds numbers. Introduction to turbulent flow.

• Chap. 7 – Turbulent flow

Present Case Study pictures

ES30D Fundamental Fluid Dynamics II

• Chap. 8 – Creeping (or Stokes) Flow. This approach is used at very low Reynolds numbers when, again, the inertial effects can be
ignored.

• Chap. 9 – Lubrication theory. Inertial forces are also negligible for these flows between surfaces with very small separations. Reynolds
lubrication theory. Slider and thrust bearings. Journal bearings. Hydrostatic bearings. Lubrication of gears and rolling bearings.

• Chap. 10 – Flow at high Reynolds number. Boundary-layer theory and its applications. This is the approach used to simplify the
equations of motion at high Reynolds numbers.

• Chap. 11 – Computational solution of the N-S equations.

• Chap. 12 – Flow control.

1.4 Revision of Relevant Topics from 1st Yr Mathematics
Some of the topics covered in the 1st Year Maths course are used in the present course. These are briefly covered below. If you are unfamiliar
with these topics you should revise your 1st Year Mathematics.

1.4.1 Calculus
1. Differentiation of a product Let f (x) and g(x) be functions of x then

d df dg
fg = g+f (1.1)
dx dx dx
2. Triple products If h(x) is also a function of x then
d d(f g) dh
f gh = h + fg (1.2)
dx | dx
{z } dx
(1.1)

3. Differentiation of a function of a function Let f (g) be a function of g where g(x) is a function of x, then

df df dg
= (1.3)
dx dg dx
Example– let g = sin x and f = g 3 = sin3 x then

df dg df
= 3g 2 , = cos x therefore = 3 sin2 x cos x
dg dx dx
4. Differentiation of an integral

b(x)
Z Z b
d ∂ db da
f (x, y)dy = f (x, y)dy + f (x, b) − f (x, a) (1.4)
dx a ∂x dx dx
a(x)

5. Taylor’s series If f (x) is a function of x then an approximation for f evaluated at x0 + δx is

df 1 d2 f
f (x0 + δx) = f (x0 ) + δx + (δx)2 + ... (1.5)
dx x0 2 dx2 x0
The vertical line with subscript x0 indicates that the derivative is to be evalauated at x = x0 . Most often we are interested in the
case when δx → 0 in which case all the terms after the second one on the RHS are so-called higher-order terms and normally omitted
on the grounds that they are infinitesimally small compared with the remaining terms.
 Eq. (1.5) can be extended to 2 or 3 dimensions by using partial differentiation. Thus if f (x, y) is now a function of x and y we can
write
∂f ∂f
f (x0 + δx, y0 + δy) = f (x0 , y0 ) + δx + δy + ... (1.6)
∂x (x0 ,y0 ) ∂y (x0 ,y0 )

1.4.2 Ordinary Differential Equations

6. Order The order of an ODE is determined by the order of the highest derivative. Examples
dy
+ xy + y 2 = 0 is 1st order (1.7)
dx

d2
 
2
− f =0 is 2nd order (1.8)
dR2 R2

d2 d2
  
2 2
2
− 2 2
− 2 f =0 is 4th order (1.9)
dR R dR R
7. Linear versus Nonlinear Equations The dependent variable and its derivatives do not appear in products for linear equations,
 2  
2 dy dy
e.g. y , , y
dx dx
would not be found in a linear equation. Fluid equations are inherently NONLINEAR and cannot be solved analytically.
The test is that if y1 and y2 are solutions then it must be possible to show from the ODE that ky1 , ky2 (k is an arbitrary constant)
and y1 + y2 are also solutions. Examples
dy
+ xy + y 2 = 0 is nonlinear (1.10)
dx
 2 
d 2
− f =0 is linear (1.11)
dR2 R2
 2  2 
d 2 d 2
− − f =0 is linear (1.12)
dR2 R2 dR2 R2
8. Homogeneous equations – Complementary solution The two linear examples given above are both homogeneus, meaning that there is
no term that is purely a function of the independent variable (this includes the possiblility of an arbitrary constant). For example,
the RHS is zero; if it were a function of R then the equation would be inhomogeneous. All solutions of a linear homogeneous ODE
must take a form known as the general solution consisting of n linearly independent functions of the dependent variable where n is
the order of the ODE. These linear independent functions must be solutions of the ODE and are called fundamental solutions.
Example 1
d2 y
The GS of −y =0 takes the form y = C1 y1 + C2 y2 (1.13a)
dx2
 where y1 and y2 are two linearly independent functions of x and C1 and C2 are arbitrary constants of integration. The form of the
fundamental solutions is often found by inspection, e.g. in Example 1 we try the form

y ∼ emy (1.13b)

substituting this in the ODE gives

(m2 − 1)emy = 0 → m = ±1 → y1,2 = e±y (1.13c)
Example 2

d2
 
2
The GS of 2
− 2 f =0 takes the form f = C1 f1 + C2 f2 (1.14a)
dR R
where f1 and f2 are two linearly independent functions of R and C1 and C2 are arbitrary constants of integration. In this case we will
try fundamental solutions of the form
f ∼ Rm (1.14b)
substituting this in the ODE gives

[m(m − 1) − 2]Rm−2 = 0 → m = −1, 2 → f1 = R−1 , f2 = R2 (1.14c)

9. Inhomogeneous equations – particular solution When the zero on the RHS is replaced by a function of x (or R) the ODE is said to be
inhomogeneous. In this case all solutions can be written in the form (called the general solution)

G.S. = Complementary f unction + P articular solution

where the complementary solution is the general solution to the corresponding homogeneous ODE and the particular solution is any
solution to the full inhomogeneous ODE. The PS is often found by trial and error.
Example
d2 y
−y =3 (1.15a)
dx2
The CF was found above as C1 ey + C2 e−y , to find the PS try y ∼ C (where C is a constant). Substitution in the ODE shows that
C = −3 is a solution. Therefore the general solution is

y = C1 ey + C2 e−y − 3 (1.15b)
 1.4.3 Partial Differential Equations

10. Number of independent variables

Ordinary Differential Equations have 1 independent variable (usually x or t) and one or more dependent variables. Partial Differential
Equations can have many independent variables and one or many dependent variables.
Fluid equations are Partial Differential Equations For example take t time and x space as the two independent variables. Then
the equation for transport for a quantity θ by a velocity u, is
∂θ ∂θ
+u =0 (1.16)
∂t ∂x
Or the diffusion equation for a diffusivity κ is
∂θ ∂2θ
=κ 2 (1.17)
∂t ∂x

1.4.4 Vector Algebra

Scalar product
~ If
Or dot product. Noteboldface represents vectors: ()

~ = A = (A1 , A2 , A3 )
A and B = (B1 , B2 , B3 )
X
then A · B = Ai Bi
i

From this it follows that A = |A|2 = A21 + A22 + A23 = A · A

2

and A · B = AB cos θAB

where θAB is the angle between the vectors A and B.

A · B = AB cos θAB
The following diagram shows the angle that gives
cos θAB that comes from the scalar product between two
vectors A and B in an arbitrary plane through 3D space.
Vector product
Or cross product of two vectors A and B yields a third vector C and is written two ways

C = A × B or C = A ∧ B = (A2 B3 − A3 B2 , A3 B1 − A1 B3 , A1 B2 − A2 B3 )

The second is preferred by some mathematicians because it refers to a generalisation of this concept to higher dimensions. Physicists and
engineers use the first notation.

• In general C is perpendicular to both A and B.

• That is: C · A = 0 and C · B = 0

• Assume A = (A1 , A2 , 0) and B = (B1 , B2 , 0) then C = (0, 0, C3 )

This is represented by the following figure where |C| =

|A||B|| sin θ|. Note that for a small angle between A
and B, where sin θ is small, that the magnitude of C is
small.

1.4.5 Vector Calculus and Gradient operator

The gradient operator is a vector consisting of derivatives. Defining what is sometimes called grad or del=
  X
~ ∂ ∂ ∂ ∂
∇= , , = ei
∂x1 ∂x2 ∂x3 i
∂xi

∂
(grad)i = ∇i =
∂xi
∇ is called nabla in some typesetting languages, including the one that created this document called LaTeX.
The GRADIENT is this operator applied to a scalar φ to create a vector
 
~ = ∇φ
~ = ∂φ ∂φ ∂φ
A , ,
∂x1 ∂x2 ∂x3
 ~
The DIVERGENCE is the equivalent of a scalar product using ∇

~ = ∂A1 + ∂A2 + ∂A3

~ ·A
∇
∂x1 ∂x2 ∂x3
It can be shown that if the variables (x1 , x2 , x3 ) are transformed to a new orthogonal coordinate system (x01 , x02 , x03 ) that the divergence is
invariant.
The CURL is the equivalent of a vector product using ∇ ~
 
~ × ~u = ∂ ∂ ∂
curl ~u = ∇ , , × (u1 , u2 , u3 ) =
∂x1 ∂x2 ∂x3
     
∂u3 ∂u2 ∂u1 ∂u3 ∂u2 ∂u1
− , − , −
∂x2 ∂x3 ∂x3 ∂x1 ∂x1 ∂x2

2 GOVERNING EQUATIONS
2.1 Introduction
Here we will derive the fundamental governing equations for fluid flow – the Navier-Stokes equations. These are

• obtained by applying conservation of mass and momentum to an infinitesimal rectangular control volume – see Fig. 2.1.

This is a (δx,δy) area within the streamlines on the leading edge of flow past a cylinder

• The derivation will be carried out in 2D.

• The 3D version of the equations will be given without proof.

Why physics?
Although the derivations are fairly straightforward many students find them somewhat difficult. Nonetheless it is important to have a good
appreciation of the fundamental physics involved. Why?

• In this way can you can understand how similar flow phenomena are important in many branches of engineering.

• An understanding of the derivation of the Navier-Stokes equations will help you gain physical insight into fluid motions.

• It is also an excellent example of the use of differential calculus for modelling continuously changing processes in continuum mechanics.
2.2 Continuity Equation
Why incompressible?
On this course we will assume that the flow speeds are sufficiently low for the flow to be essentially incompressible.
• Incompressibility implies that we can take the fluid density to be uniform and neglect its variation over the flow field.
• This is usually a reasonably good approximation provided the flow speeds do not exceed more than 0.3-0.5 times the speed of sound
and the temperature differences are moderate.
• Most non-aerodynamic engineering and geophysical (atmospheric-ocean) flows satisfy this condition.
• Compressible flows have near discontinuities (shocks) that are difficult to account for.

Deriving the continuity equation

Let us denote the velocity vector of the fluid flow as

~v = v = (u, v, w)
These velocity components are our dependent variables and in general v varies continuously throughout the flow field and we use
differential calculus to describe its variation, e.g. if v = v1 at x = x1 then its value at
x = x2 = x1 + δx = (x1 + δx, y1 + δy, z1 + δz)
is given by 
∂u ∂u ∂u
v2 = v1 + δv1 = u1 + ∂x
δx + ∂y
δy + ∂z
δz,

∂v ∂v ∂v
v1 + ∂x
δx + ∂y
δy + ∂z
δz,

∂w ∂w ∂w
w1 + ∂x
δx + ∂y
δy + ∂z
δz

provided δx = (δx, δy, δz) → 0. The 3D version of the continuity equation is given by
∂u ∂v ∂w
∇·v = + + =0 (2.1)
∂x ∂y ∂z
This is obtained by applying the conservation of mass with constant density to a cubic control volume with sides measuring δx×δy×δz. Note
that in the derivation only the lowest-order non-trivial terms are retained since only these will remain when we let δx = (δx, δy, δz) → 0.

Insert notes taken from lectures on the derivation in 2D, including Fig. 2.2.
2.3 Momentum Equation in PDE form
The momentum equation:

• Newton’s second law of motion

(mass times acceleration = force or rate of change of momentum equals force)

applied to a control volume in a fluid flow.

It states:

• Rate of increase in momentum within the c.v + net momentum flux out of the c.v.
= force applied to the c.v.

Similar to: conservation of mass in §2.2

Two distinct classes of force act on the fluid particles within the c.v.

(i) Body forces that act on all the fluid within the c.v. Here the (ii) Surface forces. OR Forces on the surfaces of the c.v..
only body force of interest is the force of gravity. These only act on the control surface because their effect on
the fluid inside the c.v. cancels out. They are always expressed
in terms of stress (force per unit area). Two main types of
surface force are involved namely:

Surface forces
(i) Pressure force. (Thermodynamic) pressure, p, is a stress which always acts perpendicular to the control surface and in the opposite
direction to the unit normal (in other words it always tends to compress the c.v.). Although p can vary from point to point in the
flow field it is invariant with direction at a particular point (in other words irrespective of the orientation of the infinitesimal c.v. the
pressure force on any face will be −pδA where δA is the area of the face) – see Fig 2.3.

Fig 2.3: This is a tetrahedron with unit normals indicated on the faces. The pressure is the same independent of the direction
of the normal.

p depends on velocity v and its derivatives and can be generated in the absence of viscous forces. This is expressed below in Bernoulli’s
equation.

(ii) Viscous forces. In general the viscous force acts at an angle to the face of the infinitesimal c.v., so in general it will have 3 components
acting on each face (one due to a direct stress acting normal to the face and 2 shear stresses acting tangential to the face). As an
example let us consider the top of a cubic infinitesimal c.v. (see Fig. 2.4).

(iii) Other surface forces, e.g. surface tension, can be important in some engineering applications.
Viscous stress
The stress tensor can be expressed in terms of 9 components in matrix form as:
 
σxx σxy σxz
 σyx σyy σyz 
σzx σzy σzz

This is what is termed a second-order tensor – i.e. it is a quantity that is characterized by a magnitude and two directions (c.f. a vector or
first-order tensor that is characterized by a magnitude and 1 direction).
Owing to symmetry, the off-diagonal component of the viscous stress tensor σxy = σyx , σxz = σzx , and σzy = σyz . This is known as
a symmetric tensor. Just as the components of a vector change when the co-ordinate system is changed so do the components of the
stress tensor. In most engineering applications the direct viscous stresses (σxx , σyy , σzz ) are negligible compared with the shear stresses
(σxy , σxz , σyz ). The viscous stress is generated by fluid motion and cannot exist in a stationary fluid.

Fig 2.4 shows how these com-

ponents can be associated with
the surfaces of the c.v.
Consider the action on the xz
plane normal to the vertical or
y direction. The unit normal
to this plane is ey (unit vec-
tor in the y direction) and the
two unit tangential vectors are
ex and ez (the unit vectors in
the x and z directions respec-
tively). In this case, the vis-
cous force would be given by

(σyx ex +σyy ey +σyz ez )δxδz

Similar expressions can be

written for each of the other faces.
The 3D momentum equations
When the momentum equation is applied to an infinitesimal c.v. in 3D the following 3 equations for each of the three components of force
and momentum are obtained.
 
∂u ∂u ∂u ∂u ∂p ∂σxx ∂σxy ∂σxz
ρ + u +v +w = ρgx − + + +
∂t
|{z} ∂x ∂y ∂z |{z} | {z∂x} | {z∂x } ∂y ∂z
| {z } | {z } | {z }
 
∂v ∂v ∂v ∂v ∂p ∂σyx ∂σyy ∂σyz
ρ + u +v +w = ρgy − + + +
∂t
|{z} ∂x ∂y ∂z |{z} ∂y ∂x ∂y ∂z (2.9a, b, c)
| {z } | {z } | {z } | {z } | {z }
 
∂w ∂w ∂w ∂w ∂p ∂σzx ∂σzy ∂σzz
ρ + u +v +w = ρgz − + + +
∂t
|{z} ∂x ∂y ∂z ∂z} | {z
|{z} | {z ∂x } ∂y ∂z
| {z } (iii) | {z } | {z }
(i) (ii) (iv) (v) (vi) (vii)

Following the derivation of the 2D version, it will be your responsibility to identify the terms in the 3D momentum equations.

Insert notes taken from lectures on the derivation in 2D. Includes Figs. 2.5-2.7. 2.5 is an area to which the following are
applied: gravity, pressure and viscous stress.
2.4 Distortion of fluid element by velocity, rotation and strain. Vorticity
In general the transformation comprises the following operations:

(i) Translation – movement from one position to another.

(ii) Dilation/Compression – shape invariant but volume reduces or increases. =0 because we are assuming incompressible flow.
That is the volume remains invariant from one position to another.

(iii) Distortion – change of shape keeping volume invariant.

Distortion can be decomposed into

(1) anticlockwise rotation (2) a shear strain (3) and direct shear
α−β α+β
2 2

Fig 2.8a shows how a shear with a jump

∂u
δu = δy
∂y
across a c.v. δx × δy can be transformed into a transla-
tion+rotation+shear strain. β is the angle of the shear
strain after a time δt.
 Fig 2.8b shows how a shear with a jump
∂v
δv = δx
∂x
across a c.v. δx × δy can also be transformed into a
translation+rotation+shear strain. α is the angle of the
shear strain after a time δt.
Note that the shear strains in Fig 2.8a and Fig
2.8b are the same.

Combining the two simple shears, the result is an anticlockwise rotation through angle (α − β)/2 and a shear strain of angle (α + β)/2
The angles α and β are shear angles.

Figure 2.9 This shows an elemental c.v. ABCD which initially at time t = ti is
square. After time δt has elapsed ABCD has moved and distorted into A0 B 0 C 0 D0 .

Insert notes for determining angles α and β, shear strains γxy , γxz , γyz and
the direct strains xx , yy , zz
Vorticity and Circulation
• What is vorticity?

• What is circulation?

The instantaneous rate of rotation of a fluid element is given by (α̇ − β̇)/2 – see above. This corresponds to a fundamental property of
fluid flow called vorticity which in 2D is defined as
∂v ∂u
ζ = α̇ − β̇ = −
∂x ∂y
3D vorticity is a vector given by
 
∂w ∂v ∂u ∂w ∂v ∂u
ω
~ = (ξ, η, ζ) = − , − , − (2.19a, b, c)
∂y ∂z ∂z ∂x ∂x ∂y
It can be seen that the 3 components of vorticity are twice the instantaneous rates of rotation of the fluid element about the three co-ordinate
axes. Mathematically it is given by the following vector operation

~ =∇×v
ω (2.20)

• Vortex lines can be defined analogously to stream lines as lines that are tangential to the vorticity vector at all points in the flow
field.

• Flow structures can be thought of flow structures as vortices composed of bundles of vortex lines.

• Vorticity and vortex lines in many respects are even more fundamental to understanding flow physics than are velocity and
streamlines.

The total amount of vorticity within any closed circuit C of area A within a flow field resolved in a direction perpendicular to the plane of
the circuit is called the circulation, Γ – see Fig. 2.10. Thus
Z Z
Γ= n·ω~ dA (2.21)
A

Circulation can be regarded as a measure of the combined strength of the vortex lines passing through A. Γ is a measure of the
vorticity flux carried through A by these vortex lines. Owing to Stokes theorem we can also write
Z Z Z Z I
Γ= n·ω ~ dA = n · ∇ × vdA = v · dl (2.22)
C
A A
 Figure 2.10 Circulation about a bundle of vortex lines. The circles around the bundle indicate where the circulation integral might be
determined. As the bundle expands and contracts, the radius for this integral will grow or shrink. But the integral, the circulation,
will remain invariant.

2.5 Relationship between rates of strain and viscous stresses. Newtonian Fluids
In solid mechanics the fundamental theoretical model linking the stress and strain fields is Hooke’s law which states that

STRESS ∝ STRAIN

The equivalent in fluid mechanics is the model of the Newtonian fluid for which it is assumed that

STRESS ∝ RATE OF STRAIN

Major difference between the two models.

• At best Hooke’s law is a reasonable approximation for describing small deformations of some solids, particularly structural
steel.

• The Newtonian fluid model is very accurate or the behaviour of almost all homogeneous fluids, in particular water and air.
Even for large deformations.

• non-Newtonian fluids: It does not give good results for pseudofluids formed from suspensions of particles in homogeneous fluids,
e.g. blood, toothpaste, slurries. Various Non-Newtonian fluid models are required to describe such fluids.
In simple terms the constitutive law for the Newtonian Fluid is given by
   
σxx σxy σxz ˙xx γ̇xy γ̇xz
 σyx σyy σyz  = 2µ  γ̇yx ˙yy γ̇yz  (2.23)
σzx σzy σzz γ̇zx γ̇zy ˙zz

The factor 2 is merely used for convenience so as to cancel out the factor 1/2 in the expressions (2.17) for the rate of shear strains. The
above relationship is sufficient in the case of an incompressible fluid. For a compressible fluid we should also allow for the possibility
of direct stress being generated by rate of change of volume or dilation. Thus we we need to add the following to the right-hand side of
(2.23)  
˙xx + ˙yy + ˙zz 0 0
λ 0 ˙xx + ˙yy + ˙zz 0  (2.24)
0 0 ˙xx + ˙yy + ˙zz
• µ = first coefficient of viscosity = dynamic viscosity

• λ = second coefficient of viscosity.

• ν ≡ µ/ρ = kinematic viscosity

If it is required that the actual pressure p − 13 (σxx + σyy + σzz ) in a viscous fluid be identical to the thermodynamic pressure p then it
is easy to show that

2
3λ + 2µ = 0 or λ=− µ This is often called Stokes hypothesis.
3
• Stokes hypothesis assumes that the bulk viscosity,
2
µ0 = λ + µ ' 0 (2.25)
3
This links the average viscous direct stress to the rate of volumetric strain.
This is still a rather controversial question. Bulk viscosity is of no importance in the great majority of engineering applications but
can be important for describing the propagation of sound waves in liquids and sometimes in gases also.

• Here we are assuming incompressible flow so that

∂u ∂v ∂w
˙xx + ˙yy + ˙zz = + + =0 and (2.23) is valid.
∂x ∂y ∂z

2.6 The Navier-Stokes equations

Insert notes for the final steps in the derivation of the 2D Navier-Stokes equations

The 3D version of the Navier-Stokes equations are given on the next page:
3D version of the Navier-Stokes equations
   
∂u ∂p ∂2u ∂2u ∂2u
ρ ∂t
+ u ∂u
∂x
+ v ∂u
∂y
+ w ∂u
∂z
= ρgx − ∂x
+µ ∂x2
+ ∂y 2
+ ∂z 2
(2.29a)
   
∂v ∂v ∂v ∂p ∂2v ∂2v ∂2v
ρ ∂t
+ u ∂x + v ∂y + w ∂v
∂z
= ρgy − ∂y
+µ ∂x2
+ ∂y 2
+ ∂z 2
(2.29b)
   
∂w ∂p ∂2w ∂2w ∂2w
ρ ∂t
+ u ∂w
∂x
+ v ∂w
∂y
+ w ∂w
∂z
= ρgz − ∂z
+µ ∂x2
+ ∂y 2
+ ∂z 2
(2.29c)

∂u ∂v ∂w
Add incompressibility (2.1) + + =0
∂x ∂y ∂z

N-S equations in cylindrical co-ordinates

Fig 2.11 a illustrates the cylindrical co-ordinate system Fig 2.11b shows 2D slices of the control volume.
Let the velocity components be (u, v, w) corresponding to the (r, φ, z) directions respectively.
∂u v ∂u v 2
 2
1 ∂2u 2 ∂v ∂ 2 u
  
∂u ∂u ∂p ∂ u 1 ∂u u
ρ +u + − +w = ρgr − +µ + − + − + (2.30a)
∂t ∂r r ∂φ r ∂z ∂r ∂r2 r ∂r r2 r2 ∂φ2 r2 ∂φ ∂z 2
 2
1 ∂2v 2 ∂u ∂ 2 v
  
∂v ∂v v ∂v uv ∂v 1 ∂p ∂ v 1 ∂v v
ρ +u + + +w = ρgφ − +µ + − + + + (2.30b)
∂t ∂r r ∂φ r ∂z r ∂φ ∂r2 r ∂r r2 r2 ∂φ2 r2 ∂φ ∂z 2
 2
1 ∂2w ∂2w
  
∂w ∂w v ∂w ∂w ∂p ∂ w 1 ∂w
ρ +u + +w = ρgz − +µ + + 2 2 + (2.30c)
∂t ∂r r ∂φ ∂z ∂z r∂r2 r ∂r r ∂φ ∂z 2
∂u u 1 ∂v ∂w
+ + + =0 (2.31)
∂r r r ∂φ ∂z
N-S equations in spherical co-ordinates

Fig 2.12 shows the spherical coordinate system. Let the velocity components be
(u, v, w) corresponding to the (R, θ, φ) directions respectively.

 
∂u ∂u v ∂u w ∂u v 2 +w2 ∂p
ρ ∂t
+ u ∂R + R ∂θ
+ R sin θ ∂φ
− R
= ρgR − ∂R
+
h 
∂(Rw)
  i (2.32a)
R2
µ
sin θ
∂
∂φ
1 ∂u
sin θ ∂φ
− ∂R
− ∂
∂θ
sin θ ∂(Rv)
∂R
− sin θ ∂u
∂θ
 
∂v ∂v v ∂v w ∂v uv w2 1 ∂p
ρ ∂t
+ u ∂R + R ∂θ
+ R sin θ ∂φ
+ R
− R tan θ
= ρgθ + +
hR ∂θ   i (2.32b)
µ ∂ ∂(Rv) ∂u ∂ ∂(w sin θ) ∂v
R sin θ ∂R
sin θ ∂R − sin θ ∂θ − ∂φ ∂θ
− ∂φ
 
∂w ∂w v ∂w w ∂w uw uw 1 ∂p
ρ ∂t
+ u ∂R + R ∂θ
+ R sin θ ∂φ
+ R
− = ρgφ − R sin
R tan θ
+
h  θ ∂φ   i (2.32c)
µ ∂ 1 ∂ 1 ∂v ∂ 1 ∂u ∂u
R ∂θ R sin θ ∂θ
(w sin θ) − R sin θ ∂φ − ∂R sin θ ∂φ
− ∂R
(Rv)
 
1 ∂ 2 1 ∂ ∂w
(R u) + (v sin θ) + =0 (2.33)
R2 ∂R R sin θ ∂θ ∂φ
2.7 Properties of the Navier-Stokes Equations
It is not necessary to solve the Navier-Stokes equations in order to obtain useful information from them. This is illustrated by following
example:

Example 2.7A – Hydraulic modelling

Suppose we wished to build a hydraulic model of a lowland river system for some civil engineering project. Let L be a representative
dimension of the river, say the width at some representative point where the volumetric discharge rate is typically Q. Let us use the
following to render the variables in eqs. (2.1) and (2.29) dimensionless:

• Reference length L – so let x̄ = x/L etc. • Reference pressure ρŪ 2 – so let p̄ = p/(ρŪ 2 )
• Reference velocity Ū = Q/L2 – so let ū = u/Ū etc.
• Reference time L/Ū – so let t̄ = tŪ /L • Reference acceleration g – so let ḡx = gx /g etc.

If we substitute x = Lx̄, u = Ū ū etc. into the governing equations (2.1) and (2.29) we obtain first
∂~v̄ L ∂~v L ∂ ū ∂v̄ ∂ w̄
= 2 (2.34) so multiplying the we get + + = 0 (2.35a)
∂ t̄ U ∂t Navier-Stokes equations by U2 ∂ x̄ ∂ ȳ ∂ z̄

∂ 2 ū ∂ 2 ū ∂ 2 ū
 
∂ ū ∂ ū ∂ ū ∂ ū Lg gx ∂ p̄ µ
+ ū + v̄ + w̄ = 2 − + + + (2.35b)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ Ū g ∂ x̄ ρLŪ ∂ x̄2 ∂ ȳ 2 ∂ z̄ 2
The other two momentum equations (2.35b,c) are similar. We can see from the form of equations (2.34,2.35a,b) that for a fixed geometry
of river system the dimensionless dependent variables are
Lg µ
(ū, v̄, w̄, p̄) = fns(x̄, ȳ, z̄, t̄; , )
Ū2 ρLŪ
This tells us that in order for the flows in the model to be dynamically similar to those in the prototype, as well as requiring geometric
similarity, we must also require the two dimensionless groups
Lg µ
,
Ū 2 ρLŪ
to be the same for model and prototype. We do not need to study the full system if these groups are the same for model and
prototype.
This is equivalent to the well-known result that the Reynolds number and Froude number, i.e.
ρLŪ Ū
Re = Fr = √
µ Lg
be required to be the same for the model and prototype.
 • In practice the roughness of the wetted walls of the prototype and model also have an important effect on the fluid dynamics.

• Therefore, imposing geometric similarity must also require k/L (where k is the mean surface roughness height) to be the same
for model and prototype.

• In practice it is quite impossible to keep all three invariant in the model and prototype, since this implies that model and
prototype are identical in size.

• Usually, owing to the relatively large roughness elements, the flows in the prototype are almost independent of Reynolds
number, and only the Froude number and k/L are kept the same. Even then it is usually necessary to use a distorted model
in order for it not to be impractically large.

Example 2.7B – Aerodynamic modelling

Let us suppose that we are interested in the aerodynamic forces exerted on a structure without a free surface, far from a surface like the
ground:

• A bridge deck . • Flow around a building. • Flow around an aircraft.

Conditions:
• A free surface are absent • There are no body forces.

• Non-dimensionalise using U∞ , the air speed approaching the structure, in place of Ū , and L can be the width or span of the
structure.

• The only dimensionless parameter is the Reynolds number Re ≡ ρU∞ L/µ.

The momentum equations now take the form:

∂ 2 ū ∂ 2 ū ∂ 2 ū
 
∂ ū ∂ ū ∂ ū ∂ ū ∂ p̄ 1
+ ū + v̄ + w̄ =− + + + (2.36a)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ ∂ x̄ Re ∂ x̄2 ∂ ȳ 2 ∂ z̄ 2
• Ideally want to keep Re invariant between model and prototype to achieve dynamic similarity.

• Not usually practicable to achieve dynamic similarity in aerodynamic model tests.

• Why? If Lm << Lp , to get the same Re we need (U∞ )m >> (U∞ )p , which can rarely be achieved.

• It is rarely possible to use any other fluid but air for the model tests,

• Solution: Reynolds number similarity. Many effects are invariant once high Reynolds number is achieved. Or dependence upon
Reynolds number can be determined experimentally.
Coping with high Reynolds numbers
The N-S equations are deceptively simple in form, but at high Reynolds numbers the resulting flow fields can be exceedingly complex even
for simple geometries. Why?

• For many engineering applications the Reynolds numbers are very large, values well in excess of 106 are commonplace.

• If viscosity is very small (µ/ρ << U∞ L), could one drop the viscous terms on the RHS of eqn. (2.29)?

• This view is mistaken and one never achieves a flow field similar to the inviscid one no matter how high the Reynolds number.

• Numerical codes that claim to be inviscid (µ/ρ = ν = 0) always contain viscous effects due to imperfect numerical methods.

• Turbulent flows are inherently vortical. Viscous effects create nonzero vortex structures, where viscosity balances the
most extreme shear strains and rotation.

• Viscosity dominates exceedingly thin boundary layers adjacent to the body surface. As Re → ∞ the boundary-layer thickness,
δ → 0.
If the boundary layers remained attached to the surface they would have little effect beyond giving rise to skin-friction drag. But:

• Separation: Boundary layers separate from the surface of the body forming complex unsteady vortex structures.

• Either due to the adverse pressure gradient or because they reach the rear of the body or its trailing edge.

Example 2.7c – Flow around a circular cylinder

• Steady laminar flow remains a perfectly good solution to the N-S equations regardless of how large Re is.

• As flow speed increases there is a transition from laminar to turbulent flow.

• This essentially occurs because when the viscous effects become sufficiently weak (i.e., the Re becomes sufficiently large) it becomes
impossible to sustain steady flow.

• That is: the laminar solution becomes unstable to small disturbances when Re reaches a certain critical value that varies widely
depending on the type of flow.

Example:
• Flow around the circular cylinder, which undergoes a series of (symmetry breaking) transitions as the Re ranges from less than 1 to
> 106 – see Fig. 2.12.

• Rather paradoxically, the highly symmetrical flow field at low Reynolds number (Re < 1), when the viscous effects are strong,
looks most similar to the Re → ∞) flow field around the cylinder (although the two streamline patterns are not identical).
 • Two of the transitions, that occur to the flow around the cylinder as Re is increased from 1, are :

(i) The transition from the steady flow with the recirculating wake (Fig. 2.12b) to the unsteady periodic flow – the von
Kármán vortex street (Fig. 2.12c). This happens at Re ' 40. The steady flow becomes unstable to small vortex-like
disturbances. The flow then undergoes a transition to an unsteady vortex flow. But it is important to emphasize that this
unsteady flow is not turbulence.
(ii) Below Re ' 200 the vortex street persists to great distances downstream. Re > 200 the wake becomes unstable to 3D
disturbances. Above Re ' 400 the wake becomes truly turbulent far downstream and as the Re increases the fully
turbulent flow moves closer to the cylinder.

• At very high Re a high degree of symmetry may again be recovered, but it is quite unlike the large-scale symmetry of the inviscid
flow pattern or of Fig. 2.12a. What happens is that at very high Re the fine-scale turbulence becomes locally isotropic.

Further Reading

• Schlichting. Chaps. III, IV

• Tritton. Chaps.3, 5, 7 & 8.

• Anderson. Chaps. 15 & 18.

• Sherman. §2.1-2.6. Chap. 5.

3 EXACT SOLUTIONS OF THE NAVIER-STOKES EQUATIONS
3.1 Introduction
In general, analytical solutions to the Navier-Stokes equations cannot be found. There are, however, a number of special cases where exact
analytical solutions can be found – or at least where the equations can be reduced to a set of ordinary differential equations. A few of these
special solutions are widely used in theoretical modelling for engineering applications. Here we will look at four of them, namely
Couette flow - simple shear flow

Plane Poiseuille flow - fully developed channel flow

Axisymmetric Poiseuille flow - fully developed laminar pipe flow

Hiemenz flow - flow in the vicinity of a stagnation point or line.

The proper names attached to these cases reflect those of the scientists who first discovered the solutions.

Past years
You should have already studied the first three cases – Couette and plane and axisymmetric Poiseuille flows – in Years 1 and 2, so we will
look only briefly at these and rely on the exercises for revision.

3.2 Couette and Plane Poiseuille flows

These are illustrated in Figs. 1a and b and can be analysed with a combined approach. Both are steady flows without body forces
characterized by streamlines that are parallel to the walls. This latter feature implies that ∂p/∂y = 0 and there is only one non-zero velocity
component, u, so that the 2D N-S equations (2.28) and (2.7) reduce to (3.1a,b), whose solution is (3.2)

• For Couette flow the pressure gradient is absent and the flow is driven by moving the top wall

• the boundary conditions are

u = 0 at y = 0 u=U at y = h

• C1 = 0 and C2 = U/h so that

y
u=U (3.3)
h
• Plane Poiseuille flow is driven by a pressure gradient

• the boundary conditions

h
u=0 at y=±
2
so that
 •
dp h2 dp h2
C1 ± C2 b + =0 ⇒ C1 = − , C2 = 0
dx 8µ dx 8µ
• Thus we obtain the well-known parabolic velocity profile

h2 dp 4y 2
u=− (1 − 2 ) (y measured from centreline) (3.4)
8µ dx h

The two solutions can be combined for the case when the top wall of the channel also moves at velocity U , to give

h2 dp y
 
y y
u=U − 1− (y measured from wall) (3.5)
h 2µ dx h h

This class of solutions is illustrated in Fig. 3.2. This sort of flows are found in the theory of hydrodynamic lubrication – see Chap. 9
(ES30D).
For applications of these special flows see Exercise 3.1 where Couette flow is applied to journal bearings and viscometers. Couette flow
is a good approximation to the flow between two concentric cylinders when the gap is small. It is possible to derive an exact solution to
the Navier-Stokes equations for the case when the gap is large – see Ex. 3.2. Case Study 1 is concerned with the use of the combined
Couette and Plane Poiseuille flow to model the flow in hydrostatic bearings. Full details will be given on the web-site and an outline in the
lectures.

3.3 Axisymmetric Poiseuille or fully developed pipe flow

This was studied in some detail in the Yr. 2 course. The solution is found in a very similar way to (3.4), but the starting point is the
axisymmetric form of the N-S equations – see Exs. 2.1 & 2.5. In this way it is possible to derive Poiseuille’s law linking the flow rate Q in
the pipe to pressure gradient and viscosity, namely

πR2  dp 
Q= − (3.6)
8µ dx
See Ex. 3.4. Versions of this law are widely used in engineering. In the form of Darcy’s law, it is the basis for analysing flow in porous
media – see Chap. 5.
3.4 Hiemenz flow – 2D stagnation point flow
An example of a 2D potential flow is a stagnation point near a surface type, as illustrated in Fig. 3.3b. This is turned 90 degrees from the
the flow illustrated in Fig. 3.2a and is a flat plate.

• The flat plate case is not purely of academic interest.

• A potential flow close enough to a curved surface can be approximated as flow over a flat plate.

– Provided that the boundary-layer thickness is much smaller than the radius of the leading-edge (or nose) of a wing.
– All 2D stagnation flows at sufficiently high Reynolds numbers (thinness of boundary layer) behave in a similar way near the
stagnation point.

• Surprisingly, in the plane case, the flow obeys an exact solution of the Navier-Stokes equation. This solution can be used to:

(i) Give estimates of the boundary-layer thickness in the leading-edge region;

(ii) Be used as the starting solution for boundary-layer calculations in the case of 2D bodies with rounded noses or leading edges –
see Ex. 3.5.; and
(iii) Start a boundary-layer calculation for the whole wing.

• There is also an equivalent axisymmetric stagnation flow – see Ex. 3.6.

• The velocity profiles for both cases are depicted in Fig. 3.4.

The flat plate flow:

1. Arrives from along the y-axis. 3. Divides into two streams.

2. Impinges on a flat wall placed at y = 0. 4. And leaves in both directions.

Properties of the stagnation point.
• In the flat plate case the stagnation point is strictly a line on a plane, or a point in a 2D flow.

• Potential flow far from the surface: defined by having no rotation, that is ω
~ = 0.

• The velocity field for an inviscid potential flow far from boundaries is

u = ax v = −ay where a is a const. (3.7)

• Near a boundary the flow must satisfy the no-slip condition at the wall.

– Not only all velocities at the surface y = 0 are zero u = v = 0.

– ∂v ∂u ∂v
But because of incompressibility, =− = 0 that is u=v= =0
∂y ∂x ∂y

Here insert notes taken in lectures.

3.5 Exact solution of the Navier-Stokes equation for Hiemenz or 2D stagnation-point flow.
From Fig. 3.3:

• At the wall, the velocity component perpendicular to the wall, v = 0.

• At a large distance from the wall, v = U∞ ,

(i.e. a uniform and constant flow velocity approaching the wall).

• Therefore v does not vary with x when y = 0 or y → ∞.

• Therefore assume that v depends only on y throughout, i.e. v is a function of y alone.

Accordingly write
v = −f (y), say (3.8)
A minus sign is used in recognition of the fact that the flow is towards the wall and in the negative y direction.
∂u ∂f
From the continuity equation (2.7) = (= f 0 ), so integrating gives u = xf 0 (y) + const (3.9)
∂x ∂y | {z }
=0

The constant of integration is zero because x = 0 coincides with the stagnation point.

• Note: the form of u is similar to the potential flow solution (3.7) (u = ax), with a function of y replacing the constant a.
Make a change of variable
f (y) = βφ(y); η = αy (3.16)
to find a universal form independent of a and ν.
α and β are constants that make eq. (3.13) independent of a and ν.
p √
You will find that : α= a/ν, β = aν (3.17)

The solution for u/Ue = φ0 (where Ue = ax is the potential-flow solution at the edge of the viscous region) is given in Fig. 3.4. φ0 = 0.99
at η = 2.4.pThis can be regarded as the boundary-layer thickness: i.e. the height of the edge of the viscous region above the surface
or δ = 2.4 ν/a.
Note that the boundary-layer thickness does not vary with x. This p is characteristic of stagnation-point flow regions.
In general the boundary layer thickness, or one wall-unit, is defined as ν/dU/dy and
y
the distance from the wall in wall units is y + = p
ν/dU/dy

The corresponding result in the axisymmetric case is fairly similar (see Ex. 3.5).
Further Reading

• Houghton & Carpenter (2003) Aerodynamics for Engineering Students, Ed. 5.

• Schlichting, H. Boundary Layer Theory, 7th Ed., Chap. 5.

• Schlichting, H. & Gersten, K. (2000) Boundary Layer Theory, 8th. Ed., pp. 101 et seq., §5.1 & 5.2.

• Tritton. §9.1 & 9.2.

• Anderson. Chap. 16.

4 INVISCID APPROXIMATION. POTENTIAL FLOW
4.1 Introduction
Since the Reynolds number is very large in many engineering applications, it is natural to expect that we could obtain a good approximation
by setting Re = ∞ in the non-dimensional Navier-Stokes equations (2.34) & (2.35), thereby discarding the viscous terms. We will also drop
the body-force terms. The resulting equations are known as the Euler equations and take the form:
∂ ū ∂v̄ ∂ w̄
+ + =0 (4.1)
∂ x̄ ∂ ȳ ∂ z̄
∂ ū ∂ ū ∂ ū ∂ ū ∂ p̄
+ ū + v̄ + w̄ =− (4.2a)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ ∂ x̄
∂v̄ ∂v̄ ∂v̄ ∂v̄ ∂ p̄
+ ū + v̄ + w̄ =− (4.2b)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ ∂ ȳ
∂ w̄ ∂ w̄ ∂ w̄ ∂ w̄ ∂ p̄
+ ū + v̄ + w̄ =− (4.2c)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ ∂ z̄
These equations are used for some engineering applications, particularly in aerodynamics. However, in many respects, the 3D Euler
equations are really as difficult to solve as the Navier-Stokes equations. Furthermore it is not certain that well-behaved solutions always
exist.
Since distortion (shearing plus rotation) of fluid elements can only come about by the effects of viscosity (see §2.4 and §2.5), the vorticity
must be zero in a truly invisicid flow, i.e. the flow must be irrotational. So use of (2.19) gives
∂w ∂v ∂u ∂w ∂v ∂u (4.3)
ω
~ =0 or − = − = − =0
∂y ∂z ∂z ∂x ∂x ∂y
Equations (4.3) can be automatically satisfied by the introduction of a velocity potential Φ, such that
Equations (4.3) can be automatically satisfied by the introduction of a velocity potential Φ, such that
∂Φ ∂Φ ∂Φ
u= , v= , w= (4.4)
∂x ∂y ∂z
When (4.4) is introduced into the continuity equation (2.1) we get the Laplace Equation

∂2Φ ∂2Φ ∂2Φ

+ + 2 =0 (4.5)
∂x2 ∂y 2 ∂z
• Equation (4.5) is linear.

• Linear superposition can be used to build up complex solutions from special simple solutions

• such as the source, doublet, and potential vortex.

• This was covered in some detail for two-dimensional potential flows in the Yr. 2 course and will not be repeated here.
The Laplace equation is the canonical example of an elliptic partial differential equation.

• Physically this implies that any point in the flow field is influenced to some extent by at all other points.

• This is also true of the steady Navier-Stokes equations.

• Unlike the Navier-Stokes equations the Laplace equation is unaffected when we replace x, y or z by −x, −y or −z.

• Thus for symmetric geometries like the circular cylinder we would expect a symmetric flow field (see Fig. 4.1) for potential flow

• but for real flow which obeys the N-S we do not expect a symmetric flow field. equations.

As explained in §2.7:

• Thin regions of vorticity originate due to the effects of viscosity in the boundary layer and tend to form complex vortical structures.

• For this reason potential flow can never be a good approximation for most real flow fields at high Reynolds numbers.

• It is a particularly poor approximation for bluff bodies like the circular cylinder.

• For non-lifting streamlined bodies, on the other hand, potential flow can often be a good approximation,

• especially when combined with boundary-layer theory to account for the viscous effects acting on the body surface.

The production of lift:

• Always accompanied by the formation of trailing vortex wakes.

• Nevertheless, by the use of the concept of bound vorticity with strength determined by the Kutta condition, reasonably accurate
potential flow solutions can be found for streamlined lifting bodies such as wings.

• These can be used to predict the pressure field and thereby the lift characteristics.

• They can also be used in conjunction with boundary-layer theory to predict some of the viscous effects.
 The numerical methods available for aerodynamic design:

• Can compute the Laplace equation for the flow around whole aircraft.

• These methods are based on discretizing the aircraft surface into a large number of rectangular panels Panel Methods.

• Each panel carries a distribution of sources.

• These can be combined with distributions of vortices or doublets, where necessary, to give the aerodynamic characteristics for the
lifting part of the aircraft.

Further Reading

• Tritton. Chap 10.

• Anderson. Part II esp. Chap. 3

• Houghton & Carpenter. Chap 3. Fig. 4.1 Potential flow around a circular cylinder

5 FLOW IN POROUS MEDIA

5.1 Introduction
Flows in porous media are widespread in engineering. We will focus primarily on:

• the flow of groundwater – a Civil Engineering application.

• Porous media: imagine as many myriad, tortuous, capillary-like channels – see Fig. 5.1.

• Many channels means many boundary layers: viscous effects are strong.

• Therefore: the inertial terms of the Navier-Stokes equations can be neglected.

• The theoretical foundation is largely based on a special solution of the Navier-Stokes equations, namely Poiseuille flow (§3.3).

5.2 Physical laws relating pressure and flow rate

Adapt Poiseuille’s law (3.6). Variables:

• the average flow rate (ū) per unit area

• average flow rate (ūp ) through a pore or channel

 • porosity  = Vv /(Vs + Vv ) (where Vs and Vv are the volumes of solids and voids respectively)

•
Re2 ∆p∗
∆p∗ = Change of piezometric pressure ū = ūp ūp =
8µ `e
• Re and `e are the equivalent radius and length of the capillary-like channels. µ is viscosity and the factor 8 comes from
averaging over the area of imaginary small pipes.

• p∗ is the piezometric pressure, p∗ = p + ρgz (where z is vertical height above an arbitrary horizontal datum) instead of pressure, since
body forces are important in groundwater flows and can most easily be taken into account by using piezometric pressure.

Result:
Re2 ∆p∗
ū = ūp =  (5.1)
8µ `e
One approach is simply to write
∂p∗
ū = −C Darcy0 s Law (5.2)
∂x
where C is a constant at a particular temperature for a specific combination of porous medium and fluid. However, it is possible to derive
a more precise expression for C. There have been many attempts to do so. A widely used result is given by

∆p∗ 3 Vs2 `
ū = : Kozeny − Carman equation (5.3)
` (1 − )2 µS 2 2`e
|{z}
k

k is often known as the Kozeny constant or function. Evidently in this case

3 Vs2
C= k (5.4)
(1 − )2 µS 2

Derivation of Eqs. (5.3) and (5.4)

The key question is how to choose the equivalent radius Re so that we can adapt eq. (5.1) for a porous medium. By definition
Vv Vs 
≡ Therefore Vv = (5.5)
Vv + Vs 1−
Vv Vs 
So = (5.6)
S S 1−
where S is the total surface area. For a capillary tube of radius R and length `

Vv πR2 ` R Vv Vs 
= = Therefore Re = 2 =2 (5.7)
S 2πR` 2 S S 1−
 Substituting (5.7) into eq. (5.1) gives eq. (5.3) and hence (5.4).
Evidently the porosity  is an important property for determining the flow rate per unit area, ū. Most granular materials have values
between 0.3 and 0.5. For packings of spheres of similar sizes  lies between 0.4 and 0.55. Kozeny’s function k depends to some extent on 
and depends also on the arrangement of particles, their shape and their size range. Nevertheless, experiment shows that for many porous
media k ranges only between 4.0 and 6.0 and an average value of 5.0 is commonly adopted.

5.3 Groundwater flow equations

To fix ideas consider Fig. 5.2 which shows an unconfined aquifer. We wish to derive the equations which govern the motion of groundwater
in such a system. In many respects the derivation is very similar to that used to obtain the continuity equation (2.7). The final 3D equation
takes the form
 ∂ 2h ∂ 2h ∂ 2h  ∂h
C 2
+ 2
+ 2 = Se (5.8)
∂x ∂y ∂z ∂t
Here, as is common in Civil Engineering applications, the piezometric pressure p∗ is expressed in terms of head h = p∗ /(ρg). Se is the
specific storage which is defined as the volume of water per unit volume of porous medium that is released by a unit change in head.

Please take notes in class on the derivation of the 2D version of eq. (5.8). This includes Fig. 5.3.

Equation (5.8) is a famous equation in physics and engineering. It is known as the diffusion or heat conduction equation. For steady
flow eqn. (5.8) reduces to another famous equation – the Laplace equation. The equation for potential flow. Equation (5.8) is comparatively
easy to solve computationally – especially the steady-state form of Laplace’s equation – and fairly flexible commercial packages are available
for this.

Boundary conditions
At impermeable boundaries the flow rate must be zero which implies that ∂p∗ /∂n = 0 or ∂h/∂n = 0 where n is the direction normal to the
impermeable boundary. At other boundaries the flow rate could be specified.
5.4 Hele-Shaw Flow
Until quite recently steady 2D groundwater flow problems (e.g. see Fig. 5.4) were solved by using a Hele-Shaw apparatus. Modern computers
have made this analogue method obsolete. But, owing to its intrinsic interest and connection with lubrication theory and creeping flow, it
is worth a brief digression.
Figure 5.4 A Hele-Shaw apparatus comprised of two
glass walls separated by a thin film of fluid. A 2D profile
of the dam, or whatever is under investigation, would be
inserted and a pressure difference created between one
side and another to produce a fluid flow.

Providing the fluid film thickness h is sufficiently small, it can be shown from the Navier-Stokes equations that to a good approximation
the pressure satisfies the Laplace equation :
∂2p ∂2p
+ =0
∂x2 ∂z 2
In analogy with eq. (3.6) we can write
h2 ∂p h2 ∂p
ū = − and w̄ = −
12µ ∂x 12µ ∂z
where, in keeping with the current notation, we have introduced

1 h 1 h
Z Z
ū = udy, w̄ = wdy
h 0 h 0
Compare this with potential flow for which

∂Φ ∂Φ h2
ū = & w̄ = Evidently then, − p≡Φ
∂x ∂z 12µ
and Hele-Shaw flow is analogous to 2D potential flow, including steady groundwater flows. The flow or stream lines obtained by injecting dye
into Hele-Shaw flows correspond to those in the ‘real’ 2D groundwater flows. 12 is a geometric factor derived in my supplementary
notes

Insert notes for derivation of ū and w̄ for Hele-Shaw flow

5.5 Application to non-steady flow to wells

Figure 5.5. A well that has been sunk to a confined aquifer. T = Kb = Cb is t
Law. b is the depth of the aquifer. h0 is the inital available depth. h is the dep
up the well. The velocity of the flow up the well will not matter in this proble
very small (a spring). S = Se b is the storage, where Se is the specific storage t

If the well is pumped at a steady rate, in order for flow to occur towards the well according to Darcy’s law:

• There must exist a gradient of head (of pressure) towards the well.

• The resulting water-table form is called a cone of depression.

• The decrease in water level is called the drawdown.

• It is desired to:

– calculate the drawdown h0 − h as a function of time and radial distance from the well
– when a steady flow rate Q is drawn out of the well.

• It will be assumed:
 – that the rock surrounding the well is homogeneous, – the confined aquifer of constant vertical depth b.
– the terrain flat, and

• By simplifying the situation to one of radial symmetry and use cylindrical co-ordinates, eqn. (5.8) simplifies to (see EX5.1)
 
1 ∂ ∂h ∂h
T r =S where T = Cb and S = Se b (5.13)
r ∂r ∂r ∂t

• These quantities are respectively called the transmissivity and storage coefficient of the aquifer. S is dimensionless and T has the
dimensions of flow rate per unit width.

Insert notes for the the derivation of Eq. (5.13), the 1D axisymmetric version of the groundwater flow equation
 u 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
×1 0.219 0.049 0.013 0.0038 0.0014 0.00036 0.00012 0.000038 0.000012
×10−1 1.82 1.22 0.91 0.70 0.56 0.45 0.37 0.31 0.26
×10−2 4.94 3.35 2.96 2.68 2.48 2.30 2.15 2.03 1.94
×10−3 6.33 5.64 5.23 4.95 4.73 4.54 4.39 4.26 4.14
×10−4 8.63 7.94 7.53 7.25 7.02 6.84 6.69 6.55 6.44

Table 5.1 Values of the well function W(u) for various values of u. [Source: Adapted from E.M. Shaw (1994) Hydrology in Practice, 3rd
Ed., Chapman & Hall]

Boundary conditions
Draw an imaginary circle radius r around the well then, if b is the vertical depth of the aquifer,
∂h
Q = −2πrbū = 2πr |{z}
Cb as r → 0 (5.14)
∂r
T
The other boundary condition is straightforward, i.e.

h = h0 as r → ∞ (5.15)
This is a classic problem found in many other branches of engineering and physics.

Insert notes for the solution to Eq. (5.13) for well flow

η2 r2 S Q
If u = = one gets h0 − h = W (u) : Thiess equation (5.23)
2 4T t 4πT
Where the function W (u) is plotted in Table 5.1 and is known as the well function
The Thiess equation can be readily used to determine the drawdown for specific values of r and t provided values of S and T are known.
This is demonstrated in Example 5.1. However, this is not how it is normally used in practice. For instance, it will be noted from the
Thiess equation that a steady state is never reached. The more practical use of the eq. (5.23) is to determine the transmissivity
and storage coefficient from measurements of drawdown see Ex5.3. The solution for the drawdown is given by
Z ∞ −u
Q r2 S e
h0 − h(r, t) = W (u); where u= and W (u) = du (5.23)
4πT 4T t u u
Equation (5.23) is known as the Thiess equation and W(u) is the Well function; it is tabulated in Table 5.1.

Design Example 5.1

To calculate the drawdown in a confined aquifer at a distance of 25 m from the well after 6 hours of pumping water at a constant discharge
of 0.0311 m3 /s. The storage coefficient and transmissivity of the aquifer are S = 0.005 and T = 0.0092 m2 /s.
 Here
r2 S 252 × 0.005
u= = = 3.93 × 10−3
4T t 4 × 0.0092 × (6 × 3600)
So Table 5.1 gives approximately
4 − 3.93
W (u) = 4.95 + × (5.23 − 4.95) = 4.97
−3} |
| 4 {z
{z }
0.28
0.07

Therefore the drawdown is given by

Q 0.0311
h0 − h(r, t) = W (u) = × 4.97 = 1.337 m
4πT 4π × 0.0092
The figures overleaf are taken from:

• E.M. Shaw (1994) Hydrology in Practice, 3rd. Ed., Chapman & Hall.

• M.S. Yalin (1971) Theory of Hydraulic Models, MacMillan.

Other further Reading

• Massey. §6.8.

• Cedergren, H.R. (1967) Seepage, Drainage, and Flow Nets, Wiley.

• Freeze, R.A. & Cherry, J.A. (1979) Groundwater, Prentice-Hall.

• Harr, M.E. (1962) Groundwater and Seepage, McGraw-Hill

• Raudkivi, A.J. & Callander, R.A. (1976) Analysis of Groundwater Flow, Edward Arnold.
6 VORTICES & VORTICITY
6.1 Introduction
A dominant feature of flows around bodies at high Reynolds number 1 is that the wake forms unsteady vortical structures or vortices.
These tend to develop into complex 3D forms, of which turbulent flow is the extreme manifestation. These vortical forms are of immense
importance in engineering applications since they determine how the forces acting on a body vary with time. They also lead to strong
departures from potential flow at high Reynolds numbers. Finally, they are what makes fluid dynamics such a challenging but difficult
subject. In particular, they make accurate computational simulations of engineering flow fields very difficult, if not impossible. Here we
will briefly consider some of the well-known physical phenomena connected with the formation of vortical structures.

6.2 Von Kármán Vortex Street

Fig. 2.12 illustrates how the flow pattern around a circular cylinder varies with Reynolds number. Here it can seen that for Re > 5 the
flow separates to form a steady wake comprising of two symmetric trapped vortices. As the Reynolds number rises these vortices elongate.
When Re exceeds about 40 this pattern becomes unstable and steady flow is no longer possible. It is supplanted by the von Kármán vortex
street – see Fig. 6.1 – whereby vortices are shed periodically from either side of the cylinder.
What do we mean by unstable? Perfectly good solutions to the steady Navier-Stokes equations can be obtained computationally for
Re > 40 showing even longer steady trapped vortices. What happens in the physical world is that the flow fields corresponding to these
solutions are unstable to small disturbances. Any background noise that is inevitably present will give rise to small disturbances. These
will grow and be transformed in the process into a more stable unsteady flow pattern. In fact, the precise staggered arrangement of the
Kármán vortex street is the most stable possible. When the Reynolds number reaches 200 or so, the 2D vortices beome unstable and a 3D
pattern takes over. Turbulent flow which is also a result of instability follows soon afterwards. However, no matter how high the Reynolds
number the dominant frequency of the wake flow remains that of the 2D vortex street – see Fig. 6.1b. This is usually characterized by a
nondimensional frequency known as the Strouhal number, namely
ωd
St ≡ (' 1.3 for vortex street) (6.1)
U∞
(ω is in rad/s). Sometimes the dominant vortex-shedding frequency is known as the Strouhal frequency. All bluff bodies produce vortex
streets in a similar way to the circular cylinder.

6.3 Helmholtz instability

This occurs when two uniform streams flow past each other in opposite directions – see Figs. 6.2 and 6.4. The fact that such a flow is
unstable to small disturbances can be deduced from Bernoulli’s equation – see Fig. 6.3. In fact, it is not necessary for the two uniform
streams to be in opposite directions since any uniform flow could be superposed and leave the arguments unchanged (Fig. 6.5). Since sharp
discontinuities of velocity cannot exist in nature because viscous effects smooth them out, the flows susceptible to Helmholtz instability are
characterized by velocity profiles with points of inflexion (Fig. 6.6).
1
In practice, this implies Re greater than about 40 or so.
 The Helmholtz instability can also be explained in terms of vorticity. Consider a 2D (potential) vortex (Fig. 6.7) then the velocity
induced by the vortex at another point P is given
Γ 1
vi = (6.2)
2π r
It always acts perpendicular to the line of length r joining the vortex centre to P .
Real vortices have a small viscous core (see Fig. 6.8) within which the velocity varies approximately linearly (solid-body rotation).
Outside of the viscous core the velocity field is given by eq. (6.2).
It turns out that the flow depicted in Fig. 6.2 can be simulated by a row of point vortices (see Fig. 6.9). We now consider what would
happen if this row of vortices were disturbed as shown in Fig. 6.10. By considering the velocities induced on each other by the neighbouring
vortices we can see that there is a tendency for neighbouring vortices to rotate around each other. This is the alternative (vorticity-based)
explanation for the Helmholtz instability.

6.4 3D vortices – Biot-Savart law

Helmholtz instability is found in 3D flows. In these cases the resulting vortical forms tend to be oriented more in a streamwise direction
and often form spiral vortices. In 3D the velocity induced by a small length δ~` of vortex filament (or line) is given by (see Fig. 6.11)

Γ δ~` × (r − s)
δVi (r) = Biot − Savart Law (6.3)
4π |r − s|3
Essentially this states that the velocity induced by the infinitesimal length of vortex is perpendicular to the length and proportional to the
inverse square of the distance between the length of vortex and the point in question.

6.5 Vortex systems associated with lift generation

Lift generation has to accompanied by the generation of circulation. Thus when an aerofoil starts to move through a fluid, at the very
first instant of time the flow appears similar to potential flow with no circulation (Fig. 6.12). Over a time scale determined by viscosity
2
(∼ ν/U∞ ) the circulation develops and the bound vorticity is generated with the flow leaving the trailing edge smoothly, thereby satisfying
the Kutta condition. However, circulation is conserved (at least approximately) and initially it was zero. Accordingly a starting vortex (Fig.
6.13) is generated simultaneously with the bound vorticity and opposite in sign. This starting vortex is carried away downstream by the
flow. Every time the lift force changes, owing to a change of angle of incidence or flow speed, say, a new starting vortex is generated. So
in the case of unsteady lift-generation processes, such as for bird and insect flight, starting vortices are continuously generated. When the
starting vortex is close to the trailing edge, the fact that it is opposite in sign to the bound vortex tends to reduce the lift that is generated.
For wings, as distinct from the 2D aerofoil flows, the wake takes the form of a vortex sheet comprising vortex filaments of varying strength.
The individual filaments roll up to form two trailing vortices (Fig. 6.14) within a few chord lengths. These vortices are responsible for
inducing a flow field that generates additional drag – known as induced drag.
For conventional wings, the trailing vortices are not an essential part of the lift-generation process, but, rather, an undesirable by-
product. For unconventional low-aspect-ratio wings, such as delta wings, the flow separates from the leading edges to form two powerful
vortices over the wing (Fig. 6.15). The low-pressure regions generated by these vortices are directly responsible for the lift generation. For
delta wings and similar configurations stall, as such, does not occur. What happens at high angle of incidence is that the core of the vortices
‘burst’. This phenomenon is known as vortex breakdown (Fig. 6.16). Initially vortex breakdown is located far downstream in the wake.
As the incidence rises the position of vortex breakdown moves upstream. When it reaches the trailing edge of the wing the lift-generation
process begins to be affected.
The vortex systems for real engineering applications can be very complex. As an example, Fig. 6.17, gives a simplified schematic
illustration of the flow around a car.

6.6 Image vortices

The effect of solid surfaces on vortices can be estimated by the use of image vortices. Judicious location of these image vortices ensures
that the boundary condition of zero normal velocity can be satisfied (Fig. 6.18).

6.7 The governing equation for vorticity

It is possible without great difficulty to derive a system of equations governing vorticity from the Navier-Stokes equations. In 2D there is
only one component of vorticity (aligned in the spanwise direction) and in this case the governing equation takes the form
Dζ
= ν ∇2 ζ (6.4)
Dt |{z}
∂2ζ 2
|{z}
∂ζ ∂ζ ∂ζ + ∂ 2ζ
+u ∂x +v ∂y ∂x2 ∂y
∂t

The viscous term on the RHS represents the diffusion (spreading) of vorticity by viscous (or molecular) action. One of the consequences of
viscous diffusion is vortex reconnection which is depicted in Fig. 6.19. This shows two vortex filaments which are initially well separated.
The vortex cores spread in size owing to diffusion and eventually touch, after which reconnection ensues as depicted.
In the absence of viscosity, eqn. (6.4) states that the vorticity is merely transported by the velocity field. Thus circulations is conserved,
as mentioned above in connection with lift generation. However, in 3D flows there is an additional inviscid effect, known as vortex stretching,
which can strengthen the vorticity. For 3D flows the vorticity equation can be written in abbreviated form using vector notation:
D~ω 2
~ ·{z∇v} + ν∇
= |ω | {z ω
~} (6.5)
Dt
I II

In 3D the vorticity ω
~ = (ξ, η, ζ) is a vector and Eqn. (6.5) represents a system of 3 equations. Term II is the viscous diffusion term, as
with 2D flow. Term I represents the rate of increase in vorticity by vortex stretching. The z-component of II takes the form:
∂w
(~ω · ∇v)z = ζ
∂z
If ∂w/∂z > 0 then the fluid element or vortex filament is elongating in the direction of the vorticity. Continuity demands that it contracts
in the x and y directions. This leads to faster rotation and increased vorticity. This is thought to be the dominant mechanism for generating
the fine scales of turbulence.
Source of Figures
 • Figs. 6.1, 6.12-6.16 are taken from Houghton & Carpenter.

• Fig. 6.3 is taken from Album of Fluid Motion.

• Fig. 6.11 is taken from K. Karamcheti Principles of Ideal -Fluid Aerodynamics, Wiley, 1966.

• Fig. 6.17 is taken from W-H. Hucho & G. Sovran 1993 Aerodynamics of Road Vehicles. Annual Review of Fluid Mechanics 25,
pp.485-537.

• Fig. 6.18 is taken from Sherman.

Further Reading

• Tritton. §3.3, §6.4-6.6, Chap. 13, §17.6

• Anderson. §2.12-13, §3.18, §4.5-6.

• Schlichting. IIb