Chapter 1

Introduction

This set of notes summarises the main results of the lecture ‘Viscous Fluid Flow’ (MATH35001). Please
email any corrections (yes, there might be the odd typo...) or suggestions for improvement to M.Heil@maths.man.ac.uk.
Alternatively, see me after the lecture or in my office (Room 2.224 in the Alan Turing building).
Generally, the notes will be handed out after the material has been covered in the lecture. You can
also download them from the WWW:
http://www.maths.man.ac.uk/ ˜mheil/Lectures/Fluids/.
This WWW page will also contain announcements, example sheets, solutions, etc.

1.1 Literature
The following is a list of books that I found useful in preparing this lecture. It is not necessary to
purchase any of these books! Your lecture notes and these handouts will be completely sufficient.
Acheson, D.J. 1990 Elementary Fluid Dynamics. Clarendon Press, Oxford, 1990.
Spiegel, M. 1974 Vector Calculus. McGraw Hill (Schaum’s Outline series).
Batchelor, G.K. 1967 An Introduction to Fluid Dynamics. Cambridge.
Sherman, F.S. 1990 Viscous Flow. McGraw Hill.

McCormack , P.S. & Crane, L.J. 1973 Physical Fluid Dynamics, Academic Press.
Panton, R.L. 1996 Incompressible Flow (second edition), Wiley.
White, F.M. 1991 Viscous Fluid Flow (second edition), McGraw Hill.

1.2 Preliminaries: Index notation & summation convention

• We will denote vectors/matrices/tensors by their components relative to a set of basis vectors. E.g.
instead of writing r = r1 i + r2 j + r3 k = r1 e1 + r2 e2 + r3 e3 = (r1 , r2 , r3 ), we simply write ri and use
the convention that all ‘free indices’ (i in this case) range from 1 to 3. Similarly, we represent the
matrix  
a11 a12 a13
A =  a21 a22 a23 
a31 a32 a33
by its generic component aij and imply that the two free indices (i and j) take on all values in the
range from 1 to 3.
• In general, we only write down the generic term of any vector (or vector equation). For instance
ai = bi + ci is taken to represent the three components a1 = b1 + c1 , a2 = b2 + c2 , a3 = b3 + c3 of
the symbolic vector equation a = b + c.

1
MATH35001 Viscous Fluid Flow: Introduction 2

• Consistency check: Every term in an equation in index notation has to have the same number of
‘free indices’. For instance, the addition of two matrices can be expressed as Aij = Bij + Cij ,
whereas the equation Aij = Bik + Clm does not make sense.

1 for i = j
• Kronecker Delta: δij =
0 for i 6= j
• Summation convention: Automatic summation over repated indices. Examples are:
P3
Dot product: a · b = i=1 ai bi . Sums like this will occur very frequently and it will turn out to
be convenient
P3 to drop the summation sign and to automatically sum over any repeated index.
I.e. i=1 a i b i = ai bi = ak bk . Note that the ‘name’ of the summation index is irrelevant as it
does not appear in the final result; therefore ai bi is the same as ak bk . Summation indices are
often called ‘dummy indices’.
Matrix-vector products: A · x = b becomes Aij xj (or Aim xm = bi , say). Similarly AT · x = c
becomes Aji xj = ci (or Ajk xj = ck , say). Note that the result of the matrix-vector product is
a vector: Hence both sides of the equations have one (matching!) free index.
δij ‘exchanges’ indices: ai δij = aj .
∂ui
• Comma denotes partial differentiation: E.g. ∂xj = ui,j .

• Some differential operators in index notation:

∇ · u = div u = ui,i (1.1)

∇φ = gradφ = φ,i (1.2)

∇2 φ = φ,ii (1.3)
Chapter 2

The Kinematics of Fluid Flow

2.1 The Eulerian flow field

• Eulerian description of the flow field: The velocity u is given as a function of the position relative
to a spatially fixed coordinate system (x, y, z) = (x1 , x2 , x3 ) = xi , and of time t.
u = u(x, y, z, t) or in index notation: ui = ui (xj , t). (2.1)

• Note that at different times, different material particles will be at a given spatial position. The
(0)
particle paths (i.e. the trajectories xpi (t) of individual material particles which are at position xi
at time t = t0 ) are obtained by integrating
∂xpi (t)
= ui (xpj , t) (2.2)
∂t
subject to the initial conditions
(0)
xpi (t = 0) = xi . (2.3)

2.2 The material derivative

• The acceleration ai of the material particle that is at position xj at time t is given by
 
d ∂ui ∂ui ∂xpk (t)
ai (xj , t) = ui (xpj (t), t) = + . (2.4)
dt xp (t)=xj ∂t ∂xk ∂t
j

Comparing this to (2.2) shows that this can be written as

∂ui ∂ui ∂u
ai = + uk or symbolically a= + (u · ∇)u. (2.5)
∂t ∂xk ∂t
• The differential operator
D ∂ ∂ D ∂
= + uk or symbolically = + (u · ∇) (2.6)
Dt ∂t ∂xk Dt ∂t
is known as the ‘material (or substantial) derivative’. Given any function φ(xj , t), Dφ/Dt represents
the rate of change of φ experienced by an observer travelling with the velocity u i (xj , t).

2.3 Vorticity and the rate of strain tensor

• The velocity field can be decomposed into four fundamental ‘modes’ which correspond to the trans-
lation, rotation, shearing and dilation of small material elements contained in the flow. The velocity
in the vicinity of a certain point xk can be expressed as
ui (xk + δxk ) = ui (xk ) + ωij δxj + ij δxj , (2.7)
| {z } | {z } | {z }
rigid body translation rigid body rotation shearing and dilation

3
MATH35001: Viscous Fluid Flow: The Kinematics of Fluid Flow 4

where ωij is the antisymmetric rate of rotation tensor

 
1 ∂ui ∂uj
ωij = − (2.8)
2 ∂xj ∂xi

and ij is the symmetric rate of strain tensor

 
1 ∂ui ∂uj
ij = + . (2.9)
2 ∂xj ∂xi

• The first term in (2.7) represents a rigid body translation: If ij = ωij = 0 then all particles have
the same velocity, i.e. the fluid moves in a straight line as a rigid body.
• The physical meaning of the second term in (2.7) is revealed by rewriting ωij δxj symbolically as
a cross product in the form Ω × δx where Ω = (ω32 , ω13 , ω21 ) is the rate of rotation vector.
This is illustrated in Fig. 2.1: The differential velocity δu = u(xj ) − u(xj + δxj ) induced by a rigid
body rotation about point P with rotation rate Ω is given by δu = Ω × δx δx.

Ω X δx

P’
Ω
δx
P

Figure 2.1: Sketch illustrating the motion induced by a rigid body rotation about point P with rotation
rate Ω . In this sketch the rate of rotation vector Ω points vertically out of the paper.

The rotation rate Ω is equal to half the vorticity ω , i.e.

 
( ∂u
∂x2 −
3 ∂u2
∂x3 )
 ∂u ∂u3 
2 Ω = ω = curl u = ∇ × u =  ( ∂x3 −
1
∂x1 )  (2.10)
( ∂u
∂x1 −
2 ∂u1
∂x2 )

• The diagonal entries of the rate of strain tensor ij represent the extensional rate of strain in the
direction of the three cartesian coordinate axes, as illustrated in Fig. 2.2, e.g. Ds 1 /Dt = e11 =
∂u1 /∂x1

δs2(t+δ t)
δs2
x2

δ s1 = δ x 1 δ s1(t+δ t) = δx1 +du1/dx1 δ x1 δ t

x1

Figure 2.2: A rectangular block of fluid undergoes a purely extensional deformation which changes the
lengths of the material lines parallel to the coordinate axes.

• The off-diagonal entries of the rate of strain tensor ij represent the shear rate of strain (in fact,
they are equal to half the shear rate in the appropriate directions; see Fig. 2.3).
MATH35001: Viscous Fluid Flow: The Kinematics of Fluid Flow 5

x2 γ

x1

Figure 2.3: Sketch illustrating the shearing of an initially rectangular block of fluid at a rate Dγ/Dt =
2 e12 = (∂u1 /∂x2 + ∂u2 /∂x1 ).

2.4 The equation of continuity

• Mass conservation requires that the rate at which mass is transported over the surface ∂V of a
spatially fixed volume V must be equal to the rate of change of mass in this volume. This physical
statement can be formulated in an integral or a differential form:
• The integral form of the equation of continuity is given by
Z I
dρ
dV + ρui ni dS = 0, (2.11)
V dt ∂V

or in symbolic form Z I
dρ
dV + ρu · n dS = 0, (2.12)
V dt ∂V

where ρ is the density of the fluid (i.e. the mass per unit volume), and n is the outer unit normal
on the surface ∂V of the spatially fixed volume V (note that u · n < 0 corresponds to an inflow).
• The corresponding differential form of the equation of continuity can be derived by applying the
integral statement to an infinitesimally small block of fluid. The result is

∂ρ ∂(ρui )
+ = 0. (2.13)
∂t ∂xi
Using the material derivative introduced in (2.6), this expression can be rewritten as

Dρ ∂ui
+ρ = 0. (2.14)
Dt ∂xi

• The latter equation shows that for incompressible fluids (i.e. fluids for which the density of material
fluid elements is constant and thus Dρ/Dt = 0), the equation of continuity presents a purely
kinematic constraint on the velocity field, namely
∂ui
=0 (2.15)
∂xi
or in symbolic form
div u = 0 or ∇ · u = 0. (2.16)
Chapter 3

Stress, Cauchy’s equation and the

Navier-Stokes equations

3.1 The concept of traction/stress

• Consider the volume of fluid shown in the left half of Fig. 3.1. The volume of fluid is subjected to
distributed external forces (e.g. shear stresses, pressures etc.). Let ∆F be the resultant force acting
on a small surface element ∆S with outer unit normal n, then the traction vector t is defined as:
∆F
F
t = lim (3.1)
∆S→0 ∆S

∆F


 

 

 

 

 

 

 

 

 

 

n

 

 

  












































 ∆F 

 

 

 

 

 

 

 

 




 

 

 

 

 

 

 

 

 

 

 

 

  































 

 

 

 

 

 

 

 

 




 

 

 

 

 

 

 

 

 

 

 

 

  















 

 

 

 

 

 

 



































 

 

 

 

 

 

 ∆ 
 
 S
 
 
 
 
  












































 



 

 

 

 

 

 

 


 

 

 

 

 

 

 


 


 


 


 


 


 


  










































∆

 S n 

 

 

 

 

 

 

 

 














 
 
 
 
 
 
 
 
 
 
 
 
 
  
























Figure 3.1: Sketch illustrating traction and stress.

• The right half of Fig. 3.1 illustrates the concept of an (internal) stress t which represents the
traction exerted by one half of the fluid volume onto the other half across a ficticious cut (along a
plane with outer unit normal n) through the volume.

3.2 The stress tensor

• The stress vector t depends on the spatial position in the body and on the orientation of the plane
(characterised by its outer unit normal n) along which the volume of fluid is cut:

ti = τij nj , (3.2)

where τij = τji is the symmetric stress tensor.

• On an infinitesimal block of fluid whose faces are parallel to the axes, the component τ ij of the
stress tensor represents the traction component in the positive i-direction on the face x j = const.
whose outer normal points in the positive j-direction (see Fig. 3.2).

6
MATH35001 Viscous Fluid Flow: Stress, Cauchy’s equation and the Navier-Stokes equations 7

x3 x3
 


 
 
 
 
 
 
 
 
 
 
 
 



































 



 



 



 



 



 



 



 



 



 



 



 
  
  
  
  
  
  
  
  τ
  33
  
  


 






























τ

 22





 τ



12








τ


21







τ





  


  τ


  


  


  


  












  





  





  τ





  
23




  



  
 

11
 




 

 

 

  13


 













































 



       





















 







 

 

  

   

   

   

   

 










 



 



 



 



 

  
 














 τ

 32























 τ



 



  

 



   

 



   

 



   

 



    

 



 


















 









 


τ
32





 




 







 







 






 




























τ



31





 





31








 







      




 

 

 

 τ

 

23



τ
13






















 
 
 
 
 











 








 



 

 

 

 

 

 



τ

33
τ


 11


 


 


 τ


 21


  
























 
τ







 τ 
 22       
















 
 
 
 
 
 














12











x1 x2 x1 x2

Figure 3.2: Sketch illustrating the components of the stress tensor.

3.3 Examples for simple stress states

• Hydrostatic pressure: τij = −P0 δij ; note that ti = τij nj = −P0 δij nj = −P0 ni , i.e. the stress on
any surface is normal to the surface and ‘presses against it’ (i.e. acts in the direction opposite to
the outer normal vector) which is precisely what we expect a pure pressure to do; see left half of
Fig. 3.3
• Pure shear stress: E.g. τ12 = τ21 = T0 , τij = 0 otherwise; see right half of Fig. 3.3. This sketch
also illustrates that the symmetry of the stress tensor is related to the balance of moments: If τ 21
were not equal to τ12 (i.e. if the tangential stress acting on the vertical faces was not equal to the
tangential stress acting on the horizontal ones) then the block would rotate about the x 3 axis.

P0 T0

T0 T0
x2 P0 P0

P0 T0

x1

Figure 3.3: Simple stress states: Hydrostatic pressure (left) and pure shear stress (right).

3.4 Cauchy’s equation

• Cauchy’s equation is obtained by considering the equation of motion (‘sum of all forces = mass
times acceleration’) of an infinitesimal volume of fluid. For a fluid which is subject to a body force
(a force per unit mass) Fi , Cauchy’s equation is given by
∂τij
ρai = ρFi + , (3.3)
∂xj
where ρ is the density of the fluid. ai is the acceleration of the fluid, given by (2.5), therefore
Cauchy’s equation can also be written as
Dui ∂τij
ρ = ρFi + (3.4)
Dt ∂xj
MATH35001 Viscous Fluid Flow: Stress, Cauchy’s equation and the Navier-Stokes equations 8

or  
∂ui ∂ui ∂τij
ρ + uk = ρFi + . (3.5)
∂t ∂xk ∂xj

• Note that Cauchy’s equation is valid for any continuum (not just fluids!) provided its deformation
is described by an Eulerian approach.

3.5 The constitutive equations for a Newtonian incompressible

fluid
• In chapter 2 we derived a quantity (the rate of strain tensor ij ) which provides a mathematical
description of the rate of deformation of the fluid.
• Cauchy’s equation provides the equations of motion for the fluid, provided we know what state of
stress (characterised by the stress tensor τij ) the fluid is in.
• The constitutive equations provide the missing link between the rate of deformation and the result-
ing stresses in the fluid.
• A large number of practically important fluids (e.g. water and oil) are incompressible and exhibit
a linear relation between the shear rate of strain and the shear stresses. These fluids are known as
‘Newtonian Fluids’ and their constitutive equation is given by

τij = −pδij + 2µij , (3.6)

or, using the definition of the rate of strain tensor,

 
∂ui ∂uj
τij = −pδij + µ + , (3.7)
∂xj ∂xi

where p is the pressure in the fluid and µ is the ‘dynamic viscosity’, a quantity that has to be
determined experimentally.
• Note that there are also many fluids which do not behave as Newtonian fluids and have different
constitutive equations (e.g. toothpaste, mayonaise). Not very imaginatively, these are often called
‘Non-Newtonian Fluids’ – the behaviour of these fluids is covered in a different lecture.

3.6 The Navier-Stokes equations for incompressible Newtonian

fluids
• We insert the constitutive equations for an incompressible Newtonian fluid into Cauchy’s equations
and obtain the famous Navier-Stokes equations
 
∂ui ∂ui ∂p ∂ 2 ui
ρ + uk = ρFi − +µ 2 , (3.8)
∂t ∂xk ∂xi ∂xj

or symbolically  
∂u
ρ + (u · ∇)u = ρF − ∇p + µ∇2 u. (3.9)
∂t
Dividing the momentum equations by ρ provides an alternative form

∂ui ∂ui 1 ∂p ∂ 2 ui
+ uk = Fi − +ν , (3.10)
∂t ∂xk ρ ∂xi ∂x2j

where ν = µ/ρ is the ‘kinematic viscosity’.

MATH35001 Viscous Fluid Flow: Stress, Cauchy’s equation and the Navier-Stokes equations 9

• In combination with the equation of continuity

∂ui
=0 (3.11)
∂xi
or symbolically
∇ · u = 0, (3.12)
the three momentum equations form a system of four coupled nonlinear, partial differential equa-
tions of parabolic type (second order in space and first order in time) for the three velocity compo-
nents ui and the pressure p.
The governing equations in selected
coordinate systems

Rectangular cartesian coordinates

The rate of strain tensor
 
xx xy xz
ij =  yx yy yz 
zx zy zz
where
∂u ∂v
xx = yy =
∂x ∂y
 
∂w 1 ∂v ∂u
zz = xy = +
∂z 2 ∂x ∂y
   
1 ∂w ∂v 1 ∂u ∂w
yz = + zx = +
2 ∂y ∂z 2 ∂z ∂x

The vorticity
 
∂w ∂v ∂u ∂w ∂v ∂u
ω = curl u = − , − , − .
∂y ∂z ∂z ∂x ∂x ∂y

The Navier Stokes equations

∂u ∂u ∂u ∂u 1 ∂P
+u +v +w =− + ν∇2 u,
∂t ∂x ∂y ∂z ρ ∂x
∂v ∂v ∂v ∂v 1 ∂P
+u +v +w =− + ν∇2 v,
∂t ∂x ∂y ∂z ρ ∂y
∂w ∂w ∂w ∂w 1 ∂P
+u +v +w =− + ν∇2 w,
∂t ∂x ∂y ∂z ρ ∂z

∂u ∂v ∂w
div u = + + = 0.
∂x ∂y ∂z

The Laplace operator

∂2 ∂2 ∂2
∇2 ≡ + + .
∂x2 ∂y 2 ∂z 2

10
MATH35001 Viscous Fluid Flow: The governing equations in selected coordinate systems11

Cylindrical Polar Coordinates

Relation to Cartesian coordinates:

x = r cos ϕ,
y = r sin ϕ,
z = z

Velocity components:
u = ur , v = uϕ , w = uz

The rate of strain tensor

 
rr rϕ rz
ij =  ϕr ϕϕ ϕz 
zr zϕ zz

where
∂u 1 ∂v u
rr = + ϕϕ =
∂r r ∂ϕ r
   
∂w 1 ∂ v 1 ∂u
zz = rϕ = r +
∂z 2 ∂r r r ∂ϕ
   
1 1 ∂w ∂v 1 ∂u ∂w
ϕz = + rz = +
2 r ∂ϕ ∂z 2 ∂z ∂r

The vorticity
 
1 ∂w ∂v ∂u ∂w 1 ∂ 1 ∂u
ω = curl u = − , − , (rv) − .
r ∂ϕ ∂z ∂z ∂r r ∂r r ∂ϕ

The Navier Stokes equations

 
∂u ∂u v ∂u ∂u v 2 1 ∂P 2 u 2 ∂v
+u + +w − =− +ν ∇ u− 2 − 2 ,
∂t ∂r r ∂ϕ ∂z r ρ ∂r r r ∂ϕ
 
∂v ∂v v ∂v ∂v uv 1 ∂P v 2 ∂u
+u + +w + =− + ν ∇2 v − 2 + 2 ,
∂t ∂r r ∂ϕ ∂z r ρr ∂ϕ r r ∂ϕ
∂w ∂w v ∂w ∂w 1 ∂P
+u + +w =− + ν∇2 w,
∂t ∂r r ∂ϕ ∂z ρ ∂z
1 ∂ 1 ∂v ∂w
div u = (ru) + + = 0.
r ∂r r ∂ϕ ∂z

The Laplace operator

 
1 ∂ ∂ 1 ∂2 ∂2
∇2 ≡ r + 2 2
+ 2.
r ∂r ∂r r ∂ϕ ∂z
MATH35001 Viscous Fluid Flow: The governing equations in selected coordinate systems12

Spherical Polar Coordinates

Relation to Cartesian coordinates:

x = r cos θ,
y = r sin θ cos ϕ,
z = r sin θ sin ϕ

Velocity components:
u = ur , v = uθ , w = uϕ

The rate of strain tensor

 
rr rθ rϕ
ij =  θr θθ θϕ 
ϕr ϕθ ϕϕ

where
∂u 1 ∂v u
rr = θθ = +
∂r r ∂θ r
 
1 ∂w u v cot θ 1 ∂  v  1 ∂u
ϕϕ = + + rθ = r +
r sin θ ∂ϕ r r 2 ∂r r r ∂θ
   
1 sin θ ∂  w  1 ∂v 1 1 ∂u ∂ w
θϕ = + ϕr = +r
2 r ∂θ sin θ r sin θ ∂ϕ 2 r sin θ ∂ϕ ∂r r

The vorticity
   
1 ∂ ∂v 1 ∂u 1 ∂ 1 ∂ 1 ∂u
ω = curl u = (w sin θ) − , − (rw), (rv) − .
r sin θ ∂θ ∂ϕ r sin θ ∂ϕ r ∂r r ∂r r ∂θ

The Navier Stokes equations

 
∂u ∂u v ∂u w ∂u v 2 + w2 1 ∂P 2u 2 ∂v 2v cot θ 2 ∂w
+u + + − =− + ν ∇2 u − 2 − 2 − − ,
∂t ∂r r ∂θ r sin θ ∂ϕ r ρ ∂r r r ∂θ r2 r2 sin θ ∂ϕ
 
∂v ∂v v ∂v w ∂v uv w2 cot θ 1 ∂P 2 2 ∂u v 2 cos θ ∂w
+u + + + − =− +ν ∇ v + 2 − 2 2 − 2 2 ,
∂t ∂r r ∂θ r sin θ ∂ϕ r r ρr ∂θ r ∂θ r sin θ r sin θ ∂ϕ

∂w ∂w v ∂w w ∂w uw vw cot θ
+u + + + − =
∂t ∂r r ∂θ r sin θ ∂ϕ r r
 
1 ∂P 2 w 2 ∂u 2 cos θ ∂v
− +ν ∇ w− 2 2 + 2 + ,
ρr sin θ ∂ϕ r sin θ r sin θ ∂ϕ r2 sin2 θ ∂ϕ
1 ∂ 2 1 ∂ 1 ∂w
div u = (r u) + (v sin θ) + = 0.
r2 ∂r r sin θ ∂θ r sin θ ∂ϕ

The Laplace operator

   
1 ∂
2 2 ∂ 1 ∂ ∂ 1 ∂2
∇ ≡ 2 r + 2 sin θ + .
r ∂r ∂r r sin θ ∂θ ∂θ r2 sin2 θ ∂ϕ2
Chapter 4

Boundary and initial conditions

4.1 Initial conditions

• For time-dependent problems, an initial condition for the velocity field, i.e. ui (xj , t = 0) has to be
specified.

4.2 Boundary conditions

• Fig. 4.1 shows a selection of common boundary conditions for flow problems.

free surface

rigid impermeable moving boundary

inflow boundary
impermeable elastic moving boundary

rigid impermeable stationary boundary

Figure 4.1: Common boundary conditions.

4.2.1 Inflow/outflow boundary conditions

• In many applications, we are only interested in the behviour of the fluid in a small region (for
instance, if we want to study the ventilation in a room, it would be impractical to include the
earth’s entire atmosphere into the model. We would only model the room and treat its interaction
with the ‘rest of the world’ via inflow boundary conditions – e.g. by prescribing the wind velocity
through an open window). Hence at inflow (or outflow) boundaries we prescribe the velocity, i.e.
ui = v i , (4.1)
where vi is a prescribed function.

4.2.2 Solid surfaces

• Most solid surfaces are impermeable to fluid and the fluid ‘sticks’ to their surfaces. Hence, there is
no slip and no penetration, and the fluid particles on the wall move with the velocity of the wall:
ui = w i , (4.2)

13
MATH35001 Viscous Fluid Flow: Boundary and initial conditions 14

where wi is the (known) velocity of the impermeable wall.

• In the special case where the walls are stationary we have

ui = 0. (4.3)

4.2.3 Free surfaces

• Free surfaces occur at the interface between two fluids. Such interfaces require two boundary
conditions to be applied: (i) A kinematic condition which relates the motion of the free interface
to the fluid velocities at the free surface and (ii) a dynamic condition which is concerned with the
force balance at the free surface.

(i) The kinematic boundary condition

– The position of a free surface can always be given in implicit form as F (xj , t) = 0. For instance,
in Fig. 4.2 the height of the free surface above the x-axis is specified as y = h(x, t) and an
appropriate function F (x, y, t) would be given by F (x, y, t) = h(x, t) − y.

Fluid (2)
n(1)
Fluid (1)
y

n(2) y=h(x,t)

Figure 4.2: Sketch illustrating the conditions at a free surface formed by the interface between two fluids.

– Fluid particles on the free surface always remain part of the free surface, therefore we must
have
DF ∂F ∂F
= + uk . (4.4)
Dt ∂t ∂xk
This is the kinematic boundary condition.
– For surfaces whose position is described in the form z = h(x, y, t), the kinematic boundary
condition becomes
∂h ∂h ∂h
w= +u +v , (4.5)
∂t ∂x ∂y
where u, v, w are the velocities in the x, y, z directions, respectively.
– For steady problems, we have ∂F/∂t = 0 and the kinematic boundary condition can be written
as
ui ni = 0 or symbolically u · n = 0, (4.6)
where n is the outer unit normal on the free surface. This condition implies that there is no
flow through the free surface (but there can be a flow tangential to it!).

(ii) The dynamic boundary condition

– The dynamic boundary condition requires the stress to be continuous across the free surface
which separates the two fluids (air and water in Fig. 4.1). The traction exerted by fluid (1)
onto fluid (2) is equal and opposite to the traction exerted by fluid (2) on fluid (1). Therefore
MATH35001 Viscous Fluid Flow: Boundary and initial conditions 15

we must have t(1) = −t(2) . Since n(1) = −n(2) (see Fig. 4.2) we obtain the dynamic boundary
condition
(1) (2)
τij nj = τij nj , (4.7)

where we can use either n(1) or n(2) as the unit normal.

– On curved surfaces, surface tension can create a pressure jump across the free surface. The
surface tension induced pressure jump is given by

∆p = σκ. (4.8)

In this expression σ is the surface tension of the fluid and κ is equal to twice the mean curvature
of the free surface, i.e.
1 1
κ= + , (4.9)
R1 R2
where R1 and R2 are the principal radii of curvature of the surface (for instance, κ = 2/a for
a spherical drop of radius a and κ = 1/a for a circular jet of radius a). Surface tension acts
like a tensioned membrane at the free surface and tries to minimise the surface area. Hence
the pressure inside a spherical drop (or inside a circular liquid jet) tends to be higher than the
pressure in the surrounding medium.
– If surface tension is important, the dynamic boundary condition has to be modified to
(1) (1) (1) (2) (1)
τij nj + σκni = τij nj , (4.10)

where κ > 0 if the centres of curvature lie inside fluid (1).

4.2.4 Other boundary conditions

• Other boundary conditions can occur in special applications. For instance, the presence of an elastic
boundary leads to fluid-structure interaction problems in which the fluid velocity has to be equal
to the velocity of the elastic wall, while the elastic wall deforms in response to the traction that the
fluid exerts on it. At porous walls, the no-penetration condition no longer holds: the volume flux
into the wall is often proportional to the pressure gradient at the porous surface. Non-uniformly
distributed surfactants (substances which reduce the surface tension) can induce tangential stresses
at free surfaces, etc.

4.3 Further remarks

• For an incompressible fluid, the boundary conditions need to fulfill the overall consistency condition
I
ui ni dS = 0, (4.11)
∂V

where ∂V is the surface of the spatially fixed volume in which the equations are solved.
• If there are no free surfaces (and associated dynamic boundary conditions), the pressure is only
defined up to an arbitary constant as only the pressure gradient (but not the pressure itself) appears
in the Navier-Stokes equations.
• For initial value problems, the initial velocity field (at t = 0) already has to fulfill the incompress-
ibility constraint.

These remarks are particularly important for the numerical solution of the Navier-Stokes equations.
Chapter 5

Parallel Flows

5.1 The parallel flow equations

• The main difficulty in the solution of the Navier Stokes equations arises from their nonlinearity.
There are, however, situations in which the nonlinear terms vanish identically.
• This happens (for instance) if the flow is unidirectional. If this is the case then we can choose our
coordinate system such that the x-axis is aligned with the flow and the velocity field has the form
u = u(x, y, z, t) ex .
• Inserting this assumption into the Navier Stokes and continuity equation shows that this is only
possible if
u = u(y, z, t) ex , (5.1)
i.e. if the velocity is independent of the streamwise coordinate.
• The flow governed by the following three linear equations
 2 
∂u ∂p ∂ u ∂2u
ρ = ρFx − +µ + 2 , (5.2)
∂t ∂x ∂y 2 ∂z

∂p
0 = ρFy − (5.3)
∂y
and
∂p
0 = ρFz − . (5.4)
∂z

5.2 The parallel flow equations without body force

• If the body force vanishes (i.e. Fx = Fy = Fz = 0) it can be shown that p = p(x, t) and the pressure
gradient has to have the form
∇p = G ex (5.5)
where G is a constant. (If the pressure gradient has any other form, then no parallel flow is possible).
• In this case, the only non-trivial equation is the x-momentum equation which becomes
 2 
∂u ∂ u ∂2u
ρ = −G + µ + 2 . (5.6)
∂t ∂y 2 ∂z

16
Chapter 6

Curvilinear Coordinates

• For flows in circular (or spherical) geometries, cartesian coordinates are not the most convenient
coordinate system to work in.
• The transformation of the Navier-Stokes and continuity equations to other coordinate systems
is straightforward (if messy) and is based on a simple coordinate transformation, such as x =
r cos ϕ, y = r sin ϕ for the transformation between 2D cartesian and plane cylindrical polar coor-
dinates. Following the usual rules, we can transform differential operators to the new coordinates,
e.g.
∂2φ ∂2φ ∂ 2 φ 1 ∂φ 1 ∂2φ
∇2 φ = + = + + . (6.1)
∂x2 ∂y 2 ∂r2 r ∂r r2 ∂ϕ2
• We also need to transform vectors to the new coordinate system by decomposing them into the new
basis vectors, e.g.,
u = u x ex + uy ey = ur er + uϕ eϕ , (6.2)
where ur and uϕ are the velocity components in radial and circumferential direction.
• Note that in curvilinear coordinates, the basis vectors depend on the coordinates (e.g. e r =
(cos ϕ, sin ϕ)). Hence, any differential operator acting on a vector acts on the basis vectors as
well as the components themselves. The resulting vector then has to be decomposed into the basis
vectors. This makes the resulting expressions considerably more complicated than their equivalents
in cartesian coordinates (see the Navier Stokes equations in curvilinear coordinates in chapter 3).
• Provided we restrict ourselves to orthogonal coordinate systems (such as cylindrical and spherical
polar coordinates) we can still use the index notation and the summation convention. For instance,
in plane cylindrical polars the traction boundary condition can be written as

ti = (−pδij + 2µeij )nj where i, j represent the coordinate directions r and ϕ. (6.3)

• Thus, for instance, the r-component of the traction t = tr er + tϕ eϕ is given by

tr = −pnr + 2µ(err nr + erϕ nϕ ) (6.4)

where n = ny er + nϕ eϕ is the outer unit normal on the fluid, decomposed into the cylindrical basis
vectors and the eij are the components of the rate of strain tensor in plane cylindrical polars as
given in chapter 3.

17
Chapter 7

Dimensional Analysis and Scaling

7.1 Dimensional Analysis and Scaling

• See the copies of the OHP transparencies.

7.2 Similarity Solutions

• Many phenomena in fluid mechanics exhibit self-similar behaviour. A simple example for this is
given by a velocity profile u(y, t) which has the same ‘shape’ for all values of time t. In that case,
the solution must have the special form
 
y
u(y, t) = a(t) f , (7.1)
b(t)

where the function a(t) changes the ‘amplitude’ of u(y, t) while b(t) provides a time-dependent
scaling for the y-coordinate, making the velocity profile ‘wider’ (for b(t) > 1) or ‘narrower’ (for
b(t) < 1).
• The existence of similarity variables is also familiar from traveling wave problems in which the
solution has the form u(y, t) = f (y − U t). This solution represents a wave of shape f (y) traveling
in the positive y-direction with velocity U . The traveling wave coordinate η = y − U t plays the role
of a similarity variable.

• In general, similarity solutions are characterised by the requirement that at least one independent
variable only occurs in a certain combination with other independent variables. This often simplifies
the mathematical analysis, for instance, by transforming PDEs into ODEs.
• The search for suitable similarity variables is often aided by dimensionality considerations.
• The choice of the similarity variable is usually not unique. For instance, a function f (η) could
also be regarded as a function F (η 2 ) – obviously, both η and η 2 are perfectly acceptable choices.
Typically, one tries to keep the similarity variable linear in the spatial coordinate, as in (7.1) where
η = y/b(t).
• Similarity solutions only ‘work’ if the boundary and initial conditions can also be formulated in
terms of the similarity variable. If this is not the case, the similarity ‘solution’ might (!) still
represent a useful approximation to the exact solution.

18
Chapter 8

The Streamfunction and Vorticity

• For 2D incompressible flows, it is possible to recast the Navier-Stokes equations in an alternative

form in terms of the streamfunction and the vorticity.
• In many applications, the streamfunction-vorticity form of the Navier Stokes equations provides
better insight into the physical mechanisms driving the flow than the ‘primitive variable’ formulation
in terms of u, v and p.
• The streamfunction and vorticity formulation is also useful for numerical work since it avoids some
problems resulting from the discretisation of the continuity equation.

Unless specifically stated, all results in this chapter are restricted

to 2D incompressible flows.

8.1 The Streamfunction

• The streamfunction is defined as Z P
ψA (P ) = u · n ds, (8.1)
A
where the integral has to be evaluated along a line from the arbitrary but fixed point A to point
P. n is the unit normal on the line from A to P. We regard ψA (P ) as a function of the location of
point P.

n
y
A u

Figure 8.1: Sketch illustrating the definition of the streamfunction.

• The sketch in Fig. 8.1 shows that u · n is equal to the component of the velocity u that crosses the
line AP. Therefore ψA (P ) represents the volume flux (per unit depth in the z-direction) through
the line between A and P.

19
MATH35001 Viscous Fluid Flow: Streamfunction and Vorticity 20

• Evaluating ψA (P ) along two different paths and invoking the integral form of the incompressibility
constraint shows that ψA (P ) is path-independent, i.e. its value only depends on the locations of
the points A and P.

• Changing the position of point A only changes ψA (P ) by a constant. It turns out that for all
applications such changes are irrelevant. It is therefore common to suppress the explicit reference
to A. Hence, we regard ψA (P ) as a function of the spatial coordinates only, i.e. ψA (P ) = ψ(P ) =
ψ(x, y).
• Streamlines are lines which are everywhere tangential to the velocity field, i.e. u · n = 0, where n
is the unit normal to the streamline. Hence the streamfunction ψ is constant along streamlines.
• Note that stationary impermeable boundaries are also characterised by u · n = 0, where n is the
unit normal on the boundary. Therefore, ψ is also constant along such boundaries.
• Invoking the integral incompressibility constraint for an infinitesimally small triangle shows that ψ
is related to the two cartesian velocity components u and v via
∂ψ ∂ψ
u= and v=− (8.2)
∂y ∂x

• Similarly, in plane cylindrical polars, the velocity components are given by

1 ∂ψ ∂ψ
ur = and uϕ = − . (8.3)
r ∂ϕ ∂r

• Flows which are specified by a streamfunction automatically satisfy the continuity equation since
   
∂u ∂v ∂ ∂ψ ∂ ∂ψ
+ = + − = 0. (8.4)
∂x ∂y ∂x ∂y ∂y ∂x

• For 2D flows, the vorticity vector ω = ∇ × u only has one non-zero component (in the z-direction),
i.e. ω = ωez where
∂v ∂u
ω= − . (8.5)
∂x ∂y
Using the definition of the velocities in terms of the streamfunction shows that
   
∂ ∂ψ ∂ ∂ψ
ω= − − (8.6)
∂x ∂x ∂x ∂x

and therefore
ω = −∇2 ψ, (8.7)
2 2 2 2 2
where ∇ = ∂ /∂x + ∂ /∂y is the 2D Laplace operator.

8.2 The Streamfunction-Vorticity form of the Navier-Stokes equa-

tions
• Straightforward algebraic manipulation of the 3D momentum equations transforms them into the
vorticity transport equation
ω
Dω
= (ωω · ∇)u + ν∇2ω (8.8)
Dt
(see the separate handout for the derivation; this equation is valid in 3D).
ω /Dt, is con-
• This equation shows that the rate of change of the vorticity of material particles, Dω
ω · ∇)u; this is a familiar result from inviscid fluid
trolled by ‘vortex stretching’ (described by (ω
mechanics) and by diffusion (described by ν∇2ω ). The diffusion of vorticity only occurs in viscous
flows.
MATH35001 Viscous Fluid Flow: Streamfunction and Vorticity 21

• For 2D flows, vortex stretching is absent since u = u(x, y) ex + v(x, y) ey and ω = ω(x, y) ez and
ω · ∇)u = 0.
therefore (ω
• For 2D flows, the scalar vorticity transport equation
Dω
= ν∇2 ω (8.9)
Dt
together with the equation for the vorticity in terms of the streamfunction

ω = −∇2 ψ (8.10)

and
u = ∂ψ/∂y and v = −∂ψ/∂x (8.11)
provide the streamfunction-vorticity formulation of the Navier-Stokes equations. It consists of only
two PDEs for the scalars ω and ψ rather than the three PDEs for u, v and p in the ‘primitive
variable’ form.
• In the limit of zero Reynolds number, only one fourth-order PDE for the streamfunction ψ needs
to be solved, namely the biharmonic equation

∇4 ψ = 0, (8.12)

where
∂4 ∂4 ∂4
∇4 = 4
+2 2 2 + 4. (8.13)
∂x ∂x ∂y ∂y
This can be shown by, e.g., taking the curl of the Stokes equations.