Lecture One

Overview of Lagrangian and Eulerian Descriptions

Lagrangian Description

In the Lagrangian perspective, we follow fluid particles (material points) as they move through
the flow. Mathematically, we specify the vector field

X  X0,t  (1)

where X is the position vector of a fluid particle at some time t and X 0 is the position vector of
the particle at some reference time, say t  t0 . That is

X 0  X  X 0 , t  t0  (2)

We defined the Lagrangian velocity as the time derivative of the position vector following a fluid
particle:

dX
V (3)
dt single
particle

Eulerian Description

On the other hand, the Eulerian perspective fixates on a particular point in space, and records the
properties of the fluid elements passing through that point. Mathematically, specify

v  x, t  (4)

where v is the velocity vector at the laboratory coordinate x at time t.

One way to connect the Lagrangian and Eulerian descriptions is

v  x  X  X0,t,t  
dX
 X 0 , t . (5)
Eulerian dt
Lagrangian

Both sides of equation (5) describe the velocity of the fluid particle that was at X 0 at t0 , but is at
the position x at time t.
 Lecture One
Overview of Lagrangian and Eulerian Descriptions
Material Derivative

The material derivative gives us a way to relate the Lagrangian description to the Eulerian one.
It is defined for a quantity F (which could be a scalar or a vector) as

DF F
   v   F .
Dt t

In Cartesian coordinates, this is written as

DF F F F F
  vx  vy  vz
Dt t x,y,z x t,y,z y t ,x,z z t,x,y
fixed fixed fixed fixed

The material derivative can be used to obtain the ordinary differential equation describing the
pathline of a particle. At any instant in time, the particle inhabits some laboratory point,
described by x. At some time t, the particle position X (in Lagrangian coordinates) is therefore
equal to x. Since the material derivative tells the Lagrangian time derivative in Eulerian
coordinates, we can obtain the ODE we seek by setting

dX Dx
 . (6)
dt single Dt
particle

The right-hand side can be thought of as the rate of change of the laboratory coordinate of the
fluid particle of interest with time. Let us focus only on the x-coordinate of this particle. As
shown in class, for this case the material derivative becomes

dX Dx x x x x
   vx  vy  vz  0  vx (1)  0  0  vx . (7)
dt single Dt t x, y , z x t , y , z y t , x , z z t , x , y
particle fixed fixed fixed fixed

Similarly, the time evolution of the y-coordinate of the particle is described by

dY Dy y y y y
   vx  vy  vz  0  0  v y (1)  0  v y . (8)
dt single Dt t x, y, z x t , y , z y t , x , z z t , x , y
particle fixed fixed fixed fixed
 Geometric Interpretation of Fluid Kinematics
In Steady Shear Flow
u0 u0 ⎧
γ̇ = ˙
⎪vx = γy
H
⎨
H ∂vx vy = 0
γ̇ = ⎪
⎩
vz = 0
∂y
Fluid
Element

Steady Shear Flow

Q R vx + 12 γdy
˙
⎛ ⎞
0 0 0
dy
2 'u = ⎝γ̇ 0 0⎠
0 0 0
vx
O u(x + dx) = u(x) + dx · (e + Ω)

dx
2
vx − 12 γ̇dy
P S
≡

Steady Translation
vx
u(x + dx) = u(x)
du = 0

vx
O

vx

y'
+
y x'
+
Pure Deformation
⎛ ⎞
0 γ̇/2 0
e = ⎝γ̇/2 0 0⎠
du = dx · e 0 0 0
45◦ x
O Rotation through 45◦ (e' = C e)
⇒ Principal Axes:
⎛ √1
− √12
⎛ γ̇
0
⎞ ⎞
2 2 0 0
C= ⎝ √1
2
√1
2
0⎠ ⇒ e' = ⎝ 0 − γ̇2 0⎠
0 0 0 0 0 0
+ +
Pure Rotation
⎛ ⎞
0 −γ̇/2 0
Ω = ⎝γ̇/2 0 0⎠
0 0 0

O du = dx · Ω ⎛ ⎞
0
or ω = ⎝ 0 ⎠ = − ijk Ωjk
θ̇3 = −γ̇/2 du = 12 ω×dx γ̇

(Vorticity) = 2×(Angular Velocity)

 Consider the deformation in a (small) time dt:
'
y y x' y' y '' x' x''
'
Q B R Q B'
q
dx2 dy 2 R'
+
4 4 γ̇ dy dt dθ 1
2
γ̇ dy dt

45◦ dθ
dy A C x A' C'
'
O O
dt
dβ

P D S P D S

dx dβ ≈ tan dβ = γdt
˙

In x-y coordinate frame: deformation is simple shear :

⎛ ⎞
0 γ/2
˙ 0
e = ⎝γ/2
˙ 0 0⎠
0 0 0

• Displacement: (Q → Q' ) = vQ dt = γ̇ dy dt

• Length: P Q' = P Q 1 + (γ̇dt)2 = dy 1 + (γ̇dt)2

dβ
dβ γ̇dt
• Average Angular Velocity: θ̇3 = 1
2
dα
dt − dt = 1
2 0− dt = − γ̇2
x dα

In x' -y ' coordinate frame: deformation is extensional :

⎛ ⎞
γ/2
˙ 0
∂vx: γ̇ ∂vy: γ̇
e' = ⎝ 0 −γ/2
˙ 0⎠ ⇒ = , =−
∂x' 2 ∂y ' 2
0 0 0

Line Segment:

∂vx:
A' B ' = AB + 2 dx' dt = AB + γ̇dx' dt
∂x'
∂vy:
B ' C ' = BC + 2 dy ' dt = BC − γ̇dy ' dt
∂y '

In addition, axes rotates by dθ = − 12 γ̇dt ⇒ θ̇ = − 12 γ̇ from x' y ' → x'' y ''

Note that expressions for angular displacement are only valid for small dt such that tan dβ ≈ dβ
⇒ In the limit of finite time, the change in the (initially) perpendicular line segments QPS is:

γ̇ dy dt
d tan β = ⇒ β = tan−1 (γ̇t)
dy
β
π
2

γ̇t
 Lecture One
Advanced Fluid Mechanics

The Continuity Equation:

Conservation of Mass for a Fluid Element

(
v = vx ,v y ,vz )
δV
δV ′
δm δm

No  flow  across  streamlines  

Consider a fluid element with constant mass δm and volume δV moving in a velocity field as shown above.
The streamlines are converging and the fluid element may be advected to a new position which has a higher
speed. If we assume the most general case, in which the fluid element is compressible, then δm is fixed but
δV changes. Note that:
ρ = δm/δV (1)
Hence:
δm D (δV ) δm Dρ
δV = = δmρ−1 ⇒ =− 2 (2)
ρ Dt ρ Dt
⇒
1 D (δV ) 1 Dρ
=− (3)
δV Dt ρ Dt
We already know that the left hand side of (3) is \ · v, thus:
1 Dρ
= −\ · v (4)
ρ Dt
Alternative way to reach the same thing is:
D (δm) D (ρδV ) D (ρ) D (δV )
=0⇒ = 0 ⇒ δV +ρ =0 (5)
Dt Dt Dt Dt
Dividing by ρδV leads to:
1 D (ρ) 1 D (δV )
+ =0 (6)
ρ Dt δV Dt
Again knowing that volumetric rate of strain, the second term, is equal to \ · v gives:
1 D (ρ)
= −\ · v (7)
ρ Dt
 The Continuity Equation:
Conservation of Mass for a Fluid Element

which is the same concluded in (4). The derived equation is mass conservation for any flow (compressible or
incompressible). In the case of incompressible flows (or almost ”incompressible”-Mach numbers lower than
0.3), from ”incompressibility” we will have:

1 D (ρ)
c0 (8)
ρ Dt

which by (7) means that in incompressible flows (M a < 0.3):

\·v c0 (9)
Solid Body Rotation-Extra Notes
Whenever we have a coordinate rotation the following holds:
Imagine a vector v:
v
v= x
vy
in x − y coordinate system and want to calculate the components of v in a new coordinate
system x' − y ' which comes from θ counterclockwise rotation of x − y coordinate system (Figure
1).

vy v
� �
y x

vy� vx�
θ
vx x
Figure 1: Coordinate system x' − y ' is a θ counterclockwise rotation of x − y.

vxI cosθ sinθ v

= × x
vy I −sinθ cosθ vy
If you show the vectors in a row form rather than the column form then the matrix from of
this conversion will look different:
[ J [ J cosθ −sinθ
vxI v y I = vx v y ×
sinθ cosθ

It is easy to see that both of these matrix representations express the same set of equations for
conversion between the two coordinate systems:

vxI = vx cosθ + vy sinθ

(1)
vyI = vx (−sinθ) + vy cosθ

and if you like to get the components in the x − y coordinates from the x' − y ' coordinate then
the following holds:

vx = vxI cosθ − vyI sinθ

(2)
vy = vxI sinθ + vyI cosθ
In the solid body rotation problem we have to convert components from x − y coordinate to
r − θ coordinates:

: Object in moton
y

vy v
θ r
y
vr
vθ
θ
eθ vx x
2 er
y
� x2 +

r=
ey
θ
ex x
Figure 2: Coordinate system r − θ is a θ counterclockwise rotation of x − y. Note that here
the v vector is drawn for an arbitrary case and has both r and θ components whereas in the
rotation problem the r component is zero.

The v velocity vector in the polar coordinate for this problem is:
   
vr er 0
v= =
v θ eθ rΩeθ
Using equation (2) we can convert the velocity components from r − θ coordiantes back to the
x − y coordinates:
vx = vr cosθ − vθ sinθ = 0cosθ − rΩsinθ
vy = vr sinθ + vθ cosθ = 0sinθ + rΩcosθ
Using the fact that y = rsinθ and x = rcosθ we can simplify it to:
vx = −yΩ
vy = xΩ
Now as shown in the class after using the material derivative we can find the acceleration vector
(a = Dv/Dt) in the x − y coordinate system:
   
ax ex −Ω2 x ex
a= =
a y ey −Ω2 y ey
Now using equation (1) we can convert this into the r − θ coordinates:
 ar = ax cosθ + ay sinθ = −Ω2 xcosθ − Ω2 ysinθ
aθ = ax (−sinθ) + ay cosθ = Ω2 xsinθ − Ω2 ycosθ
Using the fact that y = rsinθ, x = rcosθ and cos2 θ + sin2 θ = 1 we can simplify it to:

ar = −rΩ2
aθ = 0

thus:    
ar er −rΩ2 er
a= =
aθ eθ 0 eθ
 Advanced Fluid Mechanics

Particle Kinematics
Lagrangian and Eulerian Frames - Material Derivatives

The stream function — which will be discussed in more detail later in the course — in cylindrical co-ordinates
(r, θ) for flow past a circular cylinder of radius a with clockwise circulation Γ is given by

a2 Γ (r)
ψ(r, θ) = U r− sin θ + ln (1)
r 2π a

a) Write the stream function ψ(x, y) in Cartesian co-ordinates, and find the components of the velocity ux
and uy in the x and y directions.

Hint: The stream function is defined in terms of the velocity components as

∂ψ(x, y)
ux =
∂y

∂ψ(x, y)
uy = −
∂x

b) Derive the ordinary differential equations that govern the particle path lines.

c) Find the equation for the particle trajectory passing through the point r = 2a, θ = 0 (or equivalently,
x = 2a, y = 0).

d) Show that a particle on the surface of the cylinder always stays on the cylinder. Find the tangential
velocity component of such a particle, and determine the stagnation points.

Hint: In cylindrical co-ordinates,

1 ∂ψ(r, θ)
ur =
r ∂θ
∂ψ(r, θ)
uθ = −
∂r

e) Sketch the stream lines for the case Γ > 4πaU . What happens as r → ∞?
 Advanced Fluid Mechanics

Problem 1.00
This problem is from “2.25 Advanced Fluid Mechanics” by Ain Sonin

Rate of change of properties measured by a probe moving through the earth’s atmosphere — plus some things
about the earth and its atmosphere.

The pressure distribution in a static, constant-temperature planetary atmosphere modeled as an ideal gas is
given by
p = p0 e−z/H (1.00a)

where z is the altitude above a reference altitude z = 0, p0 is the absolute pressure at z = 0, and
RT
H= (1.00b)
Mg

is a length scale that characterizes the atmosphere. Its value is determined by the strength of the gravitational
acceleration and the parameters that appear in the ideal-gas equation of state,
RT
p=ρ ; (1.00c)
M

R = 8.32 JK−1 mol−1 is the universal gas constant, T is the absolute temperature (taken as constant in
this model of the atmosphere), M is the molar mass of the gas (0.029 kg/mol if the gas is air), and g is
the acceleration of gravity at or near the surface of the planet. For the “Standard” isothermal model of the
earth’s atmosphere, T = 288 K, p0 = 1.02 × 105 N/m2 if z = 0 at sea level, and consequently H = 8.43 km.
Note that the distribution given above is based on the assumption that H « a, where a is the planet’s
radius.

Suppose a sounding rocket or balloon equipped with a static-pressure sensor is traveling through the atmo­
sphere with given velocity (vx , vy , vz ).

1. In terms of the given quantities and z, derive an expression for the rate of change of pressure recorded
by the rocket’s sensor.
2. Evaluate this time of change at an altitude z = 20,000 m for a rocket traveling upward through
the earth’s atmosphere with a direction of 30◦ from the vertical and a speed of 465 m/s. (Answer:
−0.273 bar/min.)
3. Suppose a rocket carries instruments that measure both the instantaneous atmospheric pressure p and
the rate of change of that pressure, dp/dt. Given the value of these two quantities at a particular time
and p0 and H, derive expressions for the rocket’s instantaneous altitude and vertical (upward) velocity.

Additional things to think about, if you are so inclined:

4. Suppose the Earth’s atmosphere is isothermal and radially symmetric around a perfectly spherical
earth with radius a = 6400 km. What is the total mass of the Earth’s atmosphere? What fraction is
this of the solid and liquid parts of the planet’s mass? (Answers: 5.35 × 1018 kg and 8.96 × 10−7 .)
 Advanced Fluid Mechanics

Kundu & Cohen 2.6, 2.7, and 2.20

These problems are from chapter 2 in “Fluid Mechanics” by P. K. Kundu, I. M. Cohen, and D.R. Dowling

• (2.6) Show that the condition for the vectors a, b, and c to be coplanar is:
εijk ai bj ck
• (2.7) Prove the following relationships:
δij δij = 3 εpqi εpqj = 2δij

• (2.20) Use Stokes theorem to prove that \ × (\φ) = 0 for any single-valued twice-differentiable scalar
(φ) regardless of the coordinate system.
Solution:
2.6
Condition for three vectors a, b, and c to be coplanar is

a · (b × c) = 0 (1)

Also we know that

b × c = (εijk bj ck ei ) (2)
From (1) and (2):
ai ei · (εijk bj ck ei ) = 0 (3)
which leads to:
εijk ai bj ck (4)

2.7
From the definition for Kronecker delta:

δij = 0 if i = j and δij = 1 if i = j (5)

Thus
δij δij = δii δii = 1 + 1 + 1 = 3 (6)
For proving the other statement (εpqi εpqj = 2δij ) we should refer to the definition of the alternating tensors:
if there is any repeating index, i.e. i = j, or j = k, or i = k then εijk = 0. This means that in εpqi εpqj the
only non-zero terms are the ones in which p,q,i, and j have four different index values. Since we only have
three values for any possible index (1,2, and 3) the mentioned condition for having non-zero terms is only
true when i = j (one can easily pick two different values of i and j (e.g. i = 1 and i = 3)and see that all the
terms turn to be zero in εpqi εpqj . Thus we will have the following:

εpqi εpqj = 0 if i = j (7)

if i = j : εpqi εpqj = 0 + εpqi εpqi = 0 + ε2pqi (8)

Knowing that for any value of i there are only two remaining non-zero terms in the right hand side of (8)
which are either −1 or 1, we will have:
2 2
if i = j : εpqi εpqj = 0 + (−1) + (1) = 2 (9)

From (7) and (9):

εpqi εpqj = 2δij (10)
2.20
From Stokes’ theorem:
(\ × u) · dA = [Link] (11)
A C

Replacing u by \φ will lead to:

(\ × (\φ)) · dA = \φ.ds (12)
A C

Now considering the right hand side of (12), for the line integral of a gradient vector we have the following:
b
\φ · ds = φ (b) − φ (a) (13)
a

Using (13) for the closed integral over the curve C, we will have:

\φ.ds = φ (a) − φ (a) (14)

C

In which a can be any arbitrary point on the curve C.

Using (14) and also having (12) for any arbitrary area, A, one can conclude that \ × (\φ) = 0
 Advanced Fluid Mechanics

Panton 3.12
This problem is from “Incompressible Flow” by Ronald L. Panton

Write the following formulas in Gibbs’s notation using the symbol \. Convert the expressions to Cartesian
notation and prove that the equations are correct.

div(φv) = φ div v + v . grad φ

div(u × v) = v · curl u − u · curl v

curl(u × v) = v · grad u − u · grad v + u div v − v div u
Solution:

a.
div(φv) = φ divv + v · gradφ (1)
Using Gibbs Notation we can rewrite equation 1 as:

\ · (φ v) = φ \ · v + v · \φ (2)

In order to write the equation in index notation, starting from left hand side we have:
∂ ∂ φvj ∂ φvj
\ · (φv) = ( ei ) · (φ vj ej ) = (ei · ej ) = δij (3)
∂xi ∂xi ∂xi
We know that δij is only non zero when i = j, therefore:

∂φvj ∂φvi
δij = (4)
∂xi ∂xi
where i is the summation variable. Then for the first term on the right hand side, following the same method
as above:
∂ ∂vj ∂vj ∂vi
φ \·v =φ ( e ) · vj e j = φ (e · e ) = φ δij = φ (5)
∂xi i ∂xi i j ∂xi ∂xi
And for the second term:
∂φ ∂φ ∂φ ∂φ
v · \φ = vi ei · e = vi e · e = vi δij = vi (6)
∂xj j ∂xj i j ∂xj ∂xi

And thus equation 1 in index notation has a form of:

∂φvi ∂vi ∂φ
=φ + vi (7)
∂xi ∂xi ∂xi
Now in order to prove equation 1, we start from the left hand side of equation 7 and use the chain rule to
open the derivative:
∂φvi ∂vi ∂φ
=φ + vi (8)
∂xi ∂xi ∂xi
Multiplying the last term by δii = 1:
∂vi ∂φ ∂vi ∂φ ∂vi ∂φ
φ + vi δii = φ + vi (e · e ) = φ + vi ei · e (9)
∂xi ∂xi ∂xi ∂xi i i ∂xi ∂xi i
By multiplying the last term with δii = 1 we are able to get to the dot product of two vector quantities.
Also we know that in index notation:
∂vi
=\·v (10)
∂xi
∂φ
e = \φ (11)
∂xi i
Therefore substituting back from equations 10 and 11 into 9 we get:
∂φvi ∂vi ∂φ
\ · (φv) = =φ + vi ei · e = φ\ · v + v · \φ (12)
∂xi ∂xi ∂xi i
and equation 1 is proved.
b.
div(u × v) = v · curlu − u · curlv (13)
Equation 13 can be written in Gibbs notation as:

\ · (u × v) = v · \ × u − u · \ × v (14)

In index notation, the left hand side can be written as:

∂ ∂uj vk ∂uj vk
\ · (u × v) = ( e ) · uj vk εjkl el = ( ) εjkl ei · el = εjkl δil (15)
∂xi i ∂xi ∂xi
δil is only non zero when i = l. Thus:
∂uj vk ∂uj vk
εjkl δil = εjki (16)
∂xi ∂xi
Therefore:
∂uj vk
\ · (u × v) = εjki (17)
∂xi
As for the right hand side:

v·\×u=
∂
v i ei · e × uk ek =
∂xj j
∂uk
vi e · (ej × ek ) =
∂xj i
∂uk
vi e · εjkl el =
∂xj i
∂uk
vi εjkl δil =
∂xj
∂uk
vi εjki (18)
∂xj

u·\×v =
∂
ui ei · e × vk ek =
∂xj j
∂vk
ui ei · (e × ek ) =
∂xj j
∂vk
ui εjkl ei · el =
∂xj
∂vk
ui εjkl δil =
∂xj
∂vk
ui εjki (19)
∂xj
Therefore, equation 13 in index notation is written as:
∂uj vk ∂uk ∂vk
εjki = vi εjki − ui εjki (20)
∂xi ∂xj ∂xj
In order to prove equation 13, we start from the left hand side and use the chain rule to open the derivatives:
∂uj vk ∂uj ∂vk
\ · (u × v) = εjki = εjki vk + εjki uj (21)
∂xi ∂xi ∂xi
Then multiplying the first term by δkk and the second term by δjj ( δkk = δjj = 1 does not add anything to
the equation, however it helps in creating the dot product needed to prove the equations) :

∂uj ∂vk ∂uj ∂vk

εjki vk + εjki uj = εjki vk δkk + εjki uj δjj (22)
∂xi ∂xi ∂xi ∂xi
We know that εjki = −εikj . Thus:

∂uj ∂vk
εjki vk δkk + εjki uj δjj =
∂xi ∂xi
∂uj ∂vk
εijk vk δkk − εikj uj δjj =
∂xi ∂xi
∂uj ∂vk
εjki vk ek · ek − εikj uj ej · ej =
∂xi ∂xi
∂uj ∂vk
εjki ek · vk ek − εikj ej · uj ej (23)
∂xi ∂xi
From the definition of curl in index notation we know:
∂uj
\×u= εijk ek (24)
∂xi
and
∂vk
\×v = εikj ej (25)
∂xi
Therefore,
∂uj ∂vk
εjki ek · vk ek − εikj ej · uj ej = (26)
∂xi ∂xi
(\ × u) · v − (\ × v) · u (27)

Since dot product is commutative, we can rewrite equation 27 as:

(\ × u) · v − (\ × v) · u = v · (\ × u) − u · (\ × v) (28)

And thus equation 13 is proved.

c.
curl(u × v) = v · grad u − u · grad v + u · div v − v · div u (29)
Equation 29 in Gibbs notation is presented as:

\ × (u × v) = v · \ u − u · \ v + u \ · v − v \ · u (30)

For the index notation, starting from the left hand side of equation 29:
∂
\ × (u × v) = e × (uj ej × vk ek ) =
∂xi i
∂uj vk
ei × (ej × ek ) =
∂xi
∂uj vk
ei × (εjkl el ) =
∂xi
∂uj vk
εjkl ei × el =
∂xi
∂uj vk
εjkl εilm em (31)
∂xi
For the first term on the right hand side we have:

v·\ u=
∂uj
vi ei · ( e e )=
∂xk j k
∂uj
vi ∂ e · (ej ek ) (32)
∂xk i
Using the identity c · (a b) = (a · b)c:

∂uj
vi ∂ e · (ej ek ) =
∂xk i
∂uj
vi ∂ (e · e ) e =
∂xk i j k
∂uj
vi ∂ δij ek =
∂xk
∂ui
vi ∂ e (33)
∂xk k
Same with the rest of the terms:

v·\ u=
∂vj
ui ei · ( e e )=
∂xk j k
∂vj
ui e · (ej ek ) =
∂xk i
∂vj
ui (e · e ) e =
∂xk i j k
∂vi
ui e (34)
∂xk k

u \·v =
∂
u i ei ( e · v k ek ) =
∂xj j
∂vk
ui e (e · e ) =
∂xj i j k
∂vk
ui e δjk =
∂xj i
∂vj
ui e (35)
∂xj i

v \·u=
∂
vi ei ( e · v k ek ) =
∂xj j
∂uk
vi e (e · e ) =
∂xj i j k
∂uk
vi e δjk =
∂xj i
∂uj
vi e (36)
∂xj i
Thus equation 29 in index notation is given as :

∂uj vk ∂ui ∂vi ∂vj ∂uj

εjkl εilm em = vi ∂ e − ui e + ui e − vi e (37)
∂xi ∂xk k ∂xk k ∂xj i ∂xj i
In order to prove this equation, starting from the left hand side (equation 31), we use the identity shown in
equation 3.3.5 in Panton to change the alternating tensor into the Kronecker delta. Also, using the fact that
εilm = εlmi and εjkl = εljk we have:

∂uj vk
εjkl εilm em =
∂xi
∂uj vk
εljk εlmi em =
∂xi
∂uj vk
(δmj δik − δmk δij ) em =
∂xi
∂vk ∂uj
(uj + vk )(δmj δik − δmk δij ) em =
∂xi ∂xi
∂vk ∂vk ∂uj ∂uj
(uj δmj δik − uj δmk δij + vk δmj δik − vk δmk δij ) em (38)
∂xi ∂xi ∂xi ∂xi
The first term in equation 38 is only non zero when m = j and i = k.
∂vk
uj δmj δik em =
∂xi
∂vi
uj e =
∂xi j
∂vi
ui ej =
∂xi
u \·v (39)

The second term is only non zero when m = k and i = j:

∂vk
−uj δmk δij em =
∂xi
∂vk
−ui e δii =
∂xi k
∂vk
−ui e (e · e ) =
∂xi k i i
∂vk
−ui e · (ek ei ) =
∂xi i
∂vk
−ui ei · e e =
∂xi k i
−u · \ v (40)

Here we used the identity c · (a b) = (a · b) c to change ek (ei · ei ) into the dot product of a vector and a
Vector Calculus Panton 3.12

tensor. The third term is only non zero when m = j and i = k:

∂uj
vk δmj δik em =
∂xi
∂uj
vi e δii =
∂xi j
∂uj
vi e (e · e ) =
∂xi j i i
∂uj
vi e · (ej ei ) =
∂xi i
∂uj
vi ei · e e =
∂xi j i
v·\ u (41)

And the fourth term is only non zero when m = k and i = j:

∂uj
−vk δmk δij em =
∂xi
∂ui
−vk e =
∂xi k
∂ui
−vk ek =
∂xi
−v \ · u (42)

Putting equations 39, 40, 41 and 42 back into equation 38 and equating it with equation 31, we get:

\ × (u × v) = v · \ u − u · \ v + u \ · v − v \ · u (43)

Therefore, equation 29 is proved.

D
 Advanced Fluid Mechanics

Particle Kinematics
Lagrangian and Eulerian Frames - Material Derivatives

The stream function — which will be discussed in more detail later in the course — in cylindrical co-ordinates
(r, θ) for flow past a circular cylinder of radius a with clockwise circulation Γ is given by

a2
� �
Γ r
ψ(r, θ) = U r − sin θ + ln (1)
r 2π a

a) Write the stream function ψ(x, y) in Cartesian co-ordinates, and find the components of the velocity ux
and uy in the x and y directions.

Hint: The stream function is defined in terms of the velocity components as

∂ψ(x, y)
ux = (2)
∂y
∂ψ(x, y)
uy = − (3)
∂x
b) Derive the ordinary differential equations that govern the particle path lines.

c) Find the equation for the particle trajectory passing through the point r = 2a, θ = 0 (or equivalently,
x = 2a, y = 0).

d) Show that a particle on the surface of the cylinder always stays on the cylinder. Find the tangential
velocity component of such a particle, and determine the stagnation points.

Hint: In cylindrical co-ordinates,

1 ∂ψ(r, θ)
ur = (4)
r ∂θ
∂ψ(r, θ)
uθ = − (5)
∂r

e) Sketch the stream lines for the case Γ > 4πaU . What happens as r → ∞?
Continuum and Kinematics Particle Kinematics

Solution:

a2
 
Γ r
a) We have ψ(r, θ) = U r − sin θ + ln
r 2π a

To convert to Cartesian coordinates, we make the following transformations:

p y x
r → x2 + y 2 ; sin θ → p ; cos θ → p
2
x +y 2 x + y2
2

! p !
2
p a y Γ x2 + y 2
∴ ψ(x, y) = U x2 + y 2 − p p + ln
2
x +y 2 2
x +y 2 2π a
a2
 2
x + y2
  
Γ
= U 1− 2 y + ln (6)
x + y2 4π a2

To find ux (x, y) and uy (x, y), we use the equations 2 and 3. Upon carrying out the differentiation we have

a2 (y 2 − x2 )
 
Γ y
ux = U 1 + 2 2 2
+ (7)
(x + y ) 2π x + y 2
2

2
2a xy Γ x
uy = − 2 2 2
U+ (8)
(x + y ) 2π x + y 2
2

b) A pathline is the locus of points through which a particle of a fixed identity has traveled

Let X(t; x0 , y0 ) and Y (t; x0 , y0 ) be the Lagrangian coordinate position of a fluid particle. x0 and y0 are
parameters that refer to the initial position of the particle. We now have
dX
= ux (X, Y ) = ux (X(t; x0 , y0 ), Y (t; x0 , y0 ))
dt
dY
= uy (X, Y ) = ux (X(t; x0 , y0 ), Y (t; x0 , y0 ))
dt
We also have the initial conditions X(t = 0; x0 , y0 ) = x0 and Y (t = 0; x0 , y0 ) = y0 . These first order linear
differential equations can be solved to find the pathlines.

Another way to arrive at the above equations is to use the material derivative. We know that
DF ∂F
= + v · ∇F (9)
Dt ∂t
For consistency, the velocity expressed in the Lagrangian frame should equal the velocity expressed in the
Eulerian frame at time t, and hence v = ux êx + uy êy

Setting F = x, we get  
Dx ∂x ∂x ∂x
= +ux +uy
Dt ∂t x,y ∂x
|{z} ∂y
| {z } |{z}
1 0
0

Dx Dy
Therefore = ux and similarly, = uy
Dt Dt
Continuum and Kinematics Particle Kinematics

c) We know that the value of the stream function must be constant along a streamline. Therefore, we first
find the value of the stream function at the point (r = 2a, θ = 0) and we get

Γ
ψ(r, θ)|(2a,0) = ln 2 (10)
2π

Thus, any particle (r = 2a, θ = 0) must have the same value of the stream function all along its trajectory.
Hence, the equation of the trajectory of a particle located at (r = 2a, θ = 0) is

a2
 
Γ r
ψ(r, θ) = r − sin θ + ln =0
r 2π 2a

d) We go back to the stream function in cylindrical polar coordinates (equation 1), and using equations 4
and 5, we calculate ur and uθ to be

a2
 
ur = U 1 − 2 cos θ (11)
r
a2
 
Γ
uθ = −U 1 + 2 sin θ − (12)
r 2πr

Clearly, at r = a, ur = 0.

A formal way to show that a particle on the cylinder always stays on the cylinder is to check that r(t) = a
dr
solves the differential equation = ur , and satisfies the initial condition r(t = 0) = a. An informal way is
dt
just to check that ur = 0 whenever r = a, and that ur = 0 implies that the value of r will not change, i.e. it
always stays equal to a.

e) To sketch the velocity profiles, let us look at uθ on the surface of the cylinder (we know that ur (r = a) = 0)

Γ
uθ = −2U sin θ − (13)
2πa
uθ 2U Γ
⇒ θ˙ = =− sin θ − (14)
a a 2πa2

Let us plot θ˙ as a function of θ for the cases Γ < 4πaU and Γ > 4πaU , which is shown in Fig. 1.

It can be seen from equations 2 and 3 that as r → ∞ (x, y → ∞), the velocity field is purely horizontal with
magnitude U .

Case 1: Γ < 4πaU

The points A and B correspond to the points at which θ˙ = 0 on the surface of the cylinder. At these points,
the sign of θ˙ changes, and hence the angular velocity reverses direction. To find these points, we simply find
the points where uθ = 0 at r = a, (i. e)

Γ
2U sin θ =
2πa
Γ
⇒ sin θ = −
2πa

Since we know that ur = 0 at r = a, and also uθ = 0 at points A and B, these points are stagnation points.
Continuum and Kinematics Particle Kinematics

Figure 1: The variation of the angular velocity θ̇ at the surface of the cylinder. When Γ < 4πaU , A and B
are stagnation points. When Γ > 4πaU , there are no stagnation points on the surface of the cylinder.

Case 2: Γ > 4πaU

To find the stagnation points, we first set ur = 0

a2

� 
�
ur = U 1 − 2 cos θ = 0
r
⇒ cos θ = 0 or r = a

But at r = a, uθ =6 0 (this can be verified by setting r = a in equation 5), so there is no stagnation point on
the surface of the cylinder. This means that the stagnation point lies away from the surface of the cylinder.
We also know that this stagnation point must lie along the vertical line passing through the center of the
cylinder because cos θ = 0 ⇒ θ = 90◦ or θ = 270◦ (θ = 0 is the horizontal line through the center of the
cylinder).

To find this stagnation point, we use cos θ = 0 (sin θ = ±1)and set uθ = 0 solve for r.

a2

� 
�
Γ
uθ = 0 ⇒ U 1 + 2 = (15)
r 2πr

We have used sin θ = +1 in the above equation because sin θ = −1 has no real root. Solving this quadratic
equation we find that
1  p 
r= Γ + Γ2 − (4πaU )2 (16)
4πU

The other root lies inside the cylinder and is therefore discarded. A sketch of the flow profiles for the cases
Γ < 4πaU , Γ = 4πaU , Γ > 4πaU , is given on Page 181 of the course textbook, and has been reproduced
here for convenience.

D
Continuum and Kinematics Particle Kinematics

Figure 2: Different regimes of flow past a circular cylinder with circulation. Adapted from Fluid Mechanics
4th ed., P. K. Kundu and I. M. Cohen, Academic Press, 2008.
 Advanced Fluid Mechanics

Problem 1.00
This problem is from “2.25 Advanced Fluid Mechanics” by Ain Sonin

Rate of change of properties measured by a probe moving through the earth’s atmosphere — plus some things
about the earth and its atmosphere.

The pressure distribution in a static, constant-temperature planetary atmosphere modeled as an ideal gas is
given by
p = p0 e−z/H (1.00a)

where z is the altitude above a reference altitude z = 0, p0 is the absolute pressure at z = 0, and
RT
H= (1.00b)
Mg

is a length scale that characterizes the atmosphere. Its value is determined by the strength of the gravitational
acceleration and the parameters that appear in the ideal-gas equation of state,
RT
p=ρ ; (1.00c)
M

R = 8.32 JK−1 mol−1 is the universal gas constant, T is the absolute temperature (taken as constant in
this model of the atmosphere), M is the molar mass of the gas (0.029 kg/mol if the gas is air), and g is
the acceleration of gravity at or near the surface of the planet. For the “Standard” isothermal model of the
earth’s atmosphere, T = 288 K, p0 = 1.02 × 105 N/m2 if z = 0 at sea level, and consequently H = 8.43 km.
Note that the distribution given above is based on the assumption that H « a, where a is the planet’s
radius.

Suppose a sounding rocket or balloon equipped with a static-pressure sensor is traveling through the atmo­
sphere with given velocity (vx , vy , vz ).

1. In terms of the given quantities and z, derive an expression for the rate of change of pressure recorded
by the rocket’s sensor.
2. Evaluate this time of change at an altitude z = 20,000 m for a rocket traveling upward through
the earth’s atmosphere with a direction of 30◦ from the vertical and a speed of 465 m/s. (Answer:
−0.273 bar/min.)
3. Suppose a rocket carries instruments that measure both the instantaneous atmospheric pressure p and
the rate of change of that pressure, dp/dt. Given the value of these two quantities at a particular time
and p0 and H, derive expressions for the rocket’s instantaneous altitude and vertical (upward) velocity.

Additional things to think about, if you are so inclined:

4. Suppose the Earth’s atmosphere is isothermal and radially symmetric around a perfectly spherical
earth with radius a = 6400 km. What is the total mass of the Earth’s atmosphere? What fraction is
this of the solid and liquid parts of the planet’s mass? (Answers: 5.35 × 1018 kg and 8.96 × 10−7 .)
5. Show that 99% of the Earth’s atmosphere’s mass resides below an altitude of 39 km.
6. If the atmosphere heats up by 10◦ C, by how much will the absolute pressure at sea level change?
(Answer: it will not change at all.)
Solution:

1. Take material derivative of (1.00a).

=0
Dp
Dt
= 8
∂p
∂t
+ v · 'p =
∂p ∂z
·
∂z ∂t
p0
= − e−z/H ·
H
∂z
∂t
∂z
where = vz .
∂t

Dp vz p0 −z/H
=− e . (1.00d)
Dt H

2. Calculate the z component of the given velocity, 465 m/s, vz = 465 cos 30◦ . Plug in the values for vz ,
z, H, and p0 into Eq. (1.00d):

Dp 2
(z = 20,000 m) ≈ −454.36 N/m s ≈ −0.273 bar/min.
Dt

(Hint: 1 bar ≡ 105 N/m2 .)

Dp
3. Given: p0 , H, p,
Dt
Unknown: z, vz
Equations: (1) and (1.a) ,
obtain an expression for z in terms of H, p0 , and p by rewriting (1.00a):
p z
ln( )=−
p0 H

� �
p0
∴ z = H ln .
p
Take Eq. (1.00d) and solve for vz :

Dp vz
= − · p0 e−z/H
Dt H P � �
=p
Dp vz p
= −
Dt H

H Dp
∴ vz = − · .
p Dt

4. Rewrite (1.00c) and combine it with (1.00a) to solve for ρ:

M
ρ= p0 e−z/H . (1.00e)
RT

M p0
Massatm = ρ dV = e−z/H dV
V RT V

where dV = 4πr2 dz = 4π(a + z)2 dz.

 Perform the integration by parts twice to solve the following integral:

4πM p0 ∞ −z/H
Z
Massatm = e (z + a)2 dz
RT 0
4πM p0
= (Ha2 + 2H 2 a + 2H 3 ) (1.00f)
RT
≈ 5.35 × 1018 kg.

(Hint on Integration by parts udv = uv − vdu When integrating by parts the first time, let
u = (z + a)2 and dv = e−z/H dz. For the second time, let u = (z + a) and dv = e−z/H dz.)
The mass of the earth is known as 6 × 1024 kg.
Massatm
∴ ≈ 8.92 × 10−7 .
Massearth

5. Perform the same integration as in (d ) but with [0 39,000] as limits.

6. The absolute pressure at sea level is a measure of the total weight of the atmosphere that is sitting
on top. Because the total atmospheric mass does not change, the pressure remains constant as well.
Mathematically, if one integrates dpdz = ρg on both sides from 0 to ∞ in terms of dz, RHS becomes
∞
0
ρdz, which is constant. Therefore, LHS, p(∞) − p(0), should also be constant. Since p(∞) is zero
by definition, p(0), or the absolute pressure at sea level, becomes constant, independent of temperature
change.