Advanced Fluid Dynamics

MAK 503E
Dr. Hasan Gunes
[email protected]
http://atlas.cc.itu.edu.tr/~guneshasa
INTRODUCTION
DEFINITION OF A FLUID
A fluid is a substance that deforms continuously under the application of a shear
(tangential) stress no matter how small the shear stress may be.

F/A≡ τ
t0 t2
t1

t2 > t1 > t0

No-slip condition

Mechanics of fluids
1. Inviscid flow: viscosity assumed to be zero –
simplify analyses but meaningful result needed
2. Viscous flow: viscosity important
Solid deforms when a shear stress is applied, but it does not deform
continuously.

True fluids: all gases, common liquids, water, oil, gasoline, alcohol, ...
Non-true fluids: high-polimer solutions, emulsions, toothpaste, egg white etc.
Rheology: general study of flow and deformation of materials.

Continuum Fluid Mechanics:

Composed molecules in constant motion
Actual molecular structure molecules & gaps among them

Calculate/measure Replace the actual molecular

macroscopic fluid properties, structure by a hypothetical
e.g. viscosity continuous medium
Continuous medium means
a. There are no gaps (in the fluid) or empty spaces
b. Each fluid property is assumed to have a definite value at each point in space
ρ = ρ ( x, y, z , t ) ; T ( x, y, z , t ) ; V ( x, y, z , t ) ; p( x, y, z , t )
c. All the mathematical functions entering the theory are continuous functions except
possibly at a finite number of interior surfaces seperating regions of continuity.

ρ 1 , µ1

ρ 2 , µ2

ρ 1 , µ1

d. Derivatives of the functions are also continuous too if they enter the theory

Knudsen number, Kd = λ / L << 1

λ : mean free path of molecules
L : characteristics length
Ex. What is the density of fluid at point C ?.

y
Volume, V
of mass, m
y0

C
Volume, δV
of mass, δm
mean density = ρ =
m
∀ ∀
x
x0
In general mean density, ρ is not
equal to density at point C.
z0

δm
z

ρ ≡ lim
δ∀→δ∀′ δ∀
 δm
Volume so small, molecules
cross into and out of CV

δ∀

δ∀
δ∀′ Still have large
enough molecules

ρ = ρ ( x, y , z , t )
for consistent result

scalar field
e.g. 1m3 air ≈ 2.5x1025 molecules
Similarly, velocity at point C defined as the instataneous velocity of the fluid particle

Fluid particle: a small mass of fluid of fixed identity of volumeδ∀′

which, at a given instant, is passing through point C.

V : velocity field ; V ( x, y , z , t ) : vector field

V =ui + v j + wk , u, v, w: components of velocity vector-scalar
u = u ( x, y , z , t )
v = v ( x, y , z , t )
w = w( x, y, z , t )

Steady/unsteady flow:
Properties at every point in a flow field do not change with time.
Time-independent flow, stationary flow.

∂η ∂ρ
=0 = 0 , or ρ = ρ ( x, y , z )
∂t ∂t
,

Incompressible flow: ρ = const.

Compressible flow: ρ = variable
 = M ≤ 0.3 → gases also incomp. → V ≅ 100 m/s air at standart conditions
V
a
a = speed of sound

Vref Lref ρ
Re =
µ

Small Re → laminar: no macroscobic mixing, fluid flow in layers on

laminae (smooth flow pattern)
Large Re → turbulent: time-dependent, 3-D , chaotic motion
(random) macroscopic mixing

Basic Equations:
1. Conservation of mass (continuity eq.)
2. Conservation of momentum (Newton’s second law)
3. Conservation of energy (First law of thermodynamics)
4. 2nd law of thermodynamics
5. Equation of state ρ= ρ(p,T)
Unknowns ρ,u,v,w,p,T,s total number of equations ?
VISCOSITY •ability of a fluid to flow freely
Density : measure of the “heaviness” of a fluid.
Viscosity (µ) : measure of the “fluidity” of a fluid.

δl
Force, δFx
A A’ D D’ Velocity, δu

No slip
δy
y
δα

x
B C No slip

e.g. water & oil approx. have same density but behave differently when flowing

Fluid element at time, t : ABCD

Fluid element at time, t + δt : A’BCD’
 τ δ Fx dFx
τ yx = lim =
δ Ay →0 δ A
y x
plane direction
acts that stress
acts
y dAy
on

δ Ay : area of fluid element in contact with the plate

δα dα
=
δ t →0 δ t
Deformation rate = lim

δ l = δ uδ t
dt

δl
For small δα , tanδα ≅ δα = ⇒ δ l = δαδ y = δ uδ t
δy
δα δ u dα du
= ⇒ =
δt δ y
τ yx
dt dy
Subject to shear stress,

fluid experiences a time rate of deformation (shear rate) as du/dy

τ yx
du
∼ Newtonian fluid, linear relationship
dy
τ yx =µ
du
µ: [Pa.sec]=[kg/m-sec] (absolute viscosity)
dy
du
: velocity gradient
dy
= 0 ⇒ τ yx = 0 (µ ≠ 0)
du
No relative motion :
dy
Example: Couette Flow
U0

h y

x
 ⎛ y⎞ τ =µ =τ y x = µ i>0
u = U0 ⎜ ⎟ i
du U0
⎝h⎠
Shear stress
dy surface direction
h
Stress sign convention
Stress comp. Plane Direction
+ + +
+ - -
- - +
- + -

Which direction does the stress act?

Depends on the surface we are looking at.

Positive outward normal on

(+)
U0 y-surface (+)

(-) (-)
y Fluid
(+)
x
(-) (-)
(+)
 Bingham plastic,
toothpaste, mayonnaise
τ τ

Crude oil Shear thining, polimer

solutions, latex paint, blood

Water (60 °F)

Water (100 °F)
Shear tickening,
water-sand
Air (60 °F) mixture

0 du/dy 0 du/dy
Time rate of
deformation

Bingham plastic: can withstand a finite shear stress without motion not fluid
But once the yield stress is exceeded it flows like a fluid not a solid
Non- true fluids
 du
: finite
dy

τ yx finite ; u must vary continusly across the flow,

no abrupt change between adjoining elements of fluids.
Consider solid boundary : No-slip B.C.

With respect to size of fluid molecules, solid surfaces, no matter how well polished, have
irregularities, i.e cavities filled by fluid.
Fluid immediately in contact with the boundary, has the same speed with it.
Viscous fluid No-slip B.C.
Perfect Fluid: τ =0
No internal resistance to a change in shape (µ=0)
Fails to predict drag of body

Seperation occurs due to APG ⎛⎜ > 0 ⎞⎟ increasing pressure gradient in flow direction
dp
⎝
dx ⎠
 Flow past a circular cylinder: Inviscid theory
2-D, ρ=constant (incompressible flow), irrotational flow
Superposition of doublet & uniform flow
Vθ
Vr
U∞
r
R

Stagnation points
∂φ 1 ∂φ
Vr = − , Vθ = − θ
∂r r ∂r
⎛ R2 ⎞ ⎛ R2 ⎞
= U ∞ ⎜1 − 2 ⎟ cos θ , = −U ∞ ⎜1 + 2 ⎟ sin θ
⎝ r ⎠ ⎝ r ⎠
Vr Vθ
radial azimuthal
velocity velocity

V = Vr er + Vθ eθ = 0
(r ,θ ) = ( R, 0), ( R, π ) stagnation points
 π
θ= ⇒ Vθ = −2U ∞ = Vθ ,max
2
At the cylinder surface, r=R , Vr = 0 , Vθ = -2U∞sinθ≠0
Violates the “no-slip” condition between solid & fluid
Pressure distribution at the cylinder surface, apply Bernoully eq. (neglect elevation dif.)
Limititations:
1. Steady
2. ρ=constant
3. frictionless flow: inviscid
4. flow along a streamline
5. gravity force

+ + gz = + + gz
P∞ U ∞2 P V2
ρ ρ 2

( ) ( )
2

P − P∞ = ρ U ∞2 − V 2 = ρU ∞2 1 − 4sin 2 θ
1 1
2 2
P − P∞
cp = = 1 − 4sin 2 θ pressure distribution on cylinder, valid for inviscid flow
ρU ∞2
1
Cp: pressure coef. [-]
2
 One side of cylinder
1 FD : drag (force): force component parallel to

FD = ∫ dFx
0 the freestream flow direction :

FD : ∫ − P dA cos θ
-1
-2

(
d A = Rdθ ir = Rdθ i cos θ + j sin θ )
Ax dAx

0 π/2 π θ
dAx = R cos θ dθ (projection in x-direction)
ir dAy = R sin θ dθ
d A = Rdθ ir
θ
y
x
R

p − p∞ = ρU ∞2
1 3 dynamic
2 pressure units
p − p∞ = − ρU ∞2
3 lower than p∞
2 (atm. press.)
 ∫ dA = (b) R ∫ dθ = (b)2π R
2π

(
FD = − ∫ ⎢ p∞ + ρ u∞2 1 − 4sin 2 θ )⎤⎥⎦R cosθ dθ
A 0

⎡
2π
1
0 ⎣
2

⎛ ⎞ sin 3θ
2π 2π

= − ⎜ p∞ + ρ u∞2 ⎟ R sin θ + 2 ρ u∞2 R =0

1
⎝ 2 ⎠ 3
FD = 0!
0 0

(Noviscosity) d’Alembert paradox: inviscid flow past immersed bodies, drag=0 ,

∫
y separation
symmetric pressure distribution
FL = − pdAy = 0 Lift is zero! !
Pressure
drops
A
In reality, large drag force! R
No symmetry
Wrt y-axis θ
wake
Wake structure: Karman
depends on Re (gets Vortex
complicated as Re ) street
cp = 0 p = p∞
cp = 1 p = p∞ + ½* ρU2∞
+1 Supercritical Re
90° 180° (turb.)
0
-1
Subcritical Re
-2 (laminar) /typical
experimental
-3 trends)

≈ 60° (theory
is valid up to)

DRAG COEFFICIENT; CD :
FD = CD ρV 2 A
1
totat
2 projected
drag frontal
force area
Power ≡ Drag * Vel.
Power, that we pay to move aircraft.
Racing cars – unload tires, reduce drag & lift

FL Drag: due to
1. Pressure forces
2. Friction forces (shear stress) flow over a flat
plate parallel to the flow
W
Inlet ducts to produce

FD = ∫ τ w dA
down force

FD = f1 ( D, V , µ , ρ )
From dimensional analysis,
CD = CD (geometry,ReL) valid
⎛ ρVD ⎞
= ⎟ = f 2 (Re)
for ρ=const. over any body
2⎜
ρV D ⎝ µ ⎠
FD
2 2
f

A2 = f3 (Re) A:cross-sectional area

 U Charact. length
ρ,µ

Transition to turb. On
Theory Friction cylinder causes “drag
stokes sol. drag %5 of crises” at Rec≈3*105
Laminar total Rec = f (roughness)
Smooth
~1.2 surface

~0.3

100 103 104 105 106 Re

Rough
surface
Drag coef. Of a circular cylinder
Rough surface: early turbulance due to roughness, drag crisses occurs earlier, dimpling
on goal balls
Freestream turbulance: drag crisses occurs earlier, similar behaviour with sphere
“dimples” to “trip” the B.L. to cause TBL.
S.P Rec ≈2*105 (smooth surface)
Re< Rec
Turbulent wake low
pressure

L.BL

Upsream of the cyl.

midsection

S.P
Re>Rec
Smaller wake

L.BL Net pressure force on cylinder is

reduced CD
Boundary-Layer Control: spin
w:angular velocity (rd/sec)
moving surface to reduce skin friction effects on BL

FL (lift force)
CD
0.6
p CL
V wake 0.4

p 0.2

1 2 4
Spiln ratio,

Weak function of Re
Low spin ratio , wD/2V ≤ 0.5 neg. lift!
Flow pattern, lift and drag coef. for a smooth spinning sphere in uniform flow.
The wake is not symmetric wrt incoming vel.
AIRFOIL:

thickness, t
angle of attack
s, span
α

U∞ Thickness ratio, t/c

Chord, L Aspect ratio , AR = s/L

U∞ , p∞
Thin B.L with no
separation

α=0 Thin wake

α<5°
Leading edge
Stream lines Trailing
LIFT edge
2-D infinite span
 Re sep. point

wake
sep. point reattachement

As long as flow reattaches its symmetric & no lift

Pstag

P∞

<0 > 0 ⇒ possible seperation

dP dP
dx dx
FPG APG
 ρU ∞2 A
1
lift=CL
2
CL : lift coefficient

CL

α
 B.L with seperation
U∞ , p∞ (stall)
faster
Broad wake

slower

α>0
10-15°
Seperation , loss of lift increase drag

FORCE
CD Optimum cruise Pstag (LIFT)
CL
Lower surface

Upper
surface
CL Pressure distribution α≠0
PROPERTIES OF A FLUID & FLOW FIELD:
1. Kinematic properties: linear velocity, angular velocity, vorticity, acceleration, and
strain rate Flow field properties
2. Transport properties: µ, k
3. Thermodynamic properties: p, ρ, T, h, s, cp, Pr, β
4. Other properties: surface tension, vapor pressure, etc

DESCRIPTION OF FLUID MOTION:

A. Lagrangian description: useful in solid mech.
B. Eulerian description: proper choice in fluid mechanics

A) specifies how an individual particle moves through space.

formulation is always time-dependent
B) specifies the velocity distribution in space and time. i.e. specifies how a particle at a
point would be.

V = V ( x, y, z , t ) defines the motion at time t at all points of space occupied by the fluid
Advantages of Euler description of fluid motion:
• no need to follow the path of particles
• in some cases, unsteady flow can be considered as steady by appropriately
selecting coordinates
 Ex:

Kinematics, Substantial derivative

1) Acceleration of a Fluid Particle in a Velocity Field: a p

Problem: how to relate the local motion of a particle to the velocity field

Study velocity field as a function of position & time, not trying to follow any specific
particle paths. But conservation laws are formulated for particles (systems) of fixed
identity; i.e., they are Lagrangian in nature.
 y
Particle path
Particle at time, t
Particle at time, t+dt

V ( x, y , z , t ) = ui + v j + wk , r = r ( x, y , z )
z

For a particle moving in velocity field at time t.

V p = V ( x, y , z , t )

= V ( x + dx, y + dy, z + dz , t + dt )
t

at time t+dt V p
t + dt
The change in particle velocity can be shown by differential calculus to be,

∂V ∂V ∂V ∂V
dV p = dx + dy + dz +
∂x ∂y ∂z ∂t
dt

Or for total acceleration of particle,

∂V dx ∂V dy ∂V dz ∂V
ap = = + + +
dV p
dt ∂x dt ∂y dt ∂z dt ∂t

∂V ∂V ∂V ∂V
u v w

ap = =u +v +w +
DV
Dt ∂x ∂y ∂z ∂t
total accel. local
of particle convective accel. accel.

D (..) Substantial derivative

Material derivative
Dt Particle derivative

∂u ∂u ∂u ∂u
= =u +v +w +
Du
∂x ∂y ∂z ∂t
3 comp. Eq. ax , p etc.
Dt
 Generalization: let A represent any property of fluid, ρ (either scalar or vector)

A = A ( x , y , z, t )
∂A ∂A ∂A ∂A
=u +v +w +
DA
Dt ∂x ∂y ∂z ∂t
local time dependent
change in property due to change in the property
motion through flow field

Substantial derivative: generalizes the rate of change of a local property of a material

particle to the movement of the particle through the flow field.

DA ∂A
( )
In vector form;

= + V .∇ A
Dt ∂t
∂ ∂ ∂
∇=i + j +k
∂x ∂y ∂z
gradient operator

( ) ∂ ∂
V .∇ = u + v + w
∂x ∂y
∂
∂z
 ω
Ex1:
Vorticity change, (vector)

Dω ∂ω ∂ω ∂ω ∂ω
=u +v +w +
Dt ∂x ∂y ∂z ∂t
Density change, ρ (scalar)

Dρ ∂ρ ∂ρ ∂ρ ∂ρ
=u +v +w +
Dt ∂x ∂y ∂z ∂t
⎛ y⎞
V = U0 ⎜ ⎟ i
Ex2:
⎝h⎠
ρ = ρ (1 + ax − bt ) = ρ ( x, t )
U0

⎡ ⎤
Dρ ⎛ y⎞ ⎢ y ⎥
= ⎜ U 0 ⎟ ρ0 a − ρ0b = ρ0 ⎢U 0 a − b ⎥
Dt ⎝ h ⎠
h
⎢⎣ motion unsteady ⎥⎦
y
h
x
Motion & Deformation of a Fluid:
In order to develop differential equations of motion for a fluid we need to understand the
general type of motion
Examine for 2-D
Four different types of motion or deformation
1. Translation
2. Rotation
3. Distortion (i.e. deformation)
• Angular deformation - shear strain
• Linear deformation (extensional strain) – dilatation
Consider each from of motion individually for time interval ∆t
1)Translation: defined by displacements u ∆t & v ∆t,
that is, rate of translation is (u,v)
y
u∆t

v∆t
v
(displacement)
∆y u

∆x
x
2) Rotation:

Motion about centroid

y due to motion normal
to x-dir. with pozitif
rotation

∂v
∆ a = ( ∆x ) ∆ t
∆

∂x
∆y
∆α
Derivation of
∆x rotation: vorticity

Rotation: average angular rotation of 2 perpendicular line elements (x&y)

1 ⎛ dα d β ⎞
ΩZ = ⎜ − ⎟
2 ⎝ dt dt ⎠
 Ω: rotation
∆α ∆a / ∆x
Ω of line x= lim
∆t →0 ∆t ∆t
= lim
∆t →0

∂v
dα ( ∆x ) ∆t / ∆x ∂v
∂ =
∆t ∂x
= lim x
∆ →

d β ∂u
dt t 0

Likewise, Ω of line y: =
dt ∂y
1 ⎛ ∂v ∂u ⎞
ΩZ = ⎜ − ⎟
2 ⎝ ∂x ∂y ⎠
Ω Z rotation about axis parallel to z-direction. i.e. angular velocity rate of rotation

Rate of rotation : angular velocity, Ω

Rate of translation : velocity , u, v

1 ⎛ ∂w ∂v ⎞ 1 ⎛ ∂u ∂w ⎞
Ω = iΩ x + jΩ y + kΩ z = ∇ × V
Similarly, rotation about x & y axes,

Ωx = ⎜ − ⎟ , Ωy = ⎜ − ⎟
1
2 ⎝ ∂y ∂z ⎠ 2 ⎝ ∂z ∂x ⎠ 2

ω = 2Ω ω = ∇ ×V ω = curlV
For conventions sake, we define vorticity of a fluid particle as,
, ,
3) Angular deformation: (shear strain)
average decrease of the angle between two lines which are initially perpendicular

( dα + d β ) : shear strain increment

1
2
1 ⎛ dα d β ⎞ 1 ⎛ ∂v ∂u ⎞
∈xy = ⎜ + ⎟= ⎜ + ⎟ ∂u ⎛ ∆y ⎞
2 ⎝ dt ⎠ 2 ⎝ ∂x ∂y ⎠ ⎜
∂y ⎝ 2 ⎠
⎟ ∆t
y
dt
Shear strain rate

1 ⎛ ∂w ∂v ⎞ 1 ⎛ ∂u ∂w ⎞ ∂v ⎛ ∆x ⎞
⎜ ⎟ ∆t
∆y/2
∈yz = ⎜ + ⎟ , ∈zx = ⎜ + ⎟ ∂x ⎝ 2 ⎠
2 ⎝ ∂y ∂z ⎠ 2 ⎝ ∂z ∂x ⎠

⎛ ∈xx ∈xy ∈xz ⎞

⎜ ⎟ ∆x/2
∈ij = ⎜ ∈yx ∈yy ∈yz ⎟
⎜∈ ∈ ∈ ⎟
Strain rate tensor
⎝ zx zy zz
⎠ x
4. Linear deformation: (extensional strain)
Which motion(s) will result in stresses? Angular & linear deformation,
relative change in dimensions of line element ∆x, ∆y : STRAIN
Result: Angular def. ~ shear stress

∂u ⎛ ∆x ⎞
Linear def. ~ normal stress
∆x + 2 ⎜ ⎟ ∆t − ∆x
∂u ⎛ ∆x ⎞ ∂x ⎝ 2 ⎠
⎜ ⎟ ∆t ε ∆ =
∂x ⎝ 2 ⎠ ∆x
xx t
y
∂u
ε xx =
∂x
ε xx ∆t: extensional strain in x-direction

∂v ⎛ ∆y ⎞
⎜ ⎟ ∆t
∂y ⎝ 2 ⎠
∆x

x
(extentional strain) dilatation or increase in volume is due to velocity derivatives

∂u ∂v ∂w
∂x ∂y ∂z
, ,
 ∂u ∂v ∂w
ε xx = , ε yy = , ε zz =
∂x ∂y ∂z
∇.V = ε xx + ε yy + ε zz
DV ∂V
Dt
=
∂t
( )
+ V .∇ V =
∂V
∂t
⎛V 2 ⎞
+ ∇ ⎜ ⎟ − V × ∇ ×V
⎝ 2 ⎠
( )
ω = ∇ ×V = 0
For irrotational flow, let φ be a continuous scalar function such that
If in the flow field irrotational flow

∂φ ∂φ ∂φ φ = φ ( x, y , z , t )
V = ∇φ = i+ j+
∂x ∂y ∂z
k

ω = ∇ × ( ∇φ ) = 0
φ
identically zero.

V = ∇φ
• For irrotational flow there exist a scalar velocity potential function such that

Vorticity generators : boundaries (velocity gradients)

• Real flows are always rotational
But Boundary Layer Theory vorticity effects are confined in a thin layer adjacent to
the boundary
Outside the boundary layer flow can be treated as irrotational.
∂ 2φ ∂ 2φ ∂ 2φ
ω = 0 V = ∇φ ∇.V = ∇. ( ∇φ ) = 2 + 2 + 2 = 0 & ρ=const.
∂x ∂y ∂z
If

∇ 2φ = 0 Laplace eq. (linear PDE)

∇.V = divV change of (velocity) field at a point; indicates linear expansion of a field
(i.e. changes parallel direction of interest)
∂u ∂v ∂w
∇.V = + + = ε xx + ε yy + ε zz
∂x ∂y ∂z
Ex: V = axi
∂
∇.V = i a×l = a
∂x
y

∂u
a:expansion at a point
ε xx =
∂x
extensional strain (linear deformation)
x
 • If ρ=const. → conservation of mass → ∇.V = 0 (no expansion)

∂v
A

∂v
v+ ∂y
dydt
∂y
dy A D

dy dy

v
dx C dx C
∂u ∂u
B
u + dx
u B
∂x ∂x
t t+dt dxdt

∂u ∂v ∂u ∂v
Rectengular element under the influence of normal stresses

dxdtdy + dydtdx +
∂x ∂y ∂x ∂y ∂u ∂v
= + = ∇.V
dxdt dydt

∂x ∂y
Rate of increase in unit area :
dxdydt
Fluid behavior is a combination of these fluid motions, so we need to be able to express
cumulative behavior mathematically.
Cauchy-Store Decomposition
Consider 2-D velocity field, V = ui + v j & fluid element moving from point P through a

distance d r = dxi + dy j

V p + dV
y
Vp

dr
P

( )
x
V = V p + dV = V p + d r. ∇V

∂V ∂V ⎡ ∂u ∂u ⎤ ⎡ ∂v ∂v ⎤
=V p + dx + dy = i ⎢u p + dx + dy ⎥ + j ⎢v p + dx + dy ⎥
∂x ∂y ⎣ ∂x ∂y ⎦ ⎣ ∂x ∂y ⎦

Consider,
∂u 1 ∂u 1 ∂u 1 ∂v 1 ∂v
dy = dy + dy + dy −
∂y 2 ∂y 2 ∂y 2 ∂x 2 ∂x
dy

split in half add&subst.

 Rearrange,
∂u 1 ⎛ ∂u ∂v ⎞ 1 ⎛ ∂u ∂v ⎞
dy = ⎜ + ⎟ dy + ⎜ − ⎟ dy
∂y 2 ⎝ ∂y ∂x ⎠ 2 ⎝ ∂y ∂x ⎠
Similarly,
∂v 1 ⎛ ∂v ∂u ⎞ 1 ⎛ ∂v ∂u ⎞
dx = ⎜ − ⎟ dx + ⎜ + ⎟ dx
∂x 2 ⎝ ∂x ∂y ⎠ 2 ⎝ ∂x ∂y ⎠
⎡ ⎤ ⎡ ⎤
⎢ 1 ⎛ ∂u ∂v ⎞ 1 ⎛ ∂u ∂v ⎞ ⎥ ⎢ 1 ⎛ ∂v ∂u ⎞ 1 ⎛ ∂v ∂u ⎞ ⎥
angular def.

∂u ∂v
linear def. rotation

V =i ⎢ u p + dx + ⎜ − ⎟ dy + ⎜ + ⎟ dy ⎥ + j ⎢ vp + dy + ⎜ − ⎟ dx + ⎜ + ⎟ dx ⎥
trans.

⎢ trans. ∂ 2 ⎝ ∂y ∂x ⎠ 2 ⎝ ∂y ∂x ⎠ ⎥ ⎢ ∂ 2 ⎝ ∂x ∂y ⎠ 2 ⎝ ∂x ∂y ⎠ ⎥
⎢⎣ ⎥⎦ ⎢⎣ ⎥⎦
x y
linear def.
rotation angular def.

or in terms of tensors,
⎡ ∂u 1 ⎛ ∂v ∂u ⎞ ⎤ ⎡ 1 ⎛ ∂v ∂u ⎞ ⎤
⎢ ⎜ + ⎟⎥ ⎢ ⎜ − ⎟⎥
⎢ ∂x 2 ⎝ ∂x ∂y ⎠ ⎥ 2 ⎝ ∂x ∂y ⎠ ⎥
+ dr ⎢
0
V =V p + d r
⎢ 1 ⎛ ∂u ∂v ⎞ ∂v ⎥ ⎢ 1 ⎛ ∂u ∂v ⎞ ⎥
⎢ ⎜ + ⎟ ⎥ ⎢ ⎜ − ⎟ ⎥
⎢⎣ 2 ⎝ ∂y ∂x ⎠ ∂y ⎥⎦ ⎢⎣ 2 ⎝ ∂y ∂x ⎠ ⎥⎦
0

= V p + d r. Eij + d r. Ω
vorticity
rate of tensor
strain
tensor
 ⎡∈xx ∈xy ⎤ ⎡ Ω xx Ω xy ⎤
Normally, write tensors (2-D)

Eij = ⎢ ⎥ Ωij = ⎢
∈
⎣ yx ∈yy ⎦ ⎣Ω yx Ω yy ⎥⎦
,

accounts for distorsion accounts for rotation

Ex: Given a shear flow, V = U

y
i , determine components of deformations & rotation
h
U

∂u
ε xx = =0
∂x
h y

1 ⎛ ∂v ∂u ⎞ 1⎛ ⎞ U Ω xx = 0 always
∈xy = ⎜ + ⎟ = ⎜ + ⎟=
U
2 ⎝ ∂x ∂y ⎠ 2⎝ ⎠ 2h 1 ⎛ ∂v ∂u ⎞ 1⎛ U⎞
0
Ω xy = ⎜ − ⎟= ⎜ − ⎟ = −
h U
∂v 2 ⎝ ∂x ∂y ⎠ 2⎝ h⎠
0
∈yy = =0
2h
∂y Ω yy = 0

1 ⎛ ∂u ∂v ⎞ U 1 ⎛ ∂u ∂v ⎞ U
∈yx = ⎜ + ⎟ = Ω yx = ⎜ − ⎟=
2 ⎝ ∂y ∂x ⎠ 2h 2 ⎝ ∂y ∂x ⎠ 2h
 Ω yx (+ ) y
rate of

∈yx
y strain

Ω xy (−)
∈xy rotation

x x

Relation Between Stresses & Rate of Strains σ ij ∼∈ij

∂u ∂v ∂w
Strain rates : symmetric second-order tensor

⎛ ∈xx ∈xy ∈xz ⎞ ∈xx = , ∈yy = , ∈zz =

⎜ ⎟ ∂x ∂y ∂z
∈ij = ⎜ ∈yx ∈yy ∈yz ⎟
⎜∈ ∈ ∈ ⎟ 1 ⎛ ∂v ∂u ⎞ 1 ⎛ ∂w ∂v ⎞
∈xy =∈yx = ⎜ + ⎟ ∈yz =∈zy = ⎜ + ⎟
⎝ zx zy zz
⎠ 2 ⎝ ∂x ∂y ⎠ 2 ⎝ ∂y ∂z ⎠
∈ij =∈ ji 1 ⎛ ∂u ∂w ⎞
∈zx =∈xz = ⎜ + ⎟
2 ⎝ ∂z ∂x ⎠
Remember:
Transport properties of fluid; µ, k , α
• Viscosity: a property of fluid: ability of a fluid to flow freely, relates applied stress to the
resulting strain rate U

h y

x
no-slip condition
u( y) = U
y
linear profile

τ yx = f (∈yx )
h
general relations will be considered later!

Or Newtonian τ yx = µ = 2 µ ∈yx = µ
For simple fluids such as water, oil, or gases relationship is linear
U du
h dy

τ yx = f (∈yx )
µ: (coef. of) viscosity [N s /m2 = Pa.s] , µ= µ(T,p) different for liquids & gases

is not linear, the fluid is non-newtonian

 ideal bingham plastic
τ Pseudoplastic – shear thinning µ as
shear τ Polimer solutions , toothpaste
stress
dilatant – shear tickening,
yield suspensions, starch,
sand
stress
Newtonian ,
µ=constant

0 du/dy
time rate of deformation,
Rigid body Є inviscid flow (ideal
fluid)

τ yx ≅ 2 K ∈nyx Power-law approx. Power-Law:

⎛ TT0 n −1 ⎞ n
n=1 Newtonian K=µ
τ yx = ⎜ ke ∈yx ⎟ ∈yx
⎝ ⎠
n<1 pseudoplastic
n>1 dilatant (kaboran)
K, n : material parameters (rheological models of blood)
DIFFERENTIAL EQUATIONS OF MOTION

Review of control volume (CV) versus differential eqs. approach

a) CV eqs. (mass, momentum, energy)
1. Global characteristics
2. Must know or assume inflow & outflow profile
3. Easy & somewhat approximate

b) Differential eqs. (mass, momentum, etc.)

1. Detailed profile charact.
2. Only boundary & initial conditions
3. Easy to very difficult
exact behaviour (for laminar flow)

Objectivies:
1. Derive mass D.E. (continuity)
2. Derive momentum D.E.
• General
• Navier-stokes (stress ~ strain)
3. Solutions
∀

Conservation of Mass: CONTINUITY EQUATION

d ∀ = dxdydz
dm = ρ d ∀
y
Diff. Element

∂
∫ ( )
ρ d ∀ + ∫ ρ V .n dA = 0
x

∂t cv

( )
∂ ( ρu ) ∂ ( ρv) ∂ ( ρ w)
z cs

div ρV = + +
∂x ∂y ∂z
CV formulation
⎡time rate of change ⎤ ⎡NET mass
⎢of the fluid mass inside⎥ + ⎢flow rate
⎤
⎥ =0
∂ρ
∂t
∂∀ + ⎡ div ρV ⎤
⎣ ⎦ ( ) d∀ = 0 , d∀ ≠ 0
⎢ ⎥ ⎢ ⎥

( )
⎢⎣the CV ⎥⎦ ⎢⎣through the CS⎥⎦
∂ρ
change of field at a point

+ div ρV = 0
∂t
a scalar equation
 ∂ρ
∂t
+ ∇. ρV = 0 ( ) valid for any coordinate system

∂u ∂v ∂w
divV = + + = ∇.V
∂x ∂y ∂z
V = ui + v j + wk
scalar

fixed systems (particles). If m is the mass and ∀ is volume of fixed particle,

II. method: particle derivative. Conservation laws are Lagrangian in nature, i.e. apply

m = ρ ∀ = const
( )
D ( .) ∂ ( .)
Conservation of mass

= + V .∇ (.)
∂t
D ( ρ∀ )
Dt

=0⇒ =0
Dm

D∀ Dρ Dρ 1 D∀
Dt Dt
ρ +∀ =0⇒ +ρ =0
Dt Dt Dt ∀ Dt
 D∀
Can relate to the fluid velocity by noticing total dilatation or normal strain-rate is
Dt
equal to the rate of volume increase of the particle.
∂u ∂v ∂w 1 D∀
+ + = ε xx + ε yy + ε zz = ∇.V = = divV
∂x ∂y ∂z ∀ Dt
Dρ
+ ρ divV = 0 or
∂ρ
∂t
( )
+ ∇. ρV = 0

∂ρ ∂ ( ρ u ) ∂ ( ρ v ) ∂ ( ρ w )
Dt

+ + + =0
∂t ∂x ∂y ∂z
Notes pg.58 : cylindrical & spherical coordinates

Simplifications:

∂u ∂v ∂w
ρ=const. : flow is said to be incompressible
divV = ∇.V = 0 + + = 0 particles of constant volume,
∂x ∂y ∂z
rectangular coor.
but shape of volume can change.
STREAM FUNCTION ψ

∂ ( ρu ) ∂ ( ρv)
2-D , steady flow : continuity

+ =0
∂x ∂y
steady compressible or unsteady incompressible

Define stream ψ function such that:

∂ψ ∂ψ
ρu = , ρv = −
∂y ∂x
∂ 2ψ ∂ 2ψ
− = 0 continuity identically satisfied.
∂x∂y ∂y∂x
ψ : first & second order der. exist & continuous

Advantage
• Continuity eq. discarded
• # of unknowns (dependent variables) reduces by one.

Disadvantage
• Remaining velocity derivatives are increased by one order???
Physical significance of ψ
∂ψ ∂ψ
ψ = ψ ( x, y ) dψ = dx + dy = − ρ vdx + ρ udy = ρV .d s = dm
∂x ∂y
• Lines of constant ψ (dψ=0) are lines across which mass flow ( dm = 0 ) is zero

They are stream lines of flow

d s = dxi + dy j
Along AB x=const.

d s = dyi
V = ui + v j = 0
ρ=const.

∂ψ
m = ∫ ρV .d s = ∫ ρ udy = ∫ dy = ∫ dψ = ψ B −ψ A
yB yB yB ψ
B

yA yA yA
∂y ψA
 ∂ψ ∂ψ
dq = udy − vdx = dy + dx = dψ
ψ+dψ ∂y ∂x
dq = dψ

∫ψ dψ = ψ
ψ2
dq C q= 2 −ψ 1 > 0
1
ψ udy
A q: volume flow rate between streamlines
y ψ1 & ψ2
- vdx
x

m=ρq
ψ1 • Difference between the constant values of
ρ defining two stream lines is the mass flow
rate (per unit depth) between the two
q
streamlines.
ψ2
Note: 2-D , ρ = const., in cylindrical coor. (r
θ plane) continuity eq.
If ψ1 > ψ2 : q= ψ1 - ψ2 (flow is to the left)
 ∂ ( rVr ) ∂Vθ
+ =0
∂r ∂θ
1 ∂ψ ∂ψ
ψ (r , θ , t ) → Vr = ,Vθ = −
r ∂θ ∂r
show that continuity eq. is satisfied!

Uses of Continuity Eq.

1. to simplify D.E. of momentum
2. To relate changes in velocity in one direction to changes in another
∂u ∂v ∂v ∂u
+ =0⇒ =−
∂x ∂y ∂y ∂x
Ex: 2-D Flow, ρ=const.

∂u ∂v
y
<0⇒ > 0 ⇒ v is away from center
∂x ∂y
x
Ex:

U0 A(x) Converging channel

U 0 A0 = UA( x)
y
u(x)

U=
A0 x U 0 A0
A( x)

∂U ⎛ 1 dA ⎞ U 0 A0 ⎛ dA / dx ⎞
= U 0 A0 ⎜ − 2 ⎟= ⎜− ⎟
∂x ⎝ A ( x ) dx ⎠ A( x ) ⎝ A( x) ⎠

∂U U ( x) ⎛ dA ⎞
U ( x)

= ⎜− ⎟
∂x A( x) ⎝ dx ⎠

∂u ∂v
⇒ <0⇒ >0⇒ <0
dA
∂x ∂y
A( x) flow toward center
dx

∂u ∂v
⇒ >0⇒ <0⇒ >0
dA
∂x ∂y
A( x) flow away from center
dx
DERIVATION of MOMENTUM D.E
Newton’s Second Law applied to fluid element

d F = dm a
applied resulting acceleration
force of particle of mass, dm

relates stresses ↔ acceleration

⇓
fluid deformation ↔ substantial derivative
velocity field terms
+
pressure & body forces
 CLASSIFICATION OF FORCES ON A FLUID
1.Body Forces: All external forces developed without physical contact.
e.g. gravity, magnetic force

dFB = ρ gd ∀ = ρ gdxdydz
→
y
d∀ dy g
x dz
dx
z
body forces are distributed throughout

2.Surface Forces: All forces exerted on a boundary by surroundings through direct

contact.

a) Normal forces – e.g. pressure forces

b) Tangential forces – e.g. shear forces

∆Fn
σ nn = lim ∆A→0 ⇒ pressure
stress = force / unit area

1.Normal stresses
∆A
∆Fs
τ ss = lim ∆A→0 ⇒ shear
∆A
2.Shear stresses
Surface stresses have
a) direction & b) surface that acts on

τ yx = shear stress
two shear forces on any surface

z Fsy
y

Fn F
Fsx
x

Example: forces on a plane

 τ y x
direction that stress acts (x in this case)
plane acts on (y in this case)

τ yy
y

τ yz
τ yx
positive outward normal on y-surface

τ xy
⎡τ xx τ xy τ xz ⎤
Stress tensor

⎢ ⎥
negative outward

τ ij = ⎢τ yx τ yy τ yz ⎥
τ xz τ xz ⎢τ τ τ ⎥
x

τ zy τ zx ⎣ zx zy zz ⎦
z
stress acting on z-plane

τ ij = τ ji
symmetric tensor
 Derivation of Momentum Differential Equation

∂τ yx
Begin by applying differential analysis to differential fluid element, (dx, dy, dz)

τ yx +
τ zx ∂y
y
dy

∂σ xx
σ xx +
x

σ xx ∂x
z
dx

∂τ zx
τ zx +
τ yx ∂z
dz

d Fb + d Fs = dm
DV
Dt
Consider x-direction forces & changes across element using truncated Taylor series
 dFbx + dFsx = ρ dxdydz
Du ax
x-dir.

∂σ xx ⎞
Dt
⎛
ρ Bx dxdydz + ⎜ σ xx + dx ⎟ dydz − σ xx dydz
⎝ ∂x ⎠
⎛ ∂τ yx ⎞ ⎛ ∂τ zx ⎞
+ ⎜τ yx + dy ⎟ dxdz − τ yx dxdz + ⎜τ zx + dz ⎟ dxdy
⎝ ∂y ⎠ ⎝ ∂z ⎠
−τ zx dxdy
⎡ ∂σ xx ∂τ yx ∂τ zx ⎤
⎢ ρ Bx + + + ⎥ dxdydz
⎣ ∂x ∂y ∂z ⎦
⎡ ∂u ∂u ∂u ∂u ⎤
= ρ dxdydz ⎢ + u +v +w ⎥
⎣ ∂t ∂x ∂y ∂z ⎦
Du
Dt
 ∂τ xy ∂σ yy ∂τ zy
Similarly y-direction and z-direction

ρ By + + + =ρ
Dv
∂x ∂y ∂z
∂τ xz ∂τ yz ∂σ zz
Dt

ρ Bz + + + =ρ
Dw
∂x ∂y ∂z Dt

∂τ ij
In vector form

ρ B + ∇.τ ij = ρ ρ Bi + =ρ
DVi
∂x j
DV
or
Dt Dt
Note: Above eqs. are general eqs. apply for any fluid.

τ ij ∼ ε ij
General simplification Newtonien fluid
linear relationship between stress & rate of strain
From solid mechanics, the relation of stress tensor to strain-rate tensor yields

τ yx = c1ε xx + c2ε xy + c3ε xz + c4ε yy + c5ε yz + c6ε zz

a linear relationship of form,

where each stress component depends on all of the six rate of

strain components,
• We make assumption of an ISOTROPIC medium
(i.e. material property independent of direction)

This reduces the number constants to two, since many of the coeefficients
are identically zero or related to each other.

Theory of elasticity Hookean Solid E - mod. of elasticity

h - Poisson ‘s ratio

λ: 2nd coef. of viscosity

Newtonien Fluid µ: coef. of viscosity

or bulk viscosity
(associated only with
volume expansion)
σ xx = − p + 2µε xx + λ (ε xx + ε yy + ε zz )
General deformation law for Newtonian fluid

∂v
(∇.V )
linear stress

σ yy = − p + 2 µε yy + λ (ε xx + ε yy + ε zz ) = − p + 2µ +λ
volume exp ansion

pressure
∂y
stresses due to
linear compressibility

σ zz = − p + 2µε zz + λ (ε xx + ε yy + ε zz )
rate of
strain

⎛ ∂v ∂u ⎞
τ xy = τ yx = 2µε xy = µ ⎜ + ⎟
⎝ ∂x ∂y ⎠
⎛ ∂u ∂w ⎞
τ xz = τ zx = 2µε xz = µ ⎜ + ⎟
⎝ ∂z ∂x ⎠
⎛ ∂w ∂v ⎞
τ yz = τ zy = 2µε yz = µ ⎜ + ⎟
⎝ ∂y ∂z ⎠
 ( )
⎛ ∂ui ∂u j ⎞
τ ij = − pδ ij + µ ⎜ + ⎟ + δ ij λ ∇.V = − pδ ij + τ ij′
⎜ ∂x ∂ ⎟
⎝ j xi ⎠
thermodynamic viscous
pressure stresses

Note the inclusion of pressure because if velocity vanishes

normal stress = - pressure (hydrostatic)

⎧ 0 , i≠ j
Fluid at rest , τ ij = ⎨ ρ B − ∇p = 0
⎩− p , i = j

Stoke ‘s hypothesis For air & most gas mixtures

λ=− µ ∇.V = 0
2 For liquids
(1845)
3
 ⎛ ∂u ∂v ∂w ⎞ ⎛ ∂u ∂v ∂w ⎞
σ xx + σ yy + σ zz = −3 p + 2µ ⎜ + + ⎟ + 3λ ⎜ + + ⎟
⎝ ∂x ∂y ∂z ⎠ ⎝ ∂x ∂y ∂z ⎠
∇ .V ∇.V

= −3 p + (2 µ + 3λ )∇.V
=0

p = − (σ xx + σ yy + σ zz )
Define: mean (mechanical) pressure, p
1
3
Mean pressure in a deforming viscous fluid is not equal to the thermodynamic
pressure but distinction is rarely important

⎛ 2 ⎞
p = p − ⎜ λ + µ ⎟ ∇.V
⎝ 3 ⎠
usually small in typical flow problems
controversial subject

normal viscous stresses

 λ+ µ =0
2 (1845)
Stokes‘ hypothesis
3

ρ = const. ∇.V = 0
σ xx + σ yy + σ zz
For liquids ;

p=−
3
Now, back to D.E. substitute for stresses from above relationship,

Consider x – dir.

∂ ⎡ ∂u 2 ⎤ ∂ ⎡ ⎛ ∂v ∂u ⎞ ⎤ ∂ ⎡ ⎛ ∂u ∂w ⎞ ⎤
ρ Bx + ⎢ − p + 2 µ − µ ∇.V ⎥ + ⎢ µ ⎜ + ⎟ ⎥ + ⎢ µ ⎜ + ⎟ ⎥ = ρ ax
∂x ⎣ ∂x 3 ⎦ ∂y ⎣ ⎝ ∂x ∂y ⎠ ⎦ ∂z ⎣ ⎝ ∂z ∂x ⎠ ⎦

Let µ = const.

∂p ∂ ⎛ ∂u 2 ⎞ ∂ ⎛ ∂v ∂u ⎞ ∂ ⎛ ∂u ∂w ⎞
ρ Bx − + µ ⎜ 2 − ∇.V ⎟ + µ ⎜ + ⎟ + µ ⎜ + ⎟ = ρ ax
∂x ∂x ⎝ ∂x 3 ⎠ ∂y ⎝ ∂x ∂y ⎠ ∂z ⎝ ∂z ∂x ⎠
 ∂p
∂x
∂ 2u 2 ∂
ρ Bx − + 2 µ 2 − µ
∂x 3 ∂x
( )
∇.V + µ
∂ 2v
∂y∂x
∂ 2u ∂ 2u
+µ 2 +µ 2 +µ
∂y ∂z
∂2w
∂z∂x
= ρ ax
1 2 3 4 5

⎡ ⎤ ⎡ ⎤
∂p ⎢ ∂ u ∂ u ∂ u⎥ ∂ ⎢ ∂u
ρ Bx − + µ ⎢ 2 + 2 + 2 ⎥ + µ ⎢
∂v ∂w ⎥ 2 ∂
+ + ⎥ − µ ( )
∇.V = ρ ax
2 2 2

∂x ∂x
⎢ half ∂y ∂z ⎥ ∂x ⎢ ∂x ∂y ∂z ⎥ 3 ∂x
⎣ of 1 3 4 ⎦ ⎣ half of 1 2 5 ⎦

∇ 2u ∇.V

∂p
ρ Bx − + µ∇ u +
∂x
2 1 ∂
3 ∂x
( )
∇.V = ρ ax

∂p 1 ∂
( )
Likewise

ρ By − + µ∇ v + ∇.V = ρ a y
∂y 3 ∂y
2

∂p
ρ Bz − + µ∇ w +
∂z
2 1 ∂
3 ∂z
( )
∇.V = ρ az
 ∂V
NAVIER STOKES EQUATION

ρ B − ∇p + µ∇ V + µ∇(∇.V ) = ρ + ρ (V .∇)V
1
∂t
2

body pressure deformation

3 convective
force force stresses due to local acceleration

for ρ = const. ∇.V = 0

compressibility acceleration

ρ B − ∇p + µ∇ V = ρ
2 DV General equation
Dt coordinate independent

∂p ⎛ ∂ 2u ∂ 2u ∂ 2u ⎞ ⎛ ∂u ∂u ∂u ∂u ⎞
Cartesian coord. x – dir.

ρ Bx − + µ ⎜ 2 + 2 + 2 ⎟ = ρ ⎜ + u + v + w ⎟
∂x ⎝ ∂x ∂y ∂z ⎠ ⎝ ∂t ∂x ∂y ∂z ⎠

g x if B = g ( gravitional acceleration )
1

1 Force acting on the fluid element as a result of viscous stress distribution

on the surface of element
Note: Cylindrical coord. pg. 60

• ρ = const. ∇.V = 0
• viscosity is constant isothermal flow. For non-isothermal flows, esp. for liquids,
viscosity is often highly temp. dependent. CAUTION

CONSERVATION OF ENERGY: THE ENERGY EQ.

1st law of Thermodinamics for a system

dEt = dQ + dW
( J / m3 )
work done on system
heat added
z

⎛ ⎞
increase of energy of the system

Et = ρ ⎜ e + V − g .r ⎟ ( J / m3 )
r

⎜ ⎟
1 2 g

⎝ ⎠
y
+
2 gz
x
Et : total energy of the system (per unit volume)
e : internal enregy per unit mass

r : displacement of particle
moving system such as flowing fluid particle
need Material derivative : time rate of change, following the particle

= +
DEt DQ DW
(J/m3.s) Energy eq. for a flowing fluid
Dt Dt Dt

⎛ De ⎞
= ρ⎜ +V − gV ⎟
DEt DV
Dt ⎝ Dt Dt ⎠

Q & W in terms of fluid properties

• assume heat transfer Q to the
y element is given by Fourier ’s

∂T
law

∂qx qx = − k
qx qx +
∂x
dx ∂x
heat flows from positive to the
wx

∂wx
x neg. temp. (decreasing temperature

wx +
q = − k ∇T
gradient)
∂x
dx dx
z (HEAT FLOW)
 W/m2
Heat flow (rate) into the element in x - dir. : qx dydz
Heat flow (rate) out of the element in x - dir. : − ⎛ q +
∂qx ⎞
⎜ x dx ⎟ dydz
⎝ ∂x ⎠
∂qx
The net heat transfer to the element in x - dir. : −
∂x
dxdydz

⎛ ∂qx ∂qx ∂qx ⎞

Hence, the net heat transfer to the element = − ⎜ + + ⎟ d ∀ = −∇.q.d ∀
⎝ ∂x ∂y ∂z ⎠

= −∇.q = ∇. ( k ∇T )
DQ
[W/m3] neglect internal heat generation
Dt

Wx = − ( uσ xx + vτ xy + wτ xz )
Rate of work done to the element per unit area on the left face
negatif because work is done on the system

(σ )
surface direction

derivation: force on the left face: − xx i + τ xy j + τ xz k dydz

 ( )( )
Rate of work done on the element by this force
= − σ xx i + τ xy j + τ xz k . ui + v j + wk dydz

= − ( uσ xx + vτ xy + wτ xz ) dydz
Similarly, rate of work done by the right face stresses is
⎛ ∂Wx ⎞
= − ⎜ Wx + dx ⎟
⎝ ∂x ⎠

= −divW = −∇.W = ( uσ xx + vτ xy + wτ xz )
• Net rate of work done on the element
DW ∂
∂x
∂
( uτ yx + vσ yy + wτ yz )
Dt
+
∂y

( ) + ( uτ zx + vτ zy + wσ zz )
∂
= ∇. V .τ ij ∂z
DW
Dt
indicial notation
 ( )
Using the indicial notation,

= ∇. V .τ ij
DW
Dt

( )
∇. V .τ ij = V . ( ∇.τ ij ) + τ ij
∂ui
expression can be decomposed into

∂x j
Exercise: Show this!

⎛ DV ⎞
Remember Newton ’s 2nd law

ρ = ρ g + ∇.τ ij ⎯⎯
→∇.τ ij = ρ ⎜ −g⎟
DV
Dt ⎝ Dt ⎠

V . ( ∇.τ ij ) = ρ ⎜ V
⎛ DV ⎞
− g .V ⎟
hence,

⎝ Dt ⎠
kinetic & potential energy
terms in energy eq .

⎛ DV ⎞ ∂u
= ρ ⎜V − g .V ⎟ + τ ij i
DW
Dt ⎝ Dt ⎠ ∂x j
Energy eq. becomes;

∂ui
ρ = ∇. ( k ∇T ) + τ ij
De First law of thermodynamics
∂x j
for fluid motion
Dt

split the stress tensor into pressure and viscous terms using

⎛ ∂ui ∂u j ⎞
τ ij = − pδ ij + µ ⎜ + ⎟ + δ ij λ divV
⎜ ∂x ∂ ⎟
⎝ ⎠
deformation law given
j xi by Stokes (1845)

τ ij′

∂ui ' ∂ui

τ ij = τ ij − pdivV
∂x j ∂x j Φ: dissipation function

From continuity eq. D ρ

+ ρ divV = 0
Dt
p Dρ D ⎛ p ⎞ Dp
pdivV = − =ρ ⎜ ⎟−
ρ Dt Dt ⎝ ρ ⎠ Dt
 h = e+
ρ
p
introducing enthalpy

ρ = + div ( k ∇T ) + Φ
Energy eq. Dh Dp dissipation function
(viscous dissipation)
Dt Dt
For Newtonian fluid
⎡ ⎛ ∂u ⎞2 ⎛ ∂v ⎞ ⎛ ∂ ⎞ ⎛ ∂ ∂ ⎞ ⎛ ∂ ∂ ⎞ ⎛ ∂ ∂ ⎞ ⎤
Φ = µ ⎢2 ⎜ ⎟ + 2 ⎜ ⎟ + 2 ⎜ ⎟ +⎜ + ⎟ +⎜ + ⎟ +⎜ + ⎟ ⎥
2 2 2 2 2
w v u w v u w
∂
⎢⎣ ⎝ ⎠x ∂
⎝ ⎠y ⎝ ∂ z ⎠ ⎝ ∂x ∂ y ⎠ ⎝ ∂ y ∂ z ⎠ ⎝ ∂ z ∂ x ⎠ ⎥⎦

⎛ ∂u ∂v ∂w ⎞
+λ ⎜ + + ⎟
2

⎝ ∂x ∂y ∂z ⎠

Φ ≥ 0 , µ ≥ 0 ; 3λ + 2 µ ≥ 0 Hookes’ hypothesis λ = −2 µ / 3

Incompressible flow, ρ = const.

ρcp = βT + ∇ ( k ∇T ) + Φ
DT Dp
Dt Dt
Constant Properties, k=const.
 dh = c p dT dh = c p dT − (1 − β T )
ρ
dp
perfect gas relation

de = cv dT
1 ⎛ ∂ρ ⎞
β = 1/ T ( perfect gas) β =−
ρ ⎜⎝ ∂T ⎟⎠ P
cp, cv : specific heats

ρcp = + k ∇ 2T + Φ
DT Dp
ideal gas
Dt Dt

if ρ = const. ∇.V = 0 → ρ c p = k ∇ 2T + Φ
DT
Dt neglected

VALID for either gas (low velocity) or liquid.

Low velocity or incompressible flow Φ 0

= α∇ 2T α = k / ρ c p : thermal diffusivity ⎡⎣ m 2 / s ⎤⎦
DT
Dt
∂T ∂T ∂T ∂T
+u +v +w = α∇ 2T
∂t ∂x ∂y ∂z
(V .∇ )T :convective terms
Example: Fully developed laminar flow down and inclined plane surface

y
Given: ρ, µ
width b=1 m
x

θ=15º
liquid h=1mm

θ
h=1mm
g

Find: velocity profile

shear stress distribution
volume flow rate (per unit depth)
average flow velocity
film thickness in terms of volume flow rate
Basic eq.s for ρ=const.
∂u ∂v ∂w
+ + =0
∂x ∂y ∂z

⎛ ⎞ ⎛ 2 ⎞
4 3

∂u ∂u ∂u ∂u ∂p ∂u ∂u ∂u
ρ ⎜⎜ + u + v + w ⎟⎟ = ρ g x − + µ ⎜⎜ 2 + 2 + 2 ⎟⎟
2 2

⎜ ∂t ∂x 5 ∂y 3 ∂z ⎟ ∂x ⎜ ∂x ∂y ∂z ⎟
⎝ 1 ⎠ ⎝ 4 3 ⎠

⎛ ⎞ ⎛ ⎞
4 4

⎜ ∂v ∂v ∂v ∂v ⎟ ∂p ⎜∂ v ∂ v ∂ v⎟
ρ ⎜ + u + v + w ⎟ = ρgy − + µ ⎜ 2 + 2 + 2 ⎟
2 2 2

⎜ ∂t ∂x 5 ∂y ∂z ⎟ ∂y ⎜ ∂x ∂y ∂z ⎟
⎝ 1 4 3 ⎠ ⎝ 4 5 3 ⎠

⎛ ∂w ∂w ∂w ∂w ⎞ ∂p ⎛ ∂2w ∂2w ∂2w ⎞

ρ⎜ +u +v + w ⎟ = ρ gz − + µ ⎜ 2 + 2 + 2 ⎟
⎝ ∂t ∂x ∂y ∂z ⎠ ∂z ⎝ ∂x ∂y ∂z ⎠

ρ=const. (incomp. flow)

1. steady flow (given)
∂
= =0
2.
∂z
Fully developed flow, so no properties vary in the x-dir. , ∂ = 0
3. No flow or variation of properties in the z-dir. w 0,

∂x
4.
 ∂v
= 0 ⇒ v = const. = c
∂y v = 0 at y = 0 ⇒ v = 0 everywhere
Cont.

∂u
5.
0 = ρ gx + µ 2
2

∂y
∂p
0 = ρgy −
∂y
Continuity u=u(y) only.

∂ 2u
→ 2
d 2u
∂y ρ gx sin θ
= − = − ρ
2
dy d 2u
µ µ
g
dy 2
sin θ
= −ρ g y + C1
µ
du
dy
sin θ y 2
u( y) = − ρ g + C1 y + C2
µ 2
B.C.s No slip

u = 0 at y = 0 → C2 = 0

= 0 at y=h
du
(zero shear stress on the liquid free surface)
dy

sin θ sin θ
0 = −ρ g h + C1 → C1 = ρ g
µ µ
h

sin θ y 2 sin θ
u( y) = − ρ g + ρg
µ 2 µ
hy or

sin θ ⎛ y2 ⎞
u( y) = ρ g ⎜ hy − ⎟
µ ⎝ 2 ⎠

τ yx = µ = ρ g sin θ ( h − y )
du
dy
Volume flow rate

Q = ∫ udA = ∫ ubdy
h

ρ g sin θ ⎛ y2 ⎞ ρ g sin θ bh3

=∫
A 0

⎜ hy − ⎟bdy =
h

0
µ ⎝ 2 ⎠ µ 3

V =Q/ A=
Q
The average velocity
bh

ρ g sin θ h 2
V=
µ 3
Solving film thickness

⎡ 3µ Q ⎤
h=⎢ ⎥ → h ∼ Q1/ 3
1/ 3

⎣ ρ θ ⎦
non-linear relation
g sin .b

h=1 mm, b=1 m, θ=15º Q = 0.846 lt/sec

Summary of Basic Equations:
• Newtonien fluid
• Stokes hypothesis
• gravity is the only body force
• Fourier ‘s law, no internal heat sources

ρ = const. ρ = var iable

∇.V = 0 ∂ρ
∂t
( )
+ ∇. ρV = 0

( )
= g − ∇ p + υ∇ 2 V
ρ
DV 1
= g − ∇p + υ∇ 2 V + υ∇ ∇.V
ρ
Dt DV 1 1

= α∇ T + Φ
Dt 3

ρcp + div ( k ∇T ) + Φ
DT 1
ρ =
2
Dh Dp
Dt
Dt Dt
perfect gas dh = c p dT
unknowns u,v,p,(T) in general h = h( p, T )
# of eq.s 4,(5) µ = µ ( p, T )
ρ = ρ ( p, T )
exact solution is possible for “simple” problems
MATHEMATICAL CHARACTER OF THE BASIC EQS.
difficulties
• equations are coupled V , p & T (temp. dep. property )
• “ “ non-linear

QUASI-LINEAR 2nd order PDE

∂ 2φ ∂ 2φ ∂ 2φ
A 2 +B +C 2 = D
∂x ∂x∂y ∂y
∂φ ∂φ
x, y , φ , ,
∂x ∂y
where coef. A,B,C,D may be non-linear functions of

but not of the second derivatives of φ.

⎧< 0 the eq. is elliptic → B.V.P

⎪
if B2 -4AC ⎨= 0 the eq. is parabolic → diffusion type (mixed BVP & IVP)
discriminant ⎪
⎩> 0 the eq. is hyperbolic → IVP
Laplace eq.: ∇ 2φ = 0 Elliptic
∂T ∂ 2 T ∂ 2 T
= 2 + 2
∂t ∂x ∂y
Heat conduction parabolic

∂ 2φ ∂ 2φ
− 2 =0
∂x ∂y
Wave equation 2
hyperbolic

Navier-Stokes eqs. are too complicated to fit into this model. can be any or mixtures
of all three depending upon specific flow and geometry.

Incompressible flow with zero convective derivatives

Let us assume a Newtonian fluid with constant ρ, µ, k.

(V .∇ )V = 0 ; (V .∇ ) T = 0
further assume convective derivatives vanish.

realistic assumption for flows with gradients of flow properties are normal to the flow
direction, duct flows.

• Φ = 0 (viscous dissipation is negligible when the flow velocity is much smaller

than the speed of sound of the fluid)
∇.V=0
µ
(1)
∂V
= g − ∇p + υ∇ 2 V υ=
ρ ρ
1
∂t
(2) (kinematic viscosity)
∂T
= α∇ 2T
∂t
(3)

• Temperature effects are confined to the energy eq.

continuity eq. & momentum eqs. are uncoupled from T.
• energy eq. is heat conduction eq. (parabolic)
temp. mixed behaviour.i.e BVP in space (x, y, z). IVP in time

Taking divergence of (2) and making use of (1),

∇.(2) ∇2 p = 0 (2' ) → Laplace eq. (Elliptic)

pure B.V.P

( )
Finally, by taking the curl (∇x) of eq. (2), can eliminate p
⎛ ∂V ⎞
∇×⎜ ⎟ = ∇ × − ∇ × ( ∇p ) + υ∇ × ∇ 2 V ω = ∇ ×V
ρ
1
⎝ ∂t ⎠
g ;

ω = curlV
fluid vorticity

0 0
 ∂ω
= υ∇ 2 ω
∂t
(2'')

∴vorticity, like temp. has parabolic behaviour

eq. (2’’) also heat-conduction eq.

∴Both vorticity & temp. have diffusion coef. or diffusivities

µ
υ=
ρ
:viscous diffusivity (m 2 /s)

α=
ρcp
k
:thermal diffusivity (m 2 /s)

• They are entirely fluid properties, not geometric or flow parameters

υ viscous diffusion rate

Pr = =
α thermal diffusion rate
 Liquid metals → low Pr (Pr<<1) ,e.g. PrHg = 0.024
air → Pr = 0.72 (gases of the order of the unity)
water → Pr = 7.0
oils → high Pr. (Pr>>1) ,e.g. Prglycerin = 12 000

Low speed viscous flow (laminar flow) past a hot wall:

viscous diffusion rate >> thermal dif. rate
n thermal effects are confined near the wall

V / V∞
V / V∞
T / T∞ T / T∞
δT
0 1.0 0 1
Liquid metals: Pr << 1 Oils : Pr >> 1
ν: shows the effect of viscosity of a fluid: momentum diffusion

as υ , the region effected by viscosity is narrower

known as Boundary Layer when υ is very small.
δ (B.L. thickness) δ ∼ υ (for laminar flow)
viscous spreading length
δ T (thermal B.L. thickness) → δ T ∼ α (for laminar flow)
thermal spreading length
δ
∼ Pr → good approximation for all B.L. flows,
δT
even at high speeds

• gases: Pr ≈ O (1) both effects are equally important

• liquids: Pr >> 1, may neglect the effect of heat conduction
U .L.ρ .c p
• Pe = Pr .Re =
k
 DIMENSIONLESS PARAMETERS IN VISCOUS FLOW

Basic flow eqs are extremely difficult to analyze

Therefore, we need to get most efficient possible form.

Buckingham Pi theorem:
dependent parameters V , p, T = f ( xi , t ,15 flow parameters)

9 fluid properties: ρ , µ , λ , k , c p , cv , l , β , σ
4 reference quantities: V0 , p0 , T0 , L
1 wall heat flux, qw
1 acceleration of gravity, g

Primary dimensions, M (mass), L (length), t (time), T (temp.)

15 – 4 = 11 dimensionless numbers which governs the viscous flows
with heat transfer
1. Re 7. Nu
2. Pr 8. Kn
3. Fr 9. Cavitation# impossible to consider all of them
at one time
4. Ec 10. We (Weber number)
5. γ =cp /c v 11. viscosity ratio λ /µ
6. Gr
Few of them are important for a given problem
determined by the non-dimensionalizing the basic eqs & B.C.s

1. From the eqs: Re, Pr, Ec, Gr, (λ/µ neglected)

Slip-flow conditions: Kn, γ

2. Wall heat transfer conditions: Nu
3.
4. Free-surface conditions: Fr, We, Cavitation #.
Non-dimensionalizing the Basic Equations:
Reference properties appropriate to the flow
ρ∞ , p∞ , T∞ x2
U∞
L

x1
τw x3

x = ; V =
* xi * V
U∞
i
L
p − p∞ T − T∞
p = ; T =
ρ∞U ∞ Tw − T∞
* *
2

ρ
t = ; ρ = ; g =
tU ∞
ρ∞
* * * g
L g∞
Star denotes dimensionless variables.
 ( )
Continuity Equation
∂ρ
+ ∇. ρ V = 0

( )
∂t
ρ∞ ∂ρ * 1 *
+ ∇ ρ∞ ρ U ∞ V = 0

( )
L ∂t
*
U∞ *
*

∂ρ *
L

+∇ . ρ V = 0
∂t
* * *
*

no dimensionless parameters: dimensional & dimensionless cont. eq. are the same

Note : 1. U 0 = µ0 / ρ0 L if there is no free stream (as in free convection)

2. Residence time L/V0 (steady flows have no characteristic time of their own)

Momentum Equation
ρ = ρ∞ = const.

3
( )
ρ g − ∇p + µ∇ 2 V + µ∇ ∇.V = ρ
1 DV
Dt
 ( ) =U
Dt D ( t * L / U ∞ )
*

=
D U∞ V *
∞
2
DV DV
L Dt *

ρ g = ρ∞ g∞ ρ g
∂ ( g ∞U ∞2 p* )
* *

∂p ρ∞U ∞2 ∂p*
∂ ( x L)
∇p ⇒ →
∂x ∂x*
*
=
L
ρ∞U ∞2
∇p= ∇ * p*

∂ 2 (U ∞ u * )
L
∂u U ∞ ∂ 2u *
( )
∇V→ 2 →
2

∂x ∂ x L ∂x
2
2
= 2 *2
* 2 L

∇ V= 2 ∇ V
2 U ∞ *2 *
L
ρ∞U µU ∞ ρ∞U
ρ∞ g∞ ρ * g * - ∇ * p* + ∇* V = ρ*
*
∞ ∞
2 2
2 * DV
L L2 L Dt *
 µ
ρ g -∇ p + ∇ V =ρ
*
Lg ∞ * * * *
ρ∞U ∞ L
*2 * * DV

U∞2
Dt *

1 1
Fr 2 Re

ρ = − ∇p + 2 ρ g + ∇V
DV 1 * * 1 2
Dt pressure f.
Fr Re
inertia force gravity force viscous force

Froude number (Fr):: determines the importance of buoyancy (important for free
surface flows)
Dynamically similar flows: dimensionless parameters & dimensionless B.C.s

⇒ kinematic similarity
should be identical.

U ∞ L ρ∞
Re =
y

µ∞ g
x
liquid

Fr = θ
h
U∞
g∞ L
Ratio of Forces Dimensionless numbers V

Inertia forces → ma → ρ∀ = ρL ∼ ρV 2 L2
dV 3 dV ds

dt ds dt

Viscous forces → τ .A

µ L = µ L = µVL
du 2 V 2
dy L
inertia forces ρV L ρVL
Re = = =
2 2

viscous forces µVL µ

ρ
Fr = =
2 2 2

gravity f. ρ L g
inertia f. V L V
2
∼ 3
gL
 pressure forces ∆p. A ∆p
Eu = = =
inertia forces ρV 2 L2 ρV 2
∆p
c p = pressure coef. cp =
ρV 2
1 dynamic pressure

ρ
2

Fr 2 = = =
2 2 2

gravity forces g ρ L
inertia forces V L V flows with the free surface
3 effects
gL
ρV 2 L2 ρV 2 L
We = = =
σL σ
inertia forces
surface tension f.

M= M=
V flow speed
inertia forces
c local sonic forces due to compressibility

ρV 2 L2 ρV 2
speed

M2 = = δp
a= = Eυ = −
Eυ
dρ ρ δ∀
dp
Eυ L2 Eυ
∀
M2 = 2
compressibility 2
modulus [Pa]
V
a
For truly incompressible flow → Eν = ∞, a=∞ ⇒ M=0

Compare compressibility modulus of water and air.

What is the pressure rise to reduce the volume of water by 4% ?

• inertia force
• viscous force
• pressure force
• gravity force
• surface tension force
• compressibility force
Example 1: Lid-driven cavity flow with periodic boundary condition

ubc = V∞ sin ωt ubc

t∗ =
tV∞
streamlines
L
H

V∞ L ρ
Aspect ratio: H/L

Re =
L µ
non-dimensional boundary condition

⎛ ωL ∗ ⎞
u =
∗
= sin ωt = sin ⎜ t ⎟
ubc
V∞ ⎝ V∞ ⎠
bc

duplication of the b.c. requires that the parameter ωL/V∞ be the same between

ωL
two flows.

Strouhal number, St=

V∞
• Need to begin with the correct equations

• Non-dimensional governing diff.eqs. are useful for numerical solution.

- scaling is simplified
- unit conversion problems are reduced
- solutions can be presented in generalized form.

Example 2: Natural (free) convection: flow is due to the variation of density

1 ⎛ ∂ρ ⎞
ρ = ρ∞ + ∆ρ ≈ ρ∞ (1 − β∆T ) β =- ⎜ ⎟
ρ ⎝ ∂T ⎠ P

β ∞ gL3 (Tw − T∞ )
Low-speed → Gr=
υ∞2
, & Pr
 ⎛ ∂T ∗ ⎞
qw = h∞ (Tw − T∞ ) = ∓ k∞ w Nu = −k ⎜ ∗ ⎟
Tw dT ∗

dx ⎝ ∂n ⎠ w
T − T∞
T∞
T∗ =
qw convective heat transfer coef.

Tw − T∞
k∞
Nu = =
k∞ (Tw − T∞ )
h∞ L qw L
h∞
k∞
n Nu is the driving parameter which effects the solution

Energy Equation

ρ cp = + ∇. ( k ∇T ) + Φ
Maxwell's relation
DT Dp
dh=c p dT + (1 − β T )
ρ
dp
Dt Dt
perfect gas → dh=cp dT
Dh
Dt

substitute the new variables,

µ ρ T − T∞
c = , k* = , ∇∗ = L∇, µ ∗ = , ρ∗ = , T* =
cp
µ∞ ρ∞
k
Tw − T∞
*

k∞
p
c p∞
 ∇ .( k ∇ T ) +
DT ∗ Dp∗
ρ ∗c p ∗ = Ec ∗ + ∗ ∗ ∗ ∗
Φ
1 Ec ∗
∗
Dt Dt Re Pr Re
ρ∞V∞ L
Re =
µ∞
inertia forces
;
viscous forces
µ∞ c p υ
Pr = = ∞ =
α ∞ thermal dif. rate
∞
viscous dif. rate
k∞

c p∞ (Tw − T∞ ) a = γ RT∞
V∞2
Eckert number , Ec=

High-speed flows → all three are important for heat transfer analyses.
Low-speed flows → ( V∞ < 30% of speed of sound )
or incompressible flows → pressure term & dissipation term is neglected
RePr → Peclet number
DT∗
constant properties → = α ∇ T
∗ ∗2 ∗
1
∗
Dt Re.Pr
ρ ∗ , µ ∗ , k ∗ , c p∗ = f ( p∗ , T ∗ )
 VORTICITY CONSIDERATIONS IN INCOMPRESSIBLE
VISCOUS FLOW: VORTICITY TRANSPORT EQ.
Vorticity vector, ω = ∇ × V = curlV
-is a measure of rotational effects

ω = 2Ω
local angular velocity of a fluid element

Navier − Stokes eqs. → constant ρ ,µ Let g = − gk z

ρ = −∇p − ρ gk + µ∇ 2 V
k
g
DV
(1)
Dt i x

( ) ( )
Use the following vector identities;
⎛V 2 ⎞
( 2)
j
V .∇ V = ∇ ⎜ ⎟ − V × ∇ × V
y

⎝ 2 ⎠

( ) ( )
ω

∇ 2 V = ∇ ∇.V − ∇ × ∇ × V ( 3)
ω
Note that ∇.V = 0 for an incomp. flow
Substitute (2) &(3) into (1)
⎡ ∂V ⎛V 2 ⎞ ⎤
ρ ⎢ + ∇ ⎜ ⎟ − V × ω ⎥ = −∇p − ρ gk + µ∇ × ω
⎣ ∂t ⎝ 2 ⎠ ⎦
(4)

∂V ⎛ ⎞
ρ + ∇⎜ p + ρ + ρ gz ⎟ = ρV × ω -µ∇ × ω
V2
∂t ⎝ ⎠
(5)
2

ω = 0, irrotational flow V = ∇φ
∂ ( ∇φ )
if
⎛ ⎞
• Bernoulli eq. is valid even for
ρ + ∇⎜ p + ρ + ρ gz ⎟ = 0
2
V
∂t
viscous fluids if flow is irrotational
⎝ ⎠
2 • difficulty

⎛ ∂φ ⎞
- potential flows do not satisfy
∇⎜ ρ + p+ρ + ρ gz ⎟ = 0
V2
⎝ ∂t
no-slip condition at a solid wall.
2 ⎠
∂φ
ρ + p+ρ + ρ gz = cons tan t
V2
∂t 2
Bernoulli eq. for unsteady incompressible flow
 VORTICITY TRANSPORT EQUATION
When dealing with a real fluid, we need an equation or eqs. to determine the

vorticity ⇒ rotational behaviour of fluid ~ angular momentum

behaviour of vorticity.

ω = 2 × angular velocity = ∇ × V → local rotation of fluid

vorticity → creation, transport, destruction, stretching etc.

To obtain an eqn. for vorticity transport, we take the curl of the N-S eqn.

⎛ DV ⎞
∇×⎜ = − ∇p + B + υ∇ V ⎟
ρ
1
⎝ Dt ⎠
2

− g ∆h

since ∇ × ∇φ =0 where φ ⇒ scalar

∴ ∇ × ∇p & ∇ × ∇ h = 0
Thus,

∇× =υ ∇ × ∇ 2 V
DV
∇ 2 ω =∇ 2 ( ∇×V )
Dt
 ⎡ ∂V ⎤
∇× =∇ × ⎢ + V .∇V ⎥
DV
⎣ ∂t ⎦

( )
Dt
⎡ ∂V 1 ⎤
=∇ × ⎢ + ∇V − V × ∇ × V ⎥
⎣ ∂t 2 ⎦
2

( ) + 1 ∇ × ∇V
( )
∂ ∇ ×V
− ∇ ×V × ∇ ×V
∂t
2
=
2

( ) ( ) ( )
ω
0

∂ω ∂ω
− ∇ × V ×ω = + V.∇ ω − ω.∇ V
∂t ∂t
=
 ( ) ( )
Thus, vorticity transport equations becomes

Dω ∂ω
= + V .∇ ω = ω.∇ V + υ∇ 2 ω
Dt ∂t viscous
time rate local convection of vorticity diffusion of
of change change vorticity by production vorticity
of vorticity velocity field term
-by stretching
and tilting of
existing
vorticity

2 − D case ω.∇V ⇒ 0 ω ⊥ ∇V flow in xy plane

⎛ ∂V ∂V ⎞
ωz k . ⎜ i+ j⎟ = 0
⎝ ∂x ∂y ⎠
• In all viscous flows vorticity is generally present and is generated by relative
motion near solid walls.
• If Re is large, vorticity is swept downstream and remains close to the wall:

ω =0
B.L.
• If flow is between two walls, e.g. duct flow

ω≠0
potential-flow model is generally not
core valid in duct flow.
 THE STREAM FUNCTION , ψ
TWO – DIMENSIONAL CONSIDERATIONS :

2 − D, ρ = const. µ =const. V=u(x,y,t)i +v(x,y,t) j

∂u ∂v
+ =0
∂x ∂y
∂u ∂u ∂u 1 ∂p ⎛ ∂ 2u ∂ 2u ⎞
+u +v = gx − +υ ⎜ 2 + 2 ⎟
∂t ∂x ∂y ρ ∂x ⎝ ∂x ∂y ⎠
∂v ∂v ∂v 1 ∂p ⎛ ∂ 2v ∂ 2v ⎞
+ u + v = gy − +υ ⎜ 2 + 2 ⎟
∂t ∂x ∂y ρ ∂y ⎝ ∂x ∂y ⎠
3 eqs. u, v, p u=u(x,y,t) , v=v(x,y,t) , p=p(x,y,t)

Let k=const. ρ cp = k ∇ 2T
DT
Dt
momentum equation is uncoupled from the energy eq.
Eliminate pressure & gravity by cross-differentiation, i.e.

( ) ( )
taking the curl of the 2-D vector momentum equation

⎛ ∂V ⎞
∇×⎜ ⎟ + ∇ × ∇ ω = ∇ × − ∇ × ( ∇p ) + υ∇ × ∇ 2 V
ρ
1
⎝ ∂t ⎠
V . g

∂ω
( )
0 0

+ V .∇ ω = υ∇ 2 ω
∂t
VORTICITY

Dω
TRANSPORT EQ.

= υ∇ 2 ω
2-D

Dt

∂ω z ∂ω z ∂ω z ⎛ ∂ 2ω z ∂ 2ω z ⎞
+u +v =υ ⎜ 2 + 2 ⎟
∂t ∂x ∂y ⎝ ∂ x ∂ y ⎠
⎛ ∂v ∂u ⎞
ωz = ⎜ − ⎟ ω = ωz k , ωx = ω y = 0
⎝ ∂x ∂y ⎠
Dω z
= υ∇ 2ω z
≡ = α∇ 2T
DT
Dt rate of dissipation of
vorticity through friction Dt
substantive variation of vorticity
 2 − D, incomp. flow, µ =const. ωz = ω
Dω
= υ∇ 2ω 2 eqs.

∂u ∂v
Dt 2 unknowns
+ =0
u, v
∂x ∂y

GOAL: Reduce the governing eq. (for incomp., 2-D, µ=const.)

to just one!

Introduce the stream function ψ (x,y,t)

∂ψ ∂ψ
; v= − (Note that ψ satisfies the continuity eq.)
∂y ∂x
u=
Dω ⎫
∂v ∂u ∂ 2ψ ∂ 2ψ = υ∇ 2ω ⎪
ω = − = − 2 − 2 = −∇ 2ψ ⎬ ω ,ψ formulation
∂x ∂y ∂x ∂y
∇ 2ψ = −ω ⎪⎭
Dt

∂ 2
∂t
( ∇ ψ ) +
∂ψ ∂
∂y ∂x
. ( ∇ 2
ψ ) − . ( ∇ 2ψ ) = υ∇ 4ψ
∂ψ ∂
∂x ∂y
(A)
viscous term
local accel. convective term
1-eq. 1 unknown (ψ)
4th order PDE (non-linear) → very complex

• B.C.s would be in terms of the first derivatives of ψ

valid for 2-D, incompressible flow, µ=const.

Example:

U∞

∂ψ ∂ψ
At infinity u=U ∞ , v=0 → =0, = U∞
∂x ∂y
∂ψ ∂ψ
= =0
∂x ∂y
At the solid surface,

- only few simple analytic solutions are known.

- numerical solutions are required
 Non-dimensionalization

ψ ∗ = ψ / V0 L → Re

∂
( ) ∂ψ ∂
( ∂ψ
)∂
( )
∗ ∗
∇ ψ + ∗ . ∗ ∇ ψ − ∗ . ∗ ∇ ψ =
∗ ∗ ∗ ∗ ∗2 ∗
∇ψ
1 4 *
∂t ∂y ∂x ∂x ∂y
2 2

∗
Re
 Study of Viscous Flows :

1) Exact solutions of N-S.

a) Analytical solutions (limited success
because of the non-linearity of the eqs.)
b) Numerical solutions

Re → 0
2) Very slow motions of viscous flow CREEPING FLOW : Re<<1

F = ma ∼ ρ L3 ∼ ρ L3 ∼ ρV 2 L2
2
dV V
dt L
dV
V
dS

Re = ⇒ Inertia forces<<viscous forces

inertia forces ∼ V 2
viscous forces ∼ V
τ .A ∼ µ A ∼ µ L ∼ µVL
du V 2
dy L
 ∇ ψ ∗ = 0 → Fourth − order linear eq.
∗4

biharmonic eq. in two dimensions

- governing relation in the theory of 2-D elasticity
- large number of practical solutions exists from solid mechanics.
See Timoshenko & Goodier (1970)

2-D, µ=const. , ρ=const.

- can be applied to creeping viscous flows

3) Limit as Re→∞
Boundary layer theory

inertia forces>>viscous forces

but cannot neglect viscous terms

∴ cannot satisfy all B.C.s simultaneously.

completely but order of eqs. reduces
 Boundary-Layer Flow: Re>>1

∂V
∂t
( )
+ V .∇ V = −∇p +
1 2
Re
∇V

∇ V is never negligible near a solid boundary because

1 2
The term
Re
the no-slip condition forces ∇ 2 V to be very large, of order Re, near
the wall ⇒ B-L.

Singular-perturbation problem Van Dyke (1964)

“Perturbation Methods in F.M.”

∞
1/Re contains the highest-order derivative in the system,
i.e. solution changes mathematical character as 1/Re
Example: Flow of a uniform stream parallel to an infinite flat plate with uniform suction:

u = 1 − evw Re y ; v = vw ; vw < 0
Exact solution for arbitrary Re.
Behaviour of solution as Re → ∞

∞ ⇒ u=1 everywhere
i) Re finite (no matter how large) u 0 as y 0

⇒ frictionless solution
ii) Re

i.e. frictionless flow with very small viscosity is not a potential flow since
a layer of finite thickness always exists where viscous effects are important.
1
small Re

y medium Re
u
x Large Re
Re ∞

y=0
vw
Low Reynolds Number : Creeping flow
Limiting case of very large viscosity: Re<<1

Exact solutions N-S. → valid for arbitrary Re, at least until instability sets in and
turbulence ensues.
Lack generality & limited

e.g. Viscous flow past an immersed body at arbitrary Re

- Direct experiment
- Numerical solution
- B-L approach
- Creeping flow approx.
pressure cannot scale with the ‘dynamic’ or inertia term ρU2 but rather must depend
Basic Assumption : Re<<1, i.e inertia forces<<viscous f.

upon a ‘viscous’ scale µU/L

Non-dimensionalize the N-S eqs. with variables

p − p∞
x = , V = , t∗ = , p∗ =
∗ ∗

µU / L
x V tU
L U L
∗

⇒ Re = −∇ p + ∇ V
DV ∗ ∗ ∗2 ∗
∗
Dt

Neglect inertia if Re<<1

i.e. Creeping Flow (or Stokes flow) assumption

( )
• Note: inertia V.∇ V is also negligible if there is no convective acceleration
e.g., fully developed duct flow (no restriction on Re).
Full N-S. for ρ=const. , µ=const. (steady flow)
⎡ ⎤
∂V
( )
ρ ⎢⎢ + V .∇ V ⎥⎥ = −∇p + µ∇ 2 V
∂t
⎢⎣ 0 ⎥⎦

(V .∇ )V → 0 (inertial force)

∇p=µ∇ 2 V Momentum eq.

Re =
inertia forces
<< 1
∇.V = 0
viscous forces

in extended form,
∂p ⎛ ∂ 2u ∂ 2u ∂ 2u ⎞
= µ⎜ 2 + 2 + 2 ⎟
∂x ⎝ ∂x ∂y ∂z ⎠
∂p ⎛ ∂ 2v ∂ 2v ∂ 2v ⎞
= µ⎜ 2 + 2 + 2 ⎟
∂y ⎝ ∂x ∂y ∂z ⎠
GOVERNING EQS for
CREEPING FLOW
∂p ⎛ ∂2w ∂2w ∂2w ⎞
= µ⎜ 2 + 2 + 2 ⎟
B.C.s are the same as N-S eqs.

∂z ⎝ ∂x ∂y ∂z ⎠
∂u ∂v ∂w
+ + =0
∂x ∂y ∂z
 Characteristics of CREEPING flows
a) Solutions are independent of density
b) Take div (∇.) of the momentum eq.

div grad p = ∇ 2 p = div ⎡⎣ µ∇ 2 V ⎤⎦ µ =const.

=µ div ⎡⎣∇ 2 V ⎤⎦

=µ∇ 2 ⎡⎣ divV ⎤⎦
∇ .V = 0

∇ p=0
∇ : Laplace operator
∂ ∂ ∂
2
∇= 2 + 2 + 2
2 2 2

∂x ∂y ∂z
Pressure satisfies the Laplace (potential) eq. & p(x,y,z) is a potential (harmonic)
function.

c) Take the curl of the momentum

(3) ∇ 2 ω = 0 , ω = ∇ × V (vorticity)
ABOVE valid for any 3-D problem
FOR 2-D CREEPING FLOW
choice #1
work with primitive variables u, v, p

work in terms of ψ (stream function)

choice #2

ω has only one non-zero component

∂v ∂u ∂ψ ∂ψ
(4) ω = − = −∇ ψ ; v= −
∂x ∂y ∂y ∂x
2
Note: u=

(3) & (4) ∇ 4ψ = 0

⎛ ∂ ∂ ⎞
∇ 4ψ = ⎜ 2 + 2 ⎟ ψ
2 2 2

⎝ ∂x ∂y ⎠
In cylindrical coord. (r,θ )
⎛ ⎞
⎜ ∂ ⎟
2

1 ∂ 1 ∂
∇ψ =⎜ + + 2 ⎟ ψ
2 2

⎜ ∂r r ∂r r ∂θ ⎟
4

⎝ Laplace 's operator in cylindrical coord. ⎠

2 2 See plane elasticity problems
Examples: low-velocity, small-scale, highly viscous flows.
a) Flow past immersed bodies: no exact solutions
b) Lubrication theory: -numerical sol.
c) Biofluid dynamics: -B.L. viscous/inviscid patching schemes
d) Hele-Shaw flows: -creeping flow approx.
e) Fully dev. duct flow:
f) Small particle dynamics
Linear P.D.E’s many solutions in closed form.
Happel & Brenner, Low Reynolds number hydrodynamics (1965)

Example: Stokes’ solution for an immersed sphere:

uθ ur motion of colloidal particles under

influence of an electric field theory
r of sedimentation

θ
study of movement of aerosol particles.

U a

creeping motion of a stream of speed U about a solid sphere of radius a

(r ,θ ) Stokes stream function relation
1 ∂ψ 1 ∂ψ
ur = = −
r sin θ ∂θ r sin θ ∂r
2
, u θ (1)

note: continuity eq. identically satisfied.

⎛ ∂ 1 ∂ cot θ ∂ ⎞
⎜ 2 + − ⎟ ψ =0
2 2 2

⎝ ∂r r ∂θ r ∂θ ⎠
Mom. eq. 2 2 2
(2)

BC 's
∂ψ ∂ψ
= = 0 (uθ = ur = 0)
∂θ ∂r
At r=a (3)
ur → U cos θ
As r → ∞ ψ → Ur 2 sin 2 θ + const.
uθ → −U sin θ
1
:

ψ (r,θ )=f(r)g(θ )
2
(4)

Substitute (4) into (2) and satisfy (3). Find the solution of Stokes (1851) for
for a creeping motion past a sphere

1 2 2 ⎛ a 3r 2r 2 ⎞
ψ = Ua sin θ ⎜ − + 2 ⎟
4 ⎝r a a ⎠
 ⎛ a 3 3a ⎞
ur = U cos θ ⎜1 + 3 − ⎟
⎝ 2r 2r ⎠
⎛ a 3 3a ⎞
uθ = U sin θ ⎜ −1 + 3 − ⎟
⎝ 4r 4r ⎠

Properties
1) Streamlines & velocities are entirely independent of fluid viscosity. True for all
all creeping flows.
2) The streamlines posses perfect fore-and-aft. symmetry.

∴convective accel. terms (neglected here) are responsible for strong flow
No wake predicted.

asymmetry typical of higher Re number flows.

3) The local velocity is everywhere retarded from its freestream value. There is
no faster region such as occurs in potential flow. (uθ = 1,5U at sphere shoulder)
4) The effect of the sphere extends to enermous distances:
At r=10a, the velocities are still about 10% below their free stream values.

With ur & uθ known, pressure can be found

∇p = µ∇ 2 V
 3µ aU
Result p=p ∞ − 2
cos θ p ∞ : uniform freestream pressure
2r
• Pressure deviation is proportional to µ & antisymmetric
•

3µU
a
θ ps − p∞ = − cos θ
Dp 2a

Dp : pressure drag

pressure (form)
Drag force
surface shear stress (friction drag)

µU sin θ
Shear stress distribution in fluid

⎛ 1 ∂ur ∂uθ uθ ⎞ ⎛ 3a 3 ⎞
τ rθ = µ⎜ + − ⎟=− ⎜ 3⎟
⎝ r ∂θ ∂r r ⎠ r ⎝ 2r ⎠
Total drag is found by integrating pressure & shear around the surface

F = − ∫ τ rθ sin θ dA − ∫ p r = a cos θ dA
π π

r =a

dA = 2π a 2 sin θ dθ
0 0

F = 4πµUa + 2πµUa = 6πµUa ←

Sphere − drag formula, Stokes (1851)
2/3 viscous force + 1/3 pressure force
F=6πµ Ua ←
Strictly valid only for Re<<1 but agrees with experiment up to about Re=1.
(predicts 10% low at Re=1)

Drag Coef. F/µ Ua=6π =const.

But customary definition

CD = A: cross-sectional area = π a 2
ρU 2 A
2F
 2 x6πµUa 12µ 24
CD = = =
ρU π a
2 2
ρUa Re
2a ρU
Re =
µ
2a=D

Notes:
• Formula introduces a Re effect where none exists
• underpredicts the actual drag when Re>1 due to a symmetrical wake forms

For Re>20
- Flow seperates from the rear surface, causing markedly increased pressure
drag

Comparison with inviscid solution

1 2 2 ⎛ a3 ⎞
ψ = Ur sin θ ⎜1 − 3 ⎟
2 ⎝ r ⎠

See handout
 Streamlines are close to
the body!

Recirculation is absent
sphere drags the entire surrounding fluid with it circulation streamlines
sphere pushes fluid out of the way
 Other three-dimensional body shapes
Happel & Brenner, Low Reynolds number hydrodynamics (1965)

Disk normal to the freestream: F=16µ Ua

Disk parallel to the freestream: F=(32/3)µ Ua

See White (1991) Viscous Fluid Flow

 EXACT SOLUTIONS OF NAVIER-STOKES EQS

A) Parallel Flows: i.e. velocity is unidirectional u ≠ 0 , v=w=0

⎛ ⎞
Generalized Beltrami Flows- ∇ × ⎜ ω × V ⎟ = 0
B) Similarity solutions

⎝ =0 ⎠
C) 3D Flow

ω V
• difficulties
- non-linear PDEs
- no general solution is known

Reference C.Y. Wang (1991) “Exact solutions of the steady-state N-S eqs.”
Ann. Rev. of Fluid Mech. Vol.23, pp. 159-177.
R.Berker (1963)

Almost all of the particular solutions are for the case of incompressible Newtonien
flow with constant transport properties,

∇.V = 0

ρ = −∇p + µ∇ 2 V
DV
Dt
ρcp = k ∇ 2T + Φ
DT
Dt
P: total hydrostatic pressure. i.e. it includes the gravity term for convenience.
∇P = ∇p − ρ g or P=p+ρ gz

( )
z

1) Convective accel. V.∇ → vanishes.

g = − gk
g

( )
2) Non-linear solutions V.∇ → does not vanish. y

u ≠ 0 , v=w=0
GROUP A: PARALLEL FLOWS

parallel flow: only one velocity component is different from zero.

∂u ∂v ∂w ∂u
i.e. all fluid particles moving in one direction.

+ + =0 ⇒ =0
∂x ∂y ∂z ∂x

⇒ u=u(y,z,t) ; v ≡ 0 , w ≡ 0
0 0

∂p
y-comp. of momentum eq.

⎛ ∂v ∂v ∂v ∂v ⎞ ∂p ⎛ ∂ 2v ∂ 2v ∂ 2v ⎞
ρ⎜ +u +v + w ⎟ = − + µ⎜ 2 + 2 + 2 ⎟ =0
⎝ ∂t ∂x ∂y ∂z ⎠ ∂y ⎝ ∂x ∂y ∂z ⎠ ∂y
 ∂p
=0 ⇒ p=p ( x,t )
∂z
z- comp. of mom. eq.

⎛ ∂u ∂u ∂u ∂u ⎞ ∂p ⎛ ∂ 2u ∂ 2u ∂ 2u ⎞
x- comp. of mom. eq.

ρ⎜ +u +v +w ⎟ = − + µ⎜ 2 + 2 + 2 ⎟
⎝ ∂t ∂x ∂y ∂z ⎠ ∂x ⎝ ∂x ∂y ∂z ⎠
∂u ∂p ⎛ ∂ 2u ∂ 2u ⎞
ρ = − +µ⎜ 2 + 2 ⎟
∂t ∂x ⎝ ∂y ∂z ⎠
(A)

∂u ∂ ( .)
Linear dif. eq. for u(y,z,t)

steady → =0 = 0 (In general)

∂t ∂t

1) Parallel flow through a straight channel

Couette Flows
a) both plates stationary

x
 y=0 → u=0
=µ 2
dp d 2u BC's
dx dy y=a → u=0

∂p
=0 ⇒ = const.
dp
∂y dx
∂u ⎛ dp ⎞ 1 ⎛ dp ⎞ 2 c1
µ = ⎜ ⎟ + c1 → u = ⎜ ⎟ y + y + c2
∂y ⎝ dx ⎠ 2µ ⎝ dx ⎠ µ
1 ⎛ dp ⎞ 2 c1 1 ⎛ dp ⎞
c2 = 0, ⎜ ⎟ + → = − ⎜ ⎟a
2 µ ⎝ dx ⎠ µ 2 ⎝ dx ⎠
0= a a c 1

a 2 ⎛ dp ⎞ ⎡⎛ y ⎞ ⎛ y ⎞ ⎤
u= ⎜ ⎟ ⎢⎜ ⎟ − ⎜ ⎟ ⎥ ← velocity profile
2

2 µ ⎝ dx ⎠ ⎢⎣⎝ a ⎠ ⎝ a ⎠ ⎥⎦

Shear stress distribution

⎛ dp ⎞ 1 ⎛ dp ⎞ ⎛ dp ⎞ ⎡ y 1 ⎤
τ yx = ⎜ ⎟ y − ⎜ ⎟a = a⎜ ⎟⎢ − ⎥
⎝ dx ⎠ 2 ⎝ dx ⎠ ⎝ dx ⎠ ⎣ a 2 ⎦
 Q = ∫ V .d A = ∫ u (ldy ) = −
l ⎛ dp ⎞ 3
⎜ ⎟a
a

12 µ ⎝ dx ⎠
Volume flow rate

1 ⎛ dp ⎞ 2
A 0

V =Q/ A= − ⎜ ⎟a
average vel.
12 µ ⎝ dx ⎠

= 0 & solve for y

du
Point of maximum vel.
dy
y=a/2 ⇒ du/dy=0
1 ⎛ dp ⎞ 2 3
y=a/2 → u=u max = − ⎜ ⎟a = V
8µ ⎝ dx ⎠ 2
b) upper plate moving with constant speed, U:

∂ 2u 1 ⎛ dp ⎞ 2 c1
=µ 2 → u= ⎜ ⎟ y + y + c2
2µ ⎝ dx ⎠ µ
dp
∂y
= f ( a, µ , U )
dx
dp

u = 0 @ y=0 → c 2 = 0
dx

U µ 1 ⎛ dp ⎞ ⎛ du ⎞
u = U @ y=a → c1 = − ⎜ ⎟a ⎜ = 0 at y=0 ⎟
a 2 ⎝ dx ⎠ ⎝ dy ⎠
After rearrangement,

Uy a ⎛ dp ⎞ ⎛ y ⎞ ⎛ y ⎞ ⎤
⎡
u( y) = + ⎜ ⎟ ⎢⎜ ⎟ − ⎜ ⎟ ⎥
2 2 vel. distribution depends on

a 2µ ⎝ dx ⎠ ⎢⎣⎝ a ⎠ ⎝ a ⎠ ⎥⎦
dp
both &U
dx
zero pressure gradient → = 0 → u=U
dp y
dx a
simple shear (Couette) flow

⎛ dp 2 µU ⎞
p = −1 ⎜ = 2 > 0⎟
⎝ dx a ⎠
y/a 1.0
=0 <0
dp dp
dx U dx
<0
dp y
dx a

>0
dp x
dx u/U
0 1 2 3

⎛ dp ⎞
P= ⎜− ⎟
a2
2 µU ⎝ dx ⎠ >0
dp
dx
dimensionless pressure gradient
 ρUa
-Results are valid for laminar flow only
⎛ ⎞
-Experiments show ⎜ for = 0 ⎟ ReC = ≈ 1500
µ
dp
⎝ dx ⎠
-Theory of lubrication

journal bearing

µ = const.
Temperature considerations
flow eqs are uncoupled from the temp.

Energy eq.

ρcp = k ∇ 2T + Φ
DT
Dt
⎛ ∂T ∂T ∂T ⎞ ⎛ ∂ 2T ∂ 2T ⎞
ρcp ⎜ +u +v ⎟= k⎜ 2 + 2 ⎟+Φ
⎝ ∂t ∂x ∂y ⎠ ⎝ ∂x ∂y ⎠
 T=T1

=0
U
dP
y' dx
x
2h

u =U
y y
x 2h
T=T0

Coordinate transformation → y=y'+h

u = [ y '+ h ] =
⎛ y'⎞
⎜1 + ⎟ ←
U U
⎝ h⎠
u ( y) → T ( y)
2h 2

⎛ du ⎞
0=k 2 + µ ⎜ ⎟
2 2
d T
dy ⎝ dy ⎠
 u ( y ) = ⎜1 + ⎟
U⎛ y⎞
T=T1

2h
y 2 ⎝ h⎠

= U / 2h
x
du
dy
T=T0

µU 2 y 2
T =− + c3 y + c4

T ( −h ) = T0 & T ( h ) = T1
2
4kh 2

+
⎡ 1 0 − ⎤ µ ⎛ ⎞
T =⎢ + ⎥ + ⎜1 − 2 ⎟
2 2
T T T T y U y
⎣ 2 2 h ⎦ 8k ⎝ h ⎠
1 0

linear temp. dist. temp. rise due to

pure conduction viscous dissipation in
the fluid
Non-dimensionalizing T, by (T1-T0)

µcp
dimensionless dissipation parameter, Brinkman number, Br

µU 2
Br = = = Pr .Ec
k (T1 − T0 ) k c p (T1 − T0 )
U2

Br → represents the ratio dissipation effects to fluid

conduction effects.
T1
+1
Br=16
0
Br=8
-1
T0 Br=0

Temperature profile
For low-speed flows, only the most viscous fluids (oils) have significant Br.

Example: U=10 m/s T1-T0=10 °C

Fluid µ (kg/m-s) k (W/m-°C) Br

Air 1.8x10-5 0.26 0.0007

Water 1.0x10-3 0.6 0.017

Mercury 1.54x10-3 8.7 0.0018

SAE 30 oil 0.29 0.145 20.0

Except for heavy oils, we neglect dissipation effects in low-speed temperature

analyses.
Rate of heat transfer at the wall
lower surface

∂T µU 2
qw = k = (T1 − T0 ) ± ⎡ W ⎤
k :⎢ 0 ⎥
k
∂y ±h
2h 4h ⎣m C ⎦
upper surface

Convective heat transfer coef. ζ W/m2°C

qw = ζ (T1 − T0 )

ζL
NuL = = Ch Re L Pr
L: characteristic length of the
flow geometry, L=2h
k
2hζ
Nu2 h = = 1±
Br
k 2
Nu = 1 pure conduction Nu ≡
convection heat transfer
conduction heat transfer
 FLOW BETWEEN ROTATING CONCENTRIC CYLINDERS

steady rotational speed (angular vel.)

ω2 ω1 parallel flow: vr = vz = 0, vθ ≠ 0
vθ , T & p must be functions only of r.
r1

r2

Equations of motion of incompressible Newtonian Fluids in cylindrical

and spherical coordinates: See appendix B, Viscous Fluid Flow, F.
White
Continuity equation:

1 ∂ 1 ∂ ∂ ∂Vθ
( rvr ) + ( vθ ) + ( vz ) = 0 =0 ( 6)
r ∂r r ∂θ ∂z ∂θ
∂
=0
∂t
steady flow,

∂
=0
∂θ
axisymmetric flow

∂
=0
∂z
cylinder infinitely long

unidirectional motion vθ = vθ ( r )

Momentum eq. in r - direction balance between centrifugal force & force

=ρ (1)
which is produced by the induced pressure
dp vθ2 field.

>0
dr r p dp
p as r
dr
streamline
 ( 2)
d 2 vθ d ⎛ vθ ⎞
θ − direction → + ⎜ ⎟=0
dr 2
dr ⎝ r ⎠

⎛ dvθ vθ ⎞
( 3)
k d ⎛ dT ⎞
⎜r ⎟+µ⎜ − ⎟
2

r dr ⎝ dr ⎠ ⎝ dr r ⎠
Energy equation 0=

viscous dissipation term

vθ = ω1r1 & T=T1

BC’s
At r=r1 p=p1
At r=r2 vθ = ω2 r2 & T=T2

Integrate eq. (2)

d ⎡ dvθ vθ ⎤
⎢ + ⎥=0 ⇒ + = c1
dvθ vθ
dr ⎣ dr r⎦ dr r

= ( rvθ ) = c1 → ( rvθ ) = c1r ⇒ rvθ = c1 + c2

1 d d r2

vθ = cr + c2 / r
r dr dr 2
Momentum eq. in θ direction

⎛ ∂Vθ ∂Vθ Vθ ∂Vθ VrVθ ∂Vθ ⎞

ρ⎜ + vr + + + vz ⎟
⎝ ∂t ∂r r ∂θ r ∂z ⎠
⎡ ∂ ⎛1 ∂ ⎞ 1 ∂ Vθ ∂ Vθ 2 ∂Vr ⎤ 1 ∂p
=µ⎢ ⎜ ( rVθ ) ⎟ + 2 2 + 2 + 2 ⎥ − + ρ gθ
2 2

⎣ ∂r ⎝ r ∂r ⎠ r ∂θ ∂z r ∂θ ⎦ r ∂θ

ρ = const.
1. steady
2.

axisymmetric in θ -dir.
3. Fully developed in z-dir.
4.
5. vr = vz = 0

∂ ⎛1 ∂
⎜ ( θ )⎟ = 0
⎞
∂r ⎝ r ∂r ⎠
rV

1 ∂ ∂
( rVθ ) = c1 → ( rVθ ) = c1r
r ∂r ∂r
rVθ = c1 + c2 → Vθ = cr + c2 / r
r2
2

Energy eq.

k d ⎛ dT ⎞ ⎛ dVθ Vθ ⎞
0= ⎜r ⎟+µ⎜ − ⎟
2

r dr ⎝ dr ⎠ ⎝ dr r ⎠
(3)

BC’s

At r=r1 vθ = ω1r1 & T=T1 p=p1

At r=r2 vθ = ω2 r2 & T=T2

r22 r12 (ω2 − ω1 )

using the B.C’s

r22ω2 − r12ω1
c= c1 =
r2 − r1 r22 − r12
2 2
,
 ⎢ ( 22 1 1 )
⎡ ⎤
Vθ (r ) = 2 2 ω − ω − ( 2 1 )⎥
ω − ω ( 4)
2 2
1 r2 r1

r2 − r1 ⎣ ⎦
2 2
r r r
r

p=p(r) ← obtain this distribution!

Eq. (1) determines the radial pressure distribution resulting from the motion.

Having found vθ(r) , it is substituted into eq. (3) to find temperature distribution

T − T1 r2 (1 − ω2 / ω1 ) ⎛ r12 ⎞ ⎡ ln ( r / r1 ) ⎤ ln ( r / r1 )
= Pr Ec ⎜1 − 2 ⎟ ⎢1 − ⎥+
⎝ r ⎠ ⎣ ln ( r2 / r1 ) ⎦ ln ( r2 / r1 )
4 2

T2 − T1 r2 − r1
4 4

µ r12ω12
Pr Ec =
k (T2 − T1 )
where Brinkman number
expressing the temp. rise due to dissipation

T − T1 ln ( r / r1 )
=
T2 − T1 ln ( r2 / r1 )
if PrEc=0 heat conduction solution
SPECIAL CASES :
case when r1 → 0 i.e. in the limit as the inner cylinder vanishes
(α=0)=r1/r2
i)

Eq.(4) Vθ (r1 → 0) = ω2 r rigid-body rotation

ω = ∇ × V = ω2
i.e. fluid rotates inside the outer cylinder as a rigid body

ω2 r2
forced vortex → ω ≠ 0

ii) single cylinder rotating in an infinite fluid (r2 → ∞, ω2 = 0)

r ω1
Vθ (r2 → ∞) =
circular flow
vθ .r = const.
2
1

∞ → vθ
ω1r1
r r 0

free vortex → ω = 0
Torque transmitted by the fluid to the cylinder of length L

M 1 = Fr1 = τ rθ 2π r1 Lr1

τ rθ
1 ⎛ 1 ∂Vr ∂Vθ Vθ ⎞
= 2 µε rθ = 2µ ⎜ + − ⎟ M 1 = 4πµ Lr ω1
2 ⎝ r ∂θ ∂r r ⎠
2
r = r1 1

=-2µω1
iii) very small clearance between the cylinders,

r2 − r1
<< 1, Let ω2 = 0
r1
Eq. (4) becomes, in the limit,

r − r1
≈ 1−
Vθ represents linear Couette flow between parallel plates

r1ω1 r2 − r1 0

r2

ω1r1 r1
Viscometry: to determine the viscosity of a fluid calculate the moment (or torque)
exerted by the cylinders upon each other.

Moment = Stress x Area x Moment arm

moments on inner and outer walls are equal since the system is in equilibrium, i.e
steady flow with no heat loss.

∴ The moment on the outer wall with the cylinder of height L.

M 2 = τ rθ 2π r2 L r2
lateral area

r = r2 0

1 ⎛ 1 ∂Vr ∂Vθ Vθ ⎞
τ rθ = 2µε rθ = 2µ ⎜ + − ⎟
r = r2
2 ⎝ r ∂θ ∂r r ⎠
2 µ 1 r2 (ω1 − ω2 ) ⎤
⎡
=− 2 ⎢ ⎥
2 2
r
can be measured r ⎣ r2 − r1
2 2
⎦

M 2 = 4πµ L 2 2 (ω2 − ω1 )
r12 r22
r2 − r1
 • if inner cylinder is at rest, ω1 = 0
torque transmitted by the outer cylinder to the fluid, M2

M 2 = 4πµ L 2 2 ω2
r12 r22
r2 − r1
Work out

ω2 = 0
CASE I : inner cylinder rotates with the outer one at rest.

CASE II : ω1 = 0

Both case I & II called Couette Flows

α =r1 / r2 ,
r2
Let the ratio of radii by
width of the annulus by s=r2 − r1
r1

current relative radius by x=r/r2

x=r/r2
 α 1 − x2
Case I → =
Vθ
VθI 1 − α 2 x
(inner rotating, outer at rest)

α ⎛x α⎞
Casse II → = 2 ⎜
− ⎟
Vθ
VθII 1 − α ⎝ α x ⎠
(outer rotating, inner at rest)

VθI = r1ω1 (peripheral velocity of the inner cylinder)

VθII = r2ω2 ( " " outer " )

1.0 α= 0.1

x ' r − r1
=
0.2
Let 0.4
0.6
s s 0.8
Vθ 1.0
Vθ I
comment on plots

x ' r − r1
0 1 s = s
Example :
Axially annular Couette flow between concentric moving cylinders

a) inner cylinder moving

b) Outer cylinder moving

See White (1991) pg. 109

 Stability of Couette Flows
All solutions up to now are exact steady flow solutions of N-S. equations.

Laminar flows smooth streamline character

ALL Laminar flows become unstable at a finite value of some critical parameter,
usually the Re.

unstable a different type of flow sets in

- sometimes still laminar (different steady solution, symmetry-asymmetry
steady time-periodic
- more often, turbulent flow: fluctuating flow regime, random

U
h

Re h = = 1500
υ
laminar flow velocity
profile Uh
0
turbulent flow (approx.)

-h
Time-averaged S-shaped profile
-varies slightly with Re
-τw (and heat transfer rate) increases by two orders of magnitude.

Taylor (1923) rotating inner-cylinder

laminar profile is valid until a critical rotation rate.
For small clearence (r2-r1)<<r1, critical value of instability is given by
Taylor number,

ω
Ta = r1 (r2 − r1 )3 12 ≈ 1700
2

υ
If Ta>1700 3-D laminar flow consisting of counter rotating pairs of vortices
Taylor vortices
 POISEUILLE FLOW:
Coutte flows flow is driven by moving walls
Poiseuille flows “ “ “ pressure gradients.

application to duct flows.

Consider steady flow of a viscous fluid in a

straight duct of arbitrary but constant shape
Flow in the entry region of a tube

entrance length, Le fully developed flow

U0
viscous
layers
potential
core
U0

core vanishes,
T.B.L. layer coalesce
L.B.L. transition
depending on Re # both developing
and developed regions remain laminar
• wall friction causes viscous layer

Fully developed flow: slightly further downstream of the coalescence, the flow
profile ceases to change with axial position
 constant shape cross-section

p : total hydrostatic pressure

p = p + ρ gz
h
= const.
dp
(v=0)
dx y
u=u(y,z)
z
x
(w=0)

Fully developed duct flow:

Entrance effect : i.e. thin initial shear layer and core acceleration

Shah & London (1978) regardless of duct shape for laminar flow

≈ c1 + c2 Re Dh c1 ≈ 0.5 , c 2 ≈ 0.05
Le
Dh
Dh : suitable “diameter” scale for the duct

Re c ≈ 2000 ⇒ L e ≤ 100 D (pipes of circular cross-section)

 x > Le velocity becomes purely axial, v = w = 0
u varies only with the lateral coord. u = u(y,z)

Governing eqs. for incompressible flow

∂u ∂v ∂w ∂u
+ + =0 ⇒
∂x ∂y ∂z ∂x
Continuity =0

Momentum eq.

∂u ∂u ∂u ∂pˆ ⎛ ∂ 2u ∂ 2u ∂ 2u ⎞
x − direction u + v + w = − + µ ⎜ 2 + 2 + 2 ⎟ (3)
∂x ∂y ∂z ∂x ⎝ ∂x ∂y ∂z ⎠
∂pˆ ⎛ ∂ 2u ∂ 2u ⎞
0= − +µ⎜ 2 + 2 ⎟
∂x ⎝ ∂y ∂z ⎠
∂pˆ ∂pˆ
y - and z -direction 0 = − = − ⇒ pˆ = pˆ ( x) only
∂y ∂z
 ∂p ⎛ ∂ 2u ∂ 2u ⎞
= µ⎜ 2 + 2 ⎟
∂x ⎝ ∂y ∂z ⎠
⇓

= const.(negative)
dp
dx
∂ 2u ∂ 2u 1 ∂ p
+ 2 = = const.
∂y ∂z µ ∂x
Basic eq. for fully-dev. duct flow.
2
(A)

∇ 2u = c
B.C. no-slip condition on duct surface uw = 0

Poisson equation if c = 0
Laplace eq.

• no general solution for arbitrary cross-section

• but many known solutions for different duct shapes.

See R. Berker (1963), and Viscous Fluid Flow F. M. White (1991)

To non-dimensionalize eq. (A) no characteristic vel. & axial length scale
h: characteristic duct width

µu
y =
∗
, z = , u* =
y z
⎛ dp⎞
*

h ⎜−
⎜ dx ⎟⎟
h h
⎝ ⎠
2

( u ) = −1
eq. (A) becomes

∇ ∗2 ∗
u * = 0 on duct boundary

The circular pipe: Hagen-Poiseuille Flow

θ
r0 r

(y,z) plane (r,θ) circle

Cylindrical coord. system is preferred
1 d ⎛ d ⎞
∇2 = ⎜r ⎟ vr = vθ = 0, v z = u
r dr ⎝ dr ⎠
1 d ⎛ du ⎞ 1 d p
⎜r ⎟ = = const.
r dr ⎝ dr ⎠ µ dx

= r + c1
dr 2µ dx
du 1 dp 2
r

r + c1 ln r + c2
4 µ dx
1 dp 2
u(r)=

u (r = r0 ) = 0
u (r = 0) = finite → c1 = 0 (ln0)
or u(r) is even function u(-r)=u(r) → c1 = 0

u (r0 ) = 0 → c 2 = −
4 µ dx
1 dp 2
r0
 u (r ) = −
4µ dx
( r0 − r )
1 dp 2 2
note:
dp
dx
<0 u>0

paraboloid about centerline

π r04 ⎛ d p ⎞
Volume flow rate, Q

Q = ∫ udA = ∫ u 2π rdr = ⎜⎜ ⎟⎟
r0

A r =0
8 µ ⎝ dx ⎠

r02 ⎛ d p ⎞ 1
mean velocity

u =Q/ A= 2 = ⎜⎜ − ⎟⎟ = umax
π r0 8µ ⎝ dx ⎠ 2
Q

The wall shear stress,

⎡ ∂u ∂vr ⎤ r0 ⎛ d p ⎞ 4µ u
τ w = −µ ⎢ + ⎥ = ⎜− ⎟⎟ =
⎜
⎣ ∂r ∂x ⎦ r = r0 2 ⎝ dx ⎠ r0
 Introduce the pipe-friction factor λ

dp λ u
− = ρ
2

Darcy-Weisbach eq.
dx 2r0 2
u 2r0 ρ
Re D ≡ ⇒ λ=
µ
64
ReC ≈ 2300
VALID FOR Laminar flow
Re

2τ w λ
Cf = = → Cf =
16
ρu
2
Re D 4

fanning friction factor

(skin friction factor)
 ( u ) = −1
Non-circular ducts:
∇ *2 *
; u* = 0 Dirichlet problem
Berker (1963) , White, Viscous Fluid Flow

rectangle, ellipse, concentric annulus, circular sector, equilateral triangle, etc.

Example: equilateral triangle y

a a
z

− d p / dx ⎛ ⎞
( )
a

u ( y, z ) = ⎜ − ⎟ −
1
2 3aµ ⎝ ⎠
2 2
z a 3 3 y z
2
4 × area
Dh = = Dh =
4A a
P wetted perimeter 3
ρ uDh
Re =
µ
 Temperature distribution in fully developed duct flow:
u(y,z) is known
energy eq. can be solved for T
T(x,y,z) if B.C. change with x.

Ex: Tw=const.
Pipe Flow

r0 u (r ) = −
4 µ dx
(
1 dp 2 2
r0 − r )

T(r) only

Energy equation: Appendix B, White (1991)

k d ⎛ dT ⎞ ⎛ du ⎞ 16 µ u r
⎜ ⎟ = − µ ⎜ ⎟ = −
2 2 2

r dr ⎝ dr ⎠ ⎝ ⎠
r 4
d r r
T (r ) = Ar 4 + c1 ln r + c2
0

A is known
at r=0 T=finite c1 = 0

µu ⎛ r ⎞ µu
T = Tw + ⎜1 − 4 ⎟ ; Tmax = Tw +
2 4 2
at r=r0 T=Tw
k ⎝ r0 ⎠ k
max. temp. rise due to dissipation

µu ⎛ r4 ⎞
Br = T = 1 + Br ⎜1 − 4 ⎟
2

⎝ r0 ⎠
*

kTw
µu
Air → u = 100 ft/sec
2

max. temp. rise ≈ 1 °F (air)

k

" " ≈ 3 °F (water)

oils µ is large dissipation important

or gas dynamics where velocities are high.
 The wall heat transfer, qw

qw = k q w = h(Tw − T0 )
dT
or
dr r = r0

4k (Tw − T0 )
qw = = =8
h(2r0 ) qw (2r0 )
k (Tw − T0 )
or , Nu=
r0 k
for viscous dissipation

Homework Find u(y,z)

z
(-a,b) (a,b) Contours of u(y,z)
can use method of eigen function
y expansion

Let a=1/2
b=1
(a,-b)
(-a,-b)
 VISCOUS FLOW NEAR A STAGNATION POINT

(region of non-zero vorticity, ω )

0 rigid boundary
B-L thickness
stagnation point (u=v=0)

• At large distances from stagnation point, the flow essentially the same as that of the
corresponding potential flow problem.

∴ can use potential flow to solve the viscous flow problem.

 2-D Flow: Plane Stagnation Flow:
∂u ∂v
+ =0
∂x ∂y
Continuity (1)

Introduce stream function ψ ( x,y )

∂ψ ∂ψ
, v= − ψ satisfies cont. eq. identically
∂y ∂x
u= (2)

it must also satisfy the 2-D eq. of motions

∂u ∂u 1 ∂p ⎛ ∂ 2u ∂ 2u ⎞
+v =− +υ ⎜ 2 + 2 ⎟
∂x ∂y ρ ∂x ⎝ ∂x ∂y ⎠
u

∂v ∂v 1 ∂p ⎛ ∂ 2v ∂ 2v ⎞
(3)

u +v = − +υ ⎜ 2 + 2 ⎟
∂x ∂y ρ ∂y ⎝ ∂x ∂y ⎠

Invıscid Flow near a stagnation point of a body

ψ = Bxy , u=Bx , v= − By → V = Bxi − By j

B positive constant proportional to U 0 / L
 +
ρ 2
P1
= +
V12 P0 V02
ρ 2
⇒ P1 = P0 − ρ x + y2
B2 2
2
( )
P0 : stagnation pressure
Inviscid flow is described above slips at the wall, i.e. u ≠ 0 at y=0
It must be modified to account viscous effects.

ψ viscous = xf ( y ) → u=xf '(y) , v= − f(y) (4a)

and

p = p0 − ρ ⎡⎣ x + F ( y ) ⎤⎦
B2 2
(4b)
2
stream vel.
Note: Continuity eq. is automatically satisfied U0
B is ∼
Substituting eqs. (4a-b) into (3)
L charact.
body length

xf '. f '+ (− f ) xf " = − (−ρB 2 x) + υ ( 0 + xf "')

ρ
1

( f ') − ff " = B 2 + υ f "' ODE

2
(5a)
dimensional
 1 ⎛ B2 ⎞
0+ ( − f )( − f ') = − ⎜ −ρ F '( y ) ⎟ + υ ( 0 − f ")
ρ⎝ 2 ⎠
ff ' = B F '− υ f " ( 5b )
1 2
2
BCs for the unknown functions f & F

u=v=0 on the wall ⇒ f = f ' = 0 at y=0

p = p0 ( x=0 ) ⇒ F ( 0 ) = 0 at y=0

( inviscid flow ) ⇒ Bx=xf ' ( ∞ )

y=0
y→∞ ⇒ u=Bx
f '(∞) = B

Can first solve eq. (5a) for f then substitute the result into eq. (5b) to get F.

- instead of solving u,v & p(x,y) or ψ(x,y) via PDE

• Note :

we solve ODE only.

eliminated x, leaving only y a single similarity variable.
To eliminate dimensional constants B&υ , eq. (5a) needs to be non-dimensionalized
as follows.
• since no body-length scale "L" for this flow
υ
=m
m2
use the proper length scale as
B s.1/s

υB = = m/s
m2 1
" " velocity "
s s

( )
Appropriate dimensionless variables then

f ( y)
η=y , φ (η ) = or ψ =xφ (η ) Bυ
B
ν Bν
6a-b)

substituting eqs. 6a-b) into ( 5a )

dφ
φ "'+ φφ "− φ '2 + 1 = 0 φ'=
dη
(7) , etc
BC’s for φ
η = 0 ⇒ φ=0, φ'=0 No analytic solution exists
η=∞ ⇒ φ'=1
Numerical integration.

Solution of eq. (7) is given first by Hiemenz (1911).

Can consider thickness of B.L. y=δ where u=0.99 U
corresponding value of the

ν
potential flow

δ ≈ 2.4 ∼ ν
B

η= y
B
ν

F ' ( 2.4 ) = 0.99 → 2.44 ≅ δ

B
ν
dη
u = xf ' ( y ) = x Bνφ ' ( η ) = xB φ ' ( η ) = u

f ( y)
dy

; φ ( η) =
Ui inviscid vel. profile

η= y
B
ν Bν
v = − f ( y ) = − Bνφ ( η ) u, v satisfy the no-slip condition

= φ ' ( η)
u
Ui

φ ∝ψ
1

φ' ∝ u
u
Ui

φ" ∝ τ

η=y
B
0 2.4 ν
1 2 3
Fig. 1
 φ "'+ φφ "+ 1 − φ '2 = 0
Numerical Solution

3rd order ODE ⇒ 3 1st order ODE ‘s

Runge-Kutta numerical integration

Let Y (1) = φ " dY (1)

= Y ( 2 ) ∗ Y ( 2 ) − 1 − Y ( 3) ∗ Y (1)
Y ( 2) = φ '
dη
dY ( 2 )
Y ( 3) = φ = Y (1)
dη
dY ( 3)
= Y ( 2)
dη
Solve for Y (1) , Y ( 2 ) & Y ( 3)

φ ( 0 ) = φ ' ( 0 ) = 0 ⇒ Y0 ( 2 ) = Y0 ( 3) = 0 i.e. Y ( 2 ) = Y ( 3) = 0 at η=0

Initial conditions:

φ' ( ∞ ) = 1 ⇒ Y (2) → 1 as η → ∞
how large is “infinity” η→∞
answer → when φ" becomes very small, say <10-5

= φ ' ( η) → 1 ⇒ φ ( η ) ≅ η + const.
asymptotic analysis

as η becomes large
u
Ui
∴ φ "'+ φφ "+ 1 − φ '2 = 0

φ"'
vanishes

∴ ≅ −η φ" ≈ e - η2 / 2

φ"
or

φ '' < 10−5 if η > 4.8 = "infinity"

φ" ≈ 1 to 1.5 ← guess Y (1) = 1 ∼ 1.5 at η=0

4th order R-K method

φ'=u/U i
φ" ( 0 ) = 1.23259 = τ w|
y =0
Numerical Solution

η φ'
0.1 0.11826
0.2 0.22661
.
.
2.4 0.99055
2.8 0.99705
3.0 0.99843
 y

= 0.99
δ
u
U
x
B-L behaviour : no-slip condition creates a low-vel. region which merges smoothly

outer flow → called "free stream" velocity , U ( x )

with the outer inviscid flow

Stagnation flow: U=u ( x,∞ ) = Bx

= 0.99 = φ ' ( η = 2.4 )

accelerating freestream pressure
u
⇒ favorable pressure gradient
decreases in the flow direction
Ui

η=2.4= ( y=δ )
B
ν
 ν
δ ≈ 2.4 = const.
B
because thinning due to stream acceleration, U=Bx,
exactly balances the thickening due to viscous diffusion.

e.g. if U=cxm

the B-L will grow with x if m<1

will become thinner with x if m>1

Pressure distribution: exhibits B.L. behaviour

p = p0 − ρ ⎡⎣ x + F ( y ) ⎤⎦ p=p ( x,y )
B2 2

∂p
2
= − ρ B x = − ρU
dU U=Bx
∂x
2
Bernoulli eq. gradient parallel to the
dx
∂p ff '+ νf " 2
F ' ( y ) , F' ( y ) =
wall satisfies Bern. eq.
= −ρ
B2
∂y
= − ρ ( f f '+νf ")
2 1 B2
small if υ is small
typical to laminar boundary layers
u = xf ' ( y )
Wall shear stress

f ( y)
v = − f ( y ) = − Bνφ ( η) φ ( η) = , η=y
B
Bν ν
⎛ ∂u ∂v ⎞ ∂v
τ w = µ ⎜ + ⎟ , v = v( y ) → = 0
⎝ ∂y ∂x ⎠ y =0 ∂x
∂u ∂u ∂η
=U i φ′′ ( η)
∂y ∂η ∂y ν
B
=

Β
τ w = µ Ui φ
ν
"
0

=U i φ0" µρ B
τ w ∼ U i wall shear stress is proportional to freestream vel.)
2τ w 2φ′′(0)
Cf = = Re x =
ρU
Ux
ν
2 Friction factor varies inversly with the
Re x square root of the local Reynolds number
Common in L.B. Layers.
Stagnation-point problem: Temperature distribution
• velocities are known

u = xf ' ( y ) η=y
B
ν
f ( y)
v = − f ( y ) φ ( η) =
Bν

A similarity solution exists if the wall & stream temperatures Tw and T∞

are constant.
* A realistic approximation in typical heat transfer problems
Goldstein (1938)

⎛ ∂T ∂T ⎞ ⎛ ∂ 2T ∂ 2T ⎞
ρcp ⎜ u +v ⎟=k⎜ 2 + 2 ⎟
⎝ ∂x ∂y ⎠ ⎝ ∂x ∂y ⎠
 T − Tw
θ ( η) = , η=y
B
T∞ − Tw ν
with Tw & T∞ being constant, the fluid temp T=T ( y ) only.

= − Bνφ ( η ) . (T∞ − Tw ) = − B (T∞ − Tw ) φ ( η )

∂T ∂T ∂θ B ∂θ
= − f ( y)
∂y ∂y ∂η ν ∂η
v
∂η
∂y

= (T∞ − Tw ) 2
∂ 2T ∂ 2θ B
∂y 2
∂η ν

− ρ c p B (T∞ − Tw ) φ = k (T∞ − Tw )
∂θ B ∂ 2θ
α=
ρ cp
k
∂η ν ∂η2
d 2θ dθ ν µc p
+ φ ( η ) =0 = = const.
dη dη α
2
Pr Pr=
k

BC ' s θ ( 0) = 0 & θ ( ∞ ) = 1
φ ( η) → stream function is known from flow problem Fig.1
Tabulated values are known

IVP RK4 subroutine

But has an exact solution. (Linear eq.)

⎡ η ⎤
∫0 d η exp ⎢-Pr ∫0 φ ds ⎥
η

θ ( η) = ∞ ⎣
( )
⎦ = θ = θ Pr
⎡ ⎤
∫0 d η exp ⎢-Pr ∫0 φ ds ⎥
η

⎣ ⎦

thermal boundary layer, δ T , is the point where θ ≈ 0.99

 1.0
10 1 0.3 0.1

T − Tw
θ=
T∞ − Tw 0.01

Pr=0.001

η=y
B
ν
0
1 2 8

δu
≈ Pr 0.4 (Power law curve fit)
δT
ν
velocity B.L. is thicker than thermal B.L. because viscous
diff. exceeds conduction effects.

α
Pr=

Pr > 1 → δ > δ T
ν>α thermal diff.
viscous diff.
 Heat transfer at the wall is computed from Fourier’s Law

= −k (T∞ − Tw ) = − k (T∞ − Tw )
∂T dθ
qw = − k G ( Pr )
B B
∂y y =0
d η η=0 ν ν
⎡ η ⎤
where G ( Pr ) = ∫ d η exp ⎢-Pr ∫ φ ds ⎥
∞
−1

0 ⎣ 0 ⎦

Y ( 3) = φ
dY ( 4 )
RK 4 Formulation:

= Y ( 3) Y ( 4 ) = ∫ φd η
dη
dY ( 5 )
= exp ( − Pr∗ Y ( 4 ) )
dη
Y ( 5 ) ≡ G ( Pr )
 10
Pr G (plane flow)
G(Pr) 0.01 0.076
1 0.1 0.22
1 0.57
0.1 10 1.339
100 2.986
0.01 1 10 100 1000 Pr
1000 6.529

G ≈ 0.57 Pr 0.4
BC ' s θ ( 0 ) = 0 ⇒ Y0 ( 4 ) = 0
dθ dθ
+ φ ( η ) =0 ( II ) θ(∞) = 1 Y0 ( 5 ) = 1
2

dη dη
OR Pr

Y ( 4 ) = θ ( η)
2

dY ( 5 )
Let
dθ
Y ( 5) = ⇒ = − Pr∗ Y ( 3) .Y ( 5 ) ( 4)
dη dη
dY ( 4 )
= Y (5) ( 5)
dη
Y (1) = φ "
Y ( 2 ) = φ' → solve for Yi ( η ) i=1,...5
Y ( 3) = φ
RK4 Routine depends on Pr

1 ∂p
B2 x = − p=-ρB2 + K ( y )
2
x
ρ ∂x 2
1 ∂p
+ B2 y = − B2 y = − K ' ( y ) K' ( y ) = -ρB2 + K1
2
1 y
ρ ∂y ρ 2

p = − B2 ( x2 + y2 ) + C
ρ
x=0 , y=0 p=p0 stagnation pressure

p = p0 − B 2 ( x 2 + y 2 )
ρ
2
c=p0

p = p0 − ( u 2 + v 2 )
ρ
2
Bernoulli Eq.
2
 ⎡1 ⎤
ψ =Bxy ⎢ m 2 ⎥
y
⎣s ⎦
u =Bx v=-By

u ≠ 0 slips on the wall

m/s 1/s m
x

p = p0 − ρ ( x + F ( y ) )
y=0

ψ =xf ( y ) u=xf ' ( y ) v=-f ( y )

B2 2

∂u ∂v
f '( y) − f '( y) = 0
2
+ =0
∂x ∂y
C.E

xf ' + ( − f ) xf " = − ( −ρB 2 x ) + ν ( 0 + xf "' )

1
ρ
2

f '2 − ff " = B 2 + νf "' ODE

xf '2 + ( − f ) xf " = +ν ( 0 + xf "')

f '2 − ff " = B 2 + νf "'
1 ⎛ B2 ⎞
xf ' ( y ) .0 + ( − f )( − f ') = − ⎜ −ρ F ' ⎟ + ν ( 0 − f ")
ρ⎝ 2 ⎠
ff '= B F '− νf "
1 2
2
u = Bx y→∞ f '(∞) = B
u=0, f ' ( 0 ) = 0
v=0, f ( 0 ) = 0
y=0

x = 0⎞
⎟ F ( 0) = 0
y = 0⎠
p=p0

: [ m]
ν
η=
y
ν
proper length scale
B

ψ
νB : [ m / s ]
B
= φ ( η)
νB x
velocity scale

f = Bνφ ( η ) df ∂
=
( )
Bνφ ( η ) ∂η
= Bνφ ' B / ν
dy ∂η ∂y
∂2 f
B 2 Β = ( Bφ ' ) = Bφ "
2
B 2 φ '2 − BνφΒφ" =B +ν φ"'
d B
ν ν ∂y 2
dy ν
∂3 f d ⎛ B ⎞
= ⎜ φ ⎟ = φ
∂y 3 dy ⎜⎝ ν ⎟⎠
B B
φ'2 -φφ"=1+φ"' ν ν
B " B "'
η=0 u=0=xBφ' φ' ( 0 ) = 0
v=0=- Bνφ φ ( 0) = 0

η→∞ u=xBφ' ( η → ∞ ) = xB φ' ( η → ∞ ) = 1

Fig s.11 ( Schlichting )

u=xf ' ( y ) = xB φ ' ( η)

v=-f ( y ) = − Bνφ ( η)
Ui inviscid profile
Unsteady Motions of a Plate :
y

u=u(y,t) v=0

0 x

Parallel flow, v=w=0

• initially both the plate & fluid are at rest u(y,0)=0 for y>0
• the plate is jerked into motion in its own plane
• no-slip at the plate : u(y=0,t)=U(t) for t>0

Two-cases
1- U=constant (Stoke’s First Problem)
2- U(t)=U0cosωt (Stoke’s Second Problem)
 Steady oscillation of the plane at Ucosωt.

→ =0 → p=const.
dp
zero pressure gradient
dx
governing P.D.E. reduces

∂u ∂ 2u
=ν 2 (1-D heat conduction )
∂t ∂y

CASE I: U=const.
⎧0 for t = 0
u ( y = 0, t ) = ⎨
⎩U for t > 0
u ( y, t ) = finite
∂u ⎛ ∂ 2u ∂ 2u ⎞
= ν⎜ 2 + 2 ⎟
∂t ⎝ ∂x ∂y ⎠
Unsteady heat conduction equation

Carslaw and Jaeger (1959) – Conduction of heat in solids

⎛ y ⎞ ⎛ y ⎞
= 1 − erf ⎜ ⎟ = ⎜ ⎟
Solution methods u
⎝ 2 νt ⎠ ⎝ 2 νt ⎠
-Laplace transforms erfc
-Similarity methods U
erf ( β ) = ∫e
β
2 − x2

π
dx
0

Similarity Solutions :
• applicable for non-linear problems

similarity solutions exist for:

parabolic P.D.E with two independent variables when there is no geometric length

⇒ give ODE
scale in the problem.

1
y erfx
increasing time

erfcx
u x
t1 t2 U 2
t=0 1
Eg: At t=t1 → u=0.5U at y=y1
At t=t 2 → u=0.5U at y=y 2

u ( y, t )
Expect solution exists in the form

= f ( η) where η=α
y

η ( y, t ) : similarity variable
U tn

α : constant of proportionality:
will be found later to make η dimensionless.
u ( y, t ) = Uf ( η)
when η=const. ⇒ u=const. (y ∼ t )
n

and η as the independent variable.

IF similarity solution exists, it will result in an ODE with f as the dependent variable
 ∂u ∂ 2u
=ν 2 (1) , Let u=Uf ( η)
∂t ∂y
PDE

η=α n
y
t
∂u ∂u ∂η η
= =U ( ) n+1
− α = −
df y
∂t ∂η ∂t dη
n Un f'
t t
∂u ∂u ∂η α α
= = Uf ' n = U n f '
∂y ∂η ∂y t t
∂ 2u ∂ ⎛ ∂u ⎞ ∂ ⎛ ∂u ⎞ ∂η α α α2
= ⎜ ⎟= ⎜ ⎟ = U n f '' n = U 2n f ''
∂y 2
∂y ⎝ ∂y ⎠ ∂η ⎝ ∂y ⎠ ∂y t t t
Substituting above expressions to (1)

η α2
− U n f ' = ν U 2n f ''
t t
To eliminate t (hence obtain an ODE for f) Let n=1/2
∴ for n=1/2 a similarity solution is obtained

η
να f "+ ηf ' = 0
2
→ f "+ f '=0
2να
1
2 constant [-]
2

to make η dimensionless; u=Uf ( η) use ν & U ( if necessary )

η=α α : constant of proportionality

y
1/2
t

, ν= ⇒ α∼ → η= or η = [−]
y m m2 1 y y
ν νt 2 νt
∼
t s s

2ν α 2 = → α= → f "+2ηf '=0 f ( η)
1 1
2 2 ν

( ) η η =
y
2 νt
A f "+2 f '=0 ODE

u ( 0,t ) = U ( t>0 ) → f ( η=0 ) = 1

BCs

u ( y,t=0 ) = 0 ( y ≥ 0 ) → η → ∞ f ( η) → 0
 = −2η → ( ln f ') = −2η → ln = −η2
f" d f'
f' dη A

→ f ( η ) = A∫ e
η

f ' = Ae −η −ξ 2
dξ + B
2

f ( 0 ) = 1 ⇒ B=1
0

π
f ( η → ∞ ) = 0 ⇒ 0=A ∫ e
∞
−ξ 2
dξ + 1 = A +1
0
2

A=−
2
π
u ( y, t )
= f ( η) = 1 − ∫ dξ
η
−ξ
η=
2 y
π 2 νt
2
e
U
⎛ y ⎞
=Erfc ( η) = Erfc ⎜
0

⎟
⎝ 2 ν t ⎠

Complementary error function Erfc ( η) = 1 − ∫e

η
−ξ 2
dξ
2
π 0
 Erf ( η) = ∫e
η
−ξ 2
dξ
2
π
Error function

Numerical values of the Erfc ( η) is given in Tables. (pg. 103)

0

Erfc ( 0 ) = 1.0
Erfc ( 0.1) = 0.88754
Erfc ( 4.0 ) = 0.00000001
Erfc ( ∞ ) = 0

⎛ y ⎞
=f ( η) =Erfc ( η) =Erfc ⎜ ⎟
u
U ⎝ 2 νt ⎠
 η=
y
2 νt
1.8

1.0 Erfc ( η ) At η 1.82 → u/U=0.01

u/U
0.5 1.0

Viscous effects are confined to a region next to the plate where η ≤ 1.82. Outside
∂u
∂y
the region, the vel. gradient is too small to cause a significant shear stress.

In other words, vorticity dissipatesup to the point η ≈ 1.82.

Beyond this point vorticity is negligible.
This region called Boundary Layer.
It is customary to define the shear layer (thickness) (B-L thickness) as the point
where the wall effect on fluid drops to 1%.

∴ = 0.01 = f ( η ≅ 1.82 )
u
U
y=δ =2 νt 1.82 ≅ 3.64 νt

Ex : Air at 20 °C ν = 1.5×10-5 m 2 / s δ ≈ 11 cm after 1 minute

Water at 20 °C ν = 10-6 m 2 / s δ ≈ 3.64 10-6 × 60 2.8 cm
 STOKES’ SECOND PROBLEM:
u(0,t)
y

x 2π t

u ( 0, t ) = U cos ωt
u ( y, 0 ) = 0 ( at rest ) 2π
T=
u ( ∞, t ) = 0 ω

• differs from Stokes’ first problem by only in B.C. at y=0 oscillatory vel.
T = T0 cos ωt
Ex:cyclic ( daily or seasonal ) variations in surface temp.
∂T ∂ 2T
=α 2 T ( y,t )
∂t ∂y
determine temp. variations in ground
u ( y, t ) = f ( y ) eiωt ( 2)
Fluid is also to oscillate with the same frequency, so let,

eiωt = cos ωt + i sin ωt

substitute ( 2 ) into the gov. eq.
i= -1

∂u ∂ 2u
= ν 2 ⇒ iω f ( y ) eiωt = νf " ( y ) eiωt
∂t ∂y
iω
f" ( y ) − f ( y) = 0
ν
ODE

Characteristic eq.
ω ωi
α2 −i = 0 ⇒ Roots: α1,2 = ±
ν ν
General sol.

f ( y ) = Ae
ωi ωi

+ Be
−
ν ν
y y

u ( ∞, t ) → 0 ⇒ B=0
 1+ i
i=
2

f ( y ) = Ae
ω ω ω
(1+ i ) y
= Ae
− − y −i
2ν 2ν 2ν
y
e
ω
=k
2ν
u ( y, t ) = f ( y ) eiωt = Ae − ky e − iky e − iωt
=Ae − ky ei( ωt − ky )
=Ae − ky ⎡⎣cos ( ωt − ky ) + i sin ( ωt − ky ) ⎤⎦

u ( y, t ) is real ⇒ need to take real part of f ( y ) eiωt

u ( 0,t ) = U cos ωt → A=U ← Thus, vel. distribution

To evaluate A use B.C.

ω
u ( y,t ) = Ue − ky cos ( ωt − ky )
2ν
where k=
amplitude
phase shift
• velocity of fluid decreases exponentially as the distance from the plate ( y ) increases.
ω
2ν
rate of decrease k= , will be faster for higher frequency and smaller viscosity.

• fluid oscillates in time with the same frequency as the input freq. in boundary.
• amplitude of oscillation Ue-ky = g ( y )
max. amp. at y=0 → U
• phase shift cos ( ωt-ky )

specific time
instantaneous time

U
Can define thickness of oscillating shear layer, δ as again where velocity amplitude
drops to 1% of U. ∴u/U=0.01.

= 0.01 = e − ky =e − kδ
=e −4.6
⇒ δ=
u 4.6
U y =δ k
ω ν
k= ⇒ δ ≅ 6.5 δ∼ ν
2ν ω

Ex : For air at 20 C , with a plate frequency of 1 Hz ( ω=2π rd/sec )

remember characteristic of laminar flows

δ ≈ 10 mm
The wall shear stress at the oscillating plate
⎛ ∂u ⎞ ⎛ π⎞
τ w = τ y =0 = µ ⎜ ⎟ = U ρωµ sin ⎜ ωt − ⎟
⎝ ∂y ⎠ y =0 ⎝ 4⎠
⎛π π ⎞
shear stress lags the max. vel. by 135 ⎜ + ⎟
⎝2 4⎠
Since governing eq. is linear, the method of superposition is applicable. Hence,
superposing several oscill. of diff. freq. & ampl. the sol. for arbitrary periodic motion
of plate can be obtained.
 ( V.∇ )V ≠ 0
UNIFORM FLOW OVER A POROUS WALL
• non-linear inertia terms are not zero
• but linearized to permit a closed form solution
Example: steady, fully-dev. flow over a plate with suction

fully dev.

y p=const.
u=u(y)

• suction on surface is sometimes used to prevent boundary layers from separating.

I. solution with primitive variables, u, v, p

∂u ∂v
+ =0 ⇒
∂x ∂y
v=const.

∂u ∂u ⎛ ∂ 2u ∂ 2u ⎞ 1 ∂p
+v = ν⎜ 2 + 2 ⎟ −
∂x ∂y ⎜ ∂x ∂ ⎟ ρ ∂x
⎝ ⎠
x: u
y

v= − V ( uniform )
0

∂p
=0 u ( y=0 ) = 0
∂y
y:
u ( y → ∞) = U
B.C.s

−V =ν 2
du d 2u
dy dy

+ =0 α + α = 0 ⇒ α =0 & α = −
d 2u V du V V
ν dy ν ν
2
2
dy

u ( y ) = A + Be
⎛V ⎞
−⎜ ⎟ y
⎝ν⎠
• note if blowing instead of suction v=V ⇒ u ( y ) = A + Be
V
ν
y

y→∞→u→∞
not physically possible

u ( 0 ) = 0 → A+B=0
u would be unbounded
at large y
⎛ν⎞
u ( y → ∞) = U = A B.L. thickness δ ∼ ⎜ ⎟
⎝V ⎠
⎡ − y⎤
u ( y ) = U ⎢1 − e ν ⎥
V

⎣ ⎦

II. solution using Vorticity transport eq.

∂ω
( )
Vorticity Transport in 2-D

+ V .∇ ω = υ∇ 2 ω
∂t
∂ω z ∂ω z ∂ω z ⎛ ∂ 2ω z ∂ 2ω z ⎞
+u +v =υ ⎜ 2 + 2 ⎟
∂t ∂x ∂y ⎝ ∂ x ∂ y ⎠
 ∂ωz ∂
steady → =0 & ⇒ 0 fully-dev. flow
∂t ∂x
∂ω z ∂ 2ω z
=υ ⇒ ωz = ωz ( y )
∂y ∂y
v 2

∂
→ & v= − V=const. ( from continuity )
d
∂y dy
dωz d 2ω z
−V = υ
dy dy 2
convection of viscous diffusion
vorticity toward away from plate

V d ω z d 2ω z
plate

− =
ν dy dy 2
dω
Integrate once, − ω z = z + c1
V
ν dy
ωz = 0 at y → ∞ ⎫
⎪
dωz ⎬ ⇒ c1 = 0
=0 at y → ∞ ⎪
B.C.
dy ⎭
 dωz
− dy = yields, ωz = c2 e
−
V
ν
ωz
V
ν
y

but ωz = 0 at y → ∞, 0=c 2 e −∞ ⇒ c 2 ??
Now, let
∂v ∂u
ωz = − =−
du
∂x ∂y dy

ωz = − = c2 e ν
− y
V
du
dy
ν −Vν y
u = c2 e + c3

u ( y → ∞ ) = U → U=c 2 .0 + c3 → c3 = U
V

ν
u ( y = 0 ) = 0 ⇒ 0=c 2 + U → c 2 = −U
y
V
V ν
⎡ − y⎤
u=U ⎢1 − e ν ⎥ &
V

⎣ ⎦

ωz = −U e
V −Vν y
− ωz
ωz is max at y=0
UV
ν ν
terms in eq.
∂ωz
−V = −U 2 e → convection
V 3 −Vν y
∂y ν
∂ 2 ωz
ν 2 = −U 2 e → diffusion
V 3 −Vν y
∂y ν
ν
Note ∼ length ⇒ if V ⇒ vorticity moves toward wall
V
⇒ vorticity moves away from wall
ν → dif. term ( temdency of shear layer to grow due to viscous diffusion )
if V

V → convective term ( toward the wall )

As usual, define the B.L. thickness to be the point where
ν ⎛ν⎞
u=0.99U → δ =4.6 = ϑ ⎜ ⎟
V ⎝V ⎠
Eg . Air at 20 C , if V=1 cm/s, δ ≈ 7 mm
• For a plate with a leading edge (x=0), a Laminar shear layer would grow and
4ν U
x ≅ 2 U=10 m/s, V=1 cm/s x ≅ 6m
approach this constant value.
it is estimated by Iglisch (1944)
V
Ex2: Flow through & between porous plates

= const.
dp
h Flow
dx y
x

B.C. u=0, v= − V at y=0, h

2-D , N-S eqs. -V
∂u ∂u ∂u 1 ∂p ⎛ ∂ 2u ∂ 2u ⎞
+u +v =− + ν⎜ 2 + 2 ⎟
∂t ∂x ∂y ρ ∂x ⎜ ∂x ∂ ⎟
⎝ y ⎠
∂v ∂v ∂v 1 ∂p ⎛ ∂ 2v ∂ 2v ⎞
+u +v =− + ν⎜ 2 + 2 ⎟⎟
∂t ∂x ∂y ρ ∂y ⎜ ∂x ∂y
⎝ ⎠
 ⎛ ∂ ∂p
= 0 ⇒ P=f ( x )
⎞
Assume

2) Fully-dev. flow ⎜ → ⎟
1) Steady

⎝ ∂x ⎠ ∂y
0

∂u ∂v
+ =0 v=const.= − V
∂x ∂y
Cont.
2

−V =− +ν 2 → + = = const.
d 2u d 2u V du 1 dp
ν dy µ dx
du 1 dp
ρ dx 2

dα V
dy dy dy

=α ⇒ + α=
µ dx
du 1 dp
dy ν
Let
dy
DE solution α =α part + α hom

α part = , α part = A
1 dp
ρV dx

α hom = c1 e
−
V
ν
y
,

α= = c1 e + ⇒
− y
V
ν
du 1 dp
dy ρV dx
 ν −Vν y 1 dp
u ( y ) = −c1 e + y + c2
V ρV dx
B.C.s u=0 at y=0 & y=h
⎡ − y ⎤
h dp ⎢ y 1 − e ν ⎥
V

−
ρV dx h ⎢ − h ⎥
u=
⎢⎣ 1 − e ν ⎥⎦
V

if no suction ⇒ V=0, then

u V →0 =
2 µ dx
1 dp 2
( y − hy ) parabolic vel. profile

τ V =0 = ( 2 y − h)
1 dp
2 dx
⎡ V −Vν y ⎤
− e
du µ h dp ⎢ ν 1⎥
τ =µ = ⎢ + ⎥
dy ρV dx ⎢ h⎥
1− e
− h

⎢⎣ ⎥⎦
V
ν
 ⎡ V ⎤
µ h dp ⎢ − ν 1⎥
τw =µ = ⎢ + ⎥
du
ρV dx ⎢ h⎥
1− e
y =0
− h
V
y =0 ν
⎣ ⎦
dy

⎡ V −Vν h ⎤
⎢ − e
µ h dp ν 1⎥
τw = ⎢ + ⎥
ρV dx ⎢ h⎥
1− e
y =h
− h

⎢⎣ ⎥⎦
V
ν
 dp
If we have fully-dev. flow, shear stress & is related

τw ∑F =0
dx
L
for fully-dev. flow
y =h

=P1 h − P2 h + τ w .L − τ w .L = 0

(
⇒ ( P1 − P2 ) h = L τ w − τ w )
y =h y =0

h P2
P1

( P1 − P2 ) = − L dx
y =0 y =h

τw
dp

( P1 − P2 )
y =0

=−
dp
dx L

P1

P2