Ship Hydrodynamics

S. K. Bhattacharyya
Department of Ocean Engineering
Indian Institute of Technology, Madras
What is Marine Hydrodynamics?
Why study Marine Hydrodynamics?

Fluid Mechanics of ‘water’: Hydrodynamics

of ‘air’ : Aerodynamics

Engineering application areas of hydrodynamics

Hydrology: river flow, irrigation, dam, reservoir, canal etc.

Coastal Engg.: coastal processes, sediment transport,

coastal erosion, wave field in coastal zone,
loading on coastal structures

Marine Hydrodynamics: marine vehicles and offshore

platforms
 Ocean waves

Wave interaction with

fixed, floating & compliant
structures (seakeeping for vehicles)

Steady motion of vehicles

Marine (resistance)
Hydrodynamics
Control of vehicles
(maneuverability)

Devices, e.g. propeller, rudder,

fin, hydrofoil, nozzle
(propulsion)
 Theoretical

Marine Hydrodynamics Computational

(CFD)

Experimental
(model testing)

Study of “Marine Hydrodynamics” provides a sound basis

for understanding and insight into a wide range of
complex fluid dynamic problems that are of interest to
the naval architect and the ocean engineer.
In “Marine Hydrodynamics”:

u The fluid medium, water, is treated as incompressible.

u However, it can be either inviscid OR viscous.

u When water is treated as viscous, it is treated as

a Newtonian fluid.

u The flow can be either irrotational OR rotational.

Combinations:
Inviscid-Irrotational-Incompressible (potential theory)
Inviscid-Rotational-Incompressible (circulation and vorticity)
Viscous-Rotational-Incompressible (friction and boundary layer)
Why Incompressible?

Liquids do not expand to fill an available ‘volume’,

unlike gases, which are obviously compressible.

Incompressible means that the volume of a

given mass of fluid does not change with time. In other words,
density of a given mass of fluid is constant w.r.t. time.

Obviously, under pressure, a change of density is bound

to occur. The effect is neglected as being very small.
So what we mean is: for problems of Marine Hydrodynamics
water is ‘virtually incompressible’.
To study the propagation of ‘underwater sound’,
essential in Physical Oceanography, detection by
SONAR, underwater explosion, underwater vibration
of structures and associated noise, water must be
taken as compressible.

Mach No. (M = u/c) is a measure of compressibility of

fluids (liquid and gas).
c (air) = 300m/s, c(water) = 1500m/s
u (air) ~ c(air) in aerodynamics
u (water) << c(water) in hydrodynamics
\M(air) = O(1) M(water) <<1
Hence, water is virtually incompressible.
 Now, the physical constant associated with the measurement
of compressibility is the 'Bulk Modulus' (B), which is the ratio of
change in pressure to the fractional volume compression:
Dp
B=- (negative sign indicates
DV / V
compression)
'Compressibility' (K) is inverse of the bulk modulus:
DV / V p
K =-
Dp
V
V’

For water, B = 2.2´109 N/m2 (or Pa) at 20°C and 1 bar pressure, a
value to remember for most engineering applications.
Speed of sound in water:
2 1/ 2
B 2.2 ´10 N / m
9 2
é (kg.m / s ) / m ù
2
c= = = 1483ê ú = 1483 m / s
r 1000 kg / m 3
ë kg / m 3
û
 Variation of B (in GN/m2) with temperature and pressure is as
follows:

0°C 20°C 50°C 90°C 150°C

1 bar 2.012 2.205 2.287 2.122 -
100 bar 2.067 2.274 2.356 2.198 1.709

300 bar 2.184 2.398 2.494 2.329 1.867

1000 bar 2.618 2.825 2.935 2.790 2.412
 ! !
LagrangianQ p ( x , t ), Q p º v p , p p , r p etc.
particular particle over time
difficult to apply
Description of flow ! !
Eulerian Q( x , t ), Q º v, p, r etc.
at fixed point in space
over time
easier to apply

We will always adopt Eulerian description of flow

where the point of interest and frame of reference are
fixed. It is a field description involving pressure (p),
velocity (v: a vector) and density (r).
 Preliminaries
! ! ! ! ! ! ! ! ! ! !
Velocity field: v = v ( x , t ); v = ui + vj + wk ; x = xi + yj + zk
or u = u ( x, y, z , t ); v = v( x, y, z , t ); w = w( x, y, z , t )
!
Steady flow: ¶v
=0 Velocity depends on x alone
!
¶t
!
2D flow: v =< u( x, y, t ), v( x, y, t ), 0 >

Streamline: At a given time, the line everywhere tangent to

velocity vector, v, i.e. has same direction as v.
x = x( s ); y = y ( s ); z = z ( s ) is obtained by solution
dx / ds dy / ds dz / ds
of = = at given t.
u v w
Rate of change “following” the fluid
(substantial or material or co-moving or convected or
Lagrangian or total derivative: D/Dt)
!
f ( x, y, z , t ); f º v OR r etc.
¶f
: time rate of change of f at a fixed x, y, z.
¶t
Df
Time rate of change of f " following" the fluid, , is
Dt
Df d dx dy dz
= f {x(t ), y (t ), z (t ), t}; = u; = v; = w
Dt dt dt dt dt
 Chain rule gives:
Df ¶f dx ¶f dy ¶f dz ¶f
= + + +
Dt ¶x dt ¶y dt ¶z dt ¶t
¶f ¶f ¶f ¶f
= +u +v + w
¶t ¶x ¶y ¶z
¶f !
= + (v .Ñ) f
¶t
D ¶ " ¶ ¶ ¶ ¶
= + v .Ñ = + u + v + w
Dt ¶t ¶t ¶x ¶y ¶z
" "
Dv ¶v " !
\ Fluid acceleration is = + (v .Ñ)v
Dt ¶t
Convective accln.

Lagrangian accln. Eulerian accln.

 Illustration
A boat is measuring surface water temperature T(x,y,t) in a
!
channel with surface flow velocity v ( x, y, t ) = (vx , v y ).

• Boat is anchored: It is measuring ¶T

¶t

• Boat is drifting: It is measuring ¶T + (v!.Ñ)T

¶t
!
• Boat is moving with velocity U ( x, y, t ) = (U x ,U y )
!
relative to the water: It is measuring ¶T + ({v! + U }.Ñ)T
¶t
 Steady Flow
!
¶v
= 0 signifies steady flow; an Eulerian concept.
¶t
! ! !
¶v Dv ¶v ! ! ! !
If = 0; then = + (v .Ñ)v = (v.Ñ)v ¹ 0
¶t Dt ¶t
In other words, acceleration following the particle, is only convective

Incompressible Flow
Dr
=0 signifies incompressible flow; a Lagrangian concept.
Dt
Density of a particle does not change as it moves, but
different particles can have different densities.

D r ¶r ! ¶r !
= + (v .Ñ) r = 0; or, = -(v .Ñ) r
Dt ¶t ¶t
 ¶r signifies that densities of all particles, as they arrive at a
= 0 particular point, is constant. Thus if different particles had
¶t different densities, these densities must change when to make
this condition possible.

r = constant does not admit flow of more than one incompressible fluid.
All particles have equal densities.

In an incompressible flow where Dρ /Dt = 0 , the density ρ in general

a) may vary in time at a particular point in the flow field √
b) must be a constant at all points in the flow field ´
c) may vary in space but not in time throughout the flow field ´
In steady flow, the rate of change of f following a fluid element
! !
is (v .Ñ) f . To explain this, let ns be a unit vector parallel to
a the streamlines and has same sense as the flow. Then

! !! ! ¶f
v .Ñf = v ns .Ñf = v s is distance along a streamline.
¶s ! !
v and ns are collinear.
¶f
: rate of change of f with distance along a streamline.
¶s
!
So, multiplying it by flow speed v evidently gives
time rate of change following the fluid element along that
streamline. As a consequence:
!
(v.Ñ) f = 0 : f is constant along a streamline.
But it does not say whether f may be a different constant along
some other streamline.
 Methods of Visualizing Fluid Flows

Streamline: tangent to fluid velocity vector everywhere

at a given time. It is, therefore, an instantaneous
pattern.

Streakline: locus of particles which have earlier passed

through a given point in space. It is, therefore an
integrated pattern.
A streakline is formed by injecting dye into the
fluid at a fixed point in space. As time marches on,
the streakline gets longer and longer, and
represents an integrated history of the dye streak.
Pathline: actual path (trajectory) traversed by a given (marked)
particle. It is, therefore, an integrated pattern.
It depicts the history of where the particle had been.

Timeline: a set of particles that form a line segment at a

given time instant. It is, therefore, an integrated
pattern.
It follows the location of a line of fluid particles and it
distorts with time.

• For steady flow, streamlines, streaklines, and pathlines

are all identical.
• For unsteady flow, these three patterns can be quite different.
Gradient: scalar field (pressure) to vector field (force)

! æ ¶p ! ¶p ! ¶p ! ö
dF = -ç i + j + k ÷dx dy dz
è ¶x ¶y ¶z ø
!
dF ! æ ¶p ! ¶p ! ¶p ! ö Force per unit volume
= f = -ç i + j+ k÷
dx dy dz è ¶x ¶y ¶z ø or force intensity

! æ ! ¶ ! ¶ ! ¶ö
f = -grad p = -Ñp ç Ñ = i + j +k ÷
è ¶x ¶x ¶x ø
 Divergence (or Gauss) theorem

ì f1 ( x, y, z ) ü
! ! ï ï
f ( x ) = í f 2 ( x, y, z )ý is a differentiable vector field
ï f ( x, y , z ) ï
î 3 þ
! ! ! ¶f1 ¶f 2 ¶f 3
divergence of f : div f = Ñ. f = + +
¶x ¶y ¶z
The theorem states that
! !!
ò Ñ. f dV = ò f .n dS
V S

Flux through surface S

S is a closed surface enclosing
volume V. n is unit outward
normal to S.
! ! ! !
F ( x, y, z ) = Fx i + Fy j + Fz k

¶Fz ¶Fz æ f2 ¶F ö
òV ¶z dV = Vò ¶z dz dy dx =òR çç òf ¶z dz ÷÷ dy dx
z

è 1 ø
!!
= ò { Fz ( x, y, f 2 ) - Fz ( x, y, f1 )} dy dx S 2 : dy dx = cos g 2 dS 2 = k .n2 dS 2 (" g 2 acute)
R
!!
=ò Fz ( x, y, f 2 ) dy dx - ò Fz ( x, y, f1 ) dy dx S1 : dy dx = - cos g 1 dS1 = -k .n1 dS1 ("g 1 obtuse)
R R

!! ìï !! üï !! !! !!
= ò Fz k .n2 dS 2 - í- ò Fz k .n1 dS1 ý = ò Fz k .n2 dS 2 + ò Fz k .n1 dS1 = ò Fz k .n dS
S2 îï S1 þï S2 S1 S

¶Fz !! ¶Fx !! ¶Fy !!

or, ò dV = ò Fz k .n dS Similarly, ò dV = ò Fx i .n dS ; ò dV = ò Fy j .n dS
V
¶z S V
¶x S V
¶y S

æ ¶Fx ¶Fy ¶Fz ö ! ! ! !

Adding, ò ç + + ÷ dV = ò ( Fx i + Fy j + Fz k ).n dS
Vè
¶x ¶ y ¶z ø S
! !!
or, ò ÑF dV = ò F .n dS
V S
 (0, 0, 1)

(-1, 0, 0) Box : x0 £ x £ x1 ;
y0 £ y £ y1 ;
(0, 1, 0) z0 £ z £ z1
(0, -1, 0)

(1, 0, 0)
" " "
(0, 0, -1) ìnx ü ìcos( n , x) ü ì cos( n , i ) ü
" ï ï ï " ï ï " " ï
n = ín y ý = ícos( n , y )ý = ícos( n , j ) ý
ï n ï ï cos( n", z ) ï ïcos( n", k! )ï
î zþ î þ î þ
! x1 y1 z1 æ ¶f ¶f 2 ¶f 3 ö
òV Ñ. f dV = òx0 òy0 òz0 çè ¶x + ¶y + ¶z ÷ø dx dy dz
1
(1)
Consider the 1st integral:
¶f1 x1 y1 z1 ¶f y1 z1 x = x1
òV ¶x dV = òx0 òy0 òz0 ¶x dx dy dz = òy0 òz0 f1 x= x0 dy dz
1

ì1ü ì f1 ( x1 , y, z ) ü ì- 1ü ì f1 ( x0 , y, z ) ü
y1 z1 ï ï ï ï y1 z1 ï ïï ï
= ò ò í0ý.í f 2 ( x1 , y, z )ý dy dz + ò ò í 0 ý.í f 2 ( x0 , y, z )ý dy dz
y0 z 0 y0 z 0
ï0ï ï f ( x , y, z ) ï ï 0 ï ï f ( x , y, z ) ï
î þî 3 1 þ î þî 3 0 þ
! ! (2)
= ò f1 cos( n , i ) dS
S
Similarly, the 2nd integral is:
¶f 2 x1 y1 z1 ¶f x1 z1 y = y1
òV ¶y dV = òx0 òy0 òz0 ¶y dx dy dz = òx0 òz0 f 2
2
y = y0
dx dz

ì0ü ì f1 ( x, y1 , z ) ü ì 0 ü ì f1 ( x, y0 , z ) ü
x1 z1 ï ï ï ï x1 z1 ï ïï ï
= ò ò í1ý.í f 2 ( x, y1 , z )ý dx dz + ò ò í- 1ý.í f 2 ( x, y0 , z )ý dx dz
x0 z 0 x0 z 0
ï0ï ï f ( x, y , z ) ï ï 0 ï ï f ( x, y , z ) ï
î þî 3 1 þ î þî 3 0 þ
! ! (3)
= ò f 2 cos( n , j ) dS
S
 And the 3rd integral is
¶f 3 ! !
òV ¶z dV = òS f3 cos(n, k ) dS (4)

(2), (3) & (4) in (1) gives

! æ ¶f1 ¶f 2 ¶f 3 ö
òV Ñ. f dV = Vò çè ¶x + ¶y + ¶z ÷ø dV
! ! ! ! ! !
= ò { f1 cos(n , i ) + f 2 cos(n , j ) + f 3 cos(n , k )} dS
S
!!
= ò f .n dS Proved
S
Divergence integral = flux integral
In hydrodynamics, the divergence theorem states the physical fact
that in order to ensure conservation of mass, the density within a
region of space (i.e. volume) can change only if matter flows into or
away from the region through its boundary.
 Stress at a point

!
DA ® 0 Þ DC ® 0

i: dir. of normal
j: direction of stress

s xx , s yy , s zz : Normal stresses t xy , t yz , t xz : Shear stresses

æ s xx t xy t xz ö
ç ÷
ç t yx s yy t yz ÷ : Stress Tensor sii , tij (i, j = x, y, z )
çt t zy s zz ÷ø ! ! ! !
è zx æ DF DF ö æ dF dF ö
(s nn , t ss ) = lim ç n , n ÷ = ç n , n ÷
DA®0 è DA DA ø è dA dA ø
tij = t ji (Symmetry of stress tensor)
 Stress at a point

Consider inviscid fluid motion. Newton’s law on prismatic

mass gives:
dx dy dz dx dy dz
- s yy dx dz + s nn ds dz sin a - g =r ay
2 2
g: specific weight; r: density; ay:acceleration component
dx Neglect product of 3 infinitesimals in comparison
Use sin a =
ds to product of 2 infinitesimals to get s yy = s nn
Similarly, snn = s xx = s yy \Stress at a point is a scalar quantity.
 Stress at a point
B

C
A

Consider viscous fluid motion. Shear stresses will be present.

In the infinitesimal tetrahedron, apply Newton’s law in the
direction normal to the inclined face (ABC) of the tetrahedron.
To do this, the direction cosines of normal to plane ABC, and
hence of snn, are (l, m, n). Thus areas of faces are:

OCB = (ABC) l; OAB = (ABC) m; OCA = (ABC) n (1)

 Stress at a point
B

C
A

OCB = (ABC) l; OAB = (ABC) m; OCA = (ABC) n (1)

Equilibrium in the direction of snn :
s nn ABC - s xxOCB l - t xyOCB m - t xz OCB n Neglect gravity
- s yyOAB m - t yz OAB n - t yxOAB l and inertia as being
one order too small
- s zz OCA n - t zxOCA l - t zy OCA m = 0
Newton’s law along snn:

s nn ABC - s xxOCB l - t xyOCB m - t xz OCB n

- s yyOAB m - t yz OAB n - t yxOAB l
Neglect gravity
- s zz OCA n - t zxOCA l - t zy OCA m = 0 and inertia as being
(2) one order too small
s xxOCB (force in x direction) l (Ðbetween ABC and x)
t xyOCB (force in y direction) m (Ðbetween ABC and y)

(1) in (2): s nn = s xxl 2 + s yy m 2 + s zz n 2 + 2(t xylm + t yz mn + t zx nl )

Given any plane, normal stress on it may be found from

stress tensor.
Newton’s law along snn:

s nn ABC - s xxOCB l - t xyOCB m - t xz OCB n

- s yyOAB m - t yz OAB n - t yxOAB l
Neglect gravity
- s zz OCA n - t zxOCA l - t zy OCA m = 0 and inertia as being
(2) one order too small
s xxOCB (force in x direction) l (Ðbetween ABC and x)
t xyOCB (force in y direction) m (Ðbetween ABC and y)

(1) in (2): s nn = s xxl 2 + s yy m 2 + s zz n 2 + 2(t xylm + t yz mn + t zx nl )

Given any plane, normal stress on it may be found from

stress tensor.
 Sum of orthogonal normal stresses is an invariant (i.e
constant for any orientation of xyz axes).
The average normal stress is called bulk stress:
1
s = (s xx + s yy + s zz ) : independent of direction, i.e scalar
3
For inviscid fluid, all normal stresses are equal and as a result
must equal bulk stress.

In fluid normal stress can only be negative.

Bulk stress magnitude = Thermodynamic pressure -s = p

Conservation of mass: Continuity Eqn.

Equation of momentum: Euler Eqns.

(or Newton’s 2nd law) (inviscid flow)

Navier Stokes Eqns.

(viscous flow)

Conservation of energy
(or 1st law of thermodynamics)
System: An identified quantity of matter

A system may change its shape, size, volume, surface area,

position, temperature but must always consist of same
matter.

By keeping the mass of a system constant, conservation of

mass is automatically ensured.

Newton’s 2nd law is essentially a statement on linear

momentum of an element of mass.
! D !
dF = (dm v )
Dt
First law of thermodynamics, a statement of conservation of
energy, stipulates that energy must at all times be conserved.
It accounts for energy leaving, entering or accumulating in a
system or control volume.
Stored energy: kinetic energy Þ motion of mass
potential energy Þ position of mass
internal energy Þ molecular and atomic
energy associated with
internal fields of the mass
Transition energy: heat Þ transition of one mass to other
due to difference in temperature
work Þ transition from/to one mass to
Q - W = DE = E t2 - E t1 other when external forces,
acting on a system, move
dE = dQ - dW through a distance
Q: heat added to, W: work done by, E: total stored energy
•Application of first law of thermodynamics may be assumed
to be satisfied if Newton’s second law is obeyed. Application
of both lead to the same result.

•In the absence of friction (i.e. inviscid fluid) and heat transfer
the second law of thermodynamics does not impose any
additional restriction. It is intrinsically satisfied.

•In the absence of heat transfer, but with friction (i.e.viscous fluid),
the second law of thermodynamics does not impose any
additional restriction as long as direction of energy transfer
(e.g. friction should retard the motion and not accelerate it)
in real systems is retained.

•In Marine Hydrodynamics, we are not usually concerned with

heat transfer problems, and hence the first and second laws
of thermodynamics may be assumed to be automatically
satisfied as long as conservation of mass and momentum
equations are satisfied.
Continuity Equation
S: Closed surface enclosing volume V
!
n : unit outward normal to S
!
n v!.n! : component of velocity
!
along the normal n
V S

Fluid enters V over some part of S and leaves V over

some other part of S.
\volume of fluid leaving through a small surface
!!
element dS in unit time is v .n d S [(m/s) ´ m 2 ´ 1s ]
Net volume rate at which fluid is leaving V must be zero
for incompressible fluid (density constant and flow is
continuous).
 !!
\ ò v .n dS = 0 For incompressible fluid, conservation of
S mass imply conservation of volume.
By divergence theorem:
!! ! ¶F j
òS F .n dS = Vò Ñ.F dV OR òS Fj n j dS = Vò ¶x j dV
!
ò dV = 0
V
Ñ.v

This must be true for all regions V, since V is arbitrarily

chosen. As a result:
! ¶u ¶v ¶w Continuity Equation
Ñ.v = + + =0
¶x ¶y ¶z (conservation of mass)
for incompressible fluid
 For compressible fluid, mass flow rate out of S equals rate
at which mass is decreasing within V:
!! ¶r
\ ò rv .n dS = - ò dV
S V
¶t

Following the same argument:

¶r !
+ Ñ.(rv ) = 0
¶t Continuity Equation
¶r ¶ (ru ) ¶ (rv) ¶ (rw) (conservation of mass)
or + + + =0 for compressible fluid
¶t ¶x ¶y ¶z
 Continuity equation for compressible flow

¶r !
+ Ñ.( r v ) = 0
¶t
¶r ! !
or, + v .Ñr + rÑ.v = 0
¶t
Dr !
or, + rÑ.v = 0
Dt
Therefore,
Dr ! ! Dr
= 0 Þ Ñ.v = 0 OR Ñ.v = 0 Þ =0
Dt Dt

In other words, continuity equation for incompressible flow is given by

! Dr
Either Ñ.v = 0 OR =0
Dt
 Dr
= 0 does not mean, however, that r = constant everywhere in the flow.
Dt
It means that density of fluid particles are conserved, that is, density of a fluid
particle does not change as it moves in the flow. There could always be
different fluid particles with different densities, say oil and water which do
not mix, in a flow.
!
¶v
= 0 implies “steady flow”.
¶t
In other words, if a flow is observed from a fixed position, all fluid particles
have same acceleration while crossing that fixed position.
However, velocity can change as a particle move to a different position. In
steady flow, convective accelerations can be present, as shown below.
! !
Dv ¶v ! ! ! !
= + (v .Ñ)v = (v .Ñ)v
Dt ¶t
! !
Dv ¶v
\ = 0 and = 0 are not equivalent.
Dt ¶t
 Conservation of mass using Lagrangian Approach

òòò r d a d b d c = òòò rd x d y d z
V0
0
V

d xd y d z = J d ad bd c òòò r d a d b d c = òòò r J d a d b d c
V0
0
V

¶x ¶x ¶x
¶a ¶b ¶c
¶ ( x, y, z ) ¶y ¶y ¶y
J= = r0 - r J = 0
¶ (a, b, c) ¶a ¶b ¶c
¶z ¶z ¶z
¶a ¶b ¶c
x = x(a, b, c, t ); y = y (a, b, c, t ); z = z (a, b, c, t )
dx dy dz
u= ,v = ,w =
dt dt dt
¶u ¶ æ dx ö d æ ¶x ö
= ç ÷= ç ÷
¶a ¶a è dt ø da è ¶a ø
¶x ¶x ¶x
¶a ¶b ¶c
¶ ( x, y, z ) ¶y ¶y ¶y
J= =
¶ (a, b, c) ¶a ¶b ¶c
¶z ¶z ¶z
¶a ¶b ¶c
¶u ¶u ¶u ¶x ¶x ¶x ¶x ¶x ¶x
¶a ¶b ¶c ¶a ¶b ¶c ¶a ¶b ¶c
dJ ¶y ¶y ¶y ¶v ¶v ¶v ¶y ¶y ¶y
= + + = J1 + J 2 + J 3
dt ¶a ¶b ¶c ¶a ¶b ¶c ¶a ¶b ¶c
¶z ¶z ¶z ¶z ¶z ¶z ¶w ¶w ¶w
¶a ¶b ¶c ¶a ¶b ¶c ¶a ¶b ¶c
 ¶u ¶u ¶u ¶u ¶x ¶u ¶y ¶u ¶z ¶y ¶z
+ +
¶a ¶b ¶c ¶x ¶a ¶y ¶a ¶z ¶a ¶a ¶a
¶y ¶y ¶y ¶u ¶x ¶u ¶y ¶u ¶z ¶y ¶z
J1 = = + +
¶a ¶b ¶c ¶x ¶b ¶y ¶b ¶z ¶b ¶b ¶c
¶z ¶z ¶z ¶u ¶x ¶u ¶y ¶u ¶z ¶y ¶z
+ +
¶a ¶b ¶c ¶x ¶c ¶y ¶c ¶z ¶c ¶c ¶c
¶u ¶x ¶y ¶z ¶u ¶y ¶y ¶z ¶u ¶z ¶y ¶z
¶x ¶a ¶a ¶a ¶y ¶a ¶a ¶a ¶z ¶a ¶a ¶a
¶u ¶x ¶y ¶z ¶u ¶y ¶y ¶z ¶u ¶z ¶y ¶z
= + +
¶x ¶b ¶b ¶c ¶y ¶b ¶b ¶c ¶z ¶b ¶b ¶c
¶u ¶x ¶y ¶z ¶u ¶y ¶y ¶z ¶u ¶z ¶y ¶z
¶x ¶c ¶c ¶c ¶y ¶c ¶c ¶c !""
¶z ¶c"#"""
¶c ¶$
c
!"""#"""$ =0
=0

¶x ¶y ¶z
¶a ¶a ¶a
¶u ¶x ¶y ¶z ¶u
= = J
¶x ¶b ¶b ¶c ¶x
¶x ¶y ¶z
¶c ¶c ¶c
 ¶u ¶v ¶w
J1 = J ; J2 = J ; J3 = J
¶x ¶y ¶z
dJ æ ¶u ¶v ¶w ö
\ = J1 + J 2 + J 3 = J ç + + ÷
dt è ¶x ¶y ¶z ø
Dr ! dr æ ¶u ¶v ¶w ö
+ rÑ.v = 0 Þ +rç + + ÷=0
Dt dt è ¶x ¶y ¶z ø
dr 1 dJ dr dJ d (r J )
Þ +r =0 Þ J +r =0 Þ =0
dt J dt dt dt dt
Integrating: r J = r 0
Euler Equations
1. Force exerted by surrounding fluid on surface element d S is
!
- pnd S
(-sign since normal is outward and force inward)
and hence net force on this fluid volume is
!
- ò pn dS = - ò Ñp dV
S V

By divergence theorem
2. Force on a fluid volume due to body force per unit mass X.
" ! ! !
X = Xi + Yj + Zk
Special case: X = Y = 0, Z = - g
! Body force is only
X = -Ñjb ; jb = gz due to gravity and
is in -z direction
! !
Dv
Force is: rX dV Inertia is: r dV
Dt
Force equilibrium (Newton’s 2nd law)
! ! Special case
Dv
r dV = -Ñp dV + rX dV
Dt
!
Dv Ñp ! p
or =- + X = - Ñ( + jb ) Euler Equation
Dt r r (conservation of momentum)
Euler Equation
(conservation of momentum of inviscid fluid)
!
Dv Ñp ! p
=- + X = - Ñ ( + jb )
Dt r r Special case
In expanded form:
¶u ¶u ¶u ¶u 1 ¶p 1 ¶p
+u +v +w = - +X =-
¶t ¶x ¶y ¶z r ¶x r ¶x
¶v ¶v ¶v ¶v 1 ¶p 1 ¶p
+u +v +w = - +Y = -
¶t ¶x ¶y ¶z r ¶y r ¶y
¶w ¶w ¶w ¶w 1 ¶p 1 ¶p
+u +v +w =- +Z =- -g
¶t ¶x ¶y ¶z r ¶z r ¶z
 Incompressible viscous flow: preliminaries

For a well-ordered (or laminar) flow, where particles move in

straight, parallel lines, Newton’s viscosity law,
for certain fluids called Newtonian fluids, is
¶v ¶v
tµ or t = µ
¶n ¶n
Shear stress (t) on the interface tangent to the flow direction
is proportional to distance-rate of change of velocity in the
direction normal to the interface. The proportionality constant
(µ) is “coefficient of viscosity” (dimension: Ft/L2=M/(Lt))

More simply put: shear stress is proportional to velocity gradient

 dx dx
dq dy dt du du
tµ = = = ; t=µ
dt dt dy dy dy
t

Viscous stresses in simple shear flow

Water at 20°C:
coefficient of viscosity: µ = 10 -3 N.s/m2 (1 centipoise)
(or absolute viscosity)
1 poise = 100 centipoise = 0.1 N.s/m2
µ (air) = 1.8´10 -5 N.s/m2
density: r = 1 gm/cm3; 1000 kg/m3
n (air) = 1.5´10 -5 m2/s
kinematic viscosity: n= µ / r = 0.01 cm2/s (0.01 Stoke); 10-6m2/s
 Separated flow past hydrofoil

Behavior of fluid with small µ is completely different from

fluid with µ = 0 due to boundary layer and its separation.
 Newton’s viscosity law to Stokes’ viscosity law

Stokes viscosity law gives relations between fluid stresses

and velocities, applicable to general laminar flows of
Newtonian fluids.

¶g xy
Shear stress µ rate of shear strain: t xy = µ Stokes’ hypothesis
¶t
Compare with t xy = Gg xy of solid mechanics G: Shear modulus
For general 3D flow:
¶g xy ¶g xz ¶g yz
t xy = µ ; t xz = µ ; t yz = µ
¶t ¶t ¶t

¶g xz ! !
= q 2 - q1
¶t
¶w ¶u
[w + dx] - w [u + dz ] - u
¶x ¶w ¶z ¶u
q! 1 = - =- q! 2 = =
dx ¶x dz ¶z
1 ! ! 1 æ ¶u ¶w ö
w y = (q1 + q 2 ) = ç - ÷ Average angular velocity in y
2 2 è ¶z ¶x ø
 Angular velocity component j of a fluid element
= average of angular velocities of two concurrent
orthogonal line segments in the element normal to direction j.
Similarly other components. Summarizing:
1 æ ¶w ¶v ö 1 æ ¶u ¶w ö 1 æ ¶v ¶u ö
wx = ç - ÷; w y = ç - ÷; wz = ç - ÷
2 è ¶y ¶z ø 2 è ¶z ¶x ø 2 è ¶x ¶y ø

¶g xz ! ! ¶u ¶w
\ = q 2 - q1 = +
¶t ¶z ¶x
¶g xz æ ¶u ¶w ö
\ t xz = µ = µç + ÷
¶t è ¶z ¶x ø
Finally, Stokes’ viscosity law for shear stress:
æ ¶u ¶v ö æ ¶v ¶w ö æ ¶u ¶w ö
t xy = µç + ÷; t yz = µç + ÷; t xz = µç + ÷
è ¶y ¶x ø è ¶z ¶y ø è ¶z ¶x ø

Stokes’ viscosity law for normal stress

æ ¶u 2 ! ö ¶u
s xx = µç 2 - Ñ.v ÷ + s = 2µ + s
è ¶x 3 ø ¶x
æ ¶v 2 ! ö ¶v
s yy = µç 2 - Ñ.v ÷ + s = 2µ + s
è ¶y 3 ø ¶y
æ ¶w 2 ! ö ¶w
s zz = µç 2 - Ñ.v ÷ + s = 2µ + s
è ¶z 3 ø ¶z
-------------------- ----------------
Compressible Incompressible
 Solid Mechanics Fluid Mechanics
G (shear modulus) µ (coefficient of viscosity)
displacement components velocity components

æ ¶x 2 ! ö æ ¶u 2 ! ö
s xx = Gç 2 - Ñ.S ÷ + s s xx = µç 2 - Ñ.v ÷ + s
è ¶x 3 ø è ¶x 3 ø
! ! ! ! ¶x ¶x
S = x i + h j + z k ; e xx = etc.; u = etc.
¶x ¶t
 Navier Stokes’ Equations for Incompressible Flow

Consider only stresses those lead to a y- force.

! D ! !
dF = (v dm) : Newton' s law in terms of momentum v dm
Dt !
! Dv
dF = dm (dm = rdV )
Dt
! ! ! ! ! !
¶ ¶ ¶ ¶
or df + Xdm = dmæç + u + v + w ö÷
v v v v (1)
è ¶t ¶x ¶x ¶x ø
! !
f : resultant surface force; X : Body force per unit mass
 From figure:
¶s yy ¶t zy ¶t xy
df y = dx dy dz + dz dx dy + dx dy dz
¶y ¶z ¶x
æ ¶s yy ¶t zy ¶t xy ö (2)
=ç + + ÷ dV
è ¶y ¶z ¶x ø
Dv æ ¶s yy ¶t yz ¶t xy ö
(2) in (1) for y - eqn.: r = rY + ç + + ÷
Dt è ¶y ¶z ¶x ø
Similarly x and z equations are
Du æ ¶s xx ¶t xz ¶t xy ö
r = rX + ç + + ÷
Dt è ¶x ¶z ¶y ø
Dw æ ¶s zz ¶t zy ¶t xz ö
r = rZ + ç + + ÷
Dt è ¶z ¶y ¶x ø
 Compact forms
! ! ¶tij
Dv Dvi
r = rX + Ñ.t; r = rX i +
Dt Dt ¶x j

To continue, substitute Stokes’s viscosity law relations

in (2):
ì ¶s æ ¶ 2u ¶ 2 v ¶ 2 w öü
df y = dV í + µÑ v + µç
2
+ 2+ ÷ý
î ¶y è ¶x ¶y ¶y ¶z ¶y øþ
æ ¶s 2 ö
=0
= ç + µÑ v ÷ dV Incompressible
è ¶y ø
(3)
 !
Dv 1 ¶s µ 2 1 ¶p
(3) in (1) for y - eqn.: =Y + + Ñ v =Y - + nÑ 2 v
Dt r ¶y r r ¶y
n: kinematic viscosity
Similarly x and z equations. Finally

Du 1 ¶p
=X- + nÑ 2 u
Dt r ¶x
Dv 1 ¶p N-S Equations
=Y - + nÑ 2 v
Dt r ¶y (Incompressible flow)
Dw 1 ¶p
=Z- + nÑ 2 w
Dt r ¶z
Usually: X = Y = 0; Z= -g

! !
Dv 1 2! Vector form of N-S Equations
= X - Ñp + n Ñ v
Dt r
 Bernoulli’s Equation (Steady flow of ideal fluid)

Momentum (Euler eqn.) is

! !
Dv ¶v ! ! Ñp ! p
= + (v .Ñ)v = - + X = - Ñ ( + jb )
Dt ¶t r r

It can be recast as
!
¶v ! ! p 1 !2
+ (Ñ ´ v ) ´ v = - Ñ( + v + jb )
¶t r 2

If the flow is steady: ! ! p 1 !2 (1)

(Ñ ´ v ) ´ v = - ÑH ; H = + v + jb
! =0
! ! ! ! r 2
v.(Ñ ´ v ) ´ v = - (v.Ñ) H \(v.Ñ) H = 0
H is constant along a streamline.
 1 !2
with jb = gz; p + rv + rgz = const along a streamline
2
1 !2
with jb = 0; p + rv = const along a streamline
2
!
From (1): if Ñ ´ v = 0; then ÑH = 0
H is independent of x, y, z and t. \ H is constant everywhere.
! æ ¶w ¶v ö ! æ ¶u ¶w ö ! æ ¶v ¶u ö ! ! ! !
Ñ ´ v = ç - ÷i + ç - ÷ j + ç - ÷k = 2(wx i + w y j + wz k )
è ¶y ¶z ø è ¶z ¶x ø è ¶x ¶y ø
! 1 !
w = (Ñ ´ v ) Angular velocity vector
2
! !
Ñ ´ v = 0 Þ w = 0 Irrotational flow
 1 !2
\ p + rv + rgz = const everywhere for irrotational flow
2

• 3 momentum equations (partial differential equations) reduces

to a single equation (algebraic relation) along a streamline.

• Necessary requirements of Bernoulli’s equation are

# stationary flow
# inviscid (with or without body forces)
# incompressible
# valid along a streamline (rotational) OR everywhere (irrotational)
Flow through a converging-diverging nozzle

A1, p1,V1 A2 , p2 ,V2

Assumption: We assume that the flow is quasi 1

dimensional, therefore: p = p ( x ) , V = V ( x ) , etc.

Continuity r1V1 A1 = r2V2 A2 ( r1 = r2 ) Þ V1 A1 = V2 A2

1 1
Bernoulli p1 ( x ) + rV12 ( x ) = p2 ( x ) + rV22 ( x )
2 2

2 ( p2 - p1 )
Eliminate V2 Þ V12 =
æ æ A ö2 ö
r ç1 - ç 1 ÷ ÷
ç è A2 ø ÷
è ø
The pressure coefficient for incompressible flow is
p - p¥ p - p¥
C p := =
q¥ 1
r¥V¥2
2
2
1 1 p - p¥ æV ö
Bernoulli: 2
p + rV 2
= p¥ +
2
r¥V¥2 Þ Cp =
1
= 1- ç ÷
r¥V¥2 è V¥ ø
2

2
æV ö
Cp = 1- ç ÷
è V¥ ø
Vorticity (ideal fluid)
!
¶v ! ! p 1 !2
+ (Ñ ´ v ) ´ v = - ÑH ; H = + v + jb Euler Equation
¶t r 2
! ! !
¶v ! !
+ W ´ v = - ÑH (W = 2w is the vorticity vector )
¶t
! ! !
¶v
Ñ ´ + Ñ ´ (W ´ v ) = - Ñ ´ (ÑH ) = 0 take curl, curl grad = 0
¶t
! ! !
¶ (Ñ ´ v )
+ Ñ ´ (W ´ v ) = 0
¶t
!
¶W ! !
+ Ñ ´ (W ´ v ) = 0
¶t = 0: Continuity eqn
!
¶W ! ! ! ! ! ! ! ! div curl = 0
+ (v .Ñ)W - (W.Ñ)v + WÑ.v - v Ñ.W = 0
¶t
! !
¶W ! ! ! ! DW ! !
+ (v .Ñ)W = (W.Ñ)v or = (W.Ñ)v Vorticity eqn.
¶t Dt
 !
DW ! !
= (W.Ñ)v
Dt
! !
For 2D flow : v = [u ( x, y, t ), v( x, y, t ), 0)]; W = (0, 0, W z )
! ! !
¶v
\ (W.Ñ)v = W z =0
¶z
!
DW
\ = 0 Vorticity of individual elements
Dt conserved
! !
For 2D steady flow : (v .Ñ)W = 0 Vorticity is constant along
a streamline
Vorticity (viscous fluid)
!
¶v ! ! ! 1 2!
+ (v .Ñ)v = X - Ñp + nÑ v N-S Equation
¶t r
Take curl of this equation and a similar procedure as that
for ideal fluid will yield:
!
DW ! ! !
= (W.Ñ)v + nÑ W 2
Vorticity (transport) eqn.
Dt

Convection Diffusion
! !
DW ¶W ! ! !
In 2D: = + (v .Ñ)v = nÑ W only diffusion
2

Dt ¶t
 Vortex Line, Vortex Tube

•Draw a line in the flow joining points where W is constant.

This is a vortex line.

At a particular time:
•A vortex line is everywhere tangent to W (vorticity vector).

•A vortex tube (or filament) is a bundle of vortex lines.

A vortex line x = x(s), y = y(s), z = z(s) is obtained from the

solution of
dx / ds dy / ds dz / ds
= =
Wx Wy Wz

at a particular time.
Implications of vorticity equations

!
DW ! ! Ideal fluid Þ 1. Vorticity is ‘convected’ with the fluid
= (W.Ñ)v
Dt and it does not diffuse.
2. Fluid particles containing vorticity
will always contain it.
3. Conversely, fluid particles with zero
vorticity will always have zero
vorticity.
4. If vorticity is zero in a flow at one
time instant, vorticity will remain
zero at all future times.
Vortices are ‘persistent’ flow structures. Only can be
‘diffused’ in ‘real’ fluid and it ‘takes time’.
 Circulation

‘Draw’ an imaginary loop in a fluid medium and integrate

the velocity vector around the loop. What we get is called
‘circulation’.
! !
G = ò v .dl
C

In other words, circulation is the line integral about a

closed path of the tangential velocity component
along the path at time t .
 Stokes theorem

Consider circulation around an infinitesimal rectangular path

whose plane is parallel to xy plane.
Line A-B: ds º dx contribution to circulation: u dx

Line B-C: ds º dy contribution to circulation: [v+(¶v/ ¶x)dx] dy

Line C-D: ds º (-)dx contribution to circulation: - [u+(¶u/ ¶y)dy] dx

Line D-A: ds º (-)dy contribution to circulation: - v dy

 Summing up:
! ! æ ¶v ¶u ö !
dG = ò v .dl = ç - ÷ dx dy = 2w z dx dy = (Ñ ´ v ) z dS (1)
ABCD è ¶x ¶y ø

Break up the finite plane area S into infinitesimal area elements

and apply the above equation to each area element in same
sense (clock or anti-clockwise). Line integration will cancel
everywhere except at the outer boundary C . Then

! ! !
G = ò v .dl = òò (Ñ ´ v ) z dS (2)
C S
-------------------------- Stokes theorem in 2D
To generalize, consider a hypothetical curved surface
bounded by C, a noncoplanar curve. Then (1) becomes

! ! !
dG = (Ñ ´ v ) n dS = (Ñ ´ v ) dS
and (2) becomes
! ! ! ! ! !
G = ò v .dl = òò (Ñ ´ v ) dS = òò W dS
C S S

|--------------------------| Stokes theorem in 3D

Velocity field could be replaced by any other field in Stokes

theorem.
! ! ! ! ! !
C
ò F .dl = òò (Ñ ´ F ) dS = òò (Ñ ´ F ).n dS
S S
 Kelvin’s circulation theorem
“For inviscid fluid, circulation around a path which moves
so as always to touch the same particles of flow must be
constant.” DG
In other words: =0
Dt

! !
DG D ! ! Dv ! ! D ( dl )
=
Dt Dt C ( t ) ò v .dl = ò
C (t )
Dt
. dl + ò v .
C (t )
Dt
=0
! ! ! !
Dv 1 1
òC Dt . d l = -
rC ò Ñ p .d l - ò
C
Ñ j b .d l = -
rCò dp - ò djb = 0 Euler Eqn.
C
!
D ( dl ) ! Here is how.
= dv
Dt
!
! D ( dl ) ! !
òC Dt Cò = = òC + + = òC + + )=0
2 2 2
v . v .d v (u du v dv w dw ) d (u v w
 Helmholtz’s vorticity theorems

! !
Net flux of vorticity through a closed surface is zero: òò W.n dS = 0
S

Hence, flux into face 1 = flux out of face 2 (of vortex tube)
! ! ! ! Stokes' theorem

òò W.e dS = òò W.e dS Þ G1 = G2
S1 S2
•Strength of a vortex tube (G) is constant along its length.
Helmholtz’s 1st theorem
A vortex tube or line cannot begin or end in a fluid, it must form a loop
(i.e. close on itself) or extend to infinity or begin or end on a fluid
boundary (a rotating solid surface).

•In the absence of body force, pressure and viscous torques,

a vortex tube or line entail the same fluid or line. Thus vorticity
is convected with fluid and does not diffuse.
Helmholtz’s 2nd theorem
Vortex tube is a material tube, vortex line is material line.

•Average vorticity in vortex tube is inversely proportional

to its cross sectional area. For incompressible flow, it is
proportional to length of the tube.
Helmholtz’s 3rd theorem
Density of a finite set of vortex lines is proportional to the local
vorticity.
Common nondimensional groups in Marine Hydrodynamics

Consider an infinitesimal fluid element of size (dx, dy, dz).

æ ¶v ¶v ö 3æU Uö
(Inertia force) FI = (rdx dy dz )ç + v ÷ » rL ç + U . ÷
è ¶t ¶x ø èt Lø
rL3U
= + rU 2 L2 » rU 2 L2 ( L3 << L2 )
t
(Gravity force) Fg = rg dx dy dz » rgL3

(Pressure force) Fp = p dx dy = p dS » pL2

¶v
(Shear or viscous force) Fs = µ dx dy » µUL
¶z
(Surface tension force) Fs = sdx = sL
Groups are:
FI U 2 æ U ö
» = Fr2 çç Fr = ÷÷ Froude number
Fg gL è gL ø

FI rUL UL Reynolds number

» = = Re
Fs µ n
Cavitation number
Fp p æ p¥ - pv p¥ ö
» º d or Eu çç d = 1 ; Eu = 1 ÷
2 ÷
FI rU 2
è 2 rU
2
2 rU ø
Euler number
FI rU 2 L Weber number
» =W
Fs s
 æ ¶v ¶v ö 3æU Uö
(Inertia force) FI = (rdx dy dz )ç + v ÷ » rL ç + U . ÷
è ¶t ¶x ø èt Lø
rL3U
= + rU 2 L2 = FIE + FIC
t

Eulerian inertia Convective inertia

FIE L fL
» = = St ( f : frequency of vortex shedding, Hz)
FIC Ut U
Strouhal number
Incompressible Potential (Inviscid & Irrotational) Flow

Inviscid: µ = 0 (Ideal fluid)

Incompressible: r is conserved
!
Irrotational: w ! (no angular velocity)
= 0 = Ñ´v
!
Ñ.v = 0 (1): continuity equation
!
Ñ´v = 0 (2): irrotationality condition
! é ¶f ¶f ¶f ù
v = Ñf êu = ¶x ; v = ¶y ; w = ¶z ú (3): assume a velocity
ë û potential f
(3) automatically implies (2) [! Ñ ´ Ñf = 0 ]. So, existence of a
velocity potential implies irrotational flow and vice-versa.
 (3) in (1):
Ñ.(Ñf) = Ñ 2 f = 0
Potential Equation or
¶ f ¶ f ¶ f
2 2 2
Laplace Equation
or 2 + 2 + 2 = 0
¶x ¶y ¶z

f = f( x, y, z, t ) Potential can be steady or unsteady

• All inviscid, incompressible and irrotational flows can be described by a
potential which satisfies the potential (or Laplace) equation.
• Converse is true. All solutions of the Laplace equation generate valid
inviscid, incompressible and irrotational flows.
• The Laplace equation is linear and hence the principle of superposition
may be used. If f1 and f2 are solutions of the Laplace equation,
f = c1 f1+c2 f2 is also a valid solution.
• Complex flow paterns can be generated by a suitable combination of
elementary flows.
 Laplace equation in cylindrical coordinates (r, q, z)

1 ¶ æ ¶f ö 1 ¶ 2
f ¶ 2
f
Ñ 2f = ç r +
÷ 2 2 + =0
r ¶r è ¶r ø r ¶q ¶z 2

Laplace equation in spherical coordinates (r, q, j)

1 ì¶ æ 2 ¶f ö ¶ æ ¶f ö ¶ æ 1 ¶f öü
Ñ f= 2
2
í ç r sin q ÷ + ç sin q ÷ + ç ÷ý = 0
r sin q î ¶r è ¶r ø ¶q è ¶q ø ¶j è sin q ¶j øþ
Stream function for 2D incompressible, potential flow
¶y ¶y (1)
Stream function: y ( x, y , t ) Þ u = , v=-
¶y ¶x

y as in (1) satisfies continuity equation:

! ¶u ¶v ¶ æ ¶y ö ¶ æ ¶y ö
Ñ.v = + = ç ÷ + ç - ÷ = 0
¶x ¶y ¶x è ¶y ø ¶y è ¶x ø

For y to satisfy irrotationality condition:

¶v ¶u ¶ æ ¶y ö ¶ æ ¶y ö æ ¶ 2y ¶ 2y ö
wz = - = ç - ÷ - ç ÷ = -ç + ÷ = -Ñ 2
y=0
¶x ¶y ¶x è ¶x ø ¶y è ¶y ø è ¶x
2
¶y ø
2

or ¶ 2
y ¶ 2
y
Ñ y= 2 + 2 =0
2
y satisfies Laplace equation
¶x ¶y
Recall that
Streamline: At a given time, the line everywhere tangent to
velocity vector, v, i.e. has same direction as v.
x = x( s ); y = y ( s ); z = z ( s ) is obtained by solution
dx / ds dy / ds dz / ds
of = = at given t.
u v w

In 2D: dx = dy Þ u dy - v dx = 0 (2)
u v
Now, ¶y ¶y
dy = dx + dy = -v dx + u dy = 0 by (1) & (2)
¶x ¶y

Hence, y is constant along a streamline. Hence, the name

‘stream function’.
 ¶f ¶y ¶f ¶y (1)
From f and y relations Þ = ; =-
¶x ¶y ¶y ¶x
Cauchy-Riemann equations of
complex variable theory

¶f ¶f
f known: y = ò dy + f ( x) = - ò dx + g ( y )
¶x ¶y
Comparing these two forms, it is usually easy to get y by inspection.
Orthogonality of f and y
¶y ¶y æ dy ö ¶y / ¶x
dy = dx + dy = 0 Þ ç ÷ =-
¶x ¶y è dx ø y =C1 ¶y / ¶y
¶f ¶f æ dy ö ¶f / ¶x ¶y / ¶y
df = dx + dy = 0 Þ ç ÷ =- = by (1)
¶x ¶y è dx ø f= K1 ¶f / ¶y ¶y / ¶x
æ dy ö ´ æ dy ö = -1
ç ÷ ç ÷ where C1 and K1 meet
è dx ø y =C1 è dx ø f= K1

Hence proved.
 Bernoulli’s Equation for potential flow: pressure

Momentum (Euler eqn.) is

! !
Dv ¶v ! ! Ñp ! p
= + (v .Ñ)v = - + X = - Ñ ( + jb )
Dt ¶t r r

It can be recast as (take jb= gz)

!
¶v ! ! p 1 !2 (irrotational)
+ (Ñ ´ v ) ´ v = - Ñ( + v + gz ) (1)
¶t r 2
!
Noting: ¶ v ¶ (Ñf) ¶f
= =Ñ
¶t ¶t ¶t
æ ¶f p 1 ! ö
(1) becomes: Ñç + + v 2 + gz ÷ = 0
è ¶t r 2 ø
 ! ! ! !
Integrating after taking dot product with dr (= i dx + j dy + k dz )

æ ¶f 1 ! 2 ö
p = -rç + v + gz ÷ + C (t ) Bernoulli’s equation
è ¶t 2 ø
Where C(t) is an arbitrary function of time.
There are several ways of writing the velocity squared term:
2 2 2
!2 ! ! 2 æ ¶f ö æ ¶f ö æ ¶f ö
v = v .v = u + v + w = Ñf.Ñf = Ñf = ç ÷ + ç ÷ + ç ÷
2 2 2

è ¶x ø è ¶y ø è ¶z ø
 Boundary conditions for potential flow problems
1. On a fixed body boundary SB, which is impermeable (also called
impervious), there should be no flux across the boundary.

¶f ¶y
=0=
¶n S B ¶s S B

Equivalent mathematical forms are:

!! ! ¶f
v .n = Ñf.n = = 0 on S B
¶n
¶ ¶ ¶ ¶ !
where = nx + n y + nz and n = [nx n y nz ]T
¶n ¶x ¶y ¶z
2. When the body boundary SB itself is under motion, with a velocity
!
U B , then the no flux condition becomes:
Kinematic BC
!! ! ¶f ! !
v .n = Ñf.n = = U n = U B .n on S B
¶n
3. The incident flow in x direction at large distances from the
body is simply
¶f ¶y
= U¥ =
¶x y ®¥
x ® ¥ ¶y x ®¥
y ®¥
4. Pressure is prescribed on a boundary. Most important
example of this boundary is free surface .
Dynamic BC
æ ¶f 1 2 ö
p = -rç + Ñf + gz ÷ + C (t ) known on S F
è ¶t 2 ø
nonlinear
 SF

Ñ 2f = 0 SB

æ ¶f 1 2 ö
- rç + Ñf + gz ÷ + C (t ) prescribed
¶f è ¶t 2 ø
prescribed
¶n
 Navier-Stokes Equations
Dynamic Similarity
! !
Dv 1 2!
= X - Ñp + n Ñ v Vector form (1)
Dt r
Expresses equilibrium between body force, surface
forces (pressure and shear) and inertia force.

Body forces may be ignored if there is no free surface

and there is no inhomogeneous distribution of density,
i.e motion of homogeneous fluid without free surface.
Pressure should be interpreted as ‘over at rest pressure’

For steady flow without body force, the N-S equation is

! ! 1 2!
(v .Ñ)v = - Ñp + nÑ v (2)
r
Nondimensionalize the equation with reference values:
! !
v ¢ = v / U ; ( x¢, y¢, z¢) = ( x, y, z ) / l; p¢ = p / P (3)
! ! P n 2!
¢ ¢
(3) in (2): (v .Ñ).v = - Ñp +¢ Ñ v¢
rU 2
UL
! ! P 1 2! (4)
or, (v ¢.Ñ).v ¢ = - Ñp ¢ + Ñ v ¢
rU 2
Re

Clearly, Reynolds’ number becomes the parameter for

dynamic similarity.
 Boundary Conditions

“No slip” condition on solid, impervious boundaries:

vn = vt = 0 on solid walls
vn: normal component of velocity
vt: tangential component of velocity

In potential flow, the tangential velocity can exist, i.e.

fluid can slide on the wall. Only normal velocity is zero
on a solid wall.
Boundary Layer Equations
Consider 2D flow with small viscosity about a cylinder
of slender cross-section.

Except in immediate neighborhood of the cylinder surface,

the velocities are of the order of free-stream velocity V¥.
As a result, streamlines deviate only a little from those in
potential (frictionless) flow.
Fluid does not ‘slide’ over the
wall (as in potential flow) and ‘no slip’ condition is
satisfied by the fluid by adhering to the wall. The transition
of zero tangential velocity at the wall to ‘full’ value at some
distance from the wall occurs in a very thin layer, called
‘boundary layer’.

In boundary layer, velocity gradient normal to the wall,

¶u/¶y, is very large and though viscosity µ is small, the
shear stress, t = µ ¶u/¶y, may assume large values.

In region beyond boundary layer, velocity gradients are

not large and small viscosity does not cause any
appreciable shear stress and hence the flow is
essentially frictionless (potential).
 Boundary layer thickness

! ! 1 2!
2D
¶u ¶ æ ¶u ö
(v .Ñ)v = - Ñp + nÑ v Þ ru » µ ç ÷
r ¶x ¶y è ¶y ø
Friction comparable to inertia
rU 2 U nl d 1
»µ 2 Þd» OR d µ n and µ
l d U l Re
BL thickness negligibly small w.r.t the linear dimension
of the body.
d << l
Definition of BL thickness

It is somewhat arbitrary due to ‘asymptotic’ nature of

transition of velocity from BL to the flow ‘outside’.
For the purpose of definition, however, BL thickness is
defined as the distance from the wall where the velocity
differs by 1% from the external velocity.
The ‘displacement thickness’, d1, another measure of
BL thickness is defined as

¥
Ud1 = ò (U - u ) dy
y =0

It indicates the distance by which the

external streamlines are shifted away
from the wall due to the presence of BL.
Nondimensional Equations: Order of magnitude analysis
¶u ¶u ¶u ¶p 1 æ ¶ 2u ¶ 2u ö x-eqn.
+u +v = - + ç 2 + 2 ÷
¶t ¶x ¶y ¶x Re è ¶x ¶y ø
1 1
1 1 1 d d2 1
d d2
¶v ¶v ¶v ¶p 1 æ ¶ 2 v ¶ 2 v ö
+u +v = - + ç 2 + 2 ÷ y-eqn.
¶t ¶x ¶y ¶y Re è ¶x ¶y ø
1
d 1 d d1 d 2
d
d
¶u ¶v
+ = 0 continuity eqn. d << 1 nondim. bl. thk.
¶x ¶y
1 1 boundary
u = v = 0 for y = 0; u = U ¥ for y ® ¥ conditions
1. u nondimensionlized by V¥ (u ~ 1) and x nondimensionlized
by a characteristic length (L) of the body, chosen such that
¶u/ ¶x ~1. Hence ¶2u/ ¶x2 ~1.

2. !¶u/ ¶x ~1; from continuity equation, !¶v/ ¶y ~1.

3. At wall, v = 0 and in the BL, y ~ d. !¶v/ ¶y ~1, \v ~ d.

4. ! v ~ d, \ ¶v/ ¶x ~ d & ¶2v/ ¶x2 ~ d.

5. Assume nonsteady acceleration ¶u/ ¶t is of same order as

convective acceleration u ¶u/ ¶x. This excludes very
sudden accelerations as occur in large pressure waves.
\ ¶u/ ¶t ~1.
6. !¶v/ ¶y ~1 & y ~ d; \ ¶2v/ ¶y2 ~1/ d.

7. !u (parallel to wall, ~1) increases from 0 at wall to 1

(free stream velocity) across the BL thickness (y ~ d),
one has ¶u/ ¶y ~1/ d & ¶2u/ ¶y2 ~1/ d2.

8. Now, with all the orders of magnitude in x- equation,

it follows that for viscous forces to be of same order as
inertia forces, one requires 1/Re ~ d2.

9. In x- equation, ¶2u/ ¶x2 can now be neglected in

comparison to ¶2u/ ¶y2 .
10. From y-equation, it can be concluded that ¶p/¶y ~ d.
On integration, it will give a pressure increase across
BL of the order d2. Thus, pressure normal to BL is
practically constant and may be assumed to equal that
at the outer edge of BL imposed by frictionless flow.

11. At outer edge of BL, u becomes U(x,t), a known

(prescribed) function. Viscous terms vanish and the
flow is governed by
¶U ¶U 1 ¶p ¶U 1 ¶p
+U =- (unsteady) U =- (steady)
¶t ¶x r ¶x ¶x r ¶x
The steady flow gives Bernoulli’s equation on integration:
1
p + rU 2 = constant
2
Finally, the y- equation may be deleted as of no consequence,
the x-equation is simplified by dropping ¶2u/ ¶x2 term and
continuity equation is retained.

Writing in dimensional form,

¶u ¶u ¶u 1 ¶p ¶ 2u
+u +v = - +n 2 Prandtl’s BL
¶t ¶x ¶y r ¶x ¶y equations
¶u ¶v (unsteady)
+ =0
¶x ¶y
BCs : y = 0 : u = v = 0; y ® ¥ : u = U ( x, t )

In addition, a velocity profile within BL has to be prescribed

over whole xy region at t = 0 and the problem is that of
evolution of BL over time.
 ¶u ¶u 1 ¶p ¶ 2u
u +v = - +n 2 Prandtl’s BL
¶x ¶y r ¶x ¶y
equations
¶u ¶v (steady)
+ =0
¶x ¶y
BCs : y = 0 : u = v = 0; y ® ¥ : u = U ( x)

In addition, a velocity profile within BL at the initial section,

say x = x0 has to be prescribed by a function, say, u(x0, y)
and the problem is that of further change of velocity profile with x.
 Boundary layer separation

Under ‘adverse pressure gradient’ at some region of the wall,

fluid particles ‘retards’ at some stage since their small kinetic
energy does not allow them to penetrate region of high
pressure. As a result, BL deflects sideways from the wall and
then separates from it and moves into the main stream.
 In frictionless region:
D to E: 1. Pressure decreases
2. Fluid accelerates
3. Favourable pressure
gradient (dp/dx < 0)
4. Pressure ® KE

E to F: 1. Pressure increases
2. Fluid decelerates
3. Adverse pressure
gradient (dp/dx > 0)
4. KE ® Pressure

Particles arrive at F with same velocity as in D.

Particles very close to the wall in BL is subjected to ‘pressure’ of the flow
‘outside’. Due to large friction forces in BL, particles loses large amount of KE
(i.e. slows down) in D to E path and as a result, the remainder KE is not sufficient
to overcome the ‘pressure hill’ ( i.e. increasing pressure) in E to F path. Hence,
such particles cannot move fully from E to F path and somewhere in between,
their motion gets arrested and they starts to move in the opposite direction. This
reverse motion results in vortex.
Reverse motion just started Reverse motion penetrates fwd

Reverse motion to vortex Vortex enlarges in size

 A clockwise vortex about to separate
from cylinder before moving downstream.

This completely changes pressure

distribution in the wake. The eddying
region behind the cylinder produces
suction, causing large pressure drag
on the cylinder.

Idealized Karman Vortex Street: Diagrammatic

At larger distance into the wake, a regular
pattern of vortices is found, and they move
alternately clockwise and anticlockwise. This is “Karman Vortex Street”.
The vorices are unstable in general w.r.t small disturbances parallel
to themselves. Neutral equilibrium is achieved at h/l = 0.281.
Vortex street moves at a forward velocity (u) which is lower than
free stream velocity (U).

A regular Karman vortex street is observable in the range

60 < Re < 5000.

To keep up this vortex street, a supply of KE is required which

manifests itself as a drag force (pressure or form drag) on the
cylinder.

The frequency at which vortices are shed (f in Hz) is such that

it always keeps a dimensionless frequency (Strouhal Number)
constant:
fD D: cylinder diameter
St =
U U: Free-stream velocity
and Strouhal Number is a function of Reynolds number).
 UD
Vortex shedding depend upon Re: Re =
n
Re Regimes

Range Flow Description

Re £ 5 No flow separation.
Flow separates from back of
5 £ Re £ 45 cylinder & a symmetric vortex pair
forms in near wake.

Longl. spacing between vortices

increases linearly with Re and
becomes about 3D at Re » 45.

Wake instability sets in.

45 £ Re £ 150
One of the vortices breaks away.

Laminar wake of staggered

vortices of opposite sign is
formed, i.e. vortex street is
laminar.

Vortices breaking away from the

150 £ Re £ 300 wake become turbulent but BL
remains laminar.
Subcritical range. Laminar BL
300 £ Re £ 1.5 ´ 105 separates at 80° aft of LE of
cylinder.

Vortex shedding is strong and

periodic.
Transitional range. BL becomes
1.5 ´ 105 £ Re £ 3.5 ´ 10 6 turbulent.
Separation point moves to 140°.

Drag coefficient shows sharp

drop (to 0.3).
 Re ³ 3.5 ´ 10 6 Supercrtical range. Turbulent BL.

Regular shedding re-establishes.

Shedding persists at very high

Re, say Re » 1011.
• Longl. To lateral spacing of an ideal staggered vortex street:
h
= 0.281 .
l
h
• = 0.18 for Re = 100 to 500.
l
l
• nearly constant for most vortex streets.
D
• Typically, 4.8 < l / D < 5.2
• Downstream in the wake, vorticity decreases and staggered
pattern expands laterally.
•BL without separation (attached flow) Þ Viscous drag
(Streamlined body: fish or hydrofoil or ship)
More sensitive to Re.
Related to surface area exposed to the flow.
When the pressure losses are small and the overall drag is
mainly due to viscous drag, the body is streamlined.

•BL with separation (separated flow) Þ Viscous drag

+
Pressure (or Form) drag
(Bluff body: circular cylinder, sphere)
Less sensitive to Re.
Related to cross-sectional area and shape of the body.
When the viscous drag is small and the total drag is
dominated by pressure losses, the body is bluff.

•Whether the body is streamlined or bluff depends entirely

on the shape of the body.
Boundary layer separation (contd.)
At y = 0 : Þ u = v = 0 :
¶u ¶u 1 ¶p ¶ 2u ¶ 2u dp
u +v = - +n 2 Þµ 2 =
¶x ¶y r ¶x ¶y ¶y y =0 dx
¶ 3u
\ 3 =0
¶y y =0 Near wall, curvature of velocity profile depends
on pressure gradient

Adverse pressure gradient: pressure increasing in the direction

dp
of flow, i.e. >0
dx
Flow with decreasing pressure (accelerated flow): dp < 0
dx

Flow with increasing pressure (decelerated flow): dp

>0
dx
dp ¶ 2u ¶ 2u ¶u
<0Þ 2 <0Þ 2 <0Þ >0
dx ¶y wall ¶y BL thk. ¶y
dp ¶ 2u ¶ 2u ¶ 2u
>0Þ 2 >0 & <0Þ 2 = 0 (PI)
dx ¶y wall ¶y large dist.
2
¶y some y

Point of separation: ¶u
=0
¶y y =0
 dp
<0
dx

dp
>0
dx

¶u
>0 =0 <0
¶y y =0
 Skin Friction

Shear stress at wall: ¶u

t0 = µ
¶y y =0

Viscous(or skin friction) drag in 2D flow:

l l
¶u
D f = b ò t0 cos f ds = bµ ò dx
s =0 x =0
¶y y =0
Hence, wall velocity gradient is required to calculate
frictional drag.
Boundary layer along a flat plate at zero incidence

•Steady flow
•Zero pressure gradient
dp
=0
dx
¶u ¶u ¶ 2u 1. Seek similarity solutions as there is
u +v =n 2 no preferred length. In BL:
¶x ¶y ¶y
y
¶u ¶v h= [u ( y ) is function of a single
+ =0 g ( x)
¶x ¶y
variable h] (1)
BCs : y = 0 : u = v = 0
u
y = ¥ : u = U¥ = h(h) (2)
U¥
In other words, velocity profiles u(y) for varying x are made
identical by scale factor U¥ for u and g(x) for y.
2. Continuity equation is satisfied by taking stream function y(x, y):

¶y ¶y
u= , v=- (3a,b)
¶y ¶x
3. Integrate (3a):
¥ h

y= ò u( y) dy + c( x) = U
y =0
¥ g ( x) ò h( s) ds + c( x)
0
(4)

4. Since, the plate itself is a streamline in this uniform frictionless flow,

y = 0 at y = h = 0 Þ c( x) = 0 (5)

5. Hence, from (4):

y = U ¥ g ( x) f (h) with f (0) = 0 (6)

 ¶y ¶y ¶h æ ¶f öæ 1 ö
6. (6) in (3a) gives: u = = = çU ¥ g ÷ç ÷ = U ¥ f ¢
¶y ¶h ¶y è ¶h øè g ø
or u = U ¥ f ¢ (7)
where we have used (1) and (6). Compare (7) and (2) to get h = f ¢.
7. From (3b) we get (making use of (1) & (6)):

¶y æ ¶g ¶f ¶h ö æ ¶g ¶f ¶h ¶g ö
v=- = -U ¥ ç f +g ÷ = -U ¥ç f + g ÷
¶x è ¶x ¶h ¶x ø è ¶x ¶h ¶g ¶x ø
ì æ y ö ü æ y ö
= -U ¥ í g ¢f + gf ¢ç - 2 ÷ g ¢ý = -U ¥ ç g ¢f - f ¢ g ¢ ÷ = U ¥ g ¢(hf ¢ - f )
î è g ø þ è g ø
or v = U ¥ g ¢(hf ¢ - f ) (8)
u = U¥ f ¢
v = U ¥ g ¢(hf ¢ - f )
¶u y
= -U ¥ f ¢¢ 2 g ¢
¶x g
¶u f ¢¢ (9)
= U¥
¶y g
¶ 2u U ¥
= 2 f ¢¢¢
¶y 2
g
8. (9) in N-S equations gives, ¶u ¶u ¶ 2u
after some algebra, u +v =n 2
¶x ¶y ¶y

U ¥ gg ¢
f ¢¢¢ + ff ¢¢ = 0
n
Since we are seeking similarity solutions, we want an ode of f
as a function of h. Hence set
n
gg ¢ = constant = (convenient choice)
2U
g2 nx
so that the solution is = + k (k is a constant)
2 2U ¥
and the ode becomes 2 f ¢¢¢ + ff ¢¢ = 0 (10)
9. On y = 0, velocity changes from U¥ at x < 0 to 0 in x > 0, clearly
a singular point where one should have

¶u æ U ¥ f ¢¢ ö
ç= ÷®¥ at x ® 0
¶y è g ø
nx
and this gives k = 0 and hence g ( x) = (11)
U¥
10. (11) in (1) and (6) gives

U¥
h= y y = nxU ¥ f (h) (12)
nx
11. BCs : y = 0 : u = v = 0 Þ h = 0: f = f ¢ = 0
y = ¥ : u = U¥ h ® ¥ : f ¢ =1
where we have used (7) and (8).
 2 f ¢¢¢ + ff ¢¢ = 0 [ f = f (h)] ode
U¥
h= y y = f nxU ¥
nx
h = 0: f = f ¢ = 0
B.C
h ® ¥ : f ¢ =1

Transformation of pde (N-S equation) to ode for stream function.

The ode is a 3rd order nonlinear equation. Three B.Cs sufficient
to solve this problem.
 From table:
u U¥ nx
(= f ¢) @ 0.99 Þ h = 5 = d Þd=5
U¥ nx U¥
nx d 5
d=5 =
U¥ l Rl
BL thickness Nondimensional BL thickness
¶u U¥ U¥
t0 ( x) = µ = µU ¥ f ¢¢(0) = 0.332µU ¥
¶y y =0 nx nx
t0 ( x) n 0.664
cf = 1 = 0.664 = Skin friction coefficient
2 rU ¥
2
U¥ x Rx
l l
U¥ dx U¥
D = b ò t0 ( x) dx = 0.332 bµU ¥ ò =0.332 bµU ¥ ´ (-2 l )
0
n x =0 x n
= 0.664b U ¥3 µrl Skin friction (or friction drag)
 2D 1.328
cf = = ( A = 2bl )
Considering both sides of the plate:
1
2 rAU 2
¥ Rl
2 D = 1.328b U ¥3 µrl
Therefore: skin friction µ U ¥3 / 2 ; l 1/ 2 Blasius formulae

¥
æ u ö nx
d1 = ò ç1 - ÷ dy = 1.7208 » 0.34d displacement thickness
0è
U¥ ø U¥
¥
u æ u ö nx
d2 = ò ç1 - ÷ dy = 0.664 » 0.133d momentum thickness
0
U¥ è U¥ ø U¥
Momentum integral equation of BL (von Karman equation)

¶u ¶u 1 dp ¶ 2u (1)
u +v =- +n 2
¶x ¶y r dx ¶y In BL (steady flow)
¶u ¶v
+ =0 (2)
¶x ¶y

Outside BL, u ® U ¥ (x) and viscous effects can be neglected.

Hence, from (1)
dU ¥ 1 dp (3)
U¥ =-
dx r dx
Integrate from wall y = 0 to y = h where h is everywhere
outside BL.
h h
æ ¶u ¶u dU ¥ ö µ ¶u t0 ¶u
ò0 çè u ¶x + v ¶y - U ¥ dx ÷ø dy = r ¶y 0 = - r (t0 = µ
¶y y =0
) (4)
 Replace v from continuity equation (2) in (4):
h
æ ¶u ¶u y ¶u dU ö t0
ò0 çè u ¶x - ¶y ò0 ¶x dy - U ¥ dx ÷ø dy = - r
ç ¥ ÷ (5)

Integrate by parts

h
æ ¶u y ¶u ö h
¶u
h
¶u
h
¶u
ò0 çè ¶y ò0 ¶x ÷ø
ç ÷ ¥ò ò (ò
h
dy dy = U dy - u dy dy = u 0 = U ¥ - 0 = U ¥ )
0
¶x 0
¶x 0
¶y
2 1
so that (5) becomes
h
æ ¶u ¶u dU ¥ ö t0
ò0 çè 2u ¶x - U ¥ ¶x - U ¥ dx ÷ø dy = - r
Rearranging
h h
¶ dU ¥ t0
ò0 ¶x {u (U ¥ - u )}dy + dx ò0 (U ¥ - u )dy = r (6)
 Both the integrals vanish outside BL (U¥- u = 0) and
hence h can be replaced by¥ :
¥ ¥
¶ dU ¥ t0
ò0 ¶x ¥
{u (U - u )}dy +
dx 0ò (U ¥ - u ) dy =
r
(7)

Using the definitions of displacement and momentum thicknesses

¥ ¥
1 1
d1 = ò
U¥ 0
(U ¥ - u )dy and d 2 = 2 ò u (U ¥ - u )dy (8)
U¥ 0
in (8), we get
d dU ¥ t0
(U ¥ d 2 ) + d1U ¥
2
=
dx dx r

which is the Momentum Integral Equation of 2D, steady BL.

This is valid for both laminar or turbulent BL.
 d dU ¥ t0 3 unknowns, 1 equation!!
(U ¥ d 2 ) + d1U ¥
2
=
dx dx r

Assume velocity profile shape using t0 as a parameter.

Then we have 1 equation in 1 unknown.
¶u ¶ æ ¶f ö ¶ æ ¶f ö ¶ æ ¶f ¶h ö ¶ æ ¶f ¶h ¶g ö
= U¥ ç ÷ = U¥ ç ÷ = U¥ ç ÷ = U¥ ç ÷
¶x ¶x è ¶h ø ¶h è ¶x ø ¶h è ¶h ¶x ø ¶h è ¶h ¶g ¶x ø
¶ æ y ö y
= -U ¥ ç f ¢ 2 g ¢ ÷ = -U ¥ f ¢¢ 2 g ¢
¶h è g ø g
¶u ¶ æ ¶f ö ¶ æ ¶f ö ¶ æ ¶f ¶h ö
= U¥ ç ÷ = U¥ ç ÷ = U¥ ç ÷
¶y ¶y è ¶h ø ¶h è ¶y ø ¶h è ¶h ¶y ø
¶ æ 1ö f ¢¢
¢
= U¥ ç f . ÷ = U¥
¶h è g ø g
¶ 2u ¶ æ f ¢¢ ö U ¥ ¶ 2 æ ¶f ö U ¥ ¶ 2 æ ¶f ¶h ö
= U¥ ç ÷ = 2 ç ÷= 2 ç ÷
¶y 2
¶y è g ø g ¶h è ¶y ø g ¶h è ¶h ¶y ø
U ¥ ¶ 2 æ ¶f 1 ö U ¥
= 2 ç
. ÷ = 2 f ¢¢¢
g ¶h è ¶h g ø g
temperature ( C) surface tension (erg cm-2)
-8 77.0
-5 76.4
0 75.6
5 74.9
10 74.22
15 73.49
18 73.05
20 72.75
25 71.97
30 71.18
40 69.56
50 67.91
60 66.18
70 64.4
80 62.6
100 58.9
 Introduction to Turbulent Flow

Turbulence refers to unstable flow of a liquid or gas, characterized by

complex, unsteady, random, eddying motion which occurs at ‘high’
Reynolds numbers.

Turbulent flow has a broad spectrum of eddy sizes with a

spectrum of fluctuation frequencies and thus the rotational motion
can be thought of as a randomly distributed vortex elements with
highly unsteady vorticity vectors aligned in all directions. The inertial
forces associated with eddies are larger than the forces which tend to
damp out the eddies.

Most flows occurring in nature are turbulent.

Laminar flow is always the exception.
A precise definition of turbulence, however, does not exist.
Turbulence is a feature of fluid flow, not that of fluid.
The principal characteristics of turbulence are
•High Reynolds number flow:
Associated with high inertia forces and relatively small viscous forces.
•Irregular or random flow:
Therefore, only statistical methods are used in turbulence calculations.
•Diffusive flow:
Rapid mixing and hence increased rates of momentum, mass, and
heat transfer occur.
3D fluctuations of vorticity :
3D and rotational flow with high levels of fluctuating vorticity.
•Dissipative flow:
Requires continuous supply of energy.
•Continuum level:
Governed by equations of continuum (fluid) mechanics,
not molecular level theories.
 Drag coefficient of circular cylinder

Drag crisis

Drag crisis at about Re= 3´105 is a consequence of transition

in BL where point of separation moves downstream and
decreases wake width considerably and BL becomes turbulent
ahead of separation point.
Drag crisis
 Mean time averages for steady turbulent flow
t + Dt
1
f =
Dt ò f dt
t
(mean time average)

Dt large enough to make f independent of t.

f = f + f¢
t + Dt t + Dt t + Dt
1 1 1
f =
Dt ò ( f + f ¢) dt =
t
Dt ò
t
f dt +
Dt ò f ¢ dt
t
t + Dt t + Dt t + Dt
1 1 1
= f
Dt ò
t
dt +
Dt ò
t
f ¢ dt = f +
Dt ò f ¢ dt
t
t + Dt
1
Dt òt
\ f ¢ dt = 0 Þ f ¢ = 0 (1a)
¶g ¶g ¶
g + h = g + h ; Cg = Cg ; = (C : constant, : space derivative)
¶n ¶n ¶n
(1b,c,d)
If g = 0 and h = 0, gh ¹ 0 is possible.
 NS equations for mean time averaged quantities
¶u ¶u ¶u ¶u 1 æ ¶s ö
+ u + v + w = X + ç + µÑ 2u ÷
¶t ¶x ¶y ¶z r è ¶x ø
¶v ¶v ¶v ¶v 1 æ ¶s ö NS equations with
+ u + v + w = Y + ç + µÑ 2 v ÷
¶t ¶x ¶y ¶z r è ¶y ø s = -p
¶w ¶w ¶w ¶w 1 æ ¶s ö (2)
+u +v +w = Z + ç + µÑ 2 w ÷
¶t ¶x ¶y ¶z r è ¶z ø
Replace each term by a sum of mean time average + fluctuating
components. For example:
¶u ¶ (u + u¢) ¶u ¶u¢ ¶u ¶u¢
u = (u + u¢) =u +u + u¢ + u¢
¶x ¶x ¶x ¶x ¶x ¶x
Take time average of LHS and RHS of above using (1):
¶u ¶u ¶u ¶u¢ ¶u ¶u¢
u =u + u ´ 0 + 0´ + u ¢ =u +u ¢ etc.
¶x ¶x ¶x ¶x ¶x ¶x
 Finally (2) reduces to
æ ¶u ¶u ¶u ö æ ¶s 2 ö æ ¶u ¢ ¶u ¢ ¶u ¢ ö
rç u +v + w ÷ = rX + ç + µÑ u ÷ - rç u ¢ + v¢ + w¢ ÷
è ¶x ¶y ¶z ø è ¶x ø è ¶x ¶y ¶z ø
æ ¶v ¶v ¶v ö æ ¶s 2 ö æ ¶v¢ ¶v¢ ¶v¢ ö
rç u + v + w ÷ = rY + ç + µÑ v ÷ - rç u ¢ + v¢ + w¢ ÷
è ¶x ¶y ¶z ø è ¶y ø è ¶x ¶y ¶z ø
æ ¶w ¶w ¶w ö æ ¶s ö æ ¶w¢ ¶w¢ ¶w¢ ö
rç u +v +w ÷ = rZ + ç + µÑ w ÷ - rç u
2
¢ +v ¢ +w¢ ÷
è ¶x ¶y ¶z ø è ¶z ø è ¶x ¶y ¶z ø
Uniform Parallel Flow
f = const. u = U ¥ , v = 0 Continuity equation satisfied
U¥
y ¶f ü
u= = U¥ ï
¶x ï
ý Þ f( x, y ) = U ¥ x + constant
¶f
v= =0 ï
y = const. ¶y ïþ
x
¶y
u= = U ¥ üï
¶y ï
ý Þ y ( x, y ) = U ¥ y
¶y
v=- = 0ï
¶y ïþ
2D Source
f = f(r ) = 0; axisymmetric solution

1 ¶ æ ¶f ö 1 ¶ 2
f Volume flow rate through
Ñ f=
2
çr ÷ + 2 2 = 0
r ¶r è ¶r ø r ¶q a circle of radius R is:
d æ df ö Integrate
df Þ Q = (2pR)ur (r = R) = 2pc1
çr ÷ = 0
dr è dr ø Þ r
dr
= c1
Q
df c1 Integrate c1 =
2p
=
dr r Þ f = c1 ln(r ) + c2
Q is source strength
¶f c1 1 ¶f Q > 0: SOURCE
ur = = ; uq = =0
¶r r r ¶q Q < 0: SINK

Q Q
f= ln r ur = uq = 0
2p 2pr
 1 ¶y Q Integrate Qq
ur =
r ¶q 2pr
= Þ y (r , q) =
2p
+ f (r )

¶y ¶f
uq = - =0 Þ = 0 Þ f (r ) = c
¶r ¶r
Qq
\ y (r , q) = y (q) =
2p uq
ur
y = constant Þ rays (q = constant) f = const. r
y q
f = constant Þ circles (r = constant)
x y = const.
3D Source
f = f(r , q, j)
1 é¶ æ 2 ¶f ö ¶ æ ¶f ö ¶ æ 1 ¶f öù
Ñ f= 2
2
ê ç r sin q ÷ + ç sin q ÷ + ç ÷ú = 0
r sin q ë ¶r è ¶r ø ¶q è ¶q ø ¶j è sin q ¶j øû
¶æ 2 ¶f ö Integrate
¶f
ç r sin q ÷ = 0 Þ r sin q = c1 (q)
2

¶r è ¶r ø ¶r
c1 (q) c1 (q)
f(r , q) = ò 2 dr = - + c2 (q)
r sin q r sin q
¶f c c1 (q)
= 0 Þ f(r ) = - (i.e c2 (q) = 0 & = c)
¶q r sin q
¶f c 1 ¶f 1 ¶f
ur = = 2 uq = =0 uj = =0
¶r r r ¶q r sin q ¶j
 Volume flow rate through a sphere of radius R is:
c
m = (4pR )ur ( R) = (4pR ) 2 = 4pc
2 2

R
m m is source strength
c= m > 0: SOURCE
4p
m m m < 0: SINK
f(r ) = - ur =
4pr 4pr 2
 2D Vortex If c = 0, there is no flow at all.
f = f(q) If c ¹ 0, then G ¹ 0 and hence
flow cannot be irrotational!!
1 ¶ æ ¶f ö 1 ¶ f 2
Thus Ñ 2f = 0 cannot hold!!
Ñ f=
2
ç r +
÷ 2 2 = 0
r ¶r è ¶r ø r ¶q This contradiction is due to the
d 2f Integrate
fact that vorticity is zero
= 0 Þ f = c1q + c2 = cq everywhere except at origin,
dq 2

¶f 1 ¶f c where it is infinitely large.

ur = = 0; uq = = All contours enclosing the origin
¶r r ¶q r
2p
will have nonzero circulation.
! c
G = - ò v .dl = - ò rdq = -2pc uq
0
r
ur
Plane can be cut to prevent
contours around origin.
 G
G = -2pc f = cq = - q
2p
1 ¶y ¶y G G
ur = =0 uq = - =- Þ y = ln r
r ¶q ¶r 2pr 2p
G
y = ln r : y = constant Þ circles (r = constant)
2p
G
f = - q: f = constant Þ rays (q = constant)
2p
Uniform flow + source flow

A
Q
2V¥
q =p B q =0

Q C
y=
2
 P ( r ,q )

q1 q q2
b b

r sin q r sin q
tan q1 = tan q 2 =
b + r cosq r cosq - b
 Doublet flow

+Q: Source; -Q: Sink Distance between source & sink = l

P
Q Q
y(r , q) = (q1 - q2 ) = - Dq
Dq 2p 2p
r
Take limit l ® 0 & Q ® ¥
q1 q2
such that Ql = k (constant)
Q !"
"#"" $ -Q Dq sin q
l Lt
l ®0 l
=
r
æ Q ö æ k Dq ö k sin q
y doublet = Lt ç - Dq ÷ = Lt ç - ÷=-
l ®0 è 2p ø l ®0 è l 2p ø 2pr
 ¶y k sin q 1 ¶f
= = - (= -ur )
¶r 2pr 2
r ¶q
¶f k sin q k cos q
or, =- Þ f=
¶q 2pr 2pr
sinq y
Streamlines : y = constant Þ =c Þc= 2
r x + y2
2 2
æ 1ö æ1ö
or, x + ç y - ÷ = ç ÷
2

è 2c ø è 2c ø
Streamlines are circles with centre at (0, 1/(2c)) and radius 1/(2c).
Streamlines are oriented in x- direction.

For tilted doublet:

k k
y=- sin(q - a) f= cos(q - a)
2pr 2pr
 Flow over a circular cylinder

x = r cos q y = r sin q Solution is

1 ¶ æ ¶f ö 1 ¶ 2
f æ a2 ö
Ñ f=
2
çr ÷ + 2 2 = 0 f = U ç r + ÷ cos q
r ¶r è ¶r ø r ¶q è r ø
f = Ux = Ur cos q as r ® ¥ BC at far field: uniform flow
¶f
= 0 at r = a BC: no flow across cylinder boundary
¶r
Consider uniform flow in x + doublet at origin oriented in x
k æ k ö
f = Ur cos q + cos q = U ç r + ÷ cos q
2pr è 2pUr ø
æ a2 ö k
= U ç r + ÷ cos q using a =
2

è r ø 2pU
\Flow past a circular cylinder is obtained from superposition
of two simpler flows: parallel flow and a doublet.

k æ a2 ö
y = Ur sin q - sin q = U ç r - ÷ sin q
2pr è r ø
 Flow over a circular cylinder with circulation

Flow over cylinder (uniform flow + doublet)+ vortex

k G G
y = Ur sin q - sin q + ln r - ln a
2pr 2p 2p
æ a2 ö G r k
= U ç r - ÷ sin q + ln using a =
2

è r ø 2p a 2pU
 1 ¶y æ a2 ö G r
ur = = U ç1 - 2 ÷ cos q + ln
r ¶q è r ø 2p a
¶y æ a2 ö G
uq = - = -U ç1 + 2 ÷ sin q -
¶r è r ø 2pr
G
at r = a Þ ur = 0, uq = -2U sin q -
2pa
Pressure coefficient is given by
u 2
(u ) + (uq )
2 2
2G sin q æ G ö
2

Cp = 1- 2 = 1- r = 1 - 4 sin 2
q + +ç ÷
U U 2
paU è 2paU ø
Bernoulli equation
1 2 1 1 1
p + ru = p¥ + rU Þ p = p¥ + r(U - u ) = p¥ + rC pU 2
2 2 2

2 2 2 2
p - p¥
or, C p = 1
2 rU 2
 y
p

a dq
dq
a x

2p 2p
D = ò - pa cos qdq L = ò - pa sin qdq
0 0

Drag Lift
D D L L
CD = 1 = CL = 1 =
2 rU 2
.2 a.1 rU 2
a 2 rU 2
.2 a.1 rU 2
a
Drag coefficient Lift coefficient
 2p 2p
æ 1 ö
D = - ò pa cos qdq = - ò ç p¥ + rU 2C p ÷a cos qdq
0è ø
0
2
2p
é 1 ì 2 G sin q æ G ö
2
üù
= - ò ê p¥ + rU í1 - 4 sin q -
2 2
-ç ÷ ýú a cos qdq = 0
0 ë
2 î paU è 2paU ø þû
Integrals involving cos q, sin2 q cos q and sin q cos q are zero.

This is d’Alembert’s paradox.

“ An inviscid fluid offers no resistance to steady translational
motion of a rigid body when flow is everywhere irrotational”.
2p 2p
æ 1 ö
L = - ò pa sin qdq = - ò ç p¥ + rU 2C p ÷a sin qdq
0è ø
0
2
é
2p
1 ì 2G sin q æ G ö
2
üù
= - ò ê p¥ + rU í1 - 4 sin q -
2 2
-ç ÷ ýú a sin qdq
0 ë
2 î paU è 2paU ø þû
Integrals involving sin q, sin3 q are zero. Integrals involving sin2 q ¹0
 2p
1 2G sin 2
q 1 2Gp
L = - rU 2 a ò - dq = rU 2 a. = rUG
2 0
paU 2 paU

! !
L = rUG = rU ò v.dl Kutta-Joukowski Theorem

L
U
Unsteady Flow over a circular cylinder

Solution is
æ a2 ö (1)
f(r , q, t ) = U (t )ç r + ÷ cos q
è r ø
 Apply Bernoulli’s equation on cylinder surface (r = a) and at a
point large distance upstream(r = l, q = 0; l >> a):

é p(r , q) ¶f ù é p (r , q) ¶f ù
ê r + gz + 2 (ur + uq ) + ¶t ú = ê r + gz + 2 (ur + uq ) + ¶t ú r =l
1 2 2 1 2 2

ë û r =a ë û q=0
\Difference of wall pressure and free-stream pressure is
ìïæ u 2 + u 2 ö æ u 2
+ u 2
ö ¶f ¶f üï
æ ö æ ö
p (a, q) - p (l ,0) = ríç r q
÷ -ç r q
÷ +ç ÷ -ç ÷ ý
ïîè 2 ø qr ==l0 è 2 ø r = a è ¶t ø q=0 è ¶t ø r = a ïþ
r =l

From (1): (2)

¶f æ a2 ö 1 ¶f æ a2 ö
ur = = U ç1 - 2 ÷ cos q; uq = = -U ç1 + 2 ÷ sin q (3a)
¶r è r ø r ¶q è r ø
¶f æ a 2 ö dU (3b)
= ç r + ÷ cos q
¶t è r ø dt
 (3a, b) in (2)
ì1 2 dU dU ü
p(a, q) - p(l ,0) = rí U (1 - 4 sin q) - 2a
2
cos q + l ý
2p
î 2 dt dt þ
D = ò - p (a, q)a cos qdq
0 =0 (steady part: d’Alembert’s paradox)
2p
1
= - ò ìí rU 2 (1 - 4 sin 2 q) + p (l ,0)üýa cos qdq
0î þ
2
2p 2p
=0
dU 2 dU
+ òr 2a cos 2 qdq + ò r la cos qdq
0
dt 0
dt
dU dU dU dU
= 2rpa 2 = 2rV = CM rV = (1 + Ca )rV
dt dt dt dt
V: cylinder volume (per unit length)
rV: cylinder (displaced) mass (per unit length)
CM: inertia coefficient (= 2) Ca: added mass coefficient (= 1)

Added mass of cylinder = Mass of cylinder

 Lifting Surfaces

Applications
Flight of birds and aircraft
Hydrofoil boat, sailing ship, rudder, propeller, yacht keel, anti-roll fins.

Lifting surface is a thin, streamlined body which moves

through a fluid medium at small angle of attack and
thereby generates a fluid dynamic ‘lift’ force in the
direction normal to the forward movement.
Chord length: distance from leading edge (LE) to trailing edge (TE)
Chord line: straight line connecting LE and TE
Camber: measure of the curvature of the foil.
Mean camber line: line halfway between the upper surface and
lower surface and intersects the chord line
at LE and TE.
Between LE and TE, the camber line can curve
above or below the chord line.
Camber value: highest value above or below chord line divided by
chord length.
NACA 4412: 4: max camber as % c (chord) (i.e. 0.04c = yc)
4: position of max. camber in 1/10 of c (i.e. xc = 4c/10)
12: max. thickness as % c (i.e. 0.12c)
Mean camber line: one parabola from LE to (xc , yc) and another
parabola from (xc , yc) to TE.
Thickness distribution formula:

y = ±(t / 0.2)(0.2969 x 0.5 - 0.126 x - 0.3537 x 2 + 0.2843 x 3 - 0.1015 x 4 )

 FLOW AROUND A FOIL
1. Kutta-Joukowski theorem:
Lift (L) µ circulation (G)

2. Value of G can be assigned arbitrarily.

3. Solution of flow around a cylinder:

There are 2 stagnation points
along the foil.

4. Positions of stagnation points are

determined by the value of G around
the foil profile.
A particular value of G that moves the rear stagnation point (v = 0)
exactly on the trailing edge is the only one that is physically
realizable. This condition, which fixes a value of G by geometrical
considerations alone, is the Kutta condition.
Kutta condition fixes the value of circulation and thereby makes it
possible to evaluate the lift of a foil using essentially the same
techniques that were used to obtain lift generated by a circular
cylinder.

All flow fields, each due to a different value of circulation, are valid
solutions of the flow around a foil. Out of these, the Kutta condition
chooses one which represents the best actual flow, i.e. the flow
leaves the trailing edge in a tangential manner.

‘Kutta condition’ is central to the theory of lifting surfaces.

Physical argument for Kutta condition
1. Do a ‘cold start’ with a small velocity to the fluid, initially at rest and
hence irrotational (disregard viscous effects).
2. It must continue to remain irrotational and hence flow field around
the foil will have zero circulation (Kelvin’s theorem).
3. It will also have 2 stagnation points:
• One on the lower face of the foil (consider angle of attack in next,
figure) close to the leading edge. [upper face in earlier figure]
• One on the upper face, close to the trailing edge.

4. A fluid particle on the lower face has to travel along the

profile, make a sharp U-turn at the trailing edge, go upstream
on the upper face until it reaches the stagnation point and
then, eventually, leave the profile.
5. It is contrary to our expectation of the behavior of a real fluid.
Viscosity will come into play to damp the sharp velocity gradient
along the profile Þ separation of boundary layer Þ wake formation.

6. Wake creation leads to shedding of anti-clockwise vorticity from the

trailing edge (above figure) as below. [clockwise in first figure]

7. But circulation along a curve enclosing both vortex (shown on right

side, called starting vortex) and foil must remain zero, this leads to
a clockwise circulation around the profile (shown on left side).
8. This nonzero circulation around the profile will move the
stagnation points, specifically the rear stagnation point would
move towards the trailing edge.
Potential flow with circulation
Potential flow w/o circulation

Increase in
9. Vortex circulation Downstream shift of
shedding around foil rear stagnation point

This sequence continues until the rear stagnation point reaches

the trailing edge. At that instance, the sharp velocity gradient
disappears and shedding of vorticity stops.

10. This, then, ensures ‘equilibrium’ of the system and it freezes

the value of the circulation around the foil.
11. In steady state problem (much after cold start), starting vortex
may be disregarded as it has progressed infinitely far downstream
and will be dissipated by viscous diffusion. So, one is left with
irrotational flow around foil with
a net circulation around it.

Starting
Vortex

typical image for an

airfoil at a negative
angle of attack
 y Linearised 2D hydrofoil
y = yu(x)

x
l/2
-l/2
y = yl(x)
BV problem
!
1. Velocity vector at infinity equals free stream velocity (= -Ui )
2. If perturbation velocities are (u, v), then total velocity vector
in 2D flow is (u-U, v).
3. Velocity vector is tangential to foil surface.
4. Velocity vector is ‘finite’ at TE (this is Kutta condition).
Let (u, v) = Ñf (f is perturbation potential)
F = -Ux + f (F is total velocity potential)

steady (free stream) perturbation

 Ñ 2 F = Ñ 2 f = 0 (in fluid domain)
Ñf < ¥ (at TE) : Kutta condition Ñf ® 0 (at infinity)
¶F ¶x ¶f
= 0 (on foil surface) Þ 0 = -U +
¶n ¶n ¶n
¶f
or = Unx (on foil surface)
¶n
This BC on foil surface can be linearised by assuming:
“Foil is thin and nearly horizontal”.

This means: yu and yl are small compared to chord length l. Thus,

yu yl dyu dyl
<< 1; << 1; yu¢ = << 1; yl¢ = << 1
l l dx dx
For sufficiently small yu¢ , u << U
 In order to linearise the foil surface BC:
v dyu
= (from geometry) on y = yu (upper surface)
u - U dx
v u - U dyu u dyu dyu dyu u
or, = = - =- (! » 0)
U U dx U dx dx dx U
+ dyu (a)
or, v( x,0 ) = -U (on y = 0 + )
dx
- dyl
Similarly, v( x,0 ) = -U (on y = 0 - ) (b)
dx
A fundamental feature of lifting flows is that solution must
admit discontinuities ‘across’ x-axis. This is what Eqs. (a)
& (b) reflect.
y
Ñ 2f = 0 ¶f / ¶y = -Uyu¢
Ñf < ¥
x
TE
Linearised problem
¶f / ¶y = -Uyl¢
 The ‘half thickness’ distribution, by definition, is given by
T ( x) = 12 { yu ( x) - yl ( x)}
The camber distribution, by definition, is given by
C ( x) = 12 { yu ( x) + yl ( x)}
yielding yu = C + T ; yl = C - T
+ dyu
v( x,0 ) » -U æ dC dT ö
dx Þ ±
v( x,0 ) » -U ç ± ÷ (on y = 0 ± )
dyl è dx dx ø
-
v( x,0 ) » -U
dx
Eqs. (a) & (b)
Decomposition into ‘thickness problem’ and ‘camber problem’

± ¶f ± dT ± ¶f ± dC
v( x,0 ) = ( x,0 ) = !U v( x,0 ) = ( x,0 ) = -U
¶y dx ¶y dx
Thickness problem Camber problem
 Solution to thickness problem
Since camber is zero, it refers to the problem of a symmetric foil
with zero angle of attack. Thus, it does not contribute to lift or
moment.
x U

dT
v = -U
dx
dT
v =U
Ñ 2 f = 0 ( y ¹ 0) dx
dT
v( x, y = 0 ± ) = !U Jump in v
dx
Thickness problem can be solved with a distribution of sources
along x-axis.
 Solution to camber problem
Since thickness is zero, it refers to the problem of flow past mean
camber line.
C (x) U
x
dT
v = -U
dx

Ñ 2 f = 0 ( y ¹ 0)
± dC
v( x, y = 0 ) = -U (no jump in v)
dx
jump in u : u + ¹ u - Þ p + ¹ p - Þ Lift
Ñf < ¥ at TE ( x = -l / 2, y = 0)
Camber problem can be solved with a distribution of vortices
along x-axis.
 y f = f(q) = cq
g (x) Vortex strength
. x ur =
¶f
¶r
= 0; uq =
1 ¶f c
=
r ¶q r
x !
2p
c G
G = ò v .dl = ò rdq = 2pc Þ f = q
0
r 2p
1 -1 æ y ö Recall
df( x, y ) = g (x)dx tan ç ÷ -1 y
2p è x -xø ! q = tan
x-x
l/2
1 -1 æ y ö
f( x, y ) = ò g (x) tan ç ÷dx
-l / 2
2p è x -xø
¶f ¶ ì l/2
1 -1 æ y ö ü dC
= í ò g (x) tan ç x
÷ ý
d = -U
¶y y =0 ¶y î-l / 2 2p è x - x ø þ y =0 dx

Given dC / dx, find g ( x) ® G ® Lift : Analysis problem

Given Lift ® G ® g ( x), find dC / dx : Design problem
 g æ y ö
f= tan -1 ç ÷ Velocity potential at (x,y) due to a vortex
2p è x -xø of strength g at (x,0)
l/2
¶f g -y 1 g ( x) y
u= =
¶x 2p ( x - x) + y
2 2
Þ u ( x, y ) = - ò
2 p - l / 2 ( x - x) + y
2 2
dx

l/2
¶f g x-x 1 g (x)( x - x)
v= =
¶y 2p ( x - x) + y
2 2
Þ v ( x, y ) = ò
2 p - l / 2 ( x - x) + y
2 2
dx

l/2
1 g ( x) e
u ( x, ± e ) = ! ò
2 p - l / 2 ( x - x) + e
2 2
dx e<<l, e>0

l/2 l/2
1 e 1 é -1 x - x ù
» ! g ( x) ò dx = ! g ( x) ê tan úû -l / 2
2p -l / 2
( x - x ) 2
+ e 2
2 p ë e
g ( x)
»! Horizontal velocity along the cut
2
 l/2
1 g ( x)
v( x,0 ± ) = ò
2p -l / 2 x - x
dx Vertical velocity along the cut

This integral is interpreted in the sense of Cauchy principal value:

b
f ( x) ì x - e f ( x) b
f ( x) ü
òa x - x dx º lim íò
e ®0 î a x - x
dx + ò
x+e
x-x þ
dxý
 1 !2
p + rv + rgz = const everywhere for irrotational flow
2
Recall
p - p¥ = (- 2 rv ) - (- 2 rvU ) = - 2 r(v - U 2 )
1 !2 1 2 1 !2

= - 12 r{(u - U ) 2 + v 2 - U 2 } = - 12 r(u 2 + v 2 - 2uU )

» ruU (linearising)
l/2
L = ò ( p - p¥ )n y dS = ò {( p ( x , 0 -
) - p ¥ ) - ( p ( x , 0 +
) - p¥ )}dx
-l / 2
l/2 l/2

ò ò
- + - +
= { p ( x , 0 ) - p ( x , 0 )}dx =rU {u ( x , 0 ) - u ( x , 0 )}dx
-l / 2 -l / 2

g ( x)
" u ( x ,0 ± ) = !
2
l/2 l/2
L = rU ò {u ( x,0 - ) - u ( x,0 + )}dx = rU ò g( x)dx = rUG (Lift)
-l / 2 -l / 2
 Moment wrt mid-chord y dL = rUg ( x)dx
l/2
M = rU ò xg( x)dx
-l / 2

M x = rU ò uxdx
M = Lxcp
xcp: Centre of pressure x L M
cp CM = 1
2 rU 2 2
l
L
Analysis problem CL = 1
2 rU 2
l
g at TE = 0 (Kutta condition)
Given C ( x), hence find dC / dx
From v( x,0) = -UdC / dx find v( x,0)
l/2
1 g ( x)
ò
±
Solve v( x,0 ) = dx to obtain g (x)
2p -l / 2 x - x
This is an integral equation since g is in the integrand.
 General solution of this integral equation is:
2 ì l/2
v(x,0) (l / 2) 2 - x 2 Gü
2 í ò
g ( x) = dx + ý
p (l / 2) - x î-l / 2
2
x- x 2þ
l/2
where G = ò g(x)dx
-l / 2

However, since G is unknown, the above solution lacks

uniqueness. To make it unique, one needs to use the
Kutta condition on TE.

! (l / 2) 2 - x 2 ® 0 as x ® ±l / 2 (LE & TE)

the general solution yields infinite vortex strengths at both LE & TE
and this can be overcome at TE if one sets
l/2
v(x,0) (l / 2) 2 - x 2 G
ò
-l / 2
x- x
dx + = 0 at x = -l / 2 (at TE)
2
 leading to
l/2
l /2-x
G = -2 ò v(x,0) dx
-l / 2
l /2+x
and hence, from Kutta-Joukowski’s theorem, lift is given by
l/2
l /2-x
L = rUG = -2rU ò v(x,0) dx
-l / 2
l /2+x
l/2
dC l / 2 - x
= 2rU ò 2
dx (using v = -U (dC / dx)
-l / 2
dx l / 2 + x
l/2
L 4 dC l / 2 - x
CL =
2 rU l
1 2
= ò
l -l / 2 dx l / 2 + x
dx
l/2 l/2
dC
M = rU ò g = r ò - x dx
2 2 2
x ( x ) dx 2 U (l / 2)
-l / 2 -l / 2
dx
l/2
M 4 dC
2 ò
CM = 1 = (l / 2) 2
- x 2
dx
2 rU l l -l / 2 dx
2 2
 Water waves (gravity waves on free surface)
z
Free surface, z = h

Wave height H h MWL, z = 0

(crest to trough) x

d
Ñ 2f = 0

seabed, z = -d
Principal assumptions
•Fluid is homogeneous & incompressible (density constant)
irrotational & inviscid (potential flow)
•Pressure on free surface is uniform and constant.
•Seabed is horizontal, fixed and impermeable.
•Wave is two dimensional, its form is invariant in time and space.
 Free surface BC z
SF : Free surface
H h
x

d
Ñ 2f = 0

¶f ¶f
=0 Þ = 0 at z = -d
¶n ¶z
F ( x, y, z, t ) = z - h( x, y, t ) = 0 : Equation of free surface
= z - h( x, t ) = 0 : 2D wave (1)
Kinematic BC: A fluid particle on free surface continues to be on
free surface at all times.ß
Fluid moves only tangentially on free surface. ß
Total derivative of free surface is zero. If one moves
with free surface, it does not change. [DF/Dt = 0]
 DF ¶F ¶F ¶F ¶F
= +u +v +w = 0 on z = h (2)
Dt ¶t ¶x ¶y ¶z
Using
¶F ¶h ¶F ¶h ¶F ¶h ¶F
=- ; =- ; =- ; = 1 on z = h
¶t ¶t ¶x ¶x ¶y ¶y ¶z
and (1) in (2) gives
¶h ¶h ¶h ¶h ¶F ¶h ¶F ¶h ¶F
+ u + v = w OR + + = on z = h
¶t ¶x ¶y ¶t ¶x ¶x ¶y ¶y ¶z
This is kinematic BC

¶h ¶F ¶h ¶F
+ = on z = h [2D wave]
¶t ¶x ¶x ¶z
¶h ¶F
= on z = 0 [2D linear wave]
¶t ¶z
Dynamic BC: Pressure on free surface is atmospheric pressure.

¶F 1 2
Pa + rgh + r + r ÑF = C (t )
¶t 2
Taking atmospheric pressure to be constant in
space and time(gauge pressure):
¶F 1 2
rgh + r + r ÑF = C (t ) This is dynamic BC
¶t 2

¶F 1 2
gh + + ÑF = 0 [free surface is datum]
¶t 2
¶F
gh + = 0 [Linearised] KE of fluid is small in
¶t comparison to other
mechanical energies.
Valid only when h is ‘small’.
 Linearised dynamic BC gives free surface as

1 ¶F
h=- at z = 0
g ¶t

How small is ‘small amplitude’?

Particle orbits nearly circular in deep water waves

Radius »H/2
c In one cycle of period T,
H h
particle travels a distance pH
with average velocity pH/T (= c).

L = 39m, T = 5s H1= 3m H2= 0.3m

Linearised equations
c= 1.885m/s 0.1885m/s
apply upto H/L = 0.02.
gh = gH/2 = 14.715m2/s2 1.4715m2/s2
In this case: H<0.78m
(1/2)v2 = c2/2 = 1.777m2/s2 0.01777m2/s2
(gh)/((1/2)v2) = 8.281 82.81
H/L = 0.076 0.0076
 ¶h ¶F Kinematic BC
=
¶t ¶z on z = 0: Free surface SF
¶F
gh + = 0 Dynamic BC
¶t

Combining both BCs: 1 ¶ 2 F ¶F

+ = 0 on z = 0
g ¶t 2
¶z
Boundary Value Problem (BVP)
Ñ 2 F = 0 in fluid domain
¶F
= 0 on seabed ( z = - d )
¶z
1 ¶ 2 F ¶F
+ = 0 on free surface z = 0
g ¶t 2
¶z
 (1)
Ñ 2 F = 0 in fluid domain
¶F (2)
= 0 on seabed ( z = - d )
¶z
1 ¶ 2 F ¶F
+ = 0 on free surface z = 0 (3)
g ¶t 2
¶z
Assume wave - like solution as F = f ( z ) sin(kx - wt ) (4)
k = 2p / L (wave number), L (wave length)
w = 2p / T (wave frequency in rad/s), T (wave period)
(4) in (1): ¶ 2
f
- k 2
f =0 (5)
¶z 2

General solution of (5) is f ( z ) = Ae kz + Be - kz (6)

(6) in (4): F = ( Ae kz + Be - kz ) sin( kx - wt ) (7)
F = ( Ae kz + Be - kz ) sin( kx - wt ) (7)
¶F
= k ( Ae kz - Be - kz ) sin( kx - wt ) (a)
¶z
¶F
= -w( Ae kz + Be - kz ) cos(kx - wt ) (b)
¶t
¶ 2F - kz
= - w 2
( Ae kz
+ Be ) sin( kx - wt ) (c)
¶t 2

¶F
= k ( Ae - kd - Be kd ) = 0 Þ Ae - kd - Be kd = 0 (8) Seabed BC
¶z z = - d
é ¶F 1 ¶ 2 F ù w2
ê ¶z + g ¶t 2 ú = k ( A - B ) sin( kx - wt ) - g ( A + B ) sin( kx - wt ) = 0
ë û z =0 Free surface BC
or, w2 ( A + B ) - gk ( A - B ) Þ (w2 - gk ) A + (w2 + gk ) B = 0 (9)
 (8)
Ae - kd - Be kd = 0
(w2 - gk ) A + (w2 + gk ) B = 0
This image cannot currently be displayed.
(9)

For nontrivial solution

w2 - gk w2 + gk e kd
- e - kd
sinh kd
= 0 Þ w = gk kd
2
- kd
= gk = gk tanh kd
e - kd - e kd e +e cosh kd

w2 = gk tanh kd Dispersion relation

2p 2p L w
Using w = ; k= and c = = in dispersion relation :
T L T k
gT 2 gT g
L= tanh kd c= tanh kd = tanh kd
2p 2p k
Wave length Wave celerity
 Deep vs shallow water waves
d
kd = 2p
L
d gT 2
For large (or kd ), tanh kd » 1 Þ L = (= 1.56T 2 meters)
L 2p
d
For small (or kd ), tanh kd » kd Þ L = T gd ; c = gd
L
d 1 d 1
Deep water : > Shallow water : < (error < 1%)
L 2 L 20
 Velocity potential

Ae-kd - Bekd = 0 (8)

- kd E in F = ( Ae kz + Be - kz ) sin( kx - wt ) (7)
Ae = Be =
kd

2
gives
E kd kz - kd - kz e k ( d + z ) + e -k ( d + z )
F = (e e + e e ) sin(kx - wt ) = E sin(kx - wt )
2 2
= E cosh(k (d + z )) sin(kx - wt )
1 ¶F Ew H
h=- = cosh kd cos(kx - wt ) = cos(kx - wt )
g ¶t z =0 g 2
H g H gw H gw
or E = = =
2 w cosh kd 2 w2 cosh kd 2 gk tanh kd cosh kd
Hw
=
2k sinh kd
 F = E cosh(k (d + z)) sin(kx - wt )
Hw
E=
2k sinh kd
Hw cosh(k (d + z ))
\F = sin( kx - wt )
2k sinh kd
 Group Velocity
H H
h = cos k ( x - ct ) + cos k ¢( x - c¢t ) k » k ¢, c = c¢
2 2
æ k - k¢ kc - k ¢c¢ ö æ k + k ¢ kc + k ¢c¢ ö
= H cosç x- t ÷ cosç x- t÷
è 2 2 ø è 2 2 ø
æ k - k¢ kc - k ¢c¢ ö 4p 4p
Wave 1 : cosç x- t÷ Þ L = ,T =
è 2 2 ø k - k¢ kc - k ¢c¢
æ k + k¢ kc + k ¢c¢ ö 4p 2p 4p 2p
Wave 2 : cosç x- t÷ Þ L = » ,T = »
è 2 2 ø k + k¢ k kc + k ¢c¢ kc
Wave 2 is almost same as component waves.
Wave 1 has large length and period compared to component waves.
Propagation speed of the envelope of the composite wave train
is called group velocity (cg).
4p 4p
L= ,T =
k - k¢ kc - k ¢c¢
L kc - k ¢c¢ d(kc) d (kc) dw dc dc
cg = = = = = =c+k = c-L
T k - k¢ dk dk dk dk dL
gL 2pd dc
Using c = tanh in cg = c - L gives
2p L dL
1 ì 4pd 1 ü
c g = í1 + ýc = nc
2î L sinh(4pd / L) þ
1
n = : deep water n = 1 : shallow water
2