Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
FLUID DYNAMICS
Developed & Produced by EXCEL BOOKS PVT. LTD., A-45 Naraina, Phase 1, New Delhi-110 028
3
Contents
UNIT-I 5
UNIT-II 84
UNIT-III 176
UNIT-IV 235
UNIT-V 297
4
UNIT-I
Kinematics — Lagrangian and Eulerian methods. Equation of continuity. Boundary surface. Stream lines. Path lines and
streak lines. Velocity potential. Irrotational and rotational motions. Vortex lines.
Equations of Motion—Lagrange’s and Euler’s equations of motion. Bernoulli’s theorem. Equation of motion byflux method.
Equations referred to moving axes Impulsive actions. Stream function.
UNIT-II
Irrotational motion in two-dimensions. Complex velocity potential. Sources, sinks, doublets and their images. Conformal
mapping, Milne-Thomson circle theorem. Two-dimensional irrotational motion produced by motion of circular, co-axial and
elliptic cylinders in an infinite mass of liquid. Kinetic energy of liquid. Theorem of Blasius. Motion of a sphere through a
liquid at rest at infinity. Liquid streaming past a fixed sphere. Equation of motion of a sphere. Stoke’s stream function.
UNIT-III
Vortex motion and its elementary properties. Kelvin’s proof of permance. Motions due to circular and rectilinear vertices.
Wave motion in a gas. Speed of Sound. Equation of motion of a gas. Subsonic, sonic and supersonic flows of a gass.
Isentropic gas flows. Flow through a nozzle. Normal and oblique shocks.
UNIT-IV
Stress components in a real fluid. Relations between rectangular components of stress. Connection between stresses and
gradients of velocity. Navier-stoke’s equations of motion. Plane Poiseuille and Couette flows between two parallel plates.
Theory of Lubrication. Flow through tubes of uniform cross section in form of circle, annulus, ellipse and equilateral triangle
under constant pressure gradient. Unsteady flow over a flat plate.
UNIT-V
Dynamical similarity. Buckingham p-theorem. Reynolds number. Prandt’s boundary layer. Boundary layer equations in two-
dimensions. Blasius solution. Boundary-layer thickness. Displacement thickness. Karman integral conditions. Separations
of boundary layer flow.
5
FLUID DYNAMICS
UNIT-I
Basic Concepts and Definitions
(i) Let q î u ĵv k̂w, then
|q| = u 2 v2 w 2 q
u v w
D.C‟s are given by l = cos = , m = cos = , n cos
|q| |q| |q|
where l, m, n, are components of a unit vector i.e. l2 + m2 + n2 = 1
(ii) a.b ab cos , a b ab sin n̂
(iii) = î ĵ k̂ , where is a scalar and
x y z
î ĵ k̂ is a vector (operator)
x y z
u v w
(iv) div q . q , q = (u, v, w)
x y z
If q 0, then q is said to be solenoidal vector.
υ υ υ
(v) dr îdx ĵdy k̂dz, d= dx dy dz
x y z
and
= î ĵ k̂ ,
x y z
Therefore,
d = (). d r
î ĵ k̂
(vi) Curl q q
x y z
u v w
w v u w v u
= î ĵ k̂
y z z x x y
(vii) (a) Gradient of a scalar is a vector.
(b) Divergence of a scalar and curl of a scalar are meaningless.
(c) Divergence of a vector is a scalar and curl of a vector is a vector.
FLUID DYNAMICS 6
2 2 2
(viii) = 2 =
x 2 y 2 z 2
where 2 is Laplacian operator.
(ix) Curl grad = 0, div curl q = 0
(b) n̂ q dS curl q dv
S V
(xii) Green’s theorem
(a) dV d S 2 dV
V S V
= d S 2 dV
S V
(b) ( 2 2 )dV dS
V V n n
(xiii) Stoke’s theorem q dr curl q d S curl q n̂ dS
C S S
(xiv) Orthogonal curvilinear co-ordinates :
Let there be three orthogonal families of surfaces
f1(x, y, z) = , f2(x, y, z) = , f3(x, y, z) = (1)
where x, y, z are Cartesian co-ordinates of a point P(x, y, z) in space. The
surfaces
= constant, constant, = constant (2)
form an orthogonal system in which every pair of surfaces is an orthogonal
system. The values , , are called orthogonal curvilinear co-ordinates.
From three equations in (1), we can get
r
A tangent vector to the -curve ( = constant, = constant) at P is . A unit
tangent vector is
r
ê1
r
r
or h1ê1
2 2 2
r x y z
where h1 =
Similarly, ê 2 , ê 3 are unit vectors along -curve and -curve respectively such
that
r r
h 2 ê 2 , h 3ê 3
r r r
Further, dr d d d
= h1 d ê1 h 2 d ê 2 h 3d ê 3
Therefore,
(ds)2 = d r.d r h12d2 h 22d2 h 32d 2
1 1 1
(i) grad = , ,
h1 h 2 h 3
(ii) If q (q1 , q 2 , q 3 ), then
1
div q (h 2 h 3q1 ) (h 3 h1q 2 ) (h1h 2 q 3 )
h1h 2 h 3
(iii) If curl q (1 , 2 , 3 ), then
1
1 = (h 3q 3 ) (h 2 q 2 )
h 2h3
1
2 = (h1q1 ) (h 3 q 3 )
h 3 h1
FLUID DYNAMICS 8
1
3 = (h 2 q 2 ) (h1q1 )
h 1h 2
1 h 2 h 3 h 3 h1 h1h 2
(iv) 2 = .
h 1h 2 h 3 h1 h 2 h 3
The Cartesian co-ordinate system (x, y, z) is the simplest of all orthogonal co-
ordinate systems. In many problems involving vector field theory, it is
convenient to work with other two most common orthogonal co-ordinates i.e.
cylindrical polar co-ordinates and spherical polar co-ordinates denoted
respectively by (r, , z) and (r, , ). For cylindrical co-ordinates, h1 = 1, h2 =
r, h3 = 1. For spherical co-ordinates, h1 = 1, h2 = r, h3 = r sin .
1. Fluid Dynamics
Fluid dynamics is the science treating the study of fluids in motion. By the
term fluid, we mean a substance that flows i.e. which is not a solid. Fluids may
be divided into two categories
(i) liquids which are incompressible i.e. their volumes do not change when the
pressure changes
(ii) gases which are compressible i.e. they undergo change in volume whenever
the pressure changes. The term hydrodynamics is often applied to the science
of moving incompressible fluids. However, there is no sharp distinctions
between the three states of matter i.e. solid, liquid and gases.
In microscopic view of fluids, matter is assumed to be composed of molecules
which are in random relative motion under the action of intermolecular forces.
In solids, spacing of the molecules is small, spacing persists even under strong
molecular forces. In liquids, the spacing between molecules is greater even
under weaker molecular forces and in gases, the gaps are even larger.
If we imagine that our microscope, with which we have observed the molecular
structure of matter, has a variable focal length, we could change our
observation of matter from the fine detailed microscopic viewpoint to a longer
range macroscopic viewpoint in which we would not see the gaps between the
molecules and the matter would appear to be continuously distributed. We
shall take this macroscopic view of fluids in which physical quantities
associated with the fluids within a given volume V are assumed to be
distributed continuously and, within a sufficiently small volume V, uniformly.
This observation is known as Continuum hypothesis. It implies that at each
point of a fluid, we can prescribe a unique velocity, a unique pressure, a unique
density etc. Moreover, for a continuous or ideal fluid we can define a fluid
particle as the fluid contained within an infinitesimal volume whose size is so
small that it may be regarded as a geometrical point.
1.1. Stresses : Two types of forces act on a fluid element. One of them is
body force and other is surface force. The body force is proportional to the
9
FLUID DYNAMICS
mass of the body on which it acts while the surface force is proportional to the
surface area and acts on the boundary of the body.
Suppose F is the surface force acting on an elementary surface area dS at a
point P of the surface S. surface force
1.2. Viscosity : It is the internal friction between the particles of the fluid
which offers resistance to the deformation of the fluid. The friction is in the
form of tangential and shearing forces (stresses). Fluids with such property are
called viscous or real fluids and those not having this property are called
inviscid or ideal or perfect fluids.
Actually, all fluids are real, but in many cases, when the rates of variation of
fluid velocity with distances are small, viscous effects may be ignored.
From the definition of body force and shearing stress, it is clear that body force
per unit area at every point of surface of an ideal fluid acts along the normal to
the surface at that point. Thus ideal fluid does not exert any shearing stress.
Thus, we conclude that viscosity of a fluid is that property by virtue of which it
is able to offer resistance to shearing stress. It is a kind of molecular frictional
resistance.
1.3. Velocity of Fluid at a Point : Suppose that at time t, a fluid particle is at
the point P having position vector r (i.e.OP r )
Q( r Sr, t St)
Sr
r Sr P (r, t )
r
O
FLUID DYNAMICS 10
and at time t + t the same particle has reached at point Q having position
vector r δr . The particle velocity q at point P is
(r r ) r r dr
q Lt Lt
St 0 t St 0 t dt
where the limit is assumed to exist uniquely. Clearly q is in general
dependent on both r and t, so we may write
q q(r, t) q(x, y, z, t),
1.5. Flux across any surface : The flux i.e. the rate of flow across any surface
S is defined by the integral
(q n̂)dS
S
where is the density, q is the velocity of the fluid and n̂ is the outward unit
normal at any point of S.
Also, we define
Flux = density normal velocity area of the surface.
2. Eulerian and Lagrangian Methods (Local and Total range of change)
We have two methods for studying the general problem of fluid dynamics.
2.1. Eulerian Method : In this method, we fix a point in the space occupied by
the fluid and observation is made of whatever changes of velocity, density
pressure etc take place at that point. i.e. point is fixed and fluid particles are
allowed to pass through it. If P(x, y, z) is the point under reference, then x, y, z
do not depend upon the time parameter t, therefore x , y , z do not exist (dot
denotes derivative w.r.t. time t).
11
FLUID DYNAMICS
Let f(x, y, z, t) be a scalar function associated with some property of the fluid
(e.g. its density) i.e. f(x, y, z, t) = f( r , t ), where r x î y ĵ z k̂ is the position
vector of the point P, then
f f (r, t δt ) f (r, t )
Lt (1)
t δt0 δt
f
Here, is called local time rate of change.
t
2.2. Lagrangian Method :- In this case, observations are made at each point
and each instant, i.e., any particle of the fluid is selected and observation is
made of its particular motion and it is pursued throughout its course.
Let a fluid particle be initially at the point (a, b, c). After lapse of time t, let the
same fluid particle be at (x, y, z). It is obvious that x, y, z are functions of t.
But since the particles which have initially different positions occupy different
positions after the motion is allowed. Hence the co-ordinates of the final
position i.e. (x, y, z) depend on (a, b c) also. Thus
For this case, if f(x, y, z, t) be scalar function associated with the fluid, then
df f (r δr, t δt ) f (r, t )
Lt (2)
dt δt0 δt
where x , y , z exist.
df
Here is called an individual time rate or total rate or particle rate of change.
dt
Now, we establish the relation between these two time rates (1) & (2).
We have
f = f(x, y, z, t)
Therefore,
df f dx f dy f dz f
dt x dt y dt z dt t
f f f dx dy dz f
= î ĵ k̂ î ĵ k̂
x y z dt dt dt t
f
= f q
t
where
dx dy dz
q î ĵ k̂ = (u, v, w)
dt dt dt
FLUID DYNAMICS 12
Thus
df f
q f (3)
dt t
dF F
q F
dt t
Hence the relation (3) holds for both scalar and vector functions associated
with the moving fluid.
(iii) The Eulerian method is sometimes also called the flux method.
(iv) Both Lagrangian and Eulerian methods were used by Euler for studying
fluid dynamics.
(v) Lagrangian method resembles very much with the dynamics of a
particle
(vi) The two methods are essentially equivalent, but depending upon the
problem, one has to judge whether Lagrangian method is more useful
or the Eulerian.
3. Streamlines, Pathlines and Streaklines
3.1. Streamlines : It is a curve drawn in the fluid such that the direction of the
tangent to it at any point coincides with the direction of the fluid velocity
vector q at that point. At any time t, let q = (u, v, w) be the velocity at each
point P(x, y, z) of the fluid. The direction ratios of the tangent to the curve at
P(x, y, z) are dr = (dx, dy, dz) since the tangent and the velocity at P have the
same direction, therefore q dr 0
13
FLUID DYNAMICS
i.e. (u î v ĵ w k̂ ) (dx î dy ĵ dz k̂ ) 0
i.e. (vdy w dy) î ( wdx udz) ĵ (udy vdx)k̂ 0
i.e. vdz wdy = 0 = wdx udz = udy vdx
dx dy dz
u v w
These are the differential equations for the streamlines.
i.e. their solution gives the streamlines.
q2 q3
q1 q4
P3 Streamline
P2 P4
P1
3.2. Pathlines: When the fluid motion is steady so that the pattern of flow does
not vary with time, the paths of the fluid particles coincide with the
streamlines. But in case of unsteady motion, the flow pattern varies with time
and the paths of the particles do not coincide with the streamlines. However,
the streamline through any point P does touch the pathline through P. Pathlines
are the curves described by the fluid particles during their motion i.e. these are
the paths of the particles.
The differential equations for pathlines are
dr dx dy dz
q i.e. u, v, w (1)
dt dt dt dt
where now (x, y, z) are the Cartesian co-ordinates of the fluid particle and not a
fixed point of space. The equation of the pathline which passes through the
point (x0, y0, z0), which is fixed in space, at time t = 0 say, is the solution of (1)
which satisfy the initial condition that x = x0, y = y0, z = z0 when t = 0. The
solution gives a set of equations of the form
x = x(x0, y0, z0, t)
y = y(x0, y0, z0, t) (2)
z = z(x0, y0, z0, t)
which, as t takes all values greater than zero, will trace out the required
pathline.
3.3. Remarks : (i) Streamlines give the motion of each particle at a given
instant whereas pathlines give the motion of a given particle at each instant.
FLUID DYNAMICS 14
(ii) If we draw the streamlines through every point of a closed curve in the
fluid, we obtain a stream tube. A stream tube of very small cross-
section is called a stream filament.
If the co-ordinates of a particle of marker fluid are (x, y, z) at time t and the
particle coincided with the injection point (x0, y0, z0) at some time , where
t, then the time-history (streakline) of this particle is obtained by solving the
equations for a pathline, subject to the initial condition that x = x0, y = y0,
z = z0 at t = . As takes all possible values in the angle t, the
locations of all fluid particles on the streakline through (x0, y0, z0) are obtained.
Thus, the equation of the streakline at time t is given by
3.5. Remark: (i) For a steady flow, streaklines also coincide with streamlines
and pathlines.
15
FLUID DYNAMICS
4. Velocity Potential
i.e. dx dy dz dt = u dx + vdy + wdz where = (x, y, z, t)
x y z t
is some scalar function, uniform throughout the entire field of flow.
Therefore,
u= ,v ,w , 0
x y z t
But
= 0 = (x, y, z)
t
Hence
q u î v ĵ w k̂ î ĵ k̂ =
x y z
where is termed as the velocity potential and the flow of such type is called
flow of potential kind.
4.1. Theorem : At all points of the field of flow the equipotentials (i.e.
equipotential surfaces) are cut orthogonally by the streamlines.
Proof. If the fluid velocity at any time t be q = (u, v, w), then the equations of
streamlines are
dx dy dz
(1)
u v w
are such that the velocity is at right angles to the tangent planes. The curves
(1) and the surfaces (2) cut each other orthogonally. Suppose that the
expression on the left hand side of (2) is an exact differential, say, d, then
u= ,v ,w
x y z
i.e. q = to hold is
The surfaces (5) are called equipotentials. Thus the equipotentials are cut
orthogonally by the stream lines.
dx dy dz
u v w
where q = (u, v, w)
we have
From here,
(u, v, w) = , , = (cy, cx, 0)
x y z
dx dy dz
cy cx 0
i.e. x dx ydy = 0, dz = 0
i.e. x2 y2 = a2, z = K
4.4. Example. If the speed of fluid is everywhere the same, the streamlines are
straight.
dx dy dz
u v w
vx uy = constant, vz wy = constant
The intersection of these planes are necessarily straight lines. Hence the result.
4.5. Example. Find the stream lines and path lines of the particles for the two
dimensional velocity field.
x
u= , v = y, w = 0
1 t
dx dy dz
u v w
Therefore,
dx dy dz
(1+t)
x y 0
FLUID DYNAMICS 18
dx dy dz
u, v, w
dt dt dt
Therefore,
dx x dy dz
, y, 0
dt 1 t dt dt
dx dt dy
, dt, dz 0
x 1 t y
x = a(1+t), y = bet, z = c
x a
y be a ;z=c
4.6. Note. In case of path lines, t must be eliminated since these give the
motion at each instant (i.e. independent of t).
4.7. Example. Obtain the equations of the streamlines, path lines and
streaklines which pass through (l, l, 0) at t = 0 for the two dimensional flow
x t y
u= 1 , v , w = 0.
t0 t0 t0
where l and t0 are constants having respectively the dimensions of length and
time.
x y z t
X= , Y , Z ,T
l l l t0
1 1 1 1
such that dX = dx, dY dy, dZ dz, dT dt
l l l t0
Xl Yl
and u= (1 T), v , w=0
t0 t0
dx dy dz
u v w
t 0 l dX t l dY ldZ
0
Xl (1 T) Yl 0
dX dY dZ
X(1 T) Y 0
X = C2 Y(1+T) (2)
dX dY dZ
X(1 T), Y, 0
dT dT dT
dX T2
Therefore, (1 T)dT log X T log K1
X 2
FLUID DYNAMICS 20
2/2
X = K1 e TT (3)
dY dY
Y dT log Y T log K 2
dT Y
Y = K2 eT. (4)
dZ = 0 Z = constant = K3 (5)
These are the parametric equations of path lines. The path line through P(1, 1,
0) i.e. X = 1 = Y, Z = 0, T = 0 is obtained when K1 = K2 = 1, K3 = 0
T2
T
X =e 2 , Y = eT, Z = 0
Elimination of T gives.
T T T 1
T1 1 1 1 logY
X= e 2 e T 2
Y 2 = Y 2 ,Z 0
1 1
X = exp. T T 2 2 ,
2 2
Y = exp (T), Z = 0
These are the parametric equations of the streaklines true for all values of T.
At T = 0, the equations give
2
X = exp. , Y = exp(), Z = 0.
2
Eliminating , we have.
We know that the streamlines are obtained from the differential equations
q dr 0 (1)
where q is the velocity vector and r is the position vector of a liquid particle.
q
d r = 0 (2)
1
, , (dr, rd, dz) = 0
r r z
dr rd dz
. (3)
r 1 / r z
1 1
and = grad = , ,
r r r sin
1 1
, , (dr, rd, r sin d) = 0
r r r sin
4.9. Example. Show that if the velocity potential of an irrotational fluid motion
A
is = 2 cos, where (r, , ) are the spherical polar co-ordinates of any
r
point, the lines of flow lie on the surface r = k sin2, k being a constant.
dr rd
2A 1 A
3 cos 2 sin
r r r
dr 2rd dr cos
2 d
cos sin r sin
4.10. Note. In the above example, the velocity potential, in Cartesian co-
ordinates, can be written as
y
= A(x2 + y2 + z2)3/2 z. tan1 ,
x
where
x = r sin cos , y = r sin sin, z = r cos
r̂ r ˆ r sin
ˆ
1
curl q 2 r
r sin
2Mr cos Mr sin
3 2
0
=
1
r sin
2
r̂ 0 rˆ 0 r sin
ˆ (2Mr 3 sin 2Mr 3 sin ) = 0
υ 1 υ 1 υ
Now, using the relation q υ r̂ θ̂ ψ̂ , we have
r r θ r sin θ ψ
1 ˆ 1
2Mr 3 cos r̂ Mr 3 sin ˆ r̂ ˆ
r r r sin
From here,
2Mr 3 cos , Mr 2 sin , 0
r
Therefore,
d = dr d d
r
= d (Mr 2 cos )
Integrating, we get
= Mr 2 cos
dr 2 cos
d = 2 cot d
r sin
Integrating, we get
r = A sin2 , = constant
The equation = const. shows that the streamlines lie in planes which pass
through the axis of symmetry = 0.
= curl q q
is called the vortex vector or vorticity and it‟s components are (1, 2, 3),
given by
w v u w v u
1 = , 2 , 3
y z z x x y
5.2. Vortex Motion (or Rotational Motion). The fluid motion is said to be
rotational if
curl q 0
5.4. Vortexline. It is a curve in the fluid such that the tangent at any point on
the curve has the direction of the vorticity vector .
dx dy dz
i.e.
1 2 3
5.5. Vortex Tube. It is the locus of vortex line drawn at each point of a closed
curve i.e. vortex tube is the surface formed by drawing vortex lines through
each point of a closed curve in the fluid.
5.7. Circulation It is the flow round a closed curve. If C be the closed curve in
a moving fluid, then circulation about C is given by
= q dr = n̂ curl q dS n̂ ξ dS.
C S S
= υ dr dυ υ(A) υ(A) 0,
C C
where A is any point on the curve C. This shows that for an irrotational
motion, circulation is zero.
5.8. Theorem :-The necessary and sufficient condition such that the vortex
lines are at right angles to the stream lines, is
(u, v, w) = , ,
x y z
i.e. q 0, = curl q
This shows that the streamlines are at right angles to the vortex lines. Thus the
streamlines and vortex lines are at right angles to each other if the differential
equation q dr 0 is integrable.
But curl q curl() 0 . Thus the vortexlines do not exist. The equations
q dr 0 are therefore not exact.
1 q dr 0 is integrable.
q .
Therefore,
= (1 q q ) = 0
27
FLUID DYNAMICS
This shows that the directions of streamlines and vortexlines are at right angles
to each other.
6. Equation of Continuity
6.1. Equation of Continuity by Euler’s Method (Equation of conservation
of Mass): Equation of continuity is obtained by using the fact that the mass
contained inside a given volume of fluid remains constant throughout the
motion. Consider a region of fluid in which there is no inlets (sources) or
outlets (sinks) through which the fluid can enter or leave the region. Let S be
the surface enclosing volume V of this region and let n̂ denotes the unit vector
normal to an element S of S drawn outwards.
Let q be the fluid velocity and be the fluid density.
n̂
S
S
V
s
q n̂
(q n̂)t
Hence the rate at which fluid leaves V by flowing across the element S is
(q n̂) S.
Summing over all such elements S, we obtain the rate of flow of fluid coming
out of V across the entire surface S. Hence, the rate at which mass flows out of
the region V is
By Gaussdivergencethe
ρ(q n̂ )dS (ρq ) n̂dS F n̂ dS FdV.
S S
S V
FLUID DYNAMICS 28
ρ
div(ρq ) dV 0
V t
Since V is arbitrary, we conclude that at any point of the fluid which is neither
a source nor a sink,
ρ
.div(ρq) 0
t
ρ
i.e. ..(ρq ) 0 (3)
t
Equation (3) is known as equation of continuity.
Corollary (1). We know that
div (ρq) ρ div q .q. (grad)
Therefore, (3) takes the form
ρ
ρ( q) (q )ρ 0 (4)
t
D
Corollary (2). We know that the differential operator is given by
Dt
D
(q )
Dt t
Dρ
Therefore, from (4), we obtain the equation of continuity as .ρ( q) 0
Dt
29
FLUID DYNAMICS
Dρ
i.e. ρ div q 0 (5)
Dt
Corollary (3). Equation (5) can be written as
1 Dρ
div q 0
ρ Dt
D
(log ρ) div q 0 (6)
Dt
ρ
Corollary (4). When the motion of fluid is steady, then 0 and thus the
t
equation of continuity (3) becomes
div (ρq) 0 |Here is not a function of time i.e. = (x, y, z) (7)
Corollary (5). When the fluid is incompressible, then = constant and thus
Dρ
0.
Dt
The equation of continuity becomes
div q = 0 (8)
which is same for homogeneous and incompressible fluid.
Corollary (6). If in addition to homogeneity and incompressibility, the flow is
of potential kind such that q = , then the equation of continuity becomes
single word
div(.) = 0 .(.) = 0 2 = 0 (9)
which is known as the Laplace equation.
6.2. Equation of continuity in Cartesian co-ordinates :- Let (x, y, z) be the
rectangular Cartesian co-ordinates.
Let q u î v ĵ w k̂ (1)
and = î ĵ k̂ (2)
x y z
ρ
Then, the equation of continuity div(ρq) 0 can be written as
t
ρ
(ρu) (ρv) (ρw) 0 (3)
t x y z
ρ ρ ρ ρ u v w
i.e. u v w ρ 0 (4)
t x y z x y z
which is the required equation of continuity in Cartesian co-ordinates.
FLUID DYNAMICS 30
ρ
Corollary (1). If the fluid motion is steady, then 0 and the equation (3)
t
becomes
(ρu) (ρv) (ρw ) 0 (5)
x y z
Corollary (2). If the fluid is incompressible, then = constant and the
equation of continuity is q = 0
u v w
i.e. 0 (6)
x y z
Corollary (3). If the fluid is incompressible and of potential kind, then
equation of continuity is
2 = 0
2 2 2
i.e. 0, where q .
x 2 y 2 z 2
6.3. Equation of continuity in orthogonal curvilinear co-ordinates: Let (u1,
u2, u3) be the orthogonal curvilinear co-ordinates and ê1 , ê 2 , ê 3 be the unit
vectors tangent to the co-ordinate curves.
Let q q1 ê1 q 2 ê 2 q 3 ê 3 (1)
ê 2
ê 3
ê1
0
1
f ( h 2 h 3 f1 ) (h 3 h1f 2 ) (h 1h 2 f 3 ) (3)
h 1h 2 h 3 u1 u 2 u 3
where h1, h2, h3 are scalars.
31
FLUID DYNAMICS
ρ 1
(h 2 h 3ρq1 ) (h 3h1ρq 2 ) (h1h 2ρq 3 ) (4)
t h1h 2 h 3 u1 u 2 u 3
Corollary (1). When motion of fluid is steady, then equation (4) becomes
(h 2 h 3ρq1 ) (h 3 h1ρq 2 ) (h1h 2ρq 3 ) 0 (5)
u1 u 2 u 3
Corollary (2). When the fluid is incompressible, the equation of continuity is
( = const)
( h 2 h 3 q1 ) ( h 3 h 1q 2 ) (h1h 2 q 3 ) 0 (6)
u1 u 2 u 3
becomes
h 2 h 3 h1h 3 h1h 2
0 (7)
u1 h1 u1 u 2 h 2 u 2 u 3 h 3 u 3
Now, we shall write equation (4) in cylindrical & spherical polar co-ordinates.
6.4. Equation of continuity in cylindrical co-ordinates (r, , z) . Here,
u1 r, u2 , u3 z and h1 = 1, h2 = r, h3 = 1
The equation of continuity becomes
ρ 1
(rρq1 ) (ρq 2 ) (rρq 3 ) 0
t r r θ z
ρ 1 1
i.e. (rρq1 ) (ρq 2 ) (ρq 3 ) 0 (8)
t r r r θ z
Corollary (1). When the fluid motion is steady, then equation (8) becomes
(rρq1 ) (ρq 2 ) r (ρq 3 ) 0 (9)
r θ z
Corollary (2). For incompressible fluid, equation of continuity is
q
(rq1 ) (q 2 ) r 3 0 (10)
r z
Corollary (3). When the fluid is incompressible and is of potential kind, then
equation (8) takes the form
FLUID DYNAMICS 32
1
r r 0 (11)
r r r z z
where q ; is expressed in cylindrical co-ordinates.
Solution. Let P and Q be the points on the end sections of the stream filament.
P Q
s
f(s) f(s+s)
(ρσδs) (ρσ)δs
t t
(2)
Using law of conservation of mass, we have from (1) & (2)
(ρσ)δs (ρqσ)δs 0 | Total rate = 0
t s
i.e. (ρσ) (ρσq) 0 (3)
t s
which is the required equation at any point P of the filament.
6.8. Deduction :- For steady incompressible flow, (ρσ) 0 and equation (3)
t
reduces to
(q) 0 (q) 0 q = constant
s s
which shows that for steady incompressible flow product of velocity and cross-
section of stream filament is constant. This result means that the volume of
fluid a crossing every section per unit time is constant
dis tan ce volume
σq c σ c c
t t
6.9. Example. A mass of a fluid moves in such a way that each particle
describes a circle in one plane about a fixed axis, show that the equation of
continuity is
35
FLUID DYNAMICS
ρ
(ρω) 0 ,
t θ
where is the angular velocity of a particle whose azimuthal angle is at time
t.
Solution. Here, the motion is in a plane i.e. we have a two dimensional case
and the particle describe a circle y
r
x
Therefore, z = constant, r = constant
0, 0 (1)
z r
i.e. there is only rotation.
We know that the equation of continuity in cylindrical co-ordinates (r, , z) is
ρ 1 1
(rρq1 ) (ρq 2 ) (ρq 3 ) 0 (2)
t r r r θ z
Using (1), we get
ρ 1
(ρq 2 ) 0
t r θ
ρ 1
(ρrω) 0 , where q = q2 = r.
t r θ
() 0
t
Hence the result
6.10. Example. A mass of fluid is in motion so that the lines of motion lie on
the surface of co-axial cylinders, show that the equation of continuity is
ρ 1
(ρv θ ) (ρv z ) 0
t r θ z
where v, vz are the velocities perpendicular and parallel to z.
Solution. We know that the equation of continuity in cylindrical co-ordinates
(r, , z) is given by
FLUID DYNAMICS 36
ρ 1 1
(ρrvr ) (ρvθ ) (ρv z ) 0 , where q = (vr, v, vz)
t r r r θ z
Since the lines of motion (path lines) lie on the surface of cylinder, therefore
the component of velocity in the direction of dr is zero i.e. vr = 0
Thus, the equation of continuity in the present case reduces to
ρ 1
(ρvθ ) (ρv z ) 0
t r v z
Hence the result
6.11. Example. The particles of a fluid move symmetrically in space with
regard to a fixed centre, prove that the equation of continuity is
ρ ρ ρ
u 2 . (r 2 u ) 0 .
t r r r
where u is the velocity at a distance r
Solution. First, derive the equation of continuity in spherical co-ordinates.
Now, the present case is the case of spherical symmetry, since the motion is
symmetrical w.r.t. a fixed centre.
Therefore, the equation of continuity is
ρ 1
2 . (ρq1r 2 ) 0 0
t r r θ ψ
ρ 1
2 . (ρq1r 2 ) 0 ¸ where q1 u
t r r
ρ 1 ρ 2 1
2 . ur 2 .ρ. (ur 2 ) 0
t r r r r
ρ ρ ρ
u. 2 (r 2 u) 0
t r r r
Hence the result
6.12. Example. If the lines of motion are curves on the surfaces of cones
having their vertices at the origin and the axis of z for common axis, prove that
the equation of continuity is
2ρ cos ecθ
(ρq r ) q r (ρq ψ ) 0
t r r r ψ
Solution. First derive the equation of continuity in spherical co-ordinates (r, ,
) as
ρ 1
2 sin θ (ρq1r 2 ) r (ρq 2 sin θ) r (ρq 3 ) 0
t r sin θ r θ ψ
37
FLUID DYNAMICS
In the present case, it is given that lines of motion lie on the surfaces of cones,
therefore velocity perpendicular to the surface is zero i.e. q2 = 0
Therefore, the equation of continuity becomes.
ρ 1 1
2 (ρq r r 2 ) . (ρq ψ ) 0 where (q1, q2, q3)
t r r r sin θ ψ
(qr, q, q)
ρ 1 2 1
r r (ρq r ) ρq r (2r ) r sin θ ψ (ρq ψ ) 0
r r 2
2 cos ec
(q r ) q r (q ) 0
t r r r
Hence the result
6.13. Example. Show that polar form of equation of continuity for a two
dimensional incompressible fluid is
v
(ru) 0
r
cos
If u = , then find v and the magnitude of the velocity q, where q =
r2
(u, v)
Solution. First derive the equation of continuity in polar co-ordinates (r, ) in
two dimensions as
ρ 1 1
(ρrq1 ) (ρq 2 ) 0 |z0
t r r r θ
In the present case = constant
Therefore, the equation of continuity reduces to
ρ ρ
(ru) (v) 0, where q = (q1, q2, q3) (u, v, w)
r r r θ
v
i.e. (ru) 0
r
Hence the result.
μ cosθ μ cosθ v
Now u = r 0
r 2 r r 2 θ
μ cos θ v v μ cos θ
0
r 2 θ θ r2
Integrating w.r.t , we get
FLUID DYNAMICS 38
sin
v=
r2
and thus | q | q u 2 v 2
r2
6.14. Equation of Continuity by Lagrange’s Method. Let initially a fluid
element be at (a, b, c) at time t = t0 when its volume is dV0 and density is 0.
After time t, let the same fluid element be at (x, y, z) when its volume is dV
and density is . Since mass of the fluid element remains invariant during its
motion, we have
0 dV0 = dV i.e. 0 da db dc = dx dy dz
( x, y, z)
or 0 da db dc = da db dc
(a, b, c)
or J = 0 (1)
( x, y, z)
where J =
(a , b, c)
which is the required equation of continuity.
6.15. Remark. By simple property of Jacobians, we get
dJ
J q
dt
d dρ dJ
Thus (1) gives (ρJ) 0 J ρ 0
dt dt dt
d d D
J J q 0 q 0 or q 0
dt dt Dt
which is the Euler‟s equation of continuity.
7. Boundary Surfaces
Physical conditions that should be satisfied on given boundaries of the fluid in
motion, are called boundary conditions. The simplest boundary condition
occurs where an ideal and incompressible fluid is in contact with rigid
impermeable boundary, e.g., wall of a container or the surface of a body which
is moving through the fluid.
Let P be any point on the boundary surface where the velocity of fluid is q and
velocity of the boundary surface is u.
39
FLUID DYNAMICS
n̂
u
q
qu
P
The velocity at the point of contact of the boundary surface and the liquid must
be tangential to the surface otherwise the fluid will break its contact with the
boundary surface. Thus, if n̂ be the unit normal to the surface at the point of
contact, then
(q u) n̂ 0 q n̂ u n̂ (1)
where q u is the relative velocity and n̂ is a unit vector normal to the surface
at P.
The equation of the given surface is
F (r, t) F(x, y, z, t ) 0 (2)
We know that a unit vector normal to the surface (2) is given by
F
n̂
| F |
Thus, from (1), we get q F u F (3)
since the boundary surface is itself in motion, therefore at time (t + t), it‟s
equation is given by
F (r δr, t δt ) 0 . (4)
From (2) & (4), we have
F (r δr, t δt ) F(r, t) 0
i.e. F (r δr, t δt) F(r, t δt) + F(r, t δt) F(r, t) 0
By Taylor‟s series, we can have
(δr )F(r, t δt ) δt
{F (r, t)} = 0
t
F F F
F(x δx, y δy z δz) F(x, y, z) δx δy δz ...
x y z
= F(x, y, z) + δr F
δr F
F(r, t δt ) 0
δt t
Taking limit as t0, we get
dr F
. F 0
dt t
F DF
(q.)F 0 i.e. 0 (5)
t Dt
which is the required condition for any surface F to be a boundary surface
Corollary (1) If q = (u, v, w), then the condition (5) becomes
F F F F
u v w 0
t x y z
41
FLUID DYNAMICS
F
In case, the surface is rigid and does not move with time, then 0 and the
t
F F F
boundary condition is v w
u 0 i.e.( q )F 0
x y z
Corollary (2) The boundary condition
F F F F
u v w 0
t x y z
is a linear equation and its solution gives
dt dx dy dz D d
in Lagrangian view
1 u v w Dt dt
dx dy dz
u, v, w
dt dt dt
which are the equations of path lines.
Hence once a particle is in contact with the surface, it never leaves the surface.
Corollary (3) From equation (5), we have
F
q F
t
F F t
q
| F | | F |
F t
q n̂
| F |
which gives the normal velocity.
Also from (1), we get
F t
u n̂ q n̂ u n̂.
| F |
which gives the normal velocity of the boundary surface.
7.2. Example. Show that the ellipsoid
x2 y 2 z 2
kt n 1
a 2 k 2 t 2n b c
is a possible form of the boundary surface of a liquid.
Solution. The surface F(x, y, z, t) = 0 can be a possible boundary surface, if it
satisfies the boundary condition.
FLUID DYNAMICS 42
DF F F F F
u v w 0 (1)
Dt t x y z
where u, v, w satisfy the equation of continuity
u v w
q 0 i.e. 0 (2)
x y z
x2
n y
2
z
2
Here, F(x, y, z, t) k t 1 0
a 2 k 2 t 2n b c
F x 2 .2n
n 1 y
2
z
2
Therefore, 2 2 2n1 nkt
t a k t b c
F 2x F 2kt n y F 2kt n z
2 2 2n , 2 , 2 .
x a k t y b z c
x 2 2n
n 1 y
2
z
2
nk t
a 2 k 2 t 2n 1 b c
nx 2x ny 2kytn nz 2kzt n
or u 2 2 2n v 2 w 2 0
t a k t 2t b 2t c
which will hold. if we take
nx ny nz
u 0, v 0, w 0
t 2t 2t
nx ny nz
i.e. u= , v , w (3)
t 2t 2t
It will be a justifiable step if equation (2) is satisfied.
n n n
i.e. 0.
t 2t 2t
which is true.
Hence the given ellipsoid is a possible form of boundary surface of a liquid.
43
FLUID DYNAMICS
P(time t+t)
r r r
P(time t)
r
O A(t0=0)
rA
The points A, P, P are geometrical points of region occupied by fluid and they
coincide with the locations of the same fluid particle at times t0, t, t + t
respectively. Let f be the acceleration of the particle at time t when it
coincides with P. By definition
But the particle vel. at time t is q(r, t ) and at time t+t it is q(r r, t t ) .
(3)
Since r is independent of time t, therefore
q(r, t t ) q(r, t ) q
Lt (4)
t 0 t t
FLUID DYNAMICS 44
2
1
+ x y z . F(x,y,z) + ……
| 2 x y z
and
x y z (r ), where
x y z
ˆ ˆ ˆ
r x ˆi y ˆj z kˆ , i j k ]
x y z
But r is merely the displacement of the fluid particle in time t, therefore,
where R. H. S. of (4) & (7) are evaluated at P (r, t) . Hence, from (2), the
acceleration of fluid at P in vector form is given by
q
f (q )q (8)
t
8.1. Remark. We have obtained the acceleration i.e. rate of change of velocity
q . The same procedure can be applied to find the rate of change of any
physical property associated with the fluid, such as density. Thus, if F =
F (r, t) is any scalar or vector quantity associated with the fluid, it‟s rate of
change at time t is given by
DF F
(q )F
Dt t
45
FLUID DYNAMICS
D
The operator (q ) is Lagrangian and operators on R.H.S. are
Dt t
D
Eulerian since r is independent of t. is also called material derivative.
Dt
D
(q )
Dt t
q i q
fi = q jq i, j , where q i, j i
t x j
q 1
f q 2 q ( q )
t 2
FLUID DYNAMICS 46
q 1
= q 2 q , where curl q q .
t 2
We have
q q q
(q )q (q î ) (q. ĵ) (q k̂) (2)
x y z
A ( B C) (A C) B (A B) C
i.e.
(A B) C (A C) B A ( B C)
q
In particular, taking A q, B ˆi, C , we get
x
q q q
(q î ) q î q î
x x x
1 2 q
= î q q î (3)
x 2 x
Similarly,
q 1 q
(q j) ĵ q 2 q ĵ (4)
y y 2 y
q 1 q
(q k̂ ) k̂ q 2 q k̂ (5)
z z 2 z
1 q
(q )q q 2 q ĵ
2 x
1
= q 2 q q
2
Thus, from (1), we obtain
dq q 1
f q 2 q q
dt t 2
q 1
= q 2 q (6)
t 2
47
FLUID DYNAMICS
Now, let (u1, u2, u3) denote the orthogonal curvilinear co-ordinates.
Also let q (q1 , q 2 , q 3 , f = (f1, f2, f3) = (1, 2, 3), where the terms have
their usual meaning. We know that the expression for the operator in
curvilinear co-ordinates is
1 1 1
, , ,
h1 u1 h 2 u 2 h 3 u 3
f1
q1
1
t 2h1 u1
q12 q 2 2 q 3 2 ( 2 q 3 3q 2 )
q
f2 2
1
t 2h 2 u 2
q12 q 2 2 q 3 2 ( 3q1 1q 3 )
(8)
q
f3 3
1
t 2h 3 u 3
q12 q 2 2 q 3 2 (1q 2 2 q1 )
Here,
u1 r, u2 u3 z. and h1 = 1, h2 = r, h3 = 1
FLUID DYNAMICS 48
1
Therefore, , ,
r r z
and
1 q 1 q 3 q 2
1 = 3 (rq2 ) = .
r z r z
q q
2 = (q1 ) (q 3 ) 1 3
z r z r
1 q q q 1 q1
3 = (rq2 ) 1 = 2 2
r r r r r
Thus,
f1 =
q1 1 2
t 2 r
q q
q1 q 2 2 q 3 2 q 3 1 q1 3
z r
q 2 q 2 2 q 2 q1
q2
r r r
q1 q q q q q
= q1 1 q 2 2 q 3 3 q 3 1 q1 3
t r r r z r
q 2 q 22 q 2 q1
q2
r r r
q1 q q q q q 2
= q1 1 2 1 q 3 1 2
t r r z r
D d q
q1 2 q 3 , then
Dt dt t r r z
Dq 1 q 22 Du v 2
f1 =
Dt r Dt r
Dq2 q1q 2 Dv uv
Similarly, f2 = (9)
Dt r Dt r
49
FLUID DYNAMICS
Dq3 Dw
f3 =
Dt Dt
ordinates.
u1 r, u2 , u3 and h1 = 1, h2 = r, h3 = r sin
1 1
Therefore, , ,
r r r sin
and
1
1 = ( r sin q ) ( rq )
r 2 sin
3 2
1 q 3 q 2
r
= r cos q sin
3
r 2 sin
1 q q
= q 3 cos sin . 3 2
r sin
1
2 = (q1 ) (r sin q 3 )
r sin r
1 q1 q
= sin q 3 r sin 3
r sin r
1 1 q q
3 = (rq 2 ) (q1 ) = q 2 r 2 1
r r r r
Thus,
f1 =
q1 1 2
t 2 r
q q
q
q1 q 2 2 q 3 2 3 1 q 3 sin r sin 3
r sin r
FLUID DYNAMICS 50
q2 q 2 q1
q 2 r r
r
q 3 q 22 q q q
q3 q2 3 2 1
r r r r
Dq1 q 22 q 32 D q q
= , where q1 2 3
Dt r r Dt t r r r sin
Dq 1 q 22 q 32 Du v 2 w 2
i.e. f1 =
Dt r Dt r
ordinates.
F
A
F
Further, if we assume that Lt exists uniquely, then this limit is called the
A0 A
(hydrodynamic) fluid pressure at point P and is denoted by p.
8.7.Theorem :- Prove that the pressure p at a point P in a moving inviscid fluid
is same in all direction.
Proof :- Let q be the velocity of the fluid. We consider am elementary
tetrahedron PQRS of the fluid at a point P of the moving fluid, Let the edges of
the tetrahedron be PQ = x, PR = y, PS = z at time t, where x, y, z are
taken along the co-ordinate axes OX, OY, OZ respectively. This tetrahedron is
also moving with the velocity q of the local fluid at P.
Z S
z
P
y R
x
Q
Y
O
X
Let p be the pressure on the face QRS where area is s. Suppose that < l, m,
n> are the d.c.‟s of the normal to s drawn outwards from the tetrahedron.
Then,
1 yz
= y.z
2 2
Similarly,
1 zx
ms = area of face PQS = z.x
2 2
and
1 xy
ns = area of face PQR = x.y
2 2
The total force exerted by the fluid, outside the tetrahedron, on the face QRS is
FLUID DYNAMICS 52
= ps (l ˆi m ˆj n kˆ )
p
= (yz ˆi zx ˆj xy kˆ )
2
Let px, py, pz be the pressures on the faces PRS, PQS, PRQ. The forces exerted
on these faces by the exterior fluid are
1 1 1
p x yz ˆi, p y zxˆj, p z xykˆ respectively.
2 2 2
p 1
(yz ˆi zx ˆj xy kˆ ) p x yz ˆi
2 2
1 1
+ p y z x ˆj p z xy kˆ
2 2
=
1
2
(p x p)yz ˆi (p y p)zx ˆj (p z p)xy kˆ (1)
In addition to surface force (fluid forces), the fluid may be subjected to body
forces which are due to external causes such as gravity. Let F be the mean
body force per unit mass within the tetrahedron.
1 1
Volume of the tetrahedron PQRS is h s i.e. x y z, where h is the
3 6
perpendicular from P on the face QRS.
1
Thus, the total force acting on the tetrahedron PQRS is = e F x y z (2)
6
Where is the mean density of the fluid.
From (1) and (2), the net force acting on the tetrahedron is
1
2
1
p x p)yz ˆi (p y p)zx ˆj (p z p)xy kˆ Fxyz
6
Dq 1
Now, the acceleration of the tetrahedron is and the mass x y z of
Dt 6
fluid inside it is constant.
Thus, the equation of motion of the fluid contained in the tetrahedron is
53
FLUID DYNAMICS
1
2
1
(p x p)yz ˆi (p y p)zx ˆj (Pz p)xy kˆ Fxyz
6
1 Dq
= xyz . (f = ma)
6 Dt
i.e.
(pxp) l
1 1 Dq
s ˆi (p y p)m s ˆj (p z p)ns kˆ Fhs hs
3 3 Dt
i.e.
px = py = pz = p. (3)
Since the choice of axes is arbitrary, the relation (3) establishes that at any
point P of a moving ideal fluid, the pressure p is same in all directions.
9. Equations of Motion
9.1. Euler’s Equation of Motion of an Ideal Fluid (Equation of
Conservation of Momentum). To obtain Euler‟s dynamical equation, we
shall make use of Newton‟s second law of motion.
Consider a region of fluid bounded by a closed surface S which consists of
the same fluid particles at all times. Let q be the velocity and be the density
of the fluid.
Then d is an element of mass within S and it remains constant.
n̂
p
dM d dq
q d = d (1)
dt dt dt
The first type of forces are the surface forces which are due to the fluid exterior
to .
Since the fluid is ideal, the surface force is simply the pressure p directed along
the inward normal at all point of S.
The total surface force on S is
The second type of forces are the body forces which are due to some external
agent. Let F be the body force per unit mass acting on the fluid. Then F d
is the body force on the element of mass ed and the total body force on the
mass within is
F d (3)
dq
dt d Fd pd
dq
dt F p d 0
dq
F p 0
dt
i.e.
dq 1
F p (4)
dt
55
FLUID DYNAMICS
which holds at every point of the fluid and is known as Euler‟s dynamical
equation for an ideal fluid.
9.2. Remark. The above method for obtaining the Euler‟s equation of
motion, is also known as flux method.
d D
q ,
dt Dt t
q 1
(q )q F p (5)
t
1
But (q )q q 2 q, curl q
2
q 1 1
q 2 q F p. (6)
t 2
p p p
(ii) Cartesian Form. Let q (u, v, w ), F = (X,Y,Z) and p = , , ,
x y z
then equation (5) gives
u u u u 1 p
u v w X
t x y z x
v v v v 1 p
u v w Y (7)
t x y z x
w w w w 1 p
u v w Z
t x y z z
p 1 p p
p = , ,
r r z
dq du v 2 dv uv dw
,
dt dt r dt r dt
dq 1
F p. becomes
dt
du v 2
1 p
Fr
dt r r
dv vu 1 p
F (8)
dt r r
dw 1 p
Fz
dt z
p 1 p 1 p
p = , ,
r r r sin
are
dq du v 2 w 2 dv w 2 cot uv dw vw cot
, ,
dt dt r dt r r dt r
du v 2 w 2 1 p
Fr
dt r r
dv w 2 cot uv 1 p
F (9)
dt r r r
dw vw cot 1 p
F
dt r r sin
9.4. Remark :- The two equations, the equation of continuity and the Euler‟s
equation of motion, comprise the equations of motion of an ideal fluid. Thus
the equations
div(q) 0
t
q 1
and (q )q F p
t
are fundamental to any theoretical study of ideal fluid flow. These equations
are solved subject to the appropriate boundary and initial conditions dictated by
the physical characteristics of the flow.
J = 0 (1)
( x, y, z)
where J=
(a , b, c)
2x 2y 2z
x ,y 2 ,z 2
t 2 t t
F = (2)
dq 1 1
F p p (3)
dt
Its Cartesian equivalent is
2x 1 p
t 2 x x
2y 1 p
(4)
t 2 y y
2z 1 p
t 2 z z
2 x x 2 y y 2 z z 1 p
(5)
t 2 a t 2 a t 2 a a a
Similarly, we get
2 x x 2 y y 2 z z 1 p
2 2 (6)
t b t b t b
2
b b
2 x x 2 y y 2 z z 1 p
2 2 (7)
t c t c t c
2
c c
These equations (5), (6), (7) together with equation (1) constitute Lagarange‟s
Hydrodynamical Equations.
Solution. We note that the small bore of the tube permits us to ignore any
variation of velocity across any cross-section of the tube and to suppose that
the flow is unidirectional.
59
FLUID DYNAMICS
We u be the velocity along the tube and p be the pressure at a general point P at
distance x from the centre of force O. Also, let h be the distance of the centre
of mass G of the fluid, as shown in the figure.
l l
G P î
h
x
Equations of motion of the fluid are :
(i) Equation of Continuity : Here, q (u,0,0)
Therefore, equation of continuity becomes
u
0 u u(t ) (1)
x
(ii) Euler’s Equation : In this case, it becomes
u uu 1 p 1 p
X x
t x x x
u 1 p
x (2) | using (1)
t x
where x î is the body force per unit mass, being a positive constant.
du 1 dp
x (3)
dt dx
du x2 p
x C (4)
dt 2
du 1
(hl ) (h l ) 2 C
dt 2
FLUID DYNAMICS 60
du 1
(h+l) (h l ) 2 C
dt 2
du
h (5)
dt
But in the fluid motion all fluid particles move with the same velocity u and u
dh
=
dt
Equation (5) becomes
d2h
h (6)
dt 2
(D2 + ) h = 0
D2 + = 0 D = i
1 dp
x h
dx
1 dp
( h x )
dx
p (h x) 2
D (7)
2(1)
l2
. D
2
l 2
i.e. D = / +
2
Therefore, equation (7) becomes
p (h x) 2 l 2
/
2 2
=
2
(h x ) 2 l 2
= (h x l )( h x l )
2
9.7. Example. Homogeneous liquid is in motion in a vertical plane, within a
curved tube of uniform small bore, under the action of gravity. Calculate the
period of oscillation.
Solution. Let O be the lowest point of the tube, AB the equilibrium level of
the liquid and h the height of AB above O. Let and be respectively the
inclinations of the tube to the horizontal at A and B and be the inclination of
the tube at a distance s along the tube from O. Let a and b denote the arc
lengths of OA and OB respectively and suppose that at time t, the liquid is
displaced through a small distance z along the tube from its equilibrium
position.
Due to the assumption of uniform small bore the flow is unidirectional along
the tube. y
B A
z
N h
u is independent of s
Euler‟s equation of motion becomes
u u 1 p
u g sin
t s s
du u 1 dp
g sin
dt t ds
du dy 1 dp
i.e. g (2)
dt ds ds
ds
dy
dx dy
sin =
ds
du p
s gy C (3)
dt
du
(a + z) g(h z sin ) C
dt
du
(bz) g(h z sin ) C
dt
du
(a +b) g z(sin sin ) (4)
dt
Since
dz du d 2 z
u= ,
dt dt dt 2
equation (4) becomes
d2z
(a +b) gz(sin sin )
dt 2
d2z
z, (5)
dt 2
g(sin sin )
where =
ab
We observe that equation (5) represents the simple harmonic motion. It‟s
period T is given by
1
2 ab 2
T= 2 .
g(sin sin )
where q is velocity, F is the body force, p and are pressure and density
respectively.
F be conservative so that it can be expressed in terms of a body force potential
function as
F = (2)
q
When the flow is steady, then 0 (3)
t
Therefore, in case of steady motion with a conservative body force equation
(1), on using (2) and (3), gives
FLUID DYNAMICS 64
1 1
q 2 q p
2
dq q
(q.)q
dt t
dq q 1 q
or q 2 q and 0
dt t 2 t
1 1
q 2 p q (4)
2
Further, if we suppose that the liquid is barotropic i.e. density is a function of
pressure p only, then we can write
1 dp
p
Using this in (4), we get
1 dp
q 2 q . (5)
2
Multiplying (5) scalarly by q and noting that
q (q ) (q q ) 0 , we get
1 dp
q q 2 0 (6)
2
If ŝ is a unit vector along the streamline through general point of the fluid and
s measures distance along this stream line, then since ŝ is parallel to q ,
therefore equation (6) gives
sˆ is parallel to q
1 2 dp
q 0 q k sˆ
s 2
sˆ
s
Hence along any particular streamline, we have
1 2 dp
q C (7)
2
where C is constant which takes different values for different streamlines.
Equation (7) is known as Bernaull‟s equation. This result applies to steady
flow of ideal. barotropic fluids in which the body forces are conservative.
Now, if ŝ is a unit vector taken along a vortexline, then, similarly, we get
65
FLUID DYNAMICS
1 2 dp
q C along any particular vortexline. (Here, we
2
multiply scalarly by )
10.2. Remark. (i) If q 0 i.e. if q & are parallel, then streamlines and
vortex lines coincide and q is said to be Beltrami vector.
1 2 dp
q C
2
dp p
.
p 1 2
q C
2
10.3. For Unsteady Irrotational Flow. Here also, we suppose that the body
forces are conservative i.e. F =
For irrotational flow, q q 0
The equation of motion
q 1 1
q 2 q ( q ) F p (1)
t 2
1 1
q 2 p
t 2
FLUID DYNAMICS 66
1 dp
q 2 0 | Barotropic fluid.
2 t
Integrating, we get
1 2 dp
q f (t ) (2)
2 t
10.4. Example. A long straight pipe of length L has a slowly tapering circular
cross section. It is inclined so that its axis makes and angle to the horizontal
with its smaller cross-section downwards. The radius of the pipe at its upper
end is twice that of at its lower end and water is pumped at a steady rate
through the pipe to emerge at atmospheric pressure. It the pumping pressure is
twice the atmospheric pressure, show that the fluid leaves the pipe with a speed
U give by
32
U2 = gL sin ,
15
ˆ ˆ ˆ
F q ˆj i j k
x y z
g gy
y
p 1 2
Bernoulli‟s equation, q C | For water is const.
2
p 1 2
becomes q gy C (1)
2
Applying this equation of the two ends of the pipe, we get
2a
P L
a
O
L
y
y = L sin
2 1 2 1
.V gL sin U 2 (2) |for
2 2
lower end y = 0
Let a and 2a be the radii of the lower and upper ends respectively, then by the
principle of conservation of mass
(2a)2V = a2 U
U
V= (3)
4
From (2) and (3), we obtain
1 U2 1
+ gL sin U2
2 16 2
FLUID DYNAMICS 68
1 2 U2
U gL sin .
2 16
15 2
U = +gL sin.
32
32
U2 = gL sin
15
10.5. Example. A straight tube ABC, of small bore, is bent so as to make the
angle ABC a right angle and AB equal to BC. The end C is closed and the tube
is placed with end A upwards and AB vertical, and is filled with liquid. If the
end C be opened, prove that the pressure at any point of the vertical tube is
instantaneously diminished one-half. Also find the instantaneous change of
pressure at any point of the horizontal tube, the pressure of the atmospheric
being neglected.
Solution. Let AB = BC = a
A
z
M
P
u
x
u
B x Q C
When the liquid in AB has fallen through a distance z at time t, then let P be
any point in the vertical column such that
AM = z, BP = x, BM = az
If u and p be the velocity and pressure at P, then equation of motion is
u u 1 p
u g (1) |u u(x, t)
t x x
and equation of continuity is
u
0 i.e. u = u(t)
x
69
FLUID DYNAMICS
p u
i.e. ( x a z) g (3)
t
Now, we take a point Q in BC, where BQ = x and let u, p be the velocity and
pressure at Q, then
p' u '
( x 'a ) | z = 0 and g is not effecting (4)
t
Equating the pressure at B, when x = 0, x = 0, we get
u u '
(az) g a | From (3) & (4)
t t
u
= a | u = u
t
Initially, when C is just opened, then z = 0, t = 0 and we have
u u
a g a
t t 0 t t 0
u g u
i.e. g / 2 (5)
t t 0 2 t 0
p0 u
(x a ) g | p 0 (p) t 0
t 0
g
= (x a )
2
1
p0 = g(a x ) (6)
2
But when the end C is closed, the liquid is at rest and the hydrostatic pressure
at P is
1
p 0 ' g(a x' )
2
When the end C is closed, the initial pressure (hydrostatic) p2 at Q (or B or C)
is g a .
Therefore, instantaneous change in pressure
1 1
= p 2 p 0 ' ga g(a x' ) = g(a x' )
2 2
10.6. Example. A sphere is at rest in an infinite mass of homogeneous liquid of
density , the pressure at infinity being . Show that, if the radius R of the
sphere varies in any manner, the pressure at the surface of the sphere at any
time is
d 2 (R 2 ) dR
2
+
2 dt 2 dt
Solution. In the incompressible liquid, outside the sphere, the fluid velocity q
will be radial and thus will be a function of r, the radial distance from the
centre of the sphere (the origin), and time t only.
71
FLUID DYNAMICS
1 d 2
(r u ) 0 (1)
r 2 dr
q (u,0,0), u u(r, t ), ,0,0
r
1
q 2 (r 2 u).
r r
i.e. sphericalsymmetry.
p=
R
R
The pressure equation for irrotational non-steady fluid motion in the absence of
body forces is
p 1 2
q C(t )
2 t
FLUID DYNAMICS 72
p 1 2
i.e. u C(t ). (4)
2 t
where C(t) is a function of time t.
As r , p , u = f/r2 0, 0
so that C(t) = / for all t (5)
Therefore, from (2), (3), (4) and (5), we get
2
p 1 R 2 R
f / r 2 (6)
t 2 r
f d 2 R 2 2RR
2
But (R R ) R
t dt
p 1
R
R 2 1 R 2
2R R 2 R
2
p 1 R 2
2R 2 R R
2
1 2 )
= (3R 2R R (7)
2
Now,
R (2RR ) (R ) 2
d 2 (R 2 ) 2 d
2
dt dt
2R 2 ) R 2
= ( 2R R
3R 2
= 2R R
1 d 2 (R 2 ) dR
2
p = +
2 dt 2 dt
rest and there is a cavity in the form of the sphere r = a in it, show that the
cavity will be completely filled after an interval of time a5/3(10)1/2.
Solution. The motion is entirely radial and consequently irrotational and the
present case is the case of spherical symmetry. The equation of continuity is
1 d 2
2
(r u) 0 r 2 u = constant = f(t) (1)
r dr
= v (say)
On the surface of the sphere, r = R, R
Therefore,
r2 r = f(t) = R2 R
f (t ) R 2 R
R
2RR R 2 dv 2Rv 2
dt
f (t ) dv dv dR
2v 2 R 2v 2 R
R dt dR dt
dv
2v 2 Rv (2)
dR
The Euler‟s equation of motion, in radial direction, using r u , is
u u 1 p
u Fr
t r r
u f ( t ) f ( t )
But 2 2 , Fr r 7 / 3
t t r r
f (t) 1 2 p
u (3)
r 2 r 2 r 7 / 3 r
Let us assume that the cavity has radius R at time t and its velocity then is
R v. Integrating (3) over the whole liquid (r = R to r = ) at time t, we
obtain
f ( t ) 1 2
0
3 1 p
u
r R 2 v 4 r 4 / 3 R R
FLUID DYNAMICS 74
we get
f (r ) 1 2 3 1
v
R 2 4 R4 3
dv 3 1
2Rv 3v 2 | using (2)
dR 2 R4 3
dv 3
2R3v 3R 2 v 2 R 2 / 3
dR 2
d(R 3 v 2 ) 3m 2 / 3
R
dR 2
Integrating, we get
9m 5 / 3
R3 v2 = A R (4)
10
v = 0, which gives A = 9 a 5 / 3 .
When R = a, R
10
< 0 because as the cavity fills, R decreases with time.
Now, we take v = R
1/ 2
dR 9
1/ 2
a5/3 R5/3
10 3
dt R
Therefore,
1/ 2
9 0 R 3 / 2 dR
t a
10 a 5/3
R 5/3
1/ 2
6a 5 / 3 / 2
0 sin d | R = a sin i.e. R = a (sin )
2 5/3 5/3 2 6/5
=
5
3a 5 / 3
=
10
Thus,
t = a5/3 (10)1/2 .
75
FLUID DYNAMICS
t t
I Lt
t 0 t
Fdt
t t
(1)
P Lt
t 0 t
pdt
We note that finite body forces such as gravity do not contribute to the
impulsive body force I .
To determine the equation of impulsive motion, we consider the Euler‟s
equation
Dq dq q 1
(q )q F p
Dt dt t
t t 1 t t
= Lt
t 0 t
Fdt Lt t
t 0
p dt (2)
1
q 'q I P (3)
where q and q' denote respectively the fluid velocity before and after the
impulsive action.
Thus, the equation of impulsive motion is
(q 'q ) I - P (4)
P
(uu) = X
x
P
(vv) = Y
y
P
(ww) = Z
z
where
becomes
Further, if the motion is irrotational, then q , q' ' , where and
denote the velocity potential just before and just after the impulsive action,
then (5) becomes
1
q I P (7)
| In (3), put q 0 and q ' q
77
FLUID DYNAMICS
P = q (8)
P = (9)
Equations (6) and (9) give the relation between impulsive pressure P and the
velocity potential .
Equation of motion is
v' v' 1 p
v' | No body force
t r' r'
f ( t ) v' 1 p
or v'
r' 2
r ' r '
f ( t ) 1 2 p
v' C
r' 2
FLUID DYNAMICS 78
f ( t ) 1 2 p
Thus v' (2)
r' 2
f ( t ) a
/ i.e. f ( t ) (3)
a
The velocity v vanishes just after annihilation, so from (2) and (3), we get
a p a
p
r' r'
a a
p p 1
r r
where r is the radius of the inner surface and v is the velocity there.
Integrating (5), we get
P = v (r2/r) + C1
f ( t ) 1 2
v (7)
r 2
d dv dr dv dr
From (4), f (t ) (r 2 v) r 2 v.2r r 2 2rv2
dt dt dt dr dt
f ( t ) dv
rv 2v 2
r dr
Thus (7) becomes
dv 1
rv 2v 2 v 2
dr 2
dv 3 2
or rv v
dr 2
dv 2 2
2r3v 3v 2 r 2 r |Multiplying by r2
dr
d( r 3 v 2 ) 2 2
r
dr
Integrating, we get
2 3
r3v2 = r C2
3
2a 3
Since r = a, v = 0 so we find C2 =
3
2 3 3
Therefore, r3 v2 = (a r )
3
The velocity v at the surface of the sphere r = a/2, on which the liquid strikes,
is
2 a 3 (a / 2) 3 14
v2 =
3 (a / 2) 3 3
14 a 2
P= . (8)
4 3 r'
which determines the impulsive pressure at a distance r‟ from the centre of the
sphere.
FLUID DYNAMICS 80
Thus, the impulsive pressure at the surface of the sphere of radius a/2 is given
by
14 a 2
P= 7a 2 / 6
4 3 a/2
When motion is the same in all planes parallel to xy plane (say) and there is no
velocity parallel to the z-axis, i.e. when u, v are functions of x, y, t only and w
= 0, we may regard the motion as two-dimensional and consider only the
cases confined to the xy plane. When we speak of the flow across a curve in
this plane, we shall mean the flow across unit length of a cylinder whose trace
on the xy plane is the curve in question, the generators of the cylinder being
parallel to the z-axis.
For a two-dimensional motion in xy-plane, q is a function of x, y, t only and
the differential equation of the streamlines (lines of flow) are
dx dy
i.e. vdx udy = 0 (1)
u v
u v
0 (2)
x y
We note that equation (2) is the condition of exactness of (1), it follows that (1)
must be an exact differential, d(say). Thus
vdx udy = d = dx dy
x y
so that u= ,v
y x
From the above discussion, it is clear that the existence of stream function is
merely a consequence of the continuity and incompressibility of the fluid. The
81
FLUID DYNAMICS
stream function always exists in all types of two dimensional motion whether
rotational or irrotational. However, it should be noted again that velocity
potential exists only for irrotational motion whether two dimensional or three
dimensional.
sin cos ds u , v
C y x y x
dx dy dx dy
ds cos , sin
C x ds y ds ds ds
= dx
dy
C x y
= d = (B A)
C
where A and B are the values of at the initial and final points of the curve.
Thus, the difference of the values of a stream function at any two points
represents the flow across that curve, joining the two points.
12.2. Corollary. If we suppose that the curve C be the streamline, then no fluid
crosses its boundary, then
FLUID DYNAMICS 82
(B A) = 0 B = A
= î ĵ
y x
= î ĵ
y x
= ( ĵ k̂) (î k̂)
y x
= î ĵ k̂
x y
= k̂ = grad k̂
i.e. grad = (grad ) k̂ = k̂ grad
83
FLUID DYNAMICS
i.e. = k̂ (4)
Again, from (3), we note that
x x y y
0
x x y y
i.e. = 0 (5)
Thus, for irrotational incompressible two-dimensional flow (steady or
unsteady), (x, y), (x, y) are harmonic functions and the family of curves
= constant (equipotentials) and = constant (streamlines) intersect
orthogonally.
12.4. Exercise. Show that u = 2c xy, v = c(a2 + x2 y2) are the velocity
components of a possible fluid motion. Determine the stream function and the
streamlines.
12.5. Remark. We shall consider the study of two dimensional motion later
on. At present we continue discussing three dimensional irrotational flow of
incompressible fluids.
FLUID DYNAMICS 84
UNIT – II
n̂
P 1
d r
S
dS r 2
2 dS = r2d (1)
d 1
85
FLUID DYNAMICS
Now, by definition
dS
1
S
dS
dS 4r 2 S
S
1 1
2
4 S
= r 2 d d
4r S
1 1 dS
r 4 S r
d
4 S r r 2
1
2
= dS (2)
4r S r
| r2 is constant on S
Since the normal n̂ to the surface is along the radius r, therefore on S, we have
.n̂ (3)
r n
From (2) & (3), we find
1
. n̂dS
r 4r 2 S
1
= div()d | Gauss theorem
4r 2
1
= 2
2 d 0 , | 2 = 0
4r
where is the volume enclosed by the surface S.
Thus 0 = constant.
r
This shows that is independent of choice of r and hence mean value of is
same over all spherical surfaces having the same centre P. When S shrinks to
point P, then = (P)
Proof. Let P be any interior point of the fluid and Q be a neighbouring point
also lying in the fluid. Let us take the direction of x-axis along the direction of
q at P. Let qP and qQ denote the speed of particles at P & Q respectively.
2
Then q 2P
x P
2
2 2
and q Q2
x Q y Q z Q
q
P x
qP
Since 2 = 0 (2) = 0 2 0
x x
satisfies Laplace equation. Therefore, by mean value theorem
x
(corollary), cannot be maximum or minimum at P. Thus, there are points
x
such as Q in the neighbourhood of P such that
2 2
qQ qP
2 2
Q P
x x
1.6. Corollary. In steady irrotational flow, the pressure has its minimum value
on the boundary.
Equation (1) shows that p is least when q2 is greatest and by above theorem, q2
is greatest at the boundary. Thus, the minimum value of p must occur only on
the boundary.
1.7. Note. The maximum value of p occurs at the stagnation points, where q =
0.
where M and C are constants, provided that the liquid is at rest at infinity.
Proof. Suppose that the volume of fluid acrossing each of internal surfaces
contained within , per unit time, is a finite quantity say 4M (i.e. 4M
represents the flux of fluid across or S). Since the fluid velocity at any point
of is radially outwards, the equation of continuity gives
R
d = 4M (1)
R
But d = R2d
n̂
S2
S1
S3
d
SN
d
FLUID DYNAMICS 88
Therefore,
1 2
4 R
R d M
1 M
4 R
d 2
R
1 M
4 R
d 2
R
Integrating w.r.t. R, we get
1 M
4
d
R
C
where C is independent of R.
1 d M
2 C
4 R R
d
M
C
4R 2 R
M
C (2)
R
To show that C is an absolute constant, we have to prove that it is independent
of co-ordinates of centre of sphere . Let the centre of the sphere be
displaced by distance x in an arbitrary direction while keeping R constant,
then from (2),
C
(3)
x x
| R is constant
1 1
Also, dw dw
x x 4 4 x
= 0, since 0 on when R as the liquid is at rest at
x
infinity.
89
FLUID DYNAMICS
T= dS,
2 S n
1
2
= (.)d | q
S1 n̂
S3
S2
FLUID DYNAMICS 90
=
2S (. n̂ )dS = dS, where S = S0 + S1 + S2 +…+ Sn
2 S n
denotes the sum of the outer boundary surface S0 and the inverse boundaries
S1, S2,…, Sn and n̂ is unit normal to S drawn out of the fluid on each
boundary.
Also T =
2S dS, where n̂ is unit normal to S drawn inside the fluid on
n
each boundary.
Proof. Let T be the K.E., q be the fluid velocity and be the velocity potential
of the given irrotational motion. Let be the region occupied by the fluid and S
be the surface of this region, then
T=
2 q 2d = ()2 d
2
= dS
2 S n
(1)
Let T1 be the K.E. and q1 be the velocity of any other motion of the fluid
consistent with the same normal velocity of the boundary S (or consistent with
the same kinetic boundary condition)
For both the motions, the continuity equation is satisfied i.e.
q = 0 = q1 (2)
i.e. q n̂ q1 n̂
91
FLUID DYNAMICS
2
T1 T = (q12 q2) d
=
2
2q (q
1 q ) (q1 q ) 2 d
=
2 2q (q1 q )d (q1 q ) 2 d
2
2
= . (q1 q ) d + (q1 q ) 2 d (4)
From vector calculus, we have
[ (q1 q ) ] = (q1 q ) + (q1 q )
>0
T1 > T .
Hence the theorem.
FLUID DYNAMICS 92
Proof. Let be a large surface enclosing the surface (s) S and be the region
bounded by S internally and by externally.
Using the result of K.E. for finite liquids, we find that the K.E. T* for finite
region is given by
S1 S2
S3
SN
T* = dS dS
2 S n 2 n
(1)
n̂ dS 0 dS 0
SU SU n
dS dS 0 (2)
S n n
dS 0 (4)
n
= dS | Using (4)
2 S n
Hence the result
2.3. Remark. We note that the K.E. for finite and infinite liquid has the same
expression.
Proof. If possible suppose that acyclic irrotational motion is possible and let
be the velocity potential. Then, K.E. of the fluid is
T
2 () 2 d dS
2 S n
(1)
Where S is the sum of all the rigid boundaries when is finite or the sum of
internal rigid boundaries when is infinite.
Now, since the boundaries are rigid, then at every point of S, the normal
velocity is zero
i.e. 0 at each point of S (2)
n
q d = 0 q = 0 q 0 at each point of .
2 2
liquid is at rest.
Proof. If possible, assume that the motion remains irrotational, then the K.E. is
given by
1 1
T= q 2 d dS (1)
2 2 S n
q 0 at each point of S.
= constant at each point of S.
0 constant at each point of S.
n
q 0 in
there is no motion.
Thus the motion is no longer irrotational.
Since the kinetic conditions at the boundaries are satisfied by both flows,
therefore at each point of S,
1 2
(2)
n n
Let = 1 2
95
FLUID DYNAMICS
1 2
2 = 21 22 = 0 at each point of fluid. and 0 at
n n n
each point of S.
represents a possible irrotational motion.
2 (q ) 2 d dS 0
2 S n n
0
Theorem II. If the region occupied by the fluid is infinite and fluid is at rest at
infinity, prove that only one irrotational motion is possible when internal
boundaries have prescribed velocities.
Proof. If possible, let there be two irrotational motions given by two different
velocity potentials 1 & 2. The conditions on boundaries are
1 2
(1)
n n
1 2
0 | using (1)
n n n
q n̂ 0 q 0 on the surface
Also,
FLUID DYNAMICS 96
q 1 2
= q1 q 2 0 at | using (2)
2.7. Remark. The above two uniqueness theorems are useful in finding
solutions of 2 = 0 subject to prescribed boundary conditions.
P(x,y,z)
(r, , )
r
0
Y
X Q
1 2 1
(r q r ) (sin q ) 0 (1)
r r
2
r sin
1
qr = , q =
r r
97
FLUID DYNAMICS
1 2 1
r sin 0 (2)
r 2 r r r 2 sin
2 1
r (R) sin (R) 0
r r sin
2 R R
r sin 0
r r sin
1 d 2 dR 1 d d
r sin (4)
R dr dr sin d d
1 d 2 dR
r n (n 1) (5)
R dr dr
and
d d
sin n (n 1) sin 0 (6)
d d
dR
R = rm m r m1
dr
Thus (5)
r m dr
1 d 2 m1
r mr
n(n 1)
m
dr
d m1
r r m n(n 1)
FLUID DYNAMICS 98
m (m+1) rm = rm n(n+1)
(m2 + mn2n) = 0
(mn) (m+n+1) = 0
m = n or m = (n+1)
cos =
d d d d
sin
d d d d
d d
sin sin ( sin ) +n (n+1) sin 0
d d
d 2 d
sin n(n 1) 0
d d
d d
(1 cos2 ) n(n 1) 0
d d
d d
(1 2 ) n(n 1) 0 (8)
d d
Equation (8) is a Legendre‟s Equation and possesses a solution known as
Legendre Function of the first kind Pn()
Therefore,
= Pn()
Hence the general solution of (3) is of the form
(r, ) = R(r) ()
= [An rn + Bn r(n+1)] Pn (cos ) (9)
99
FLUID DYNAMICS
( complete solution is the sum of all such solutions i.e. …….)
n 0
3.1. Uniform Flow. Consider the flow which corresponds to a potential given
by (9) with
An = US1n, Bn 0, (n = 0, 1, 2, ……) | Sij is knonecker delta
Sii = 1 Sij = 0 for i j
Where U is a constant.
Since P1 (cos) = cos, equation (9) becomes
(r, ) = Ur cos Uz | z = r cos
Thus
q = = k̂ U k̂
z
which is a uniform streaming motion of the fluid with speed U along the
direction of z-axis or the axis = 0.
U cos 1 0 i.e. velocity normal to
r r r
sphere is zero at r = a
1
= + Ucos (r = a, a )
r
and
|1| 0 as r.
Hence a suitable form of function 1 is
1 = B r 2 cos
So, we assume (in view of (9)) that
B
(r, ) = Ur cos cos (1)
r2
The constant B is to be determined from the fact that there is no flow normal to
the surface r = a i.e. 0
r r a
2B 1
U cos + 3
cos 0 B Ua 3
a 2
a3
= U r 2 cos (2)
2r
Now, the uniqueness theorem II infer that the velocity potential in (2) is
unique.
The velocity components at P(r, , ), (r a), are
a3
qr = U1 3 cos
r r
1 a3
q =
r
U 2r 3 sin
1 (3)
101
FLUID DYNAMICS
1
q = 0.
r sin
(i) Stagnation Points : Stagnation points are those points in the flow where the
velocity vanishes i.e. q 0 . Thus these points are obtained by solving the
equations
a3
U1 3 cos 0
r
and (4)
a3
U1 3 sin 0
2r
1 2r 3 a 3
dr 2 cot d
r r 3 a 3
2r a 3 r 2
2 3 1
dr 2 cot d
r a r
Integrating, we get
log (r2 a3 r1) = 2 log sin + log C
FLUID DYNAMICS 102
r3 a3
log log sin2 + log C
r
Cr
sin2 = , where C 0
r a3
3
p 1
()2 = C
2
At infinity, p = p and = U k̂ , we get
p 1 2
C= U
2
Thus
1 1
p = p + U 2 ()2
2 2
p = p +
1 2 1 2 a3
2 2
2 a3
U U 1 3 cos U 1 3 sin 2
2
2 2 r 2r
| = q
1 2 a 3
2 2
a3
p = p U 1 3 cos 1 3 sin 2 1
2
(5)
2 r 2r
which gives the pressure at any point of the fluid. Of particular interest is the
distribution of pressure on the boundary of the sphere. It is obtained by putting
r = a in (5) and thus
1 2
2
a3
p = p U 1 3 sin 2 1
2 2r
1 9 1
= p U 2 sin 2 1 p U2 (49 sin2)
2 4 8
103
FLUID DYNAMICS
1
= p + U2 (9cos2 5)
8
The maximum pressure occurs at the stagnation points, where = 0 or . Thus
1
pmax = p + U2
2
(pmax. is also called stagnation pressure)
The minimum pressure occurs along the equatorial circle of the sphere where
=/2
Therefore,
5
pmin. = p U2
8
A fluid is presumed to be incapable of sustaining a negative pressure, thus
8p
pmin. = 0 U =
5
At this stage the fluid will tend to break away from the surface of the sphere
and cavitation is said to occur. i.e. a vacuum is formed.
(iv) Thrust on the Hemisphere : Now, we find the thrust (force) on the
hemisphere on which the liquid impinges, r = a, 0 /2.
pmin
The component of thrust on S is p cos S. Hence the total thrust on the
hemisphere is along ZO and is given by
a sin
a a sin l
d
a
l = r ad
FLUID DYNAMICS 104
/2
= p cos (2a sin) (ad)
0
/2
1
= (2a2) sin cos p U 2 (9 cos2 5)d
0 8
(using value of p at boundary)
1
= a2 p U 2 .
16
3.3. Sphere in Motion in Fluid at Rest at Infinity. Let a solid sphere of
radius „a‟ centred at 0 be moving with uniform velocity U k̂ in
incompressible fluid of infinite extent, which is at rest at infinity. Z-axis is the
axis of symmetry and k̂ is unit vector in this direction. (As the sphere is
moving with velocity U k̂ the relative velocity of fluid if the sphere be
considered to be at rest is U k̂ .)
2 = 0 (1)
such that = U cos, (r = a) (2)
r
and
|) 0, (r) (3)
The present case is also a problem with axial symmetry about the axis = 0, ,
so
= (r, )
Also, since P1(cos) = cos | Legendre‟s function
and the boundary condition (2) implies that the dependence of on must be
like cos, therefore has the form
B B
= Ar 2 P1 (cos ) Ar 2 cos
r r
However, to satisfy (3), it is necessary that A = 0, and then from (2), we get B
1
= Ua3.
2
105
FLUID DYNAMICS
Ua 3
= cos (4)
2r 2
Ua 3 1 Ua 3
qr = cos , q sin, q = 0,
r r3 r 2r 3
2 S n
T1 = dS (5)
Where n̂ is the outwards unit normal. But for the sphere n̂ is along radius
vector
Therefore,
n S r r a
1
= Ua cos (U cos)
2
FLUID DYNAMICS 106
1 2
= U a cos2
2
Therefore,
1 2 aU 2
2 S 2 0
T1 = U a cos2
dS cos2 (2a sin) (ad)
4
a 3 U 2
0 cos sin d
2
=
2
0
0 2
a 3 U 2 cos3
= 3
2 0
4 U
2
1
= a 2 U 2 a 3 .
3 3 4
1
= M1U 2 (6)
4
4
where M1 = a3 is the mass of the liquid displaced by the sphere.
3
Also, K.E. of the sphere moving with speed U is given by
1
T2 = M U2 (7)
2
4
where M = a3 is the mass of the sphere, being the density of the
3
material of the sphere.
Therefore, from (6) and (7), total K.E. T is given by
1 M' 2
T = T1 + T2 = M U (8)
2 2
M'
The quantity M + is called the virtual mass of the sphere.
2
3.4. Accelerating Sphere Moving in a Fluid at Rest at Infinity. The solution
derived above for is applicable when the sphere translates unsteadily along a
straight line. In the present case, we take U = U(t) and get the velocity
potential as
U ( t )a 3
= (r, , t) = cos (1)
2r 2
107
FLUID DYNAMICS
The instantaneous values of velocity components and K.E. at time t are given
by
U ( t )a 3 U ( t )a 3
qr = cos , q sin, q = 0 | similar to steady case
r3 2r 3
1 1 2
and T= M M' U (t) (2)
2 2
The pressure at any point of the fluid is obtained by using Bernoulli‟s equation
for unsteady flow of a homogeneous liquid, in the absence of body force, as
p 1 2
U f (t) (3)
2 t
1 a 3 (U r)
= (5)
2 r3
since r is the position vector of a fixed point P of the fluid relative to the
moving centre 0 of the sphere, it follows that
U (r ) (6)
t
r r
Also, since r2 = r r r r r U |using (6)
t t
= (r ) ( U k̂ )
= rU (r̂ k̂) | r r r̂
= r U cos
FLUID DYNAMICS 108
r
U cos (7)
t
Differentiating (5) w.r.t. time t and using (6) & (7), we get
1 3 U 2 cos U 3U 2
a 3 2 3 cos2
t 2 r r t r
a3 U cos U 2 3U 2 cos2 U
= 2 3 U
2 r r r3 t
Also,
U 2a 6 U 2a 6
U 2 q 2r q 2 6
cos 2
6
sin 2
r 4r
U2a 6 2 1
= 6
cos sin 2
r 4
The pressure at any point of the fluid can be obtained from equation (4).
In particular, at a point on the sphere r = a
1
t
2
Ua cos U 2 3U 2 cos2
U2
and U (4 cos2 + sin2)
4
and the corresponding pressure is given by
p p 1 1
Ua cos U 2 (9 cos2 5) (8)
2 8
The force (thrust) acting on the sphere is given by
F 0 p cos(2a sin )(ad)k̂
1 1
= 2a2 k̂ 0 p U a cos U 2 (9 cos2 5) cos sin d
2 8
=
2
a 3U k̂ 1 4 a 3 U k̂ 1 M' U
k̂
3 23 2
4
where M = a3 is mass of the liquid displaced. This shows that the force
3
acts in the direction oppositing the sphere‟s motion.
109
FLUID DYNAMICS
3.5. Equation of Motion of the Sphere. Let R be the external force per unit
mass in the direction of motion of the sphere. Let us use the result that the rate
of doing work is equal to the rate of increase in K.E.
d 1 d M' 2
Thus RU = M U ( t )
dt 2 dt 2
| From (2)
M' dU
= M U
2 dt
dU 1 dU
M R M' (9)
dt 2 dt
If the liquid is not there, then M = 0 and the equation of motion of the sphere
is
dU
M R (10)
dt
Comparing equation (9) & (10), we note that the presence of the liquid offers a
1 dU
resistance of the amount M' to the motion of the sphere
2 dt
Let R be the external force per unit mass on the sphere when there is no liquid,
then
MR = external force on the sphere in the presence of the liquid.
M' dU M M'
or M R ' R '
2 dt M
FLUID DYNAMICS 110
dU M M'
M R' R ' (12)
dt M' 1
M
2 2
This is the required equation of motion of a sphere in a liquid at rest at infinity.
From equations (10) & (12), we note that the effect of the presence of the
liquid reduces the external force in the ration : + .
2
3.6. Remark. We have already studied the impulsive actions in Unit-I, where,
we had derived the relation between the impulsive pressure P and the velocity
potential as P = . Here, we derive the expression for K.E. generated due to
impulsive action.
T=
2 S
dS
n
(1)
n̂. U i (2)
n
| q
using (2) in (1), we get
m
T= U i .S i n̂dS
2 i 1
(3)
Ri = Si n̂ P dS Si n̂ dS | P =
(4)
Thus from (3) & (4), we get
1m
T = Ui R i (5)
2 i1
111
FLUID DYNAMICS
2Ua 3 b 3
k̂
b3 a 3
Also calculate the corresponding K.E. generated by the impulsive motion.
Solution. The motion generated in the fluid is irrotational
q 2 0 which is the equation of continuity. The boundary
conditions which must satisfy, are
U cos (r = a) (1)
r
0 (r = b) (2)
r
with (r, , ) spherical
polar co-ordinates and
with = 0 along the
direction of k̂ .
The form of boundary conditions suggest a
solution of the form
= (Ar + B r 2 ) cos (3)
U cos
b
a
k̂
U k̂
Sa
Sb
Ua 3 b3
= r cos
b 3 a 3 2r 2
Ua 3 b3
= b cos
b 3 a 3 2b 2
3 2Ua 3 b cos
=
2 b3 a 3
and for the outer sphere r = b,
dS = 2 (b sin) (bd), 0
3 Ua 3 b
Thus, impulsive force, F 0 cos2 (2b 2 sin ) d k̂
2 b3 a 3
3Ua 3 b 3 k̂ 2
b 3 a 3 0
= cos sin d
2Ua 3 b 3
= k̂
b3 a 3
Hence the result
Now, if U1 , U 2 denote the velocity of spheres of radii a & b respectively and
R 1 , R 2 be the corresponding impulsive forces exerted by the fluid, then
2Ua 3 b 3
U1 U k̂, U 2 0, R 2 F k̂
b3 a 3
1
K.E., T = U i . R i
2
1
2
1
2
1
= U1 R1 U 2 R 2 U1 R1 U k̂ R1 (4)
2
Also,
R i S n̂ dS R1 S n̂ dS
i 1
R 1 k̂ S n̂ k̂() r a dS
a
113
FLUID DYNAMICS
Ua 3 b3
= S cos 3 a cos 2 (a sin) ( a d)
3
a b a 2a 2
( negative sign due to inwards normal i.e. on the inner sphere, pressure is
inwards)
Ua 3 2a 3 b 3
= 3 .2a 2 0 cos2 sin d
b a 3
2a 2
2 Ua 3 (2a 3 b3 )
=
3 b3 a 3
Thus, from equation (4), we get
1 U 2 a 3 (2a 3 b 3 )
T=
3 b3 a 3
3.9. Deduction : If we let b, then it becomes the case of a sphere of radius
„a‟ moving in an infinite liquid at rest at infinity and we get
2a 3
U 2 a 3 1 3
T = Lt
1 b 1
U 2 a 3
b 3 3
a 3
1 3
b
14 3 2 1 ' 2
= a U M1 U
43 4
4
where M1' a3 is the mass of liquid displaced by the sphere r = a
3
3.10. Example. (Motion of Two Concentric Spheres) : The space between two
spheres is filled with incompressible fluid. The spheres have radii a, b (a < b)
and move with constant speeds U, V respectively along the line of centres.
Show that at the instant when the spheres are concentric, the velocity potential
is given by
3 1 3 3 2
(a U b V)r 2 ( U V)a b r cos
3
=
b3 a 3
Also determine the impulse which is required to produce the velocity U to the
inner sphere, when outer sphere is at rest.
= U cos, r = a (1)
r
and = V cos, r = b (2)
r
where U & V are taken in the same direction.
The solution of the Laplace equation is of the form
= (Ar + B r 2 ) cos
2B
= A 3 cos
r r
and thus the boundary conditions give
2B 2B
A U, A V
a3 b3
Solving for A & B, we find
1 ( U V )a 3 b 3 a 3 U b3V
B= , A
2 a 3 b3 a 3 b3
Thus the velocity potential for this motion is
a 3 U b 3 V 1 ( U V)a 3 b 3 1
= 3 3
r cos
a b 2 a b
3 3
r 2
3 1 3 3 2
(a U b V)r 2 ( U V)a b r cos
3
=
b3 a 3
Hence the result
Impulse :- When outer sphere is at rest, then V = 0 and from equation (3), we
get
Ua 3 b3
= r cos (4)
b 3 a 3 2r 2
4 2
Let M= a be the mass of inner sphere
3
4 3
and M = a is the mass of liquid displaced by the inner sphere.
3
115
FLUID DYNAMICS
| P =
Ua 3 (2a 3 b 3 )
i.e. I = MU +
b a
3 3 0
cos2 sin d
2 a 3 U(2a 3 b 3 )
= MU +
3 b3 a 3
1 M' U(2a 3 b 3 )
= MU +
2 b3 a 3
3.11. Deduction :- If b , then it will be the case of a solid sphere moving
in an infinite liquid and
M' M'
I = MU + U M U
2 2
3.12. Remark. The problem in which we solve the Laplace equation 2 = 0
when the normal derivative of i.e. is given on the boundary, then such
n
type of problem is called a Neumann problem whereas the solution of
2 = 0 when the value of is given on the boundary, is termed as Dirichlet
problem.
FLUID DYNAMICS 116
Thus, a source is a point at which fluid is continuously created and distributed e.g. an
expanding bubble of gas pushing away the surrounding fluid. If the volume of fluid per unit
time which is emitted from a simple source at 0 is constant and equal to 4m, then m is termed
4.2. Sink : A negative source is called a sink. At such points, the fluid is
constantly moving radically inwards from all directions. Thus a simple sink of
strength m is a simple source of strength m.
qr
0 r
q.n̂ dS
S
In case of a source there is only the radial velocity i.e. q has only radial
component qr .
117
FLUID DYNAMICS
= qr (4 r2).
Thus, we get
m m
qr = (1)
r 2
r r
qr = | q (2)
r
m
=
r
which is the required expression for the velocity potential for a source.
4.4. Remarks. (i) For a simple sink of strength m, the velocity potential is =
m
r
0
M
z-axis
We shall find the velocity potential at any point P(z, , ). From P, draw on
OZ. Let OP = r, | POZ = ; OM = z
We observe that the velocity potential of the uniform stream in the absence of
source is
q Uk̂ k̂
z
U Uz
z
m
2 = (2)
r
m
= 1 + 2 = Ur cos +
r
m
= Ur cos (3)
r
m
qr = U cos 2
r r
0
1
q = = U sin 0 2
r
0
1
q = =0
r sin
m
The stagnation points (q 0) are given by U cos + 0, sin = 0 = 0
r2
or
119
FLUID DYNAMICS
m
But = 0 gives r to be imaginary = and r =
U
m
Thus there is only one stagnation point , ,0
U
OO1 h, OO 2 h, h | h |
The velocity potential at P due to the combination of source and sink at O1 and
O2 is
m m mr 2 mr1
= =
r1 r2 r1r2
m(r2 r1 ).( r2 r1 )
=
r1r2 (r1 r2 )
r2 h r
But r2 r1 2h and r2 r1 2r
r1 h r
FLUID DYNAMICS 120
m(2h).(2r) 4mh.r
Thus =
r r2 (r1 r2 ) r1r2 (r1 r2 )
2.r
= , where 2mh (1)
r1r2 (r1 r2 )
In equation (1), let us first keep a finite constant and non-zero vector, so that
= | | is a finite constant and non-zero scalar. Let h 0 along O1O.
Then m in such a way that remains the same finite non-zero constant
vector. In that case, both r1, r2r and thus under this limiting process, (1)
results in
The limiting source sink combination obtained at 0 when we keep the direction
of h fixed but let h0 and m with = 2mh remaining a finite non-zero
constant, is called a three-dimensional doublet (or dipole). The scalar quantity
is called the moment or strength of the doublet. The vector quantity = ̂
is called the vector moment of the doublet & ̂ (unit vector from 02 to 01)
determines the direction of the axis of the doublet from sink to source.
2 cos
qr =
r r3
1 sin
q =
r r3
q = 0
dr
d = 0 = constant and = 2cot d
r
r = A sin2
121
FLUID DYNAMICS
P(r,,)
U k̂
r
M Z
O
Let P be a point in the fluid having spherical polar co-ordinates (r, , ), the
direction OZ of the doublets axis being the line = 0. We shall find the
resultant velocity potential due to the combination of the uniform stream and
the doublet. We know that the velocity potential due to the uniform stream is
1 = Uz = Ur cos (1)
= 1 + 2 = (Ur + r 2) cos
2
qr = U 3 cos
r r
1
q = U 3 sin
r r
1
q = 0
r sin
Stagnation points are determined by solving.
2
U 3 cos 0, U 3 sin 0 |q0
r r
FLUID DYNAMICS 122
1/ 3
2
which are satisfied when sin = 0 and r =
U
2 1 / 3 1/ 3
,0 and 2 ,
U U
which lie on the axis of symmetry.
1/ 3
2 1
If we write r = a i.e. a = i.e. = U a3, then for the region r a, we
U 2
obtain the same velocity potential as for a uniform flow past a fixed
impermeable sphere of radius a and centre 0. Thus, for r a, the effect of the
1
sphere is that of a doublet of strength = Ua3 situated at its centre, its axis
2
pointing upstream. So the sphere can be represented by a suitably chosen
singularity at its centre.
r1 d
r r2
M
A x x B x2
x1
Let P be a point in the fluid at a distance r from this element, then the velocity
mx
potential at P due to the point source is .
r
The total velocity potential at P due to the entire line distribution AB (= 2l) is
2l dx
=m (1)
0 r
123
FLUID DYNAMICS
2l dx
=m 0
(x1 x) 2 (r12 x12 )
2l 1
log ( x x ) ( x x ) 2 (r 2 x 2 ( dx
= m
1 1 1 1 x a2
2
1
0
log x x 2 a 2
= m log (x1 x) (x1 x) 2 (r12 x12 ) 0
2l
= m log(x r ) logx
1 1 2 x 22 r12 x
2
1 | x1 2l = x1 AB = x2
x r
= m log 1 1 , where r12 x12 d 2 r22 x 22 .
x 2 r2
r1 x1 r2 x 2 r1 r2 x1 x 2
r2 x 2 r1 x1 r1 r2 x 2 x1
r1 r2 2l
=
r1 r2 2l
r r 2l
Thus, = m log 1 2
r1 r2 2l
a l
= m log (2)
a l
Expression for Velocity :- The velocity at P is given by q n̂
n
P (3)
r1=n
P
r1+r1 r2+r2
r1 r2
A B
a l 1 1 a 2lm a
Thus q m log n̂ m n̂ = 2 2 n̂ (4)
n a l a l a l n a l n
The normal at P to the a-surface bisects the angle 2 between the focal radii
AP, BP.
Now,
a 2 b2 c2 c
cosC b
2ab
c a b 2 2ab cosC
2 2
B a C
2r1 r1 = 2r1 n cos + (n) (r1)
2 2
a r1 r2
2 . = cos + cos = 2 cos
n n n
a
cos
n
m m m m
q 3
AP0 A' P0 q r̂ 3 r
(AP0 ) (A' P0 ) 3 Y r 2
r
P0
(m) O (m)
X
A(a,0,0) A(a,0,0)
x =0
m
q (AP0 A' P0 )
(AP0 ) 3
AP0 A' P0
m 2m
= 3
(2OP0 ) = (OP0 ) (AO oP0 ) (A'0 OP0 )
(AP0 ) (AP0 ) 3
2OP0
This shows that at any point P0 of the plane YY, the fluid flows tangentially to
the plane x = 0 and so there is no transport of fluid across this plane.
Let denotes the velocity potential then, at all points P0 on the plane YY, the
normal component of velocity is zero
= 0. Hence, the image of a source at A in the rigid plane YY is
n
a source at A, as required.
(ii) Image of Doublet in a Y
Plane : Consider a pair of
sources m at A and m at B(m) B(m)
B, taken close together and
on one side of the rigid
plane YY‟. The image X
A(m) A(m)
system is m at A‟, m at
B, where A & B are
respectively the reflections
of A and B in the plane Y
YY. In the limiting case,
when BA along BA in
such a way as to form a
doublet at A, we find that
the image of
127
FLUID DYNAMICS
r2 r r1
2
1
A() a 0 a A() M Z
cos1 cos 2
= (1)
r12 r22
AM OM OA r cos a
But cos1 =
r1 r1 r1
(a r cos )
cos 2 =
r2
r cos a (a r cos)
= 2
r12 r1 r2 r2
r cos a r cos a
= (3)
r13 r23
r1 r r a cos
2 r1 2r 2a cos 1
r r r1
r2 ra sin
.
r2
cos r 1 cos r 1
qr= 3 3 2 4 (r cos a ) 3 3 1 4 (r cos a )
r r2 r r2 r1 r r1
r r
(r cos a ) 1 (r cos a ) 2
1 r sin r sin 3
q = = 3
r r r1 3
r14 3
r2 4
r2
129
FLUID DYNAMICS
q = 0
When the point P lies on the plane YY or = /2, we have r12 r22 r2 + a2
and so at (r, /2, ), the velocity components are
p 1 2 p
q const ,
2
where q 0 at infinity.
1
p = p 36 2 a 2 r 2 (r 2 a 2 ) 5
2
18 2 a 2 r 2
i.e. p(r) = p 2
(r a 2 ) 5
Now,
dp
p(r) = 36 2 a 2 r(4r 2 a 2 ) (r 2 a 2 ) 6
dr
1
which gives p(r) = 0 when r = a
2
Also
a a
p 0, p' 0
2 2
a
i.e. p(r) changes sign from negative to positive when r passes through
2
a
p is minimum at r = = /2
2
FLUID DYNAMICS 130
a
i.e. at the point P0 , / 2,
2
2
a 5
a
2
a
2 2
5
Hence p is least at a distance a from the doublet and the minimum value is
2
5
9 4 1
pmin. = p 2 6
2 5 a
a a2 1 2
(r, , ) + , , 0a / r (R, , ) dR, (r > a)
r r a
a2
where r and are the inverse points w.r.t the sphere of radius a.”
r
Here, the last two terms refer to perturbation potential due to the presence of
the sphere.
r P
a
0 B A(f,0,0)
m
(r, ) =
AP
AP = r 2 f 2 2rf cos
a a 2 1 a2 / r
, (R, ) dR
r r a 0
1 / 2
am a 4 a2 m a2 / r
i.e. 2 f 2
2 f cos 0 [R2 + f2 2Rf
r r r a
cos]1/2 dR
(ma f ) m a2 / r dR
i.e.
r 2 2r(a 2 / f ) cos (a 2 / f ) 2 a 0 R 2 2Rf cos f 2
This shows that the image system of a point source of strength m placed at
distance f(> a) from the centre of solid sphere consists of a source of strength
FLUID DYNAMICS 132
ma a2
at the inverse point in the sphere, together with a continuous line
f f
m
distribution of sinks of uniform strength per unit length extending from the
a
centre to the inverse point.
(ii) Image of a doublet in a sphere when the axis of the doublet passes
through the centre of the sphere :- Let us consider a doublet AB with its axis
BA pointing towards the centre 0 of a sphere of radius a. Let OA = f, OB = f
+ f. Let A, B be the inverse points of A & B in the sphere so that
O (m) (m)
B A A B
a
a 2 a 2 a 2 f
= to the first order
f f f f
a2
= 2 f to the first order
f
The line distribution of sinks and sources from 0 to B cancel each other
m
leaving behind a line distribution of sinks of strength per unit length from
a
m m a2 a a
B to A i.e. sink of strength BA = f (mf ) at B. The
a f 2 2
f2
a f
source at B is of strength
1
ma ma f ma f
1 1 , to the first
f f f f f f
order terms
ma ma ma a
= 2 f 2
f f f f
ma a
which is equivalent to a sink at B' and a source 2 at B.
f f
a
As there is already a sink at B, therefore source and sink at B neutralize.
f2
ma ma
Finally, we are left with source at A and a sink. at B. Thus, to the
f f
first order, we obtain a doublet at A of strength
ma ma a 2
(BA) = f
f f f2
ma 3 a 3
= f .
f3 f3
If the streamlines in all the planes passing through a given axis are the same,
the fluid motion is said to be axi-symmetric. We have already considered such
flow for irrotational motion in spherical polar co-ordinates. (r, , ) in which
the line = 0 is the axis of symmetry.
Suppose the z-axis be taken as axis of symmetry, then q = 0 and the fluid
motion is the same in every plane = constant (meridian plane) and suppose
FLUID DYNAMICS 134
that a point P in the fluid may be specified by cylindrical polar co-ordinates (r,
, z). Thus, all the quantities associated with the flow are independent of .
The equation of continuity in cylindrical co-ordinates, becomes
(rqr ) (rqz ) 0
r z
i.e. (rqr ) (rqz ) (1)
r z
rqrdz r qz dr = 0 (2)
This means that (2) is an exact differential equation and let L.H.S. be an exact
differential d(say)
Therefore,
rqr dz rqz dr = d = dr dz
r z
which gives
rqz , rqr (3)
r z
dr dz
qr qz
qz dr = qr dz
1 1
dr dz
r r r z
dr dz 0
r z
d = 0
135
FLUID DYNAMICS
= constant = C
1 2 1
(r q r ) 2 (r sin q ) 0
r r
2
r sin
2
i.e. (r sin q r ) (r sin q ) (1)
r
r sin q dr qr r2 sin d = d = dr d
r
q r sin , q r r 2 sin .
r
6.2. Remark. In the above cases, the motion need not be irrotational i.e.
velocity potential may not exist. In case of irrotational motion, it can easily be
shown that the velocity potential and the Stoke‟s stream function do not
satisfy CR equations due to the fact that is not harmonic.
6.3. Stoke’s Stream Function for a Uniform Stream : Let a uniform stream
with velocity U be in the direction of z-axis such that q U k̂ . Then, from the
relations
1 1
q3 = , qr ,
r r r z
1 1
we get U= ,0
r r r z
Ur , 0
r z
FLUID DYNAMICS 136
r2
= U , where the constant of integration is found to be
2
zero.
U U
= (r sin ) 2 r 2 sin 2 .
2 2
q f (r ) r̂
m
q r̂(r 0) in spherical polar co-ordinates.
r2
m
i.e. (qr, q) = r̂ (1)
r2
1 1
qr = , q (2)
r sin
2
r sin r
m 1
2 , 0
r 2
r sin r
m sin , 0
r
= m cos .
A constant may be added to this solution and this is usually done to make =
0 along the axis of symmetry = 0. In such case,
= m (cos 1)
= m (1cos)
137
FLUID DYNAMICS
6.5. Stoke’s Stream Function for a Doublet at Origin : We assume that the
flow is due to only a doublet at origin 0 of strength . Taking the axis = 0 of
the system of spherical co-ordinates to coincide with the axis of the doublet,
we find that the velocity potential at P(r, , ) is
cos
= (r > 0) (1)
r2
2 cos 1 sin
qr = , q q = 0 (2)
r r 3
r r3
But the relations between the velocity components and the Stoke‟s stream
function are
1 1
qr = , q (3)
r 2 sin r sin r
From (2) and (3), we get
2 sin cos sin 2
,
r r r2
Integrating, we get
sin 2
=
r
6.6. Stoke’s Stream Function due to a Uniform Line Source : Let a uniform
line source of fluid extends along the streamline segment AB of length l.
Consider an element QQ of length z at a distance z ( = AQ) from A. Thus we
have a simple source of strength m z, where m is the constant source strength
per unit length of the distribution along AB.
Let QP = r, | PQB = Q, PM = d P
QM QB BM
In PQM, cos
l l bz PQ PQ
= m 0 dz ml
d (l b z) 2
2 l zb l zb
r d 2 (l b z) 2
Putting l + b z = x dz = dx
When z = 0, x = l + b,
when z = l, x = b
Therefore,
b x(dx)
=m l b ml
d2 x2
m l b
or = b
(d2 + x2)1/2 (2x) dx ml
2
l b
m d2 x2
= ml
2 1/ 2 b
= m d 2 (l b) 2 d 2 b 2 ml
= m[AP BP] mAB
= m[AP BP AB] .
As p is the only variable point, the simpler form m (APBP) can be taken for
evaluating velocity components at P. The stream surfaces are
These are confocal hyperboloids of revolution about AB, with A and B as foci.
In spherical polar co-ordinates (r, , ), the Stoke‟s stream functions for each
separate distribution are
1 2 2
1 = Ur sin (for uniform stream, q Uk̂ )
2
2
2 = sin (for doublet at origin)
r
1
(r, ) = Ur 2 / r sin 2
2
1 2 2
Ur / r sin = 0
2
1 2
sin = 0 or Ur 0
2 r
1/ 3
2
= 0, i.e. the z-axis or r = , the surface of the sphere
U
1/ 3
2
with centre 0 and radius .
U
FLUID DYNAMICS 140
Suppose that a fluid moves in such a way that at any given instant, the flow
pattern in a certain plane within the fluid is the same as that in all other parallel
planes within the fluid. Then at the considered instant, the flow is said to be
two-dimensional flow or plane flow. Any one of the parallel planes is then
termed as flame of flow.
If we take the plane of flow as the plane z = 0, then at any point in the fluid
having cartesian co-ordinates (x, y, z), all physical quantities i.e. velocity,
density, pressure etc, associated with the fluid are independent of z.
1 1 2 2
r 0 (1)
r r r r 2 2 z 2
If the flow is two dimensional and the co-ordinate axes are so chosen that all
physical quantities associated with the fluid are independent of z, then = (r,
) and (1) simplifies to
1 1 2
r 0 (2)
r r r r 2 2
g()
1d
rf ' (r) 12 f (r)g' ' () = 0
r dr r
141
FLUID DYNAMICS
r
d
rf ' (r) g' ' ()
i.e. dr (4)
f (r ) g()
Equation (6) has periodic solutions when > 0. Normally the physical
problem requires that g( + 2) = g() and this is satisfied when = n2 for
n = 1, 2, 3,….
df df dt 1 df
Also, f (r) = .
dr dt dr r dt
d 2 f d df d 1 df
and f (r) =
dr 2 dr dr dr r dt
1 d df df 1
=
r dr dt dt r 2
1 d df dt 1 df
= 2
r dt dt dr r dt
1 d 2 f 1 df
= 2 2 2
r dt r dt
FLUID DYNAMICS 142
d 2 f df
r2 f (r) =
dt 2 dt
d 2 f df df
2
n 2f 0
dt dt dt
d 2f
n 2f 0
dt 2
It‟s solution is
f = exp ( + nt) = e+nt = e t
n
r n
A special solution of (2) is obtained by linear superposition of the forms (7) &
(8) to give
f = k1 + k2t = k1 + k2 log r
and g = k3 + k4
7.3. Example. Discuss the uniform flow past an infinitely long circular
cylinder.
143
FLUID DYNAMICS
where the velocity potential of the uniform stream is Ux = Ur cos and due to
perturbation, it is Ar1 cos which 0 as r and gives rise to a velocity
pattern which is symmetrical about = 0, . (the term r1 sin is not
there since it does not give symmetric flow)
= 0, when r = a (2)
r
Thus, the velocity potential for a uniform flow past a fixed infinite cylinder is
a2
(r, ) = U cos r , r > a, 0 2 (3)
r
a2
qr = U cos1
r r2
q =
1 1 a2
U sin r U sin 1 a 2 / r 2
r r
r
qz = 0
z
We note that as r, qr = U cos, q = U sin which are consistent with the
velocity at infinity U î of the uniform stream.
7.4. Example. A cylinder of infinite length and nearly circular section moves
through an infinite volume of liquid with velocity U at right-angles to its axis
and in the direction of positive x-axis. If the section is specified by the
equation.
FLUID DYNAMICS 144
r = a(1 + cosn)
where n is positive integer and is small, show that the approximate value of
the velocity potential of the fluid is
a a
n 1
a
n 1
Ua cos cos(n 1) cos(n 1)
r r r
Solution. Let the tangent at a point P on the plane of cylinder makes angles ,
() with the radial line OP drawn from 0 as shown in the figure
UN
90
P
U
-
X
0
At large radial distances r from OZ, the fluid velocity becomes vanishingly
small.
Let us assume the velocity potential (r, ) of the form rk sin
cos
k (k = 1,
2,…..).
Thus, we seek a solution of the form
(r, ) = rk (Ak cos k + Bk sin k) (1)
k 1
At any point (r, ,z) of the fluid, the cylindrical polar velocity components are
(q = )
qr = k. Ak. r(k+1) cos k
r k 1
145
FLUID DYNAMICS
1
q = k. Ak. r(k+1) sin k
r k 1
qz = =0
z
At P on the boundary, since ( ) is the angle between the tangent and the
radius vector OP, therefore
1 dr d
cot () = (log r)
r d d
r
d
cot = [log a(1+cosn)]
d
1
= (a n sin n)
a (1 cos n)
n sin n d
cot = sin () = r (3)
1 cosn ds
dr
The normal component of velocity UN of the boundary at P is cos () =
ds
UN = U sin ( )
As there is n transport of
fluid across the surface
1+ cos n
and n breakaway from it,
so UN is also the normal
velocity component of the
fluid. n sin n
Thus,
UN = qr sin + q cos .
FLUID DYNAMICS 146
k A k a ( k 1) (1 cosn) ( k 1) [cosk(1 cosn) sin k n sin n]
k 1
=
(1 cosn) 2 ( n sin n) 2
(5)
k A k a ( k 1) (1 cosn) ( k 1) [cosk(1 cosn) n sin k sin n]
k 1
L.H.S. of (6)
= k A k a ( k 1) [1 (k 1) cosn][cosk cosk cosn n sin k sin n]
k 1
= k A a
k 1
k
( k 1)
[cos k (k 1) cos k cos n
= k A k a ( k 1) [cos k k cos k cos n n sin k sin n]
k 1
= k A a
k 1
k
( k 1) k
cos k 2 cos(n k) cos(n k)
n
cos(n k) cos(n k)
2
= k A k a ( k 1) cos k (n k) cos(n k) (n k) cos(n k) (7)
k 1 2 2
R.H.S. of (6)
147
FLUID DYNAMICS
n
= U cos cos(n 1) cos(n 1) cos(n 1) cos(n 1)
2 2
= U cos (1 n ) cos(n 1) (1 n ) cos(n 1) (8)
2
Correct to the first order of approximation, from (6), (7) & (8), comparing
coefficients of cos , cos (n1), cos(n+1), we get
A1
U= A1 = Ua2 (9) | n + k 1
a2
1 1
(n 1)A n 1a n A1a 2 (n 1) U (n 1)
2 2
In (7) cos k cos(n1)
cos (nk) cos (n1)
similarly for n+1
1 1
and (n 1)A n 1a ( n 2) A1a 2 (n 1) U (n 1)
2 2
All Ak other than A1, An1, An+1 are zero. Putting the value of these three non-
zero co-efficients in (2), we get
(r, ) = [A1r1 cos + An1 r(n1) cos (n1) +An+1 r(n+1) cos (n+1)]
a a
n 1
a
n 1
= Ua cos cos(n 1) cos(n 1) .
r r r
u= , v (2)
x y y x
Thus and satisfy the C-R equations and so W must be an analytic function
of z = x +iy
The function W = f(z) is called the complex potential of the plane flow.
W = + i and z = x +iy
W
i i = u + iv
x x x x y
W dW z dW z
But . 1
x dz x dz x
dW
Thus. u iv
dz
dW
u iv = q cos iq sin
dz
dW
Thus, speed q = u 2 v2
dz
dW
and for stagnation points, 0
dz
W = z2
Solution. We have
149
FLUID DYNAMICS
W = + i = z2 = (x + iy)2 = x2y2 + 2i xy
8.3. Complex Potential for a Uniform Stream. Let the uniform stream
advance with a velocity having magnitude U and being inclined at angle to
the positive direction of the x-axis.
dW
= u iv = U ei
dz
W = Uz ei
8.4. Line Source and Line Sink. Line source and line sink are the two-
dimensional analogues of the three-dimensional simple source and sink. Let A
be any point of the considered plane of flow and C be any closed curve
surrounding A. We construct a cylinder having its generators through the
points of C and normal to the plane of flow. Suppose that in each plane of
flow, fluid is emitted radically and symmetrically from all points on the infinite
line through A normal to the plane of flow and such that the rate of emission
from all such points as A is the same. Then the line through A is called a line
source. We may take the closed curve C to be a circle having centre A and
radius r.
An example of a line source is a long straight hose with perforations along its
length, commonly used for watering lawns for long periods of time.
8.5. Complex Potential for a Line Source. Let there be a line source of
strength m per unit length at z = 0. Since the flow is radial, the velocity has the
radial component qr only. Then the flow across a circle of radius r is (by law
of conservation of mass)
m
qr =
r
dW
u iv = qr cos i qr sin
dz
m i
= qr (cos i sin) = e
r
151
FLUID DYNAMICS
dW m m m
e i i
dz r re z
Integrating, we get
W = m log z
We can write it as
+ i = m log (r ei)
= m log r i m
= m log r, = m
r = constant, = constant
y
i.e. x2 + y2 = constant, tan1 = constant
x
Thus the equipotentials are circles and streamlines are straight lines passing
through origin.
W = m log (zz0)
For a line sink of strength m per unit length at z = z0, the complex potential is
W = m log (zz0).
8.6. Complex Potential for a Line Doublet. The combination of a line source
and a line sink of equal strength when placed close to each other gives a line
doublet. Let us take a line source of strength m per unit length at z = a ei and
a line sink of strength m per unit length at z = 0
z aei aei
= m log m log1
z z
x2 + y2 = 2k2y (2)
Family (1) have centres (k1, 0) and radii k1 and family (2) have centres (0, k2)
and radii k2
The two families are orthogonal
y
Streamlines
Equipotentials
8.8. Milne-Thomson Circle Theorem :Let f(z) be the complex potential for a
flow having no rigid boundaries and such that there are no singularities within
the circle |z| = a. Then on introducing the solid circular cylinder |z| = a, with
impermeable boundary, into the flow, the new complex potential for the fluid
outside the cylinder is given by
where z is the image of the point z w.r.t. the circle. If z is outside the circle,
then z = a2/z is inside the circle. Further, all the singularities of f(z) lie
outside C and the singularities of f(a2/z) and therefore those of f (a2/z) lie
inside C. Therefore f(a2/z) introduces no singularity outside the cylinder.
Thus, the functions f(z) and f(z) + f (a2/z) both have the same singularities
outside C. Therefore the conditions satisfied by f(z) in the absence of the
cylinder are satisfied by f(z) + f (a2/z) in the presence of the cylinder. Further,
the complex potential, after insertion of the cylinder |z| = a, is
FLUID DYNAMICS 154
= f(z) + f (z)
It follows that = 0
f(z) = Uz
Thus
f ( z) f (z) Uz U z U z
and so
f (a2/z) = Ua2/z .
With the cylinder |z| = a present, by circle theorem, the complex potential, for
the liquid region |z| a, is
W = f(z) + f (a2/z)
a2
i.e. + i = U z
z
a2
= Re(W) = U cos r | Same expression as derived earlier
r
155
FLUID DYNAMICS
a2
= Im(W) = U sin r
r
a2
The perturbation term f (a2/z) = U gives the image of the flow in the
z
cylinder. This image represents a uniform line doublet of strength Ua2 per unit
length and axis in the direction î .
In a two dimensional fluid motion, if the flow across a curve C is zero, then the
system of line sources, sinks, doublets etc on one side of the curve C is said to
form the images of line sources, sinks, doublets etc on the other side of C. To
discuss the images in two dimensions, we use complex potential.
P
Y
r2 r1
m 2 1
X
O m
A(a,0) A(a,0)
= m log r1e
i1
r2 e
i 2
mlogr r e
1 2
i (12 )
+ i = m log (r1 r2) im (1 + 2)
= m (1 + 2)
i.e. 1 = 2 1 + 2 =
which shows that y-axis is a streamline. Hence the image of a line source of
strength m per unit length at A(a, 0) is a source of strength m per unit length at
A(a, 0). In other words, image of a line source w.r.t. a plane (a stream line)
is a line source of equal strength situated on opposite side of the plane (stream
line) at an equal distance.
B m m B
m m
X
A O A
the stream line OY. Let there be line sources at the points A and B, taken very
close together, of strengths m and m per unit length. Their respective images
in OY are m at A, m at B, where A, B are the reflections of A, B in OY.
The line AB makes angle with OX . Thus A' B' makes angle () with
OX . In the limiting case, as m, AB0, we have equal line doublets at A
and A with their axes inclined at , () to OX . Hence, either of the line
doublet is the hydrodynamical image of the other in the infinite rigid
impermeable plane (stream line) x = 0
9.3. Image of Line Source in a Circular Cylinder (or in a circle). Let a line
source of strength m per unit length be present at a point z = d in the fluid; d >
a. Let us then insert a circular cylinder |z| = a in the fluid. The complex
potential in the absence of cylinder is m log (zd) and after the insertion of
cylinder, by circle theorem, we get
m m m
m
d a 2
= m log (zd) m log z
z d
Ignoring the constant term, we observe from (1) that the complex potential
represents a line source at z = d, another line source at the inverse point z =
a2/d and an equal line sink at the centre of the circle. Thus the image of a line
source of strength m per unit length at z = d in a cylinder is an equal line
source at the inverse point z = a2/d together with an equal line sink at the centre
z = 0 of the circle. Further, (1) can be written as
+ i = m log (x d) 2 y 2 1/ 2 y
i tan 1
x d
2 2
1/ 2
a y
m log x y i tan
2 1
x a /d
2
d
y
+ m log(x 2 y 2 )1/ 2 i tan 1 .| log z = log + i r = x 2 y 2 ,
x
y
= tan1
x
y 1 y 1 y
= m tan1 m tan m tan .
x d x a /d
2
x
FLUID DYNAMICS 158
y y
x d x a2 / d 1 y
= m tan1 m tan
1 y y x
x d x a 2 / d
x 2 y2 a 2
a 2 y2 x 2
y y
= m tan1 m tan 1 0 .
x x
Thus, the circular cylinder is a streamline i.e. there is no flow of fluid across
the cylinder.
e ie
zd
When the cylinder |z| = a is inserted, the complex potential, by circle theorem,
becomes
e i e i
W= 2
z d (a / z) d
e i e i z
=
z d d z a 2
d
e i z e i ( )
=
z d d z a 2
d
e i e i() a 2 e i()
= 2 (1)
zd d d z a2
d
If the constant term (second term) in (1) is neglected, then the complex
potential in (1) is due to a line doublet of strength per unit length at z = d,
159
FLUID DYNAMICS
a 2
inclined at an angle with x-axis and another line doublet of strength per
d2
unit length at the inverse point z = a2/d inclined at an angle with x-axis.
Thus the image of a line doublet of strength per unit length z = d inclined at
a 2
angle with x-axis is a line doublet of strength per unit length at the
d2
inverse point a2/d which is inclined at an angle with x-axis.
m
m
z=a2/d z =d
m m
9.5. Remark. The above two cases i.e. (iii) and (iv) alongwith „uniform flow
past a fixed infinite circular cylinder‟ are applications of Milne-Thomson circle
theorem.
9.6. Example. What arrangement of sources and sinks will give rise to the
a2
function W = log z ?
z
a2 z2 a 2
Solution. We have W = log z log
z z
y y y
= tan1 tan 1 tan 1
x a xa x
FLUID DYNAMICS 160
y y
= tan1 x a x2 a tan 1
y
y x
1 2
x a2
2xy y
= tan1 2 tan 1
2
x y a
2
x
x 2 y 2 a 2 y
2
1
2
= tan (2)
x y a x
2
where is a constant.
W = Vz + 2z/(z2 + a2)
Also show that for 0 < < 4a2V, there are no stagnation points on the
boundary and that the pressure on it is a minimum at the origin and maximum
at the points ( + a 3 , 0).
161
FLUID DYNAMICS
Solution. We know that the image of the line doublet î at point A(0, a) is a
line doublet î at point A(0, a)
V î
A(0,a) î
x
y =0
A(0,a)
y
Therefore, the complex potential of the system is
W = Vz +
z ia z ia
2z
= Vz + = Vz + 2z (z2 + a2)1
z a
2 2
dW
= V + 2(a2 z2) (a2 + z2)2
dz
dW dW
q= = V + 2 (a2x2) (a2 + x2)2
dz dx
dW
For stagnation points 0
dx
= 4 ( 4a2V)
From here, < 0 if 0 < < 4a2V, showing that the quadratic equation (1) has
no real root. Therefore there is no stagnation points on the boundary y = 0.
FLUID DYNAMICS 162
2
2(a 2 x 2 )
P is maximum when X = V 2 is minimum and conversely.
(a x 2 ) 2
1 1/2 dX
X . X = 4x (3a2x2) (a2 + x2)3 X =
2 dx
dW
2
M = Real part of C z dz
2 dz
163
FLUID DYNAMICS
Proof. Let ds be an element of arc at a point P(x, y) and the tangent at p makes
an angle with the x-axis. The pressure at P(x, y) is pds, p is the pressure per
unit length. pds acts along the inward normal to the cylindrical surface and its
components along the co-ordinate axes are
pds sin
ds
ds
pds cos
pds
O
x
dF = dX + idY
= p sin ds + ip cos ds
= ip (cos + i sin) ds
pds sin along negative x axis
pds sin along positive x axis
dx dy dx dy
= ip i ds cos , sin
ds ds ds ds
p 1 2
q constant
2
1 2
or p= q + k (2)
2
dW
Further u + iv = q cos + iq sin
dz
FLUID DYNAMICS 164
dx dy
and dz = dx + idy = i ds = (cos + i sin) ds = ei ds (4)
ds ds
i
= 2
q dz | C dz = 0
2 C
i
= 2 i
C q e ds
2
From here ;
i 2 i
X iY = C q e ds
2
i 2 2i i
= C (q e ) e ds
2
2
i dW
= C dz | using (3) & (4)
2 dz
dy dx
= C p y ds p x ds
ds ds
r dF
î ĵ k̂
x y 0
pds sin pds cos
= C p(x dx + ydy]
1
C k 2 q (xdx + ydy)
2
=
165
FLUID DYNAMICS
1
= k C d (x 2 y 2 ) C q2 (xdx + ydy)
2 2
= 2
C q (xdx + ydy) | 1st integral
2
vanishes.
dx cosds
=
2 C
q2 (x cos + y sin) ds
dy sin ds
= R.P. of
2 C
q 2 ( x iy )(cos i sin )ds
= R.P of q 2 ze i ds
2 C
e 2 2i i
= R.P of C z(q e )e ds
2
e dW
2
= R.P. of C z dz .
2 dz
(i) satisfies Laplace equation i.e. 2 = 0 at every point of the liquid.
0 and 0 at infinity.
x y
FLUID DYNAMICS 166
(iii) Along any fixed boundary, the normal component of velocity must be
zero so that = 0 i.e.
s
(iv) On the boundary of the moving cylinder, the normal component of the
velocity of the liquid must be equal to normal component of velocity of
the cylinder.
= An rn cos n + Bn rn sin
where n is any integer, An and Bn being constants. Also, all the observations
made for , are valid for velocity potential , where and satisfy CR
equations.
(i) U cos
r r a
Ucos
1
(ii) and 0 as r
r r U
x
A suitable form of is
B
(r, ) = Ar cos (1)
r
B
A 2 cos (2)
r r
B
A 2 cos U cos , (A + 0.B) = 0 for all .
a
167
FLUID DYNAMICS
B
A + = U, A = 0
a2
A = 0, B = U a2
Ua 2
Thus (r, ) = cos (3)
r
1
But (CR equation)
r r
1 Ua 2
so, 2 cos
r r
Ua 2
i.e. cos
r
Ua 2
= sin (4)
r
Ua 2
Thus W = + i = (cos i sin)
r
Ua 2 Ua 2
=
re i z
11.2. Remarks. (i) For the case of „Uniform flow past a fixed circular
cylinder‟, using circle theorem, we have obtained the complex potential as
W = f(z) + f(a2/z)
a2
= Uz + U
z
where the cylinder moves with velocity U along positive direction of x-axis. If
we give a velocity U to the complete system, along the positive direction of x-
axis, then the stream comes to rest and the cylinder moves with velocity U in
x-direction.
FLUID DYNAMICS 168
Thus, we get
a2 Ua 2
W = Uz + U Uz
z z
Ua 2
W= Uz.
z
e i
W=
z z0
which shows that the complex potential due to a circular cylinder with velocity
U along x-axis in an infinite mass of liquid is the same as the complex potential
due to a line doublet of strength = Ua2 pre unit length situated at the centre
with its axis along x-axis.
Ua 2
W = + i =
z Ut
Find the magnitude and direction of the velocity in the fluid and deduce that
for a marked particle of fluid whose polar co-ordinates are (r, ) referred to the
centre of the cylinder as origin,
1 dr d U a 2 i i a2
i e e and r sin = constant
r dt dt r r 2 r
Ua 2
W = + i =
z Ut
169
FLUID DYNAMICS
dW Ua 2 Ua 2 2i
Therefore e , z Ut = rei
dz (z Ut ) 2 r 2
Ua 2
i.e. u iv = (cos 2 i sin 2)
r2
Ua 2 Ua 2
u= cos 2, v sin 2
r2 r2
Ua 2
Therefore, q= u v 2
2 2
r
v
The direction of velocity is tan = = tan 2 = 2
u
Ua 2 Ua 2
W = Uz + U( x iy ) 2 ( x iy )
z r
Ua 2
i.e. + i = Ur (cos + i sin) + (cos i sin )
r
Ua 2 cos a2
= Ur cos + , U r sin
r r
a2
r sin = constant
r
Further,
dr Ua 2
U cos 2 cos | q
dt r r
d 1 Ua 2 sin
r U sin
dt r r2
U a 2 i
= e e i
r r 2
Ua 2 Ua 2
Thus, = cos, = sin
r r
so = U cos
r r a
Let T1 be the K.E. of the liquid on the boundary of the cylinder and T2 that of
the cylinder. Let and be the densities of material of the cylinder and the
liquid respectively. Then
T1 = ds
2 C n
2
2 0
= ad, s = a ds = ad | l = r
r r a
2 Ua 2
= cos (U cos) ad
0
2 a
U 2 a 2 2
=
2 0 cos2 d
U 2 a 2 U2 U2
= (a 2 ) M' ,
2 2 2
where M = a2 = mass of the liquid displaced by the cylinder of unit length.
1
K.E. of the cylinder, T2 = MU2, M = a2
2
1
T = T1 + T2 = (M + M) U2 (1)
2
Let R be the external force on the cylinder in the direction of motion. We use
the fact that rate of change of total energy is equal to the rate at which work is
being done by external forces at the boundary.
1 d
RU = (M + M) U2
2 dt
work done force. dis tan ce
time time
force. velocity
M M' dU
= 2U
2 dt
dU
= (M + M) U
dt
dU dU
M R M' (2)
dt dt
Equation (2) is the equation of motion of the cylinder. This shows that the
presence of liquid offers resistance (drag force) to the motion of the cylinder,
since if there is no liquid, then M = 0 and we get
dU
M R (3)
dt
R
Now, if = external force on the cylinder per unit mass be constant and
M
conservative, then by the energy equation, we get
1 R
(M + M1) U2 (M M1) r = constant (4)
2 M
where r is the distance moved by the cylinder in the direction of R. Diff. (4)
w.r.t. t, we get
dU R
(M + M) U (M M' ) U = 0
dt M
dU M M' a 2 a 2
or M R R
dt M M' a 2 a 2
FLUID DYNAMICS 172
dU
i.e. M R (5)
dt
du dv
M X, M Y (6)
dt dt
: + .
V V sin
U cos
U
The boundary conditions for the velocity potential which is the solution of
2 = 0, are (q )
(i) = U cos, r = a (1)
r
(ii) = V sin, r = b (2)
r
B D
= Ar cos Cr sin (3)
r r
B D
A 2 cos C 2 sin (4)
r r r
173
FLUID DYNAMICS
B D
U cos = A 2 cos C 2 sin
a a
B D
V sin = A 2 cos C 2 sin
b b
B D
A U, C 0
a2 a2
B D
A 0, C V
b2 b2
Ua 2 Ua 2 b 2 Vb 2 Va 2 b 2
A= 2 , B , C , D
a b2 a 2 b2 a 2 b2 a 2 b2
Ua 2 b 2 Vb 2 a 2
= r cos r sin
a 2 b 2 r a 2 b 2 r
Ua 2 b 2 Vb 2 a 2
= r cos r sin (5)
b 2 a 2 r b 2 a 2 r
1
r r
i.e. r
r
Ua 2 b 2 Vb 2 a 2
= r cos r sin
b 2 a 2 r b 2 a 2 r
Ua 2 b 2 Vb 2 a 2
= r sin r cos (6)
b 2 a 2 r b 2 a 2 r
It should be noted that the values of and given by (5) and (6), hold only at
the instant when the cylinders are on starting i.e. the initial motion.
11.6. Corollary. If the cylinders move in the same direction then the boundary
conditions are
(i) = U cos , r = a
r
(ii) = V cos, r = b
r
Using these conditions in (4), comparing co-efficients of cos and sin and
then solving the resulting equations, we get
Ua 2 Vb 2 UVa 2 b 2
A= , B , C = 0, D = 0
b2 a 2 b2 a 2
So, =
1
b a2
2
Ua 2
Vb 2
r
UVa 2 b 2
cos
r
and =
1
b a2
2
Ua 2
Vb 2
r
UVa 2 b 2
sin
r
11.7. Example. An infinite cylinder of radius a and density is surrounded
by a fixed concentric cylinder of radius b and the intervening space is filled
with liquid of density . Prove that the impulse per unit length necessary to
start the inner cylinder with velocity V is
a 2
[(+) b2 () a2]V
b a
2 2
B D
= Ar cos Cr sin (1)
r r
The boundary conditions are (q )
(i) = V cos, r = a
r
175
FLUID DYNAMICS
(ii) = 0, r = b
r
Applying these conditions in (1) and then comparing co-efficients of cos and
sin, we get
B D
A 2 V, C 2 0
a a
B D
A 2 0, C 2 0
b b
Solving for A, B, C, D, we obtain
Va 2 Va 2 b 2
A= 2 , B , C=D=0
b a2 b2 a 2
Thus, the potential (1) is
1 2 Va 2 b 2
= 2 Va r cos
b a 2 r
Now, the impulsive pressure at a point on r = a (along x-axis), is
Va 2 b 2
P = ( )r = a = 2 r cos |r a
b a 2 r
Va
= 2 (a 2 b 2 ) cos
b a 2
b2 a 2
= a2 2 V
2
b a
Now, change in momentum = the sum of impulsive forces
b2 a 2
Therefore, a2 (V0) = I a2 2 V
2
b a
b a
2 2
I = a2 V + a2 2 V
2
b a
Thus, impulse due to external forces, is
a 2 V
I= 2 [ (b2a2) + (b2 + a2)]
b a 2
a 2 V
= 2 [( + ) b2 () a2]
b a 2
UNIT – III
1. Vortex Motion
So far we have confined our attention to the cases involving irrotational motion
only. But the most general displacement of a fluid involves rotation such that
the rotational vector (vortex vector or vorticity) curl q 0 . Here we
consider the theory of rotational or vortex motion. First of all we revisit some
elementary definitions.
Lines drawn in the fluid so as at every point to coincide with the instantaneous
axis of rotation of the corresponding fluid element are called vortex lines.
Portions of the fluid bounded by vortex lines drawn through every point of an
infinity small closed curve are called vortex filaments or simply vortices and
the boundary of a vortex filament is called a vortex tube.
The quantity n̂ S is called the strength of the vortex tube. A vortex tube
with a unit strength is called a unit vortex tube.
We shall observe some important results for vortex motion which are
consequences of the following theorem due to Lord Kelvin.
= q.dr ,
C
where q is the velocity and r is the position vector of a fluid particle at any
time t.
Time derivative of following the motion of fluid is
d d d
q dr (q dr )
dt dt C C dt
177
FLUID DYNAMICS
dq d
= dr q (d r )
C dt dt
dq d dr
= dr q dq (1) (dr) d dq
C dt dt dt
dq 1 1
F p = p (2)
dt
dq 1
dr dr dr. p
dt
dq dp
i.e. .dr d (3) dr. d
dt
d dp
d q dq
dt C
1 1
= d q 2 dp
C 2
1 1
= d q 2 dp
C 2 C
A
1 dp
= q 2
2 A C
dp
= 0 (4)
C
d
0 = constant for all time
dt
Corollary (1). In a closed circuit C of fluid particles moving under the same
conditions as in the theorem,
where S is any open surface whose rim is C. To establish (5), we note that, by
Stock‟s theorem,
This shows that the product of the cross-section and angular velocity at any
point on a vortex filament is constant all along the vortex filament and for all
times.
Corollary (2). Under the conditions of the theorem, vortex lines move with the
fluid.
Proof. Let C be any closed curve drawn on the surface of a vortex tube. Let S
be the portion of the vortex tube rimmed by C. By definition vortex lines lie
on S. Thus
d D
(n̂. curl q )dS (n̂. curl q )dS 0
dt S S Dt
Corollary (3). Under the conditions of the theorem, if the flow is irrotational in
a material region of the fluid at some particular time (e.g. t = 0 or t = t 0), the
flow is always irrotational in that material region thereafter.
i.e. If the motion of an ideal fluid is once irrotational it remains irrotational for
ever afterwards provided the external forces are conservative and density is a
function of pressure p only.
179
FLUID DYNAMICS
Proof. Suppose that at some instant (t = t0), the fluid on the material surface S
is irrotational
Then, = 0 (1)
n̂. dS = 0
S
1.2. Remarks (i) The above three corollaries are properties of vortex filaments.
(iii) From the results of the theorem, we conclude that vortex filaments must
either form closed curves or have their ends on the bounding surface of
the fluid. A vortex in an ideal fluid is therefore permanent.
1.3. Vorticity Equation. Euler‟s equation of motion for an ideal fluid under
the action of a conservative body force with potential per unit mass is
Dq q 1 1
q 2 q p (1)
Dt t 2
where the vorticity curl q q . If the fluid has constant density, then
taking curl of equation (1), we get
q 1 1
q 2 (q ) = p
t 2
FLUID DYNAMICS 180
q
(q ) 0
t
( q) (q ) 0
t
(q )
t
= ( )q (q )
(q ) ( )q
t
D
i.e. ( ) q (2)
Dt
Equation (2) is called Helmholtz‟s vorticity equation. For two-dimensional motion, the
vorticity vector is perpendicular to the velocity vector q and the R.H.S. of (2) is identically
zero. Thus, for two dimensional motion of an ideal fluid, vorticity is constant.
In the case, when body force is not conservative, equation (2) becomes
D
( )q curl F
Dt
about the axis r = 0 where (r, , z) are cylindrical polar co-ordinates. The cylindrical polar
resolutes of velocity are [qr(r, z), 0, qz(r, z)]. Show that if a fluid particle has vorticity of
magnitude 0 when r = r0, its vorticity when at general distance r from the axis of
symmetry has magnitude = (0/r0)r, if any body forces acting are conservative.
181
FLUID DYNAMICS
D
( ) q (1)
Dt
Now,
r̂ rˆ ẑ
1
curl q
r r z
q r ( r , z) 0 q z ( r , z)
1
= r̂ q z (r, z) r ˆ q r (r, z) q z (r, z) ẑ(0)
r z r
q r q z
= z r . (2)
Therefore,
1
( ) r̂ ˆ ẑ
r r z
q q 1
= ˆ r z . r̂ ˆ ẑ
z r r r z
1 q r q z
= ˆ.r̂ 0 ˆ.ẑ
r z r
1 q q
Thus (.)q r z (q r r̂ q z ẑ)
r z r
qr q r q z r̂
= z r
r
FLUID DYNAMICS 182
q r q r q z ˆ r̂ ˆ ẑ
= | , 0 (3)
r z r
qr
Hence ( .) q | using (2) (4)
r
D qr
(5)
Dt r
D
r q. r̂ (6)
Dt
Dr Dr
Since r 2 r 2 r. r
Dt Dt
r Dr Dr Dr Dr Dr
. r̂. r̂.q
r Dt Dt Dt Dt Dt
D Dr
r
Dt Dt
D Dr
r
D Dr Dt Dt 0
r 0 2
Dt Dt r
D
0 const. 0
Dt r r r0
r 2 0
r
r0
2. Vorticity in Two-dimensions
q (u, v,0), , ,0
x y
v u
Therefore, q (0,0, )
x y
v u
= k̂
x y
v u
Also, =
x y
Thus k̂
d
( )q gives
dt
d
0 constant
dt
i.e. = constant.
u= , v
y x
2 2
Therefore, k̂ 2 2 k̂ 2
x y
i.e. = 2
2.1. Circular Vortex. The section of a cylindrical vortex tube whose cross-
section is a circle of radius a, by the plane of motion is a circle and the liquid
inside such a tube is said to form a circular vortex.
If is the angular velocity and a2 the cross-sectional area of the vortex tube,
then circulation
q d r curl q n̂ dS curl q d S
C S S
= dS a 2 k(say)
S
This product of the cross-section and angular velocity at any point of the vortex
tube is constant along the vortex and is known as the strength of the circular
vortex.
2.3. Different Types of Vortices. We may divide vortices into the following
four types
(i) Forced vortex in which the fluid rotates as a rigid body with constant
angular velocity.
(ii) Free cylindrical vortex for which the fluid moves along streamlines
which are concentric circles in horizontal planes and there is no
variation of total energy with radius.
(iii) Free spiral vortex which is a combination the free cylindrical vortex
and a source (radial flow)
(iv) Compound vortex in which the fluid rotates as a forced vortex at the
centre and as a free vortex outside.
If the circulation in a closed circuit is 2k, then k is called the strength of the
circulation.
W = + i = ik log z (1)
This shows that the circular cylinder is a streamline and thus equation (1) gives
the required complex potential for circulation about a circular cylinder.
When the fluid moves once round the cylinder in the positive sense, increases
by 2 and then
= 2k
dw ik
Also,
dz z
dW k
q=
dz r
i.e. k = rq
Therefore, k = q when r = 1
2 k̂
2 2
i.e. = = 2
x 2 y 2
2 1 1 2
=
r 2 r r r 2 2
1 d d
= r , for r < a
r dr dr
= 0, for r > a
d d
i.e. r = r, for r < a
dr dr
= 0, for r > a
Integrating, we find
d r2
r + A, for r < a
dr 2
= B, for r > a
We are interested in the fluid motion outside the cylinder |z| = a. Therefore,
integrating the second of the above result, we get
The constant C may be chosen to be zero. Further, for r > a, the vorticity is
zero and the fluid motion is irrotational, therefore velocity potential exists
and is related to as
1 B
r r r
= B + D
187
FLUID DYNAMICS
= B, neglecting D
Let k be the circulation while moving once round the cylinder, then
= B [(+2)] = 2B
B = k/2 = K(say)
= iK (logr + i)
k
= iK log z = i log z.
2
If the rectilinear vortex is situated at the point z = z0, then by shifting the
origin, we get
W = iK log (zz0)
If there are vortices of strengths K1, K2, …Kn situated at z1, z2,…, zn
respectively, then the complex potential is
W = m log z + iK logz
2.8. Example. Find the complex potential for the motion due to a system
consisting of a coincident line-source of strength m per unit length and line-
vortex of strength K per unit length in the presence of a circular cylinder of
radius a, whose axis is parallel to and at a distance b( > a) from the line of the
source and vortex. Show that the cylinder is attracted by a force of magnitude
Solution. We suppose the line-source and line-vortex to be at the origin, then the complex
potential is
When the circular cylinder |zb| = a (b > a) is inserted, the complex potential, by circle
theorem, becomes
a2
W = (iKm) log z + (iKm) log
z b b (2)
where
a2 a2
z b = z = b
b zb
By Blasius theorem, force on the cylinder C is given by
2
dW
X iY = [ sum of residues of within C ]
dz
(3)
2
dW iK m a2
2
iK m
Now = (4)
dz z a 2 b( z b) z b
189
FLUID DYNAMICS
2
dW a2
The only singularities of within C are at z = b and z = b since z = 0 is not
dz b
inside C.
Now,
K 2 m2
residue (z = b) = 2
Only product term of (4)
b
a2 2(K 2 m 2 )b
residue (z = b )= .
b (b 2 a 2 )
b 1
XiY = 2 (K2+ m2) b 2 a 2 b
= 2a2(K2 + m2)/b(b2a2)
Thus
The negative sign implies that the cylinder is attracted towards the origin where the spiral
vortex is situated.
2.9. Complex Potential for a Vortex Doublet. Two equal and opposite
vortices placed at small distance apart, form a vortex doublet.
z aei aei
= iK log iK log 1
z z
aei a 2 e 2i
= iK 2
.....
z 2 z
FLUID DYNAMICS 190
ie i e i ( / 2)
W=
z z
This is the required complex potential for a vortex doublet at the origin.
e i
Also, we note that the complex potential for a doublet at the origin is .
z
Thus, it follows that the complex potential of a vortex doublet is the same as
that for a doublet with its axes rotated through a right angle.
z z1 r P
= K log K log 1
z z2 r2
r2 r1
a2
W = iK log (zd) iK log d
z
a2
i.e. + i = iK log(zd) iK log z + iK log z + constant
d
(1)
191
FLUID DYNAMICS
= iK log (x d) 2 y 2
1/ 2 y
i tan 1
x d
2 2
1/ 2
iK log x y i tan
a 2 1 y
d a2
x
d
y
+ iK log x 2 y 2 i tan 1
x
a (a cos d) 2 a 2 sin 2 1/ 2
= K log
1/ 2
2
a2 2
a cos d a sin
2
| z | a z aei
x a cos,
y a sin
= K log d = constant.
This shows that the cylinder is a streamline. Thus (1) represents the complex
potential of the fluid motion. From (1), we observe that the image of a vortex
of strength K at z = d is a vortex of strength K at the inverse point z = a2/d
together with a vortex of strength K at z = 0 i.e. centre of the circle.
a2
W = Uz + U + ik log z. (1)
z
FLUID DYNAMICS 192
dW a 2 ik
q UU 2
dz z z
Ua 2 ik
U 0
z2 z
Uz2 + ikz U a2 = 0
ik k2
z= a 1 2 2
2U 4a U
Since a and U are constants, therefore the flow potential term depends very
much on the magnitude of k. We shall consider three cases.
k2
Case I. When k < 2aU i.e. 1, we put
4a 2 U 2
k2
2 2
sin2 and then
4a U
Thus the stagnation points are (a cos , a sin) and (a cos, a sin)
The stagnation points lie on the boundary of the cylinder. They lie on the
line MN below the diameter AB as shown in the fig. The velocity increases
above MN and decreases below MN.
B A
N M
(a cos,a sin) (a cos,a sin)
C
193
FLUID DYNAMICS
p 1 2
q constant
2
we observe that the pressure decreases above MN and increases below MN.
= 0, , z = + a
Therefore, MN coincides with AB and thus the stagnation points are at A and
B. Therefore we conclude that the circulation brings the stagnation points
downwards and put an upward thrust on the cylinder.
k2
Case II. When k = 2aU i.e. 1 , then sin = 1
4a 2 U 2
and thus the stagnation points coincide at C, the bottom of the cylinder.
k2 k2
Case III. When k > 2aU i.e. 2 2
> 1, then we put 2 2
= cosh2 so that
4a U 4a U
this shows that the stagnation points are inverse points w.r.t. the circular
boundary of the cylinder. One of these points lie inside and other is outside the
cylinder. The point which is inside the cylinder does not belong to the motion.
| z1 || iae | ae , outside the circle
| z 2 || iae | ae , inside the circle
We know that at the stagnation points (critical points), there are two
branches of the streamlines which are at right angles to each other. Thus the
liquid inside the loop formed at the stagnation points will not be carried by the
stream but will circulate round the cylinder
FLUID DYNAMICS 194
dW Ua 2 ik
U 2
dz z z
2
dW
= (sum of the residues of within the circle |z| = a)
dz
2
dW
By Cauchy' s Re sidue theorem as is a meromorphic function
dz
where X, Y are components of the pressure of the liquid and is the density of
the liquid
2
2 a2 2ikU a 2 k 2
2
dW
Now, U 1 2 1
dz z z z 2 z 2
The only pole inside the cylinder |z| = a is z = 0 i.e. a simple pole. The residue
at z = 0 is 2ikU
X = 0, Y = 2kU
This represents an upward thrust on the cylinder due to circulation. The lifting
tendency (k 0) is called the Magnus effect. The moment M is obtained to be
zero, since residue is zero in that case.
a2
W = U z ik log z represents a possible flow part a
z
circular cylinder. Sketch the streamlines, find the stagnation points and
calculate the force on the cylinder.
z ia
2.15. Example. Verify that W = iK log , K and a both real, is the
z ia
complex potential of a steady flow of liquid about a circular cylinder, the plane
195
FLUID DYNAMICS
y = 0 being a rigid boundary. Find the force exerted by the liquid on unit
length of the cylinder.
z ia
+ i = iK log
z ia
| z ia | ya y a
= ik log i tan 1 i tan 1
| z ia | x x
| z ia |
= K log
| z ia |
| z ia |
= constant = (say)
| z ia |
x2 + (ya)2 = x2 + (y +a)2 y = 0
dW 1 1
W = iK
dz z ia z ia
2
i dW
2 C dz
X iY = dz
2
i 1 1
i.e. X iY = iK dz
2 C z ia z ia
iK 2 1 1 2
2 C (z ia ) 2 (z ia ) 2 (z ia )(z ia )
= dz
The integrand has double poles at z = + ia. Out of these poles only z = ia lies
within C. Thus, we find residue at z = ia. It is only the last term of the
integrand which gives a non-zero contribution to the contour integral and the
appropriate residue at z = ia is
2 2 1 i
Lt (z ia )
(z ia )(z ia ) 2ia ia a
=
zia
K 2 i iK 2
X iY = ( 2i ) =
2 a a
K 2
X = 0, Y =
a
which shows that the liquid exerts a downward force on the cylinder of amount
K 2
per unit length. In case of moment M, the sum of residues is obtained
a
to be zero and thus M = 0
197
FLUID DYNAMICS
2.16. Motion of a Vortex Filament. We find the velocity of the point P(z)
due to a vortex filament K at z = z0. We know that, the complex potential is
W = iK log (zz0)
dW iK iK iK i
q i e
dz z z 0 Re R
where z z0 = Rei.
AP = R, arg (zz0) =
90
P u
K
R v
A
z0
iK i
u iv = e
Re
iK
= (cos i sin )
R
K K K
U= sin , v cos, q
R R R
v
Therefore, cot tan(90 )
u
d 2 1 d
2 = , where is a function of r only, due to
dr 2 r dr
d 2
symmetry i.e. 2 0
d
Thus, we get
1 d d
2 = r = , r < a (1)
r dr dr
d 1 A
r (2)
dr 2 r
1
qr = , q
r r
1 A
qr = 0, q = r
2 r
1
Therefore, q = r = 0 at r = 0
2
n
W = i Ks log (zzs) (3)
s 1
n K
dW s
u iv = i (4)
dz s 1 z z s
Further, the complex velocity of the vortex of strength Kr, which is produced
only by the other vortices, is
199
FLUID DYNAMICS
n K
ur i vr = i s
, where s r. (5)
s1 z r z s
so that
dW' dW iK r
ur ivr =
dz zzr dz z z r z z r
dW iK 2
z 1
dz zz1 z1 z 2
and
dW iK 1 iK 1
z 2
dz zz2 z 2 z1 z1 z 2
Therefore,
iK K iK K
K1 z 1 K 2 z 2 1 2 1 2 0
z1 z 2 z1 z 2
K1z 1 K 2 z 2 d K1z1 K 2 z 2
or 0 i.e. 0
K1 K 2 dt K1 K 2
Integrating, we get
K 1 z1 K 2 z 2
constant
K1 K 2
FLUID DYNAMICS 200
K 1 z1 K 2 z 2
The point divides AB in the ratio K2 : K1. This point remains
K1 K 2
fixed (not necessarily a stagnation point) and is called the
centroid G of the vortices at A and B.
Further
B(z2)
AG GB AB
K1
K 2 K1 K1 K 2
K2 G
K2
Therefore, AG = AB
K1 K 2 A(z1)
d
v=r =r
dt
The velocity of A is
K2
|u1 iv1| =
AB
K 2 AB K1 K 2
= . AG.
K1 K 2 (AB) 2
K1 K 2
where =
(AB) 2
3.2. Vortex Pair. A pair of vortices of equal and opposite strengths is called a
vortex pair.
Let K and K be the strengths of the two vortices at A(z = z1) and B(z = z2)
respectively. Then the complex potential is
= W1 + W2 (say)
K K
A (z=z1) B (z=z2)
dW2 iK
u1iv1 =
dz zz1 z1 z 2
dW
u2 iv2 = 1
dz zz2
iK iK
z 2 z1 z1 z 2
K
q1 = |u1 iv1| = ,
AB
K
q 2 | u 2 iv 2 | | |z1 z2| = AB
AB
z z1
Further, W = iK log
z z2
z z1 y
+ i = iK log i(1 2 ) tan 1
z z2 x
z z1 r
= K log K log 1
z z2 r2
r1
Therefore, the streamlines, = constant, are = constant.
r2
which are co-axial circles.
Thus the streamlines in case of a vortex pair are co-axial circles which have A
and B as limiting points.
FLUID DYNAMICS 202
3.3. Example. A vortex of circulation 2k is at rest at the point z = na (n > 1),
in the presence of a plane circular impermeable boundary |z| = a, around which
there is circulation 2k. Show that
1
=
n 1
2
Show that there are two stagnation points on the circular boundary z = aei
symmetrically placed about the real axis in the quadrants nearest to the vortex
given by
cos = (3n21)/2n3.
Solution. The circulation of vortex is 2k and thus the strength of vortex is k
2
a2
f a / z = ik log na
z
The complex potential, when the circular cylinder |z| = a is introduced into the
fluid, becomes f(z) + f (a2/z), by circle theorem.
Now, there is a circulation 2k around the cylinder. This is equivalent to the
line vortex at z = 0 of strength k.
a2
W = ik log (zna) ik log na +ik log z
z
a
= ik log (zna) ik log z + ik log z + ik log z + constant.
n
a
= ik log (zna) ik log z ik ( +1) log z + constant (1)
n
203
FLUID DYNAMICS
a
O(z=0)
A z A (z=na)
n
(+1)k k k
The velocity at point A is due to the motion of other two vortices (i.e.
excluding first term in (1))
Therefore,
ik ik ( 1)
dW
n a
zna a
na
dz
n
(Differentiating (1) and put z = an excluding Ist term of (1))
1
=
n 1
2
dW 1 1 1
ik
dz z na a z
z
n
dW ik ei 2n 3 cos 3n 2 1 1
usng
dz a (n 2 2n cos 1)(n 2 1) n 1
2
dW
0 for z = aei
dz
FLUID DYNAMICS 204
Thus
dW
0 2n3 cos 3n2+1 = 0
dz
3n 2 1
cos = (2)
2n 3
Now, we know that 1 cos 1 i.e. |cos| 1 therefore R.H.S. of (2) must lie
within these limits for to be real
Let us write
3n 2 1 3 1
f(n) = 3
3
2n 2 n 2n
Then
3 3
f(1) = 1 , and also, f (n) = 2
4
2n 2n
3
= 4
(1n2) < 0 for n > 1
2n
From here, we note that f (n) < 1 for n > 1. Thus for n > 1, f(n) decreases
monotonically from 1 at n = 1 to 0 as n. For all n >1, real values of are
obtained from (2). Two distinct values of are obtained for any given n > 1,
one of the values is = , where 0 < /2 and the other is = 2.
Hence the two stagnation points are symmetrically placed about the real axis in
the quadrants nearest to the vortex.
4. Vortex Rows
When a body moves slowly through a liquid, rows of vortices are sometimes
formed. There vortices can, when stable, be photographed. Here we consider
infinite system of parallel line vortices and two dimensional flow will be
presumed throughout.
4.1. Single Infinite Row of Vortices. We shall find the complex potential of
an infinite row of parallel rectilinear vortices (line vortices) of same strength K
and a distance „a‟ apart.
First, let there be 2n+1 vortices with their centres on x-axis and the middle
vortex having its centre at the origin. The vortices are placed at points
z = + na, n = 0, 1, 2, ……, symmetrical about y-axis. The complex potential
due to these vortices is
205
FLUID DYNAMICS
x
3a 2a a O a 2a 3a
z z 2 z2 z2 z2
= iK log 1 1 1 .......1
a a 2 2 2 a 2 32 a 2 n 2a 2
a
+ iK log (1)n (a2. 22 a2. 32a2………n2a2)
z
Ignoring the constant term and putting , we get
a
2 2 2
W = iK log 1 2 1 2 2 ..........
...1 2 2
n n
Making n , we find
z
W = iK log sin = iK log sin (1)
a
q =
d
W iK log zz0 | The motion is due to other vortices
dz
d z
= iK log sin iK log z
dz a z 0
FLUID DYNAMICS 206
z
cos a 1
= iK
sin
a z z
a z 0
Now, the velocity at any point of the fluid other than the vortices is given by
dW iK z
q u iv cot
dz a a
cos (x iy )
iK iK a
= cot ( x iy ) =
a a a
sin (x iy )
a
2 cos (x iy ) sin (x iy )
iK a a
=
a
2 sin (x iy ) sin (x iy )
a a
2x 2yi
sin sin
iK a a
=
a 2y 2x
cos i cos
a a
2x 2y
sin i sinh
iK a a
=
a 2y 2x
cosh cos
a a
K 2y K 2x
sinh sin
u= a a , v a a
2y 2x 2y 2x
cosh cos cosh cos
a a a a
z
Also, we have W = + i = iK log sin
a
z
and w = i = iK log sin
a
207
FLUID DYNAMICS
z z
( +i) ( i) = iK log sin iK log sin
a a
z z
2 i = iK log sin sin
a a
Streamlines, = constant, are found to be
2y 2x
cosh cos = constant.
a a
4.2. Double Infinite Row of Vortices. Let us suppose that we have a system
consisting of infinite number of vortices each of strength K evenly placed
along a line AA parallel to x-axis and another system also consisting of
infinite number of vortices each of strength K placed similarly along a
parallel line BB. Let the line midway between these two lines of vortices be
taken as the x-axis.
y
A A
z12a z1a z1 z1+a z1+2a
x
O (0,
0)
B B
z22a z2a z2 z2+a z2+2a
Let one vortex on infinite row AA be at z = z1 and one vortex on infinite row
BB be at z = z2, so that the system consists of vortices K at z = z1 + na and
vortices K at z = z2 + na, n = 1, 2, ….
(z z1 na)(z z1 na)
W = iK log
n 0 (z z 2 na)(z z 2 na)
( z z1 ) 2 n 2 a 2
= iK log 2 2
(z z 2 ) n a
2
0
z z1 (z z1 ) 2 n 2 a 2
= iK log iK log
z z2 n 1 (z z 2 ) 2 n 2 a 2
FLUID DYNAMICS 208
z z1 1 ( z z1 ) 2 n 2 a 2
= iK log 2 2
(1)
z z 2 n 1 1 (z z 2 ) n a
2
2
Now, since sin = n 2 2 real or complex,
1
n 1
(z z1 ) (z z 2 )
we get, on setting =
a a
(z z1 ) (z z1 ) ( z z1 ) 2
sin
a
a
1
2 2
n 1 n a
(z z 2 ) (z z 2 ) (z z 2 ) 2
sin
a
a
1
2 2
n 1 n a
( z z1 )
sin a
W = iK log (2)
sin (z z 2 )
a
dW
u iv = iK [cot (z z1) cot (zz2)], where = /a
dz
d
u1 iv1 = W iK log(z z1 )
dz zz1
1
= iK cot (z z 2 ) cot (z z1 )
z z1 z z
1
Since
1
cot(z z1 ) (z z ) 0 as zz1
1
209
FLUID DYNAMICS
Therefore,
u1 iv1 = iK cot (z1 z2), = /a
(4)
4.3. Karman Vortex Street. This consists of two parallel infinite rows AA
and BB of vortices of equal spacing „a‟ so arranged that each vortex of
strength K of AA is exactly above the mid-point of the join of two vortices of
BB each of strength K ; as shown in the figure
y
i b
a+ib 2a+ib 3a+ib
A A
ib
x
O
i
b
B B
a 3a 5a
ib ib ib
2 2 2
sin (z ib )
W = iK log a
a
sin z ib
a 2
a
( Similar to (2) of previous article on putting z1 = ib, z2 = ib).
2
iK 2ib
= tan
a a
FLUID DYNAMICS 210
K 2b
= tanh
a a
K 2b
u1 = tanh , v1 0 .
a a
It can be shown that each of vortices at the rows. AA and BB move with the
same velocity. This means that the vortex configuration remains unaltered at
K 2b
all times, since both AA and BB have the same velocity tanh in x-
a a
direction. Hence the street moves through the liquid with this velocity.
dW niK z n 1
u iv = n
dz z an
211
FLUID DYNAMICS
qa =
d
W iK log(z a )za
dz
d zn a n
= iK log
dz z a z a
=K
d
dz
log z n 1 a z n 2 ...... a n 1 z a
z n 1
a z n 2 ............... a n 1
z a
K (n 1) (n 2) ........ 2 1 K n 1
= =
a n a 2
Therefore, time period is given by
4a 2
i.e. T=
K(n 1)
Hence the result.
k
4.5. Remark. If we use K = i.e .k 2K , then
2
4a 2 8 2 a 2
T=
k (n 1)k
(n 1)
2
4.6. Example. Three parallel rectilinear vortices of the same strength K and in
the same sense meet any plane perpendicular to them in an equilateral triangle
of side a. Show that the vortices move round the same cylinder with uniform
2a 2
speed in time .
3K
Solution. Here, the vortices are situated at points z = re2pi/3 where p = 0, 1, 2,
A
FLUID DYNAMICS 212
A
a a
O
r 60o
B C
D a/2
a 3
From OCD, r cos 30 o r
2 2
a
r= , where r is the radius of the cylinder.
3
The complex potential of the system is
uA ivA =
d
W iK log(z r)zr
dz
d z3 r 3
= iK log
dz z r z r
=K
d
dz
log(r 2 z 2 z r ) zr
2z r 1
=K 2 r
K.
r z z z r
2
r
Therefore, if T be the time period during which the vortex A moves round the
cylinder, then
2
2r 2r 2 2 a 2a 2
T=
K/r K K 3 3K
Hence the result.
213
FLUID DYNAMICS
2 1 2
=
2
(1)
x 2 c 2 t 2
O O x + cT
x
(Profile at t) (Profile at t +T)
Thus the shape of the wave profile in (3) remains unchanged when it has moved a distance cT.
For this reason the wave profile represented by equation (2) is called a wave without change of
shape or undistorted wave. Equation (2) represents a wave which propagates with time. Such
a wave is also termed as progressive wave. Similarly, the function defined by
satisfies the wave equation (1) and it represents a disturbance moving without distortion in the
negative x-direction with speed c.
represents the superposition of a forward and a backward travelling wave, each moving with
speed c. From equation (5), we can show that
2 2
2
c (6)
t 2 x 2
where equation (6) is known as the one-dimensional wave equation and the form in (5) is its
general solution.
5.3. Wave Equations in Two and Three Dimensions. If a disturbance takes place in three
dimensions in such a way that the disturbance is constant over any plane perpendicular to the
direction of propagation, then the wave is called a plane wave and any such plane is called a
wave-front. If such a wave is travelling with speed c in a direction specified by the unit vector
n̂ = [l, m, n], then the function f(lx + my + nz ct) satisfies these requirements since the wave
fronts have equations lx + my + nz = constant at any considered time. Similarly g(lx + my +
nz + ct) would represent a disturbance travelling in the direction n̂ with the same speed.
Hence the function
represents the superposition of plane waves travelling with speeds c in the directions + n̂ .
Finding the second-order derivatives of w.r.t. x, y, z, t and using the fact that l2 + m2 + n2 = 1,
we get the wave equation in three dimensions as
2 2 2 1 2
=
2
(8)
x 2 y 2 z 2 c 2 t 2
Equation (7) gives the general solution of (8). The solution (7) can also be expressed as
= f (n̂ r ct) g(n̂ r ct) (9)
2 2 1 2
2 = (10)
x 2 y 2 c 2 t 2
where l2 + m2 = 1.
5.4. Spherical Waves. Let us consider the three dimensional wave equation
1 2
= 2
2
(12)
c t 2
2 2 1
2 sin
r 2
r r r sin
1 2 1 2
+ (13)
r 2 sin 2 c 2 t 2
2 2 1 2
r 2 r r c 2 t 2
or
2 1 2
(r ) = (r ) (14)
r 2 c 2 t 2
The general solution of the one-dimensional wave equation gives the solution of (14) for r as
x
f ( x ct) a cosm t ,
c
(16)
x
g( x ct) a cosm t
c
where a, c, m are constants. If the wave profile is either a sine or a cosine function, then the
waves are harmonic waves. Thus, (16) represents harmonic waves. Superposition of the
functions in (16), gives
mx
= 2a cos cos (mt) (17)
c
This type of disturbance is known as stationary wave, since its profile does not move. Thus at
1 cπ
p , where p is an integer. These
all times = 0 at the fixed positions where x =
2 m
pcπ
determine the positions of zero displacement, called nodes. The points where x =
m
determine the positions of maximum displacement, called antinodes. In both forms f(xct),
2π
g(x+ct), the amplitude is a. The period or periodic time in each case and also in , is ,
m
denoted by T. The angular frequency is m and the frequency in cycles per unit time,
m 1
denoted by n, is so that n = i.e. nT = 1. If, keeping t constant, we increase or decrease
2π T
2πc
x by an amount or whole-number multiple of it, then all f, g, remain unaltered. The
m
2πc
quantity is called the wavelength of the harmonic wave or of he combination of
m
2πc
harmonic waves which represents. The wavelength is denoted by . Thus, = . The
m
number of waves in a unit distance is called the wave number. If k is the wave number, then
1
k = 1 i.e. k = . Also, we observe that the amplitude of is 2a cos (m t), which varies with
λ
217
FLUID DYNAMICS
time. Further, if in the forms for f and g, the cosines are replaced by sines, then similar results
follow.
x
a exp im t , i 1 .
c
(i) Transverse : If the vibrations occur in planes at right angles to the direction of
propagation, then the waves are known as transverse waves. e.g. light waves
(ii) Longitudinal : When the vibrations occurs in the direction of propagation, then the
waves are called longitudinal waves. e.g. sound waves.
f(p, , T) = 0 (1)
For some gases, the molecules have negligible volume and there are virtually no mutual
attractions between the individual molecules. Such a gas is said to be a perfect gas and its
equation of state (1) takes the simple form
p = RT or pv = RT (2)
1
where v = is the volume of unit mass of the gas and R is a constant for the particular gas
under consideration. Let Q be the amount of heat added to unit mass of a substance so as to
produce a temperature increment T. Then the rate of increase of heat added with temperature
Q
rise is . This defines a quantity known as the specific heat of the substance, which is the
T
heat addition per unit mass of the substance required to produce unit temperature rise. The
Q
quantity may not be unique. For gases, it will depend on the manner in which the heat has
T
been supplied. We can associate a specific heat at constant pressure, denoted by C p, and a
specific heat at constant volume, denoted by Cv, which are defined as
Q Q
Cp = , Cv (3)
T p T v
FLUID DYNAMICS 218
For a perfect gas, the kinetic theory shows that Cp and Cv are constant and that
Cp/Cv = (4)
From first law of thermodynamics, the relation for dQ, for a perfect gas, using pv = RT, can be
written in the form
Cp Cv
dS = dv dp (5)
v p
dQ
where dS = .
T
Cp C
0, v 0
But
p v v p
Cp Cp
p v v p
so that
S S0 = Cp log v + Cv log p
S So
or log (pv) =
Cv
S S0
or pv = exp (6)
Cv
The quantity S is called the entropy per unit mass and dS is the entropy differential. Flows for
which S is constant are called isentropic and from equation (6), we find that they are
characterised by
1
where v = and K is a constant.
The change in a substance from a fixed state A to a fixed state B may be effected in many
ways. A change from A to B in which the temperature T is kept constant is said to be
isothermal. In case of a perfect gas, since p = RT is the equation of state, an isothermal
change would be governed by Boyle‟s law given by p k i.e. pv = constant.
219
FLUID DYNAMICS
An adiabatic change is one in which there is not heat exchange between the working
substance and its surroundings. If a change is made so that the entropy of every single particle
of the working substance remains constant, then such a change is termed as isentropic. When
the entropy of every single particle of a substance of fixed mass is the same and remains
constant in any change, then the change is said to be homentropic. The constant under
reference is the same for each considered small quantity of gas in isentropic flow but a
different constant attaches to each such quantity. For homentropic flow, however, the
constant is the same throughout the entire volume of gas.
We suppose that a small disturbance is created within a non-viscous gas such that
(i) The disturbance is propagated as a wave motion, known as a sound wave, by small
to-and-fro motions of the medium without resulting in mass transport of the medium
itself.
(ii) Before the disturbance, the fluid is at rest and thus the motion is irrotational so that a
velocity potential exists at each point of the fluid. The fluid velocity at any point is
q = .
(iii) The squares and products of all disturbances from the equilibrium state
specified by pressure p0 and density 0 can be neglected. Also q = | q |
is so small that q2 can be neglected.
(iv) The isentropic law p = k holds as a consequence of assuming that changes take
place so rapidly that heat exchanges and hence entropy changes are negligible, where
is an adiabatic constant.
We write = 0(1 + s), where s is the condensation of the medium. This is a dimensionless
quantity expressing the fractional increment of local fluid density during the disturbance over
the undisturbed density 0 of the medium. It is a function of time t and space co-ordinates (x,
y, z) if the motion is three-dimensional. The equation of continuity
(q) = 0 (1)
t
becomes
s
{(1+s)} = 0 (2)
t
If we assume that the velocity is so small in magnitude that s is negligible, then (2)
simplifies to
s 2
= (3)
t
In the absence of body forces, the equation of motion becomes
dp
= constant (4)
t
FLUID DYNAMICS 220
p0
Assumption (iv) implies that p = k, where k = . Therefore,
0
dp p
k 1 0 1
d 0
1
a 02 a 02 (1 s) 1 ~ a02,
0
γp 0
a 02 (5)
ρ0
Hence
dp dp d 2 d
. ~ a 0
d
a 02 log + constant
t
= a02 {log 0 + log (1 +s)} + constant
= a02s (7)
t
Eliminating s from (3) and (7), we have
2
= a02 2 (8)
t 2
Equation (8) is a wave type equation and shows that small disturbances are propagated in the
gas with speed
221
FLUID DYNAMICS
12 12
dp p
a0 = 0 (9)
d 0
This speed is called the speed of sound in the gas. A vibrating tuning fork would produce
disturbances propagating with such a speed. Equation (9) is obtained under isentropic
conditions. When we wish to emphasize this we write
p
a2 = .
s
(q) 0 (1)
t
and Euler‟s equation of motion is
q 1
(q )q F p (2)
t
In the case of steady motion under no body forces, (1) and (2) become
(q ) 0 (3)
1
(q )q p (4)
1 2 dp
q = constant (5)
2
In the special case of isentropic flow for which the entropy of each particle remains constant
along any streamline and for each such particle p = k so that (5) reduces to
1 2
q kr 2 d = constant
2
1 2 k 1
or q = constant
2 1
1 2 a2
or q = constant (6)
2 γ 1
FLUID DYNAMICS 222
p
where a2 =
s
In (6), the constant is same along any streamline, but unless the flow is homentropic, it will
vary from one streamline to another. The condition for constant entropy of a fluid particle in
its steady motion along a streamline is
(q ) S = 0 (7)
S
since the total rate of change of the particle‟s entropy S per unit mass is (q )S in the
t
general time-varying case. In addition, the equation of state may be taken in either of the
equivalent forms
f(p, , T) = 0 (8)
or
p = F(, S) (9)
where the forms f, F are known. Equation (9) is more convenient for discussing the cases of
isentropic and homentropic flows. Equations (3), (4), (7) and (8) or (9) are distinct equations
for determining p, , q and S. Bernaulli‟s equation is really derived from the equations of
motion, but the forms (5) and (6) are very useful. Thus the problem of determining the nature
of gas flow is solvable.
Now, since
1
(q )q q 2 q ξ ,
2
1 1
q q 2 q p (10)
2
p p
dp = d dS
S S
so that
p p
p = S
S S
S = 0 and p = a2 ,
223
FLUID DYNAMICS
1
2
a
q q 2 q
2
q + () q = 0
and so
1
q q 2 = a2 q (11)
2
Equation (11) is anther important result.
Let q be the speed of a gas at a certain location and let a be the local speed of sound, where
1/ 2 1/ 2
dp p
a = .
d
Then the local Mach number M is defined to be the dimensionless parameter M = q/a.
Case (i) : When q = a, M = 1 then the flow is said to be sonic since the speed of gas flow and
the local speed of sound are the same.
Case (ii) : When M < 1, q < a then the flow is subsonic i.e. the speed of the gas flow is less
than the local speed of sound.
Case (iii) : When M > 1, q > a then the flow is termed as supersonic i.e. the speed of gas flow
exceeds the local speed of sound.
Subsonic and supersonic flows have many different physical features. To know what type of
flow pattern is realized, we should know the Mach number. We examine these physical
features by discussing the nature of spherical sound waves in a moving stream of gas.
8.1. Theorem : Show that for subsonic flow, the spherical disturbances speed throughout the
entire field, whereas for supersonic flow, the disturbances are confined to the interior of the
cone, the region outside the cone being unaffected by the disturbances.
Proof : Let us consider a source O emitting spherical sound waves in a gas at rest. Spherical
wave fronts centred at O travel outwards from O and at time t after starting from O, the
disturbance is spread uniformly over the surface S of the sphere with centre O and radius a.t,
as shown in the figure 1.
S S
P
a.t O
Vt
FLUID DYNAMICS 224
O
a.t
Figure 1. v
Now, suppose that the gas flows with uniform velocity v past the source. Then at time t,
every particle of S is displaced through a distance v t relative to S and the disturbance which
was initially at O is now on the surface S of a sphere with centre P and radius a.t, where
OP v t. Here, M = v/a. When M < 1, v < a and O lies within S. When M > 1, v > a and O
lies outside S. We discuss these two cases in turn.
Case (i) : When M < 1, let P1, P2, P3,… denote the centres of the spherical disturbances at
times t, 2t, 3t,…, the radii of the corresponding spheres being at, 2at, 3at,….
3a
t 2at
at
P3 P2 P1 O
Figure 2
It is seen that the disturbances at times t, 2t, 3t,…, are on the boundaries of non-intersecting
spheres.
Case (ii) : When M > 1, then O lies outside the spheres as shown in the figure 3. It is seen that
the
3at
2at at
O
P3 P2 P1
Figure 3
spheres intersect. They have an envelope which is a right circular cone having vertex O and
axis OP1P2P3…..
From the above two cases, we conclude that for subsonic flow the spherical disturbances
spread throughout the entire field while for supersonic flow the disturbances are confined to
225
FLUID DYNAMICS
the interior of the cone, the region outside the cone being silent, i.e., supersonic flow is
characterised by a domain of dependence, which in the above case is the conical interior.
Corollary (1). In figure 3, let be the semi-vertical angle of the right-circular cone, then
at 2at 3at
sin =
vt 2vt 3vt
a 1
i.e. sin = .
v M
The angle is termed as the Mach angle and it is real only when M 1. It does not exist for
subsonic flow. The cone is called a Mach cone.
Corollary (2). In two dimensions, the spheres in the above models become circles and the
cone becomes pair of Mach lines or Mach waves.
8.3. Remark. If an aircraft is flying overhead at subsonic speed, any observer on the ground
will hear the disturbance once the sound waves have spread out to meet him. However, if the
aircraft is travelling at supersonic speed, disturbances will be confined to a domain of
dependence relative to the aircraft and the observer will hear noise only when he comes within
this domain. Thus, one may see a supersonic aircraft or missile travelling overhead but only
hear the sound some little while afterwards when the domain of dependence encloses him.
We have obtained that Bernaulli‟s equation for isentropic gas flow along a streamline is
1 2 a2
q = constant
2 γ 1
(1)
The L.H.S. of (1) shows that the maximum value of q, denoted by q max., occurs whenever a = 0.
Such case corresponds to gases expanding to zero pressure and is entirely theoretical i.e.
cannot be obtained in practice. Also we introduce the critical speed of sound a * which is
defined to be the value of a when q = a, and the stagnation speed of sound a0 corresponding to
q = 0 i.e. when the fluid is locally at rest. Then (1) can be written as
1 2 a2 1
q q 2max
2 1 2
1 2 a 02
a* (2)
2( 1) 1
Equation (2) gives three different forms of the constant on R.H.S. of (1).
p
Since a2 = , other forms of (2) are
FLUID DYNAMICS 226
1 2 p ( 1)p* p 0
q (3)
2 ( 1) 2( 1)* ( 1) 0
where p* is critical pressure, * is critical density for local sonic flow with q = a, whereas,
p0 is stagnation pressure, 0 is stagnation density for local condition of rest q = 0.
p RT
Cp T (4)
( 1) 1
where Cp R/(1) is the specific heat at constant pressure. Thus Bernoulli‟s equation along a
streamline can be written as
1 2
q CpT = CpT0 (5)
2
Here, T is the local temperature in Kelvins and T 0 is the stagnation temperature at a point on
the same streamline where q = 0. Dividing both sides of (5) by C pT, we get
T0 1 ( 1)
1 q2
T 2 p
1
=1+ (1) M2 (6)
2
q2
where M2 = 2
= q2/p, M being the Mach number. For isentropic flow, we have the
a
relations
( 1) 1
a 02 p 0 p T0 p 0
0
T p
(7)
a 2 0
so that
γ /(γ1)
p0 1
1 (γ 1)M 2 (8)
p 2
1/( 1)
0 1
1 ( 1)M 2 (9)
2
Other isentropic relations, which are easily found from (5), are
227
FLUID DYNAMICS
2 2
T γ 1 q γ 1 q
1 1
γ 1 a *
(10)
T0 2 a0
γ ( γ 1) γ ( γ 1)
p γ 1 q γ 1 q 2
2
1 1
2 a 0 γ 1 a *
(11)
p0
1 ( 1) 1 ( 1)
1 q 1 q 2
2
1 1
2 a 0 1 a *
(12)
0
γ /(γ 1)
p* 2
p 0 γ 1
(13)
1 /( 1)
* 2
0 1
(14)
T* 2
(15)
T0 γ 1
p* T*
~ 0.528, * ~ 0.630, ~ 0.833.
p0 T0
10. Reservoir Discharge through a Channel of Varying Section (Flow Through a Nozzle)
Let us consider a reservoir containing stationary gas at high pressure p 0, density 0,
temperature T0. An open-ended axially symmetric channel is fitted to the reservoir and we
assume that the gas discharges steadily and isentropically into the air at the open section where
the pressure is less than p0. Let the section of the channel vary so slowly that to a first order of
approximation, the velocity is constant across any section. However, the velocity varies from
section to section. Here, the flow can be considered as one-dimensional.
At a location of the channel where the cross-sectional area is A, let p be the pressure, be the
density and u be the gas speed. For steady flow, the equation of continuity across the section is
uA = constant
Differentiating, we get
FLUID DYNAMICS 228
udA + A du + uA d = 0
dA d du
0 (1)
A u
p0(high)
0 u
A
T0
Figure 1.
Bernoulli‟s equation is
1 2 dp
u constant (2)
2
dp
u du + =0 (3)
d
Putting dp = a2d in (3) and eliminating from (1) and (3), we obtain
du dA
(1M2) (4)
u A
where M = u/a i.e. the local Mach number. We discuss the following two cases.
Case I : If M < 1, equation (4) shows that a decrease in A produces an increase in u and
conversely. Thus, to accelerate subsonic flow through a channel it is necessary to decrease the
channel section A downstream of the flow.
Case II : If M > 1, equation (4) show that A and u increase or decrease together. Thus, to
accelerate supersonic flow it is necessary to widen the channel downstream of the flow.
d du
M 2 . (5)
u
229
FLUID DYNAMICS
which indicates how the fluid density varies with changing Mach number. In fact equation (5)
shows that for a given speed increment there is a density drop whose magnitude increases with
increasing Mach number. Further, for M 1, the drop in density is so large that the channel
must expand to satisfy continuity requirements.
dA
From equation (4), if 0 , then either M = 1 or du = 0. The case du = 0 is realized in
A
incompressible flow where the speed of the flow reaches a maximum at the stage when the
channel section attains a minimum area of cross-section. For compressible fluids, M may be
unity when the section A is a minimum.
To summarise the above results we may say that if, starting with subsonic flow in a channel,
we decrease the section downstream, then the flow is accelerated until the section has attained
a certain minimum at which the Mach number is unity. If beyond this minimum section we
now widen the channel, then the flow can be accelerated downstream of the section to produce
supersonic flow. This illustrates the principle of flow through a nozzle. The minimum
section is termed as throat, as shown in figure 2.
Throat
Figure 2
10.1. Maximum Mass Flow Through a Nozzle. We consider a channel which is tapering
steadily to a minimum section at the outlet. Let A be the section at the outlet of the channel
where the velocity is u, pressure p and density . Then, applying Bernoulli‟s equation along a
stream-line from the reservoir to the section A, we get
1 2 p p0
u (1)
2 1 1 0
so that
1/ 2
2p 0 p0
u= 1 (2)
( 1)0 p 0
1/
0 p 0
, (2) becomes
p
Using
1/ 2
2p ( 1) /
p
1
0
u= (3)
( 1)0 p0
m = u A = 0 u A(p/p0)1/ (4)
FLUID DYNAMICS 230
1/ 2
2p p 2 / p ( 1) /
m=A 0 0
1
(5)
1 p0 p0
To find the stationary values of m for fixed A and variable p/p 0, we write P = p/p0 so that
where k is constant
dm k( 1) (2 ) / 2
2m P P ( 1) / (7)
dP 1
dp
0 when P = {2/(+1)}/(1) = P* (8)
dP
dm dm
Also, 0 when P P* , 0 when P P* These inequalities depend on the
dP dP
assumption that for any gas > 1. From these inequalities, we conclude that m is maximum
when P = P*
γ /(γ 1)
p 2
p 0 γ 1
i.e. (9)
and
1/ 2
2 2
2 /( 1)
mmax. = A p 0 0 (10)
1 1
p*
Thus, we find that P* , where p* is the pressure at a point where M = 1. It therefore
p0
follows that for maximum isentropic mass flow, conditions at the exit plane are sonic.
1/ 2 1/ γ ( γ 1) / γ 1 / 2
m γ 1 1 p p
(γ 1)1/(γ1) 1
m max . γ 1 2 p0 p 0
(11)
Shock waves are not waves in real sense. These are plane discontinuities, pulse-like in nature
and are sometimes more appropriately called shock fronts. For the formation of shock waves,
we consider a piston which is being driven with uniform velocity u into a long open tube,
known as shock tube.
Gas moving
U Rest
u
Shock
The gas particles within the vicinity of the piston acquire the uniform velocity u but those some
way ahead of the piston are at rest. A plane normal discontinuity or shock front travels
forwards with velocity U(> u) into the virgin gas as the piston advances into the tube. The
shock is the mechanism by which the gas between it and the piston acquires the velocity u.
The existence of the shock can be detected experimentally by certain delicate kinds of
photographic methods such as shadowgraph. The velocity U > a, the local speed of sound in
the fluid. A simple physical explanation of the shock formation in this case is as follows :
Let us consider another more analytical model in which first of all we consider a sound wave
of velocity a moving into a gas at rest (figure. 1). The pressure ahead of the wave is p, then
density and the particle velocity zero. The pressure immediately behind the wave is p + p,
the density + and the particle velocity u. Here, the disturbance is assumed to be
weak so p, , u are small. Figure 2 shows the equivalent model obtained when the sound
wave is brought to rest by imposing a backward velocity a on the entire system.
p
p+p p
p+p
+ +
0
u a au a
Figure 1. Figure 2.
Let us consider figure 2, where we apply the equation of continuity across the stationary wave
to obtain
Here, we have considered the mass flux per unit time across unit area of the wave.
1
p p
But 2
a = 0 (2)
0 0
So, we have
d
du = a 1/ 2 p10/ 2 0 / 2 ( 3) / 2 d
du 2 1/ 2 p10/ 2 ( 1) / 2
u= d d /2
( 0 ( 1) / 2 )
0 1 0
2a 0
( 1) / 2
= 1 (3)
1 0
Hence
1
( 0 ) ( 1) 2 1 (1) (u/a0) (4)
2
where a0 is the speed of sound in the undisturbed gas. Now
1
a 2 p 0 0 p
a 02 p 0 p 0 0
(5)
Hence
/ 1
a = a0 (/0)(1) 2 = a0 + (1)u (6)
2
Now, each small disturbance propagates itself at a velocity equal to local speed of sound
relative to the fluid. Thus if the fluid moves with velocity u, then velocity of propagation of
disturbance
1
= u + a = a0 + (+1)u (7)
2
Hence in a short time interval , the distance moved by the disturbance is
1
(u + a) = a 0 2 ( γ 1)u (8)
233
FLUID DYNAMICS
Equation (8) shows that in a given interval , the points of high velocity move farther to the
right than those of low velocity. The type of shock wave just considered is a normal shock,
since it is perpendicular to the incident gas stream. Another type of shocks are oblique shocks
which are inclined at oblique angles to the direction of flow.
UNIT – IV
1. Stress Components in a Real Fluid
Then the components of stress parallel to the axes are defined to be nx, ny,
nz, where
δFnx dFnx
nx = lt ,
δS0 δS dS
δFny dFny
ny = lt ,
δS0 δS dS
Fnz dFnz
nz = lt .
S0 S dS
FLUID DYNAMICS 234
In the components nx, ny, nz, the first suffix n denotes the direction of the
normal to the elemental plane S whereas the second suffix x or y or z denotes
the direction in which the component is measured.
If we identify n̂ in turn with the unit vectors î, ĵ, k̂ in (OX ), (OY ), (OZ) , which
is achieved by suitably re-orientating S, we obtain the following three sets of
stress components
The diagonal elements xx, yy, zz of this array are called normal or direct
stresses. The remaining six elements are called shearing stresses. For an
inviscid fluid, we have
Here, we consider the normal stresses as positive when they are tensile and
negative when they are compressive, so that p is the hydrostatic pressure. The
matrix
σ xx σ xy σ xz
σ yx σ yy σ yz (1)
σ zx σ zy σ zz
is called the stress matrix. If its components are known, we can calculate the
total forces on any area at any chosen point. The quantities ij(i, j = x, y, z) are
called the components of the stress tensor whose matrix is of the form (1).
Further we observe that ij is a tensor of order two.
z
Z P1
.P(x,y,z)
P2 x
y
Y
O
235
FLUID DYNAMICS
X
Let be the density of the fluid. The mass x y z of the fluid element
remains constant and the element is presumed to move alongwith the fluid. In
the figure, the points P1 and P2 have been taken on the centre of the faces so
x x
that they have co-ordinates x , y, z and x , y, z respectively.
2 2
x
At P2 x , y, z , since î is the unit normal measured outwards from the
2
fluid, the corresponding force components across the parallel plane of area
yz, are
x x xy x
xx xx yz, xy yz , xz xz yz .
2 x 2 x 2 x
δx
For the parallel plane through P1 x , y, z , since î is the unit normal
2
drawn outwards from the fluid element, the corresponding components are
x x xy x
xx xx yz, xy yz, xz xz yz
2 x 2 x 2 x
The forces on the parallel planes through P1 and P2 are equivalent to a single
force at P with components
σ xx σ xy σ xz
, , x y z
x x x
together with couples whose moments (upto third order terms) are
Similarly, the pair of faces perpendicular to the y axis give a force at P having
components
σ yx σ yy σ yz
, , x y z
y y y
σ zx σ zy σ zz
, , x y z
z z z
Combining the surface forces of all six faces of the parallelopiped, we observe
that they reduce to a single force at P having components
σ xx σ yx σ zx σ xy σ yy σ zy σ xz σ yz σ zz
, , x y z,
x y z x y z x y z
Now, suppose the external body forces acting at P are [X, Y, Z] per unit mass,
so that the total body force on the element has components [X, Y, Z] x y
z. Let us take moments about î direction through P. Then, we have
yz = zy.
Similarly, we get
Thus, the stress matrix is diagonally symmetric and contains only six
unknowns. In other words, we have proved that
σ xx σ yx σ zx
x y z x y z + X x y z (1)
where (X, Y, Z) is the body force per unit mass and being the density of the
viscous fluid. As the mass x y z is considered constant, if q = (u, v, w)
be the velocity of point P at time t, then the equation of motion in the
î direction is
σ xx σ yx σ zx du
x y z x y z + X x y z = (x y z) dt
σ xx σ yx σ zx du
or ρX ρ (2)
x y z dt
du u u u u d
u v w where q
dt t x y z dt t
u u u u 1 σ xx σ yx σ zx
t
u v w
x y z
X
x y z (3)
v v v v 1 σ xy σ yy σ zy
t
u v w
x y z
Y
x y z (4)
w w w w 1 σ xz σ yz σ zz
t
u
x
v
y
w
z
Z
x y z (5)
Equations (3), (4), (5) provide the equations of motion of the fluid element at
P(x, y, z).
In tensor form, if the co-ordinates are xi, the velocity components ui, the body
force components Xi, where i = 1, 2, 3, the equations of motion can be
expressed as
u i 1
u j u i, j X i ji,j (i, j = 1, 2, 3).
t
The change in the relative position of the parts of the body under some force, is
termed as deformation. By Hooke‟s law, the stress is proportional to strain in
case of elastic bodies, while in case of non-elastic bodies the stress is
proportional to the rate of strain.
Strain is of two kinds, the normal and the shearing. The ratio of change in
length to the original length of a line element is called normal (or direct)
strain. The shearing strain measures the change in angle between two line
elements from the natural state to some standard state. We shall consider two
dimensional case and then extend it to three dimensions. Let us consider a
rectangular element ABCD of an elastic solid with co-ordinates of A as (x, y)
and length of sides as x and y in the natural state.
ξ η
B(x +x, y) goes to B(x + +x + x, y ++ x)
x x
ξ
y C
y
D
(x,y+y) (x+x,y+y) y
y
D C
239
FLUID DYNAMICS
y
2
2 B
x
x
1
1 x
A(x,y) B(x+x,y) A
x
x
(Before deformation) (After deformation)
The point D(x, y +y) goes to the point
D(x + + y, y + + y + y) .
y y
Therefore, projected lengths of AB along x and y axes are x + x and
x
η
x
x
Thus,
2 2
(A B)2 = x x x (1)
x x
A' B'AB
xx =
AB
ξ 2 η 2
(1+xx)2 (x)2 = (x)2 1
x x
2 2
ξ η
(1+xx)2 = 1
x x
ξ
xx = .
x
η
yy =
y
The shearing strain xy at the point A is the change in the angle between the
sides AB and AD. The right angle | DAB between AB and AD is diminished
by xy = 1 + 2 = tan1 + tan2, 1 & 2 being small.
x y
x y
i.e. xy =
1 x 1 y
x y
1 1
η ξ ξ η
= 1 1
x x y y
1 1 η ξ
xy = (γ xy ) , upto first order.
2 2 x y
In the case of fluids, there is no resistance to deformation but only to the time
rate of deformation. Hence in fluid dynamics the rate of change of strain with
time i.e. rate of strain is to be used in place of strain in elasticity. Thus, for
viscous fluids, replacing strains by rates of strain, the corresponding results are
obtained to be
ξ ξ u
xx = (u )
t x x t x x
v η
yy = ,v
y t
1 η ξ 1 v u
xy =
2 t x y 2 x y
u v w
xx ,yy ,zz
x y z
1 1 v u
xy ( xy )
2 2 x y
(A)
1 1 w v
yz ( yz )
2 2 y z
1 1 w u
zx ( zx )
2 2 x z
The six quantities xx, yy, zz, xy, yz, zx in (A) are called components of the
rates of strain or gradients of velocity
Further, ( x' ) (lx + my)
t t
u = lu + mv
and v = mu + lv
u v u v
or xx = l m l l m m
x x y y
u v v u
= l2 m2 lm
x y x y
FLUID DYNAMICS 242
v'
Similarly yy = m 2 xx + l2 yy lm xy
y'
u' v'
xy = = 2lm (yy xx) + (l2m2) xy.
y' x'
which are the rates of strain of the new system in terms of rates of strain in the
original system. If we put back l = cos, m = sin, then
These equations give the transformation formulae for the rates of strain.
We observe that the rate of strain is also a tensor of order two, there must exist
at least two invariants of the rate of strain to the choice of co-ordinate systems.
These can be obtained as follows.
u v
= xx + yy = div q, q (u, v) (1)
x y
(' xy ) 2
xx yy = (l2 xx +m2 yy + lmxy) (m2 xx + l2 yy lm xy)
4
1
[2lm (yy xx) + (l2 m2) xy]2
4
2xy
= (l + 2 l m + m ) xx yy
4 2 2 4
(l4 + 2l2 m2 + m4)
4
2xy
= xx yy (2)
4
243
FLUID DYNAMICS
Equation (1) shows that the divergence of the velocity vector at a given point is
independent of the orientation of the co-ordinate axes. Equation (2 is related to
the dissipation function. i.e. loss of energy due to viscosity.
Let us now consider the general case of the rates of strain in three dimensions.
The direction cosines between x, y, z and x, y, z are related as follows.
x y Z
x l1 m1 n1
y l2 m2 n2
z l3 m3 n3
and
x = l1x + l2 y + l3 z
We shall use these relations to find out the rates of strain w. r. t. the new co-
ordinates x, y, z.
FLUID DYNAMICS 244
u v w
= l1 m1 n1 l1
x x x
u v w
+ l1 m1 n1 m1
y y y
u v w
+ l1 m1 n1 n1
z z z
v u
= l12 xx + m12 yy + n12 zz + l1m1
x y
v w w u
+ m1n1 n1l1
z y x z
= l12 xx + m12 yy + n12 zz + l1 m1xy + m1 n1 yz + n1l1 zx
Similarly, we have
v '
yy = = l22 xx + m22 yy + n22 zz + l2 m2 xy + m2n2 yz + n2l2 zx
y '
w '
zz = = l32 xx + m32 yy + n32 zz + l3m3 xy + m3n3 yz + n3l3 zx
z'
w ' v'
yz = = 2l2 l3 xx + 2m2m3 yy + 2n2n3 zz
y' z'
u' w '
zx = = 2l3l1 xx + 2m3 m1 yy + 2n3n1 zz
z' x'
xx + yy + zz = (l12 + l22 + l32) xx + (m12 + m22 + m32) yy
u v w
= div q
x y z
is invariant.
Similarly,
1
xx yy + yy zz + zz xx [(xy)2 + (yz)2 + (zx)2]
4
1
= xx yy + yy zz + zz xx [(xy)2 + (yz)2 + (zx)2]
4
is also invariant.
3.2. Remark. The stress tensor ij and the rates of strain ij follow the same
rules of transformation. Thus, the three equations in (B) can also be written for
stress components so that we get the relations between the original and the new
stress components as
FLUID DYNAMICS 246
σ xx σ yy σ xx σ yy
σ 'xx cos2θ σ xy sin 2θ
2 2
σ xx σ yy σ xx σ yy
σ 'yy cos2θ σ xy sin 2θ (C)
2 2
σ xx σ yy
σ 'xy sin 2θ σ xy cos2θ
2
For viscous fluid, the following assumptions are to be made to find the
relations between the stress and the rate of strain.
(ii) The relations between stress and rates of strain are invariant w.r.t
rotation and reflection of co-ordinate axes (symmetry).
(iii) The stress components reduce to the hydrostatic pressure when all the
gradients of velocity are zero.
u
i.e. xx = p = yy = zz, xx = 0 = yy = zz.
x
But the relations between the original and the new stress components are (from
equation (C))
σ xx σ yy σ xx σ yy
σ 'xx cos2θ σ xy sin 2θ
2 2
σ xx σ yy σ xx σ yy
σ 'yy cos2θ σ xy sin 2θ (3)
2 2
σ xx σ yy
σ 'xy sin 2θ σ xy cos2θ
2
1 1 1
σ 'xx (A1 + A2) xx + (B1+B2) yy + (C1+C2) xy
2 2 2
1 1
+ (D1+D2) + (A1A2) xx cos 2
2 2
1 1
+ (B1B2) yy cos 2 + (C1C2) xy cos 2
2 2
1
+ (D1D2) cos 2 + (A3xx + B3 yy + C3xy + D3) sin 2 (4)
2
Also, the relations between the original and the new rates of strain are
A1 A A
σ 'xx = (xx + yy) + 1 (xx yy) cos 2 + 1 xy sin 2
2 2 2
B1 B B
+ (xx + yy) 1 (xx yy) cos 2 1 xy sin 2
2 2 2
FLUID DYNAMICS 248
A1 A
(1+cos 2) + 2 (1cos 2) + A3 sin 2
2 2
A1 B
= (1 + cos 2) + 1 (1cos 2) C1 sin 2 | xx
2 2
B1 B
(1+cos 2) + 2 (1cos 2) + B3 sin 2
2 2
A1 B
= (1 cos 2) + 1 (1+cos 2) + C1 sin 2 | yy
2 2
C1 C
(1+cos 2) + 2 (1cos 2) + C3 sin 2
2 2
A1 B
= sin 2 1 sin 2 + C1 cos 2 | xy
2 2
D1 D
(1+cos 2) + 2 (1cos 2) + D3 sin 2 = D1
2 2
A2 = B1 = B(say), B2 = A1 = A(say)
C2 = A3 = C1 = B3 = C(say)
A1 B1 A B
C3 = , D1 = D2 = D (say), D3 = 0
2 2
The stress components in terms of the rates of strain are now obtained to be
σ xx Axx Byy C γ xy D
σ yy Bxx Ayy C γ xy D (7)
AB
σ xy C(xx yy ) γ xy
2
Let us take the symmetry w.r.t. the y-axis. If (x1, y1) are the new co-ordinates
of the point with co-ordinates (x, y), then
x1 = x, y1 = y
i.e. u1 = u, v1 = v
x1x1 Ax1x1 By1y1 C x1y1 D
y1y1 Bx1x1 Ay1y1 C x1y1 D (8)
AB
x1y1 C(x1x1 y1y1 ) x1y1
2
The relations (7) are invariant where there is a symmetry w.r.t. any co-ordinate
transformation and so
x1x1 Ax1x1 By1y1 C x1y1 D
y1y1 Bx1x1 Ay1y1 C x1y1 D (9)
AB
x1y1 C(x1x1 y1y1 ) x1y1
2
AB AB
xy = xy = xy, where = is called the co
2 2
efficient of viscosity.
= 2 xx + B q p
q (u, v)
u v
xx yy q
x y
These are the required relations between the stress components and the rates of
strain in two dimensions.
u
σ xx 2μ xx B q p 2μ λ q p
x
v
σ yy 2μ yy B q p 2μ λ q p (10)
y
w
σ zz 2μ zz B q p 2μ λ q p
z
v u
xy = xy = ,
x y
w v
yz = yz = (11)
y z
w u
zx = zx =
x z
where B .
251
FLUID DYNAMICS
= 2 q +3 q 3p
σ xx σ yy σ zz
i.e. p
3
This shows that the mean normal stress is equal to the hydrostatic pressure (i.e.
constant)
2μ
4.1. Remarks : (i) For compressible fluids, B =
3
(ii) Equations (10) and (11) may be combined in tensor form. Thus, if xi
denote the Cartesian co-ordinates, ui the velocity components (i = 1, 2,
3), then (10) & (11) may be collectively written as
1
p = i,i, = 0 for incompressible flow,
3
2
= for compressible flow.
3
(iv) For viscous fluids, stress is linearly proportional to rate of strain. This
law is known as Newton‟s law of viscosity and such fluids are known
as Newtonian fluids.
U î
Pu î
h
O
X
FLUID DYNAMICS 252
Z
The figure shows two parallel planes y = 0, y = h, a small distance h apart, the
space between them being occupied by a thin film of viscous fluid. The plane
y = 0 is held fixed and the upper plane is given a constant velocity U î . If U is
not very large, the layers of liquid in contact with y = 0 are at rest and those in
contact with y = h are moving with velocity U î i.e. there is no slip between
fluid and either surface. A velocity gradient is set up in the fluid between the
planes. At some point P(x, y, z) in between the planes, the fluid velocity will
be U î , where 0 < u < U and u is independent of x and z. Thus, when y is
fixed, u is fixed i.e. fluid moves in layers parallel to two planes. Such flow is
termed as Laminar flow. Due to viscosity of the fluid there is friction between
these layers. Experimental work shows that the shearing stress on the moving
plane is proportional to U/h when h is sufficiently small. Thus, we write this
stress in the form
U du
yx = lim μ
h 0 h dy
v = /.
For most fluids depends on the pressure and temperature. For gases,
according to the Kinetic theory, is independent of the pressure but decreases
with the temperature.
Let the element moves with the velocity q . The inertial force on the element
is
dq
d | F ma
dt
The resultant of inertial forces (or the rate of change of linear momentum) is
253
FLUID DYNAMICS
dq
FI d (1)
dt
Let X be the body force per unit mass, then the resultant of body force is
FB Xd (2)
f fx î x fy î y fz îz
where î x , î y , î z are unit vectors, dA is the vectorial area of the element and
Px, Py, Pz are components of stress vector, given by
Px σ xx î x σ xy î y σ xz î z
Py σ yx î x σ yy î y σ yz î z | Tix= ij xj (4)
Pz σ zx î x σ zy î y σ zz î z
FS î x Px dτ î y Py dτ îz Pz dτ (6) dA n̂ dS
Let us use the law of conservation of momentum. By this law, the time rate of
change of linear momentum is equal to the total force on the fluid mass.
Equating the resultant of body and surface forces with that of inertial forces,
we obtain.
dq
d Xd î x Px d î y Py d î z Pz d (7)
dt
dq
X Px î x Py î y Pz î z (8)
dt
FLUID DYNAMICS 254
This is the required equation of motion in vector form using the values of
Px , Py , Pz , we get
σ xx σ xy σ xz
Px
x y z
σ yx σ yy σ yz
Py
x y z
σ σ zy σ zz
Px zx
x y z
and let q = (u, v, w), X = (Xx, Xy, Xz) then the equations of motion can be put
as
du xy xz
X x xx
dt x y z
dv yx yy yz
X y (9)
dt x y z
dw zy zz
X z zx
dt x y z
These are the equations of motion in terms of the stress components. (We have
also drawn these equations previously)
d
q u v w
dt t t x y z
u
σ xx 2μ λ q p
x
v
σ yy 2μ λ q p
y
w
σ zz 2μ λ q p
z
u v
σ xy μ γ xy μ
y x
255
FLUID DYNAMICS
w v
yz ,
y z
w u
zx
x z
du p u 2 u v u w
X x 2 q
dt x x x 3 y y x z z x
dv p v 2 v w v u
X y 2 q
dt y y y 3 z z y x x y
dw p w 2 w u w v
X z 2 q
dt z z z 3 x x z y y z
(10)
2μ
where = compressible fluids.
3
du p 1 u v w
X x 2 u
dt x 3 x x y z
du p 1 u v w
X x 2 u
dt x 3 x x y z
dw p 1 u v w
X z 2 w
dt z 3 z x y z
dq q
(q )q X p 2 q ( q )
dt t 3
u v w
= constant, q 0.
x y z
dq q 1
(q )q X p 2 q
dt t
dq p
i.e. X + v2 q
dt
p 2 q
(q )q q 0, X 0
t
dq q p
(q )q X
dt t
dq dq q 2 dq qq dq
î r r î r î z z
dt dt r dt r dt
(1)
where îr , îθ , îz are the unit vectors in the directions of r, , z increasing.
dz
d
dr
257
FLUID DYNAMICS
dq
FI d
dt
dq q 2 dq qq dq
= î r r î r î z z d (2)
dt r dt r dt
Pr î r σ rr îθ σ rθ î z σ rz
Pθ î r σ θr îθ σ θθ î z σ θz
Pz î r σ zr îθ σ zθ î z σ zz
1 σ
Pr (rσ rr ) (σ rθ ) (rσ rz ) θθ
r r θ z r
1 σ
Pθ (rσ θr ) (σ θθ ) (rσ θz ) + rθ
r r θ z r
1
Pz (rσ zr ) (σ zθ ) (rσ zz )
r r θ z
dq
X î r ( Pr ) î ( P ) î z ( Pz )
dt
reduces to
dq q 2 1
r X r (r rr ) (r ) (r rz )
dt r r r z r
FLUID DYNAMICS 258
rr 1 r rz rr
= Xr +
r r z r
dq qq 1
r X ( r ) ( ) (z ) 2 r
dt r r r z r
dq z 1 z zz rz
X z rz
dt r r z r
d q
where X (X r , X , X z ), qr qz
dt t r r z
The relations between the stress components and the rates of strain, in
cylindrical co-ordinates are
2μ
rr = 2 rr q p
3
2μ 2μ
= 2 q p λ
3 3
2μ
zz = 2 zz q p
3
q r 1 q q r
rr = , ,
r r r
q z q q 1 q r
zz = , r
z r r r
1 q z q q r q z
z = , rz
r z z r
Using the above relations, the equations of motion (Navier Stoke‟s equation) in
cylindrical co-ordinates become
dq q 2 p q r 2
r X r 2 q
dt r r r r 3
259
FLUID DYNAMICS
1 1 q r q θ q θ
μ
r θ r θ
+
r r
q r q z 2μ q r 1 q θ q r
μ
z z r r r r θ
+
r
dq qq 1 p 1 2 q 2q r 2
r X q
dt r r r r r 3
1 q r q θ q θ
μ
r r θ
+
r r
1 q z q θ 2μ 1 q r q θ q θ
μ
z r θ z r r θ
+
r r
dq z p q 2
X z 2 z q
dt z z z 3
1 1 q z q θ q r q z
μ μ
r θ r θ z r z r
+
μ q r q z
+
r z r
q r 1 q θ q z qr
where q .
r r θ z r
5.3. Special cases. (i) If = constant and = constant, then q = 0 and the
equations of motion are
dq q 2 p q 2 q
r X r 2 q r 2r 2
dt r r r r
dq qq 1 p 2 2 q q
r X q 2 r 2
dt r r r r
FLUID DYNAMICS 260
dq z p
X z 2qz.
dt z
2 1 1 2 2
where 2
r 2 r r r 2 θ 2 z 2
dq q 2 p
r X r
dt r r
dq qq 1 p
r X
dt r r
dq p
z X z
dt z
dθ dr dθ
qr = r cos sin sin 2 θ , q θ r ,
dt dt dt
dψ
q = r sin
dt
dq r q θ q ψ
2 2
and ar =
dt r
dq q r q q cot
2
a =
dt r r
dq ψ qrqψ q θ q ψ cot θ
a =
dt r r
dq
X p 2 q
dt
261
FLUID DYNAMICS
In spherical co-ordinates,
p 1 p 1 p
p = , , ,
r r θ r sin θ ψ
X (Xr , Xθ , Xψ )
Let us simplify
2 q ( q) ( q)
1 2 1 1 qψ
But q (r q r ) (sin θq θ ) +
r r r sin θ θ r sin θ ψ
h1 î r h 2 îθ h 3 î ψ
1
Further q r θ ψ
h 1h 2 h 3
h1q r h 2 q θ h 3q ψ
î r r îθ r sin θ î ψ
1
= 2 r θ ψ
r sin θ
qr r q θ r sin θ q ψ
1 q r
= î ( r sin q ) ( rq ) î r ( r sin q )
r 2 sin r
r
q
+ î ψ r sin θ (r q θ ) r
r θ
Then
2 2 2 q ψ
2 q î r 2 q r 2 q r 2 (sin θq θ ) 2
r r sin θ θ r sin θ ψ
q 2 q 2 cosθ q ψ
+ îθ 2 q θ 2 θ 2 2 r 2 2
r sin θ r θ r sin θ ψ
qψ 2 q r 2 cosθ q θ
+ î ψ 2 q ψ 2 2 2 2
r sin θ r sin θ ψ r sin θ ψ
2
FLUID DYNAMICS 262
dq q 2 q 2
r X p
dt r r
r
2 2 2 q
2 q r 2 q r 2 (sin q ) 2
r r sin r sin
dqθ q r q θ q 2ψ cotθ
= X 1 p
dt r r r
q 2 q 2 cos q
2 q 2 2 2 r 2 2
r sin r r sin
dq q r q q q cot 1 p
X
dt r r r sin
qψ 2 q r 2 cosθ q θ
+ 2 q ψ 2 2 2 2
r sin θ r sin θ ψ r sin θ ψ
2
1 2 1 1 2
where 2 r sin +
r 2 r r r 2 sin θ θ θ r 2 sin 2 θ ψ 2
If we put = 0 in the above equations, we get the equations of motion for ideal
fluid.
For a viscous incompressible fluid in steady flow, the Navier Stoke‟s equation
with negligible body forces, are
dq p 2 p
q v 2 q, v =
dt
u u u 1 p 2u 2u 2u
u v w v 2 2 2
x y z x x y z
263
FLUID DYNAMICS
du u
For steady case, (q )u (q )u, 0
dt t t
= u v w u
x y z
u u u
=u v w
x y z
u v w
0 (2) | q = 0
x y z
The equations (1) are non-linear 2nd order partial differential equations and
there is no known general method for solving them. However, we shall find
some exact solutions of the Navier-Stoke‟s equations in some special cases.
This is one of those cases.
v = 0, w = 0 and 0 (3)
z
u
0 u = u(y) (4)
x
y
FLUID DYNAMICS 264
u
O
p
0 p = p(x) (5)
y
The 3rd equation of equations (1) is identically satisfied and the 1st equation
gives
1 dp d2u dp d2u
0= v 2 μ 2 =v (6)
dx dy dx dy
dp
Since u is a function of y only, so is either a function of y or a constant.
dx
But from (5), p is a function of x alone.
dp
Hence is constant. i.e. pressure gradient is constant.
dx
1 dp y 2
u= + Ay + B (7)
μ dx 2
6.1. Couette’s Flow : It is the flow between two parallel planes (flat plates)
one of which is at rest and other moving with velocity U parallel to the fixed
plate. Here, the constants A and B in (7) are determined from the conditions
u = 0, y = 0
and u = U, y = h (8)
y
265
FLUID DYNAMICS
u=U
O u=0
x
1 dp h 2
B = 0, U = Ah
μ dx 2
U h dp
A= , B = 0 (9)
h 2μ dx
1 dp y 2 U h dp
u= y (10)
μ dx 2 h 2μ dx
y 2 hy dp Uy
= ()
2μ dx h
U h 2 dp y y
= y 1 (11)
h 2μ dx h h
This equation is known as the equation of Couette‟s flow. Thus the velocity
profile for Couetle‟s flow is parabolic. The flow Q per unit breadth is given by
h 1 dp y U h dp
2
h
Q= 0 u dy 0 y dy
m dx 2 h 2 dx
hU h 3 dp
= ()
2 12 dx
hU h 3 dp
= P, P (12)
2 12 dx
FLUID DYNAMICS 266
u y y y
α 1 (13)
U h h h
h 2 dp
where = (14)
2μU dx
u y αy 2
(1 α) 2 (15)
U h h
The pressure is increasing in the direction of flow and the reverse flow begins
when < 1
| y is small. i.e.
2
y is neglected
dp
If = 0 i.e. 0 , then the particular case is known as simple Couette‟s
dx
flow and the velocity is given by
u y
U h
1 h
u0 = u dy (16) | u = u(y)
h 0
1 h Uy y y
u0 = 0 Uα 1 dy
h h h h
Uh 2 h2 h3
= U
2h 2 2h 2 3h 3
267
FLUID DYNAMICS
U Uα 1 α
= U (17)
2 6 2 6
U 2 dp U h 2 dp
= P, P (18)
2 12 dx 2 12 dx
In the case of a simple Couette‟s flow, the velocity increases from zero on the
U
stationary plate to U on the moving plate such that the average velocity is .
2
When the non-dimensional pressure gradient is = 3, then from (17), we get
u0 = 0. This means that there is no flow because the pressure gradient is
balanced by the viscous force.
du U 1 2y
0 U 2 0
dy h h h
1 α
y = h (19)
2α
y
From here, 1, whenα 1
h
y
and 0, when = 1
h
1 α 1 α 1 α
u= α 1 U
2α 2α 2α
=
1 2 U
4
(ii) Shearing Stress : The shearing stress (drag per unit area) in a Couette‟s
flow is given by
du U U 2y
yx = 1 (20)
dy h h h
FLUID DYNAMICS 268
μU
= , for a simple Couette‟s flow ( = 0).
h
h
When y = , then the second term in (20) vanishes. Thus the shearing stress
2
is independent of on the line midway between the flow. The shearing stress
at the stationary plane is positive for > 1 and negative for < 1.
| y = 0 at stationary plate
The velocity gradient at the stationary plate is zero for = 1 and the shearing
stress is zero for = 1.
Further, drag per unit area on the lower and the upper plates are obtained from
(20) by putting y = 0 and y = h, as
U U U U
and
h h h h
combining the two results, drag per unit area on the two plates is
U U U h dp
i.e. ()
h h h 2 dx
μU Ph dp
i.e. ,P
h 2 dx
6.2. Plane Poiseuille Flow : A flow between two parallel stationary plates is
said to be a plane Poiseuille Flow.
The origin is taken on the line midway between the plates which are placed at a
distance h and x-axis is along this line.
h
u = 0, when y = + (21)
2
h/2
h O x
269
FLUID DYNAMICS
1 dp h 2
A = 0, B =
μ dx 8
1 dp y 2 h 2
u= (22)
μ dx 2 8
This represents a parabola and thus the laminar flow in a Plane Poiseuille Flow
is parabolic.
1 dp
y=0 y=0
μ dx
h 2 dp
Therefore , Umax = (23)
8μ dx
1 h/2
u0 = u dy
h h / 2
1 h / 2 h 2 dp 4y 2
1 dy
h h/ 2 8 dx
u0 =
h 2
2 h 2 dp 2
= Umax (24)
3 8μ d x 3
dp 8μ 8μ 3 12μ
2 u max . 2 u 0 2 u 0 (25)
dx h h 2 h
dp
This further shows that is a negative constant.
dx
(ii) Shearing Stress : The shearing stress at a plate (lower plate) for a plane
Poiseuille Flow is
yx y h du
dy
1 dp h
. .
dx 2
2 y h
2
h dp
=
2 dx
4μ
= u max . (26)
h
The local frictional (skin) co-efficient Cf is defined by
( yx ) h 2 4 u 02
Cf = u max .
u 02 2 h 2
4 3 u 0 12v 12
h 2 u 02 2 h u 0 R e
u 0h
Where Re = is the Reynolds number of the flow based on the average
v
velocity and the channel height.
7. Theory of Lubrication
The hydrodynamic theory of Couette flow can be applied in the study of
lubrication by considering an example of the slipper bearing which consists of
a sliding block moving over a stationary guide and inclined at a small angle
with respect to the stationary pad. The gap between the sliding block and the
pad is always much smaller than the length of the block and is filled with a
lubricant, usually oil. For such a case viscous forces are predominant. The
theory of lubrication was first developed by Osborne Reynolds in 1886, and the
discussion is due to Lord Rayleigh (1918).
load
slipper block
U
bearing guide
271
FLUID DYNAMICS
A
B
h1
h(x) h2
O
X
U
Z
Let (a, h1), (b, h2) and (x, h) be the co-ordinates of A, B and any point on AB.
Since the addition of a constant pressure throughout the fluid will make no
difference to the solution, so we may for convenience assume that p = 0
beyond the ends of the block. Since the inclination of the plane faces is small,
(i.e. the faces are nearly parallel) the velocity u at any point is given by
y 2 hy dp Uy
u = ( from () of previous article) and the
2μ dx h
flow Q in x-direction is
hU h 3 dp
Q= (from () of previous article)
2 12 dx
1 h 3 dp 1
hU = constant = h0U
2 12μ d x 2
dp h h0
= 6U 3
(1)
dx h
dp
where h0 is the value of hat the points of maximum pressure s.t. 0 .
dx
h 2 h1
hh1 = (x a )
ba
dh h 2 h1 h 2 h1
(2)
dx ba l
dh
where l is the length of the block and is the slope of the line AB.
dx
dp dp dx 6Ul 1 h 0
. (3)
dh dx dh h 2 h1 h 2 h 3
Integrating, we find
6Ul 1 h 0
p= C
h 2 h1 h 2h 2
3Ul h 0 2h
= C (4)
h 2 h1 h 2
This gives
2h 1 h 2 6Ul
h0 = , C
h1 h 2 (h 2 h1 )( h 2 h1 )
273
FLUID DYNAMICS
6Ul (h h1 )(h h 2 )
and thus p=
h 2 (h 22 h12 )
This suggests that p > 0 if h1 > h2 i.e. the stream contracts in the direction of
motion. P > 0 yields thrust rather than a suction. So we conclude that a
necessary condition for lubrication is that the relative motion should tend to
drag the fluid from the wider to narrower part of the intervening space i.e. the
stream should be convergent.
b h2 dx
P= a p dx h 1
p dh
dh
l h2
=
h 2 h1 h1
pdh | using (2)
6Ul 2 h2 (h h1 )(h h 2 )
=
(h 1 h 2 ) 2
h1 h 2 (h 1 h 2 )
dh (6)
h2
(h h1 )(h h 2 ) h2
h1h 2 h1 h 2
h
1 h 2
dh h1
1 2 dh
h h h
h
hh 2
= h 1 2 h1 log h h 2 log h
h h1
h
= 2(h1h2) + (h1 + h2) log 1
h2
1 h2 (h h1 )( h h 2 ) h h2
h1 h 2 h1 h 2
dh log(h1 / h 2 ) 2 1
h1 h 2
FLUID DYNAMICS 274
k 1
= log k2 , k = h1/h2.
k 1
6Ul 2 k 1
P= log k 2 k 1 (7)
h 22 (k 1) 2
U h dp
( yx ) yh | From () of previous article.
h 2 dx
and thus the total frictional force experienced by the moving fluid is
b
F= a ( yx ) yh dx
b U h h h 0
a
h
6U
2 h 3
dx | using (1)
h2 u 3h 0 l
= U 2 dh | using (2)
h1
h h h 2 h1
Ul h 2 u 1 6h1h 2
2 . dh
h 2 h1 h1
=
h h h1 h 2
2Ul k 1
= 2 log k 3 (8)
h 2 (k 1) k 1
Comparing (7) and (8) we see that the ratio F/P of the total friction to the total
load is independent of both and U, but proportional to h if the scale of h is
altered.
It has been found by Reynolds and Rayleigh that the value of k which makes P
Ul 2 Ul
a maximum is 2.2 (approx.) and that this makes P = 0.16 2 , F = 0.75 .
h2 h2
h2
For this case, F/P = 4.7 .
l
We consider a laminar flow, in the absence of body forces, through a long tube
of uniform circular cross-section with axial symmetry.
Let z-axis be taken along the axis of the tube and the flow be in the direction of
z-axis. Since the flow is along z-axis, the radial and transverse components of
velocity are absent.
r x z
Thus qr = q = 0 q (q r , q , q z )
q z
= 0 qz = qz(r) (1) | axial symmetry i.e. independent of
z
dq q 2 p q 2 q
r Xr 2 q r 2r 2
dt r r r
r
dq qq 1 p 2 q q
r X 2 q 2 r 2
dt r r r r
dq z p
X z 2 q z
dt z
d 1
where qr q qz ,
dt t r r z
and X (X r , X , X z )
In the present case 0 and qr = q = 0, X 0
t
FLUID DYNAMICS 276
p p
0 p p(z) (2)
r
p
0= 2 q z q z q z (r) and r is constant w.r.t. t .
z
dp d 2 q z 1 dq z
or q z 2
2
(3)
dz dr r dr
2 1 1 2 2
(In cylindrical co-ordinates 2 )
r 2 r r r 2 2 z 2
d 2q dq dp
r 2z z r
dr dr dz
d dq z r dp
i.e. r
dr dr dz
dq z 1 dp r 2
r A
dr dz 2
dq z 1 dp A
i.e. r
dr 2 dz r
1 dp 2
qz = r A log r B (4)
u dz
The first boundary condition is obtained from the symmetry of the flow
such that
dq z
0 on r = 0 (5)
dr
qz = 0, when r = a (6)
1 dp 2 1 dp 2
A = 0, B = a a
4 dz 4 dz
1 dp 2 2
qz = (a r ) (7)
4 dz
(i) The Max x Average Velocity : For extreme values of qz, we have
dq z
0 | qz is a function of r only
dr
a 2 dp
qmax. = (8)
4 dz
dp
where is a negative constant.
dz
From (7) and (8), the velocity distribution, in non dimensional from, is given
by
2
qz r
1
q max a
1 2 a
q0 =
a 2
0 0 qz r dr d
FLUID DYNAMICS 278
a 2 dp 1
q0 = q max .
8 dz 2
The volume of fluid discharged over any section per unit time (i.e. volumetric
flow) is defined as
a
Q= 0 qz. 2 r dr
a 4 dp 1 2 a 2 dp 1 2
Q= a a q max . (9)
8 dz 2 4 dz 2
dq z 1 dp r dp
rz = (2r )
dr 4 dz 2 dz
a dp a dp 2
(rz)r=a = . q max . (10)
2 dz 2 dz a
The local frictional (skin) co-efficient Cf for laminar flow through a circular
pipe is
( rz ) r a 2 q max
Cf =
q 02 2 a q 02 2
4 2q 0 8 1 16
=
a q 02 a q 0 R e
Where Re = 2aq0/v is the Reynolds number. When Re is less than the critical
Reynolds number, which is 2300 in this flow problem, the flow is laminar but
if Re > 2300, the flow ceases to be laminar and becomes turbulent. Thus, in
this problem, Re < 2300.
279
FLUID DYNAMICS
1 a 4
8.1. Example. Establish the formula (p1p2) for the rate of steady flow
8 l
of an incompressible liquid through a circular pipe of radius „a‟, p1 and p2
being the pressures at two sections of the pipe distant lapart. Also find the
drag on the cylinder.
dp
Solution. First we prove equation (9) and then we note that is the change
dz
in pressure per unit length and thus in the present case
dp p 2 p1
dz l
a 4 p1 p 2
Q=
8 l
F = 2al (rz)r=a
dp
= a2l
dz
= a2 (p1p2).
Let us consider the steady flow of a viscous fluid parallel to the axis in the
annular space between two co-axial cylinders of radii r1 and r2(r2> r1). The
velocity for such flow is
1 dp 2
qz = r + A log r + B (1)
4 dz
qz = 0 at r = r1 and r = r2 (2)
and
1 dp (n 2 1)r12
B= log r1 r12
4 dz log n
Thus the velocity distribution in the annular space between two co-axial
cylinders is
1 dp 2 2 (n 2 1)r12 r
qz = (r1 r ) log (3)
4 dz log n r1
2 nr1 1 dp 2 2 (n 2 1)r12 r
= 0 r (r1 r )
4 dz
log r dr d
1 log n r1
nr
2 dp 2 r 2 r 4 (n 2 1)r12 r 2 r r 2
1
= r1 log
4 dz 2 4 log n 2 r1 4 r1
=
r14 dp 2
2n 2 n 1
8 dz
4 n 2 1
(2 log n 1)n 2 1
log n
r14 dp 2 (n 2 1) 2
2n n 1 2n 2n
4 4 2
=
8 dz log n
281
FLUID DYNAMICS
r14 dp 4 (n 2 1) 2
= (n 1) (4)
8 dz log n
Q r12 dp 2 n 2 1
q0 = (n 1) (5)
(n 2 1)r12 8 dz log n
dq
( rz ) r r1 z
dr r r1
1 dp (n 2 1)r12
= 2r1
4 dz r1 log n
r1 dp (n 2 1)
= 2
4 dz log n
and
dq
( rz ) r r2 z
dr r r2
1 dp (n 2 1)r12
= 2r2
4 dz r2 log n
r1 dp (n 2 1)
= 2n
4 dz n log n
We consider the flow between two concentric rotating cylinders with radii r1, r2
(r2 > r1) having viscous fluid in between them. We assume that the flow is
circular such that only the tangential component of velocity exists. Let w1 and
w2 be the angular velocity of the inner and outer cylinders respectively.
q
= 0, q = q(r) 2
(1) r2
r1 z
where qr = qz = 0
1
dq q 2 p q 2 q
r X r 2 q r 2r 2
dt r r r r
dq qq 1 p 2 2 q q
r X q 2 r 2
dt r r r r
dq z p
X z 2 q z
dt z
p 1 p q
0, 2 q 2 0 (2)
z r r
q 2 p
(3)
r r
q 2 1 1 2 2
2q 0, 2
r2 r 2 r r r 2 2 z 2
d 2 q 1 dq q
0
dr 2 r dr r 2
283
FLUID DYNAMICS
d 2q d q
0 (4)
dr 2 dr r
Integrating, we get
dq q
2C1
dr r
dq d
r q 2c1r (r q ) 2c1r
dr dr
Integrating, we get
c2
r q = c1 r2 + c2 q = c1 r + (5)
r
q = r1 1, when r = r1
d
on the surface, v r v = r
dt
dl d
l = r r i.e. v = r
dt dt
b 2 a 2
q = r (9)
b 2 a 2 r
dp q 2 2 2 c 22
c1 r 2 2c1c 2 | using (5)
dr r r r
2 c 22 2c1c 2
= c1 r 3
r r
c2r 2 c2
p = 1 22 2c1c 2 log r c 3 (10)
2 2r
c12 r12 c 22
p1 = 2 2c1c 2 log r1 + c3
2 2r1
c12 r12 c 22
c3 = p1 2 2c1c 2 log r1
2 2r1
r 2 r12 c 22 1 1
p = p1 + c12 2c1c 2 log r
r2 r2
2 2 1 r1
dq q d q
r r
dr r dr r
d c r c 2 r
= r 1
dr r
285
FLUID DYNAMICS
d c 2c
= . r c1 22 r 32
dr r r
The expressions for the shearing stress on the outer and the inner cylinder are
2(2 1 )r12
( r ) r r2
r22 r12
2(2 1 )r22
( r ) r r1
r22 r12
dq
0, F 0, q 0 (1)
dt
p 2
0= q
Let us work with fixed co-ordinate axis ox, oy, oz with oz taken parallel to the
flow so that
q wk̂ , (3)
u v w
From equation of continuity 0,
x y z
w
we get 0 w = w (x, y) (4)
z
p p p 2w 2w
î ĵ k̂ 2 2 k̂
x y z x y
p p
0, 0 (5)
x y
p 2w 2w
and 2 2 (6)
z x y
dp 2w 2w
2 2 (7)
dz x y
2w 2w P dp
2 P (8)
x 2
y dz
subject to the condition that w vanishes on the walls of the tube for a viscous
fluid.
2w 2w
f (x, y)
x 2 y 2
at all points (x, y) of a region S in the plane ox, oy bounded by a closed curve c
and if f(x, y) is prescribed at each point (x, y) of S and w at each point of C,
then any solution w = w (x, y) satisfying these conditions is unique.
2w 2w
f (x, y) (9)
x 2 y 2
287
FLUID DYNAMICS
i.e. w1 = w2 on C
W = w1w2 (10)
2 W 2 W 2 w1 2 w1 2 w 2 2 w 2
Then, 2
x 2 y x
2
y 2 x 2 y 2
= f(x, y) f(x, y) = 0
2W 2W
2 0 in S (11)
x 2 y
Since w1 = w2 on C. C
Now, consider
W 2 W 2
I= dxdy
x y
S
W 2 W 2 2 W 2 W
= 2 W 2 dx dy
S x y x y 2
| using (11)
W W
= W W dxdy
S x x y y
FLUID DYNAMICS 288
W W
= W dy W dx , by Green‟s Theorem.
C x y
= 0, as W = 0 on C.
2
W W
2
W
Now, I = 0 0 in S which will be true only if 0,
x y x
W
= 0 at each point of S.
y
W = constant in S.
11.2. Tube having Uniform Elliptic Cross-Section : Suppose that the elliptic
cross-section of the tube has the equation
x 2 y2
1 0 (13)
a 2 b2
2w 2w P
2 (14)
x 2
y μ
x 2 y2
w = k 1 2 2 (15)
a b
2 2 P
k 2 2
a b μ
289
FLUID DYNAMICS
Pa 2 b 2
k= (16)
2μ(a 2 b 2 )
Pa2 b 2 x 2 y 2
w= 1 (17)
2μ(a 2 b 2 ) a 2 b 2
The uniqueness theorems shows that w, given in (17) is the required solution.
Q = w dx dy
S
P a 2b2 1 1
= 2
dxdy 2 x 2 dxdy 2 y 2 dxdy
2μ(a b )
2
a b
Pa 2 b 2 1 a2 1 b2
= πab πab πab
2μ(a 2 b 2 ) a2 4 b2 4
πP a 3 b 3
= (18)
4μ a 2 b 2
Q Q P a 2b2
Mean velocity =
dxdy ab 4 a 2 b 2
P a 4 x 2 y2
w = qz = 1
2μ 2a 2 a 2 a 2
P 2 P 2 2
= (a x 2 y 2 ) (a r ) ,
4μ 4μ
| x2 + y2 = r2
dp
where P =
dz
FLUID DYNAMICS 290
πP a 6 πPa 4
and Q =
4μ 2a 2 8μ
Q
mean velocity = Pa 2 8μ .
πa 2
x
x = a, y = + (19)
3
If we take
1
w = k(xa) y 2 x 2 (20)
3
x2
=k y 2 (x a ) ( x a )
3
y=x 3
60
x= a
60 30 x
0
60
y=x 3
Substituting for w in
2w 2w P
2 (21)
x 2
y μ
we obtain
2a P
k 2x (2x 2a)
3 μ
291
FLUID DYNAMICS
3P
k= (22)
4μa
3P 1
w= ( x a ) y 2 x 2 (23)
4μa 3
a x 3
Q = w ds 2 dx w dy | due to symmetry
S 0 0
3P a x 3 2 x2
= dx ( x a ) y dy
2μa 0 3
0
Pa 4
=
60 3μ
x =a x
O
27 Pa 4
then Q=
20 3 μ (2a, 0)
The simplest unsteady flow is that which results due to the impulsive motion of
a flat plate in its own plane in an infinite mass of fluid which is otherwise at
rest. This flow was first studied by Stokes and is generally known as Stokes
first problem.
FLUID DYNAMICS 292
Let x-axis be taken in the direction of motion of the plate, which is suddenly
accelerated from rest and moves with constant velocity U0. Let y-axis be
perpendicular to the plate. The motion is two-dimensional and the only non-
zero component of velocity is u, where q = (u, v, w). Further, u is a function
of y and t only. i.e. u u(y, t). The pressure in the whole space is constant.
The Navier-Stokes equations in the absence of body forces, for the present
case, become
u 2u
= v 2 , v = / (1)
t y
u U 0 at y 0
when t > 0 (3)
u 0 at y
We observe that the partial differential equation (1) is the same as the equation
of heat conduction, diffusion etc. It can be reduced to an ordinary differential
equation if we make the following substitution (principle of similarity of flow)
u
f (η) (4)
U0
y
where = (5)
2 vt
We have,
u u η f y
U 0
t η t η 4 v t 3 / 2
u u η f 1
U0
y η y η 2 v t
2 u u u η 2f 1
U0 2
y 2 y y η y y η 4vt
f 2f
(2η) 2
η η
d 2f df
i.e. 2η 0 (6)
dη 2 dη
f ''
2η log f ' η 2 log C1
f'
2
The solution of (6) is f ' C1e η
2
f C1 0η e η dη C 2
2
f() = C1 0η e η dη C 2 (8)
Using the boundary conditions (7) in (8), the constants of integration C1 and C2
are obtained to be
1 2
C2 = 1 and C1 = (9)
η2 π
0 e dη
u 2 η η2
f (η) 1 0 e dη
U0 π
= (1erf n) (10)
the integral
2 η η2
erf = 0 e dη (11)
π
is called the error function or the probability integral and tables for it are
readily available.
erf u
U0
0 0 1
1.00000 0
We observe that the velocity decreases continuously and tends to its limiting
value zero as tends to infinity. However, for all practical purposes, this value
is reached at about = 2.0 and therefore the corresponding value of y, which
we shall denote by , from (5), is
~
4 vt (12)
product of kinematic viscosity and time. If vt is small, then will be small and
once again we shall have a boundary layer flow.
2.0
1.5
y
η
2 vt 1.0
.5
O .2 .4 .6 .8
u 1.0
U0
UNIT-V
1. Dynamical Similarity
called dynamically similar system. One obvious condition is that the model
should be geometrically similar to the prototype which means that we can
obtain the actual object from the model by enlarging or contracting its size in
every direction in the same proportion. This eliminates the consideration of
boundary conditions in the discussion of dynamical similarity and so we have
to consider only the governing equations. In short, we can say that two fluid
motions are dynamically similar if with geometrically similar boundaries, the
flow patterns are geometrically similar. Further, two geometrically similar
flows are dynamically similar if forces acting at every point are similar i.e. the
forces are acting in same direction having same ratio in magnitude.
We now discuss the conditions under which the fluid motions are dynamically
similar. In other words, we have to find out those parameters which
characterize a flow problem. There are two methods for finding out these
parameters (i) inspection analysis (ii) dimensional analysis. In the first case,
we reduce the fundamental equations to a non-dimensional form and obtain the
non-dimensional parameters from the resulting equations. This procedure
should always be used when the basic differential equations for a problem are
available. In the second case, we form non-dimensional parameters from the
physical quantities occurring in a problem, even when the knowledge of the
governing equations is missing. We discuss these two methods with particular
reference to the flow of a viscous compressible fluid.
1.1. Remark. (i) Some authors do not differentiate between the two methods
and study both of them under the head of dimensional analysis.
u u u u 1 p 2u 2u 2u
u v w X 2 2 2 (1)
t x y z x x y z
p = P p (4)
297
FLUID DYNAMICS
where all primed quantities are pure numbers having no dimensions. Then,
since L/U is the characteristic time, we get
1 p 1 (Pp' ) P p'
x (Lx ' ) L x'
P LX
, , (6)
UL U 2
U2
UL
Re =
FLUID DYNAMICS 298
This is named after Osborne Reynolds who first introduced this number while
discussing boundary layer theory. This is most important of viscous force over
the inertia force. It can be easily seen from the equation of motion that the
u
inertia forces terms like u are of the order U2/L and the viscous
x
2u
forces terms like μ 2 are of the order U/L2.
x
Therefore,
inertia forces U 2 L UL UL
Re
viscousforces U L2
Thus, Reynolds number is the ratio of the inertia force to the viscous force. It
is infact a parameter for viscosity. If Re is small, the viscous forces will be
predominant and the effect of viscosity will be felt in the whole flow field. On
the other hand, if Re is large the inertial forces will be predominant and in such
a case the effect of viscosity can be considered to be confined in thin layer,
known as boundary layer, adjacent to the solid boundary. When Re is
enormously large, the flow ceases to be laminar and becomes turbulent. The
Reynolds number at which the transition, from laminar to turbulent, takes place
is known as critical Reynolds number.
Pr essure forces PL P
CP
Inertia forces U L U 2
2
i.e. CP gives the relative importance of the pressure force to the inertia force.
Usually, it is taken as unity.
1.5. Force Coefficient. The third non-dimensional number in (6) tells how to
scale body forces. This number is called force co-efficient, denoted by CF
which is similar to CP.
299
FLUID DYNAMICS
body forces X LX
Thus CF = 2
Inertia forces U L U
2
1 U2
Fr =
C F LX
This number is particularly used in cases when body forces are the
gravitational forces. Thus,
inertia forces U 2 L U 2
Fr =
gravity forces g gL
where 1, 2,…, nr are the dimensionless power products of Q1, Q2,…, Qn
taken r + 1 at a time, r being the rank of the dimensional matrix of the given
physical quantities.
Proof. Let Q1, Q2,…, Qn be n given physical quantities and let their dimensions
be expressed in terms of m fundamental units u1, u2,…, um in the following
manner
[Q1] = u1a11u a221 ...u amm1
[Q ] = u
2
a12 a 22 a m2
1 u 2 ...u m
…………………………
…………………………
[Qn] = u1a1n u a22n ...u ammn
so that aij is the exponent of ui in the dimension of Qj . The matrix of
dimensions i.e. the dimensional matrix of the given physical quantities is
written as
301
FLUID DYNAMICS
…………………
…………………
um: a m1 a m 2 …… a mn
then [] =
1 2 m
u a11 u a 21 .....u a m1 x1 u a12 u a 22 .....u a m 2
1 2 m
x2
............ u1a1n u a22 n .....u ammn
xn
In order that the product is dimensionless, the powers of u1, u2,…, um should
be zero i.e. M0, L0, T0 etc. Thus, we must have
…………………………………….
…………………………………….
a m1 + x1 + a m 2 x2 + ……+ a mn xn = 0
x1
x
AX = 0, X = 2
x
n
Now, from matrix algebra, we know the result that if there are m homogeneous
equations in n unknowns, then the number of independent solutions will be
nr, where r is the rank of the matrix of co-efficients, and any other solution
FLUID DYNAMICS 302
Thus if r is the rank of the dimensional matrix A, then the number of linearly
independent solutions of the matrix equation AX = 0 are nr. So,
corresponding to each independent solution of X, we will have a dimensionless
product and therefore the number of dimensionless products in a complete
set will be nr
1.8. Method for -products. To find out the -products in a complete set, we
adopt the following steps.
(ii) Find the rank of the dimensional matrix. If the rank is r(say), then the
number of ‟s will be nr.
(iv) Express 1, 2,…., nr a power products of these r quantities raised to
arbitrary integer exponents and one of the excluded, but different in
different ‟s, (nr) quantities.
(v) Equate to zero the total dimension of each fundamental unit in each -
product to get the integer exponents.
L, U, , X, P,
and the fundamental units in which the dimensions of all these quantities can
be expressed are mass [M], length [L] and time [T]. The above six quantities
have dimensions as follows
Quantity Dimensions
L-length [L]
U-velocity [LT1]
-density [ML3]
X-force per unit mass [LT2] force [MLT2]
P-pressure force per unit area) [ML1 T2]
-viscosity [ML1 T1]
L U e X P
M: 0 0 1 0 1 1
L: 1 1 3 1 1 1
T: 0 1 0 2 2 1
1 Lx1 U x 2 x3 X
(iv) Let 2 Lx 4 U x5 x 6 P
3 Lx 7 U x8 x9
(v) Now,
FLUID DYNAMICS 304
x3 = 0 x6+1 = 0 x9+1 = 0
x1 = 1 x4 = 0 x7 = 1
x2 = 2 x5 = 2 x8 = 1
x3 = 0 x6 = 1 x9 = 1
Thus, we get
LX
1 = L1 U2 0 X =
U2
P
2 = L0 U2 1 P =
U 2
3 = L1 U1 1 =
LU LU
which are the same dimensionless quantities obtained in equation (6) of the
inspection analysis
1.10. Remark. If we include the energy equation and equation of state in our
study, then, in the general case of viscous compressible fluid dynamics, there
are 9 physical quantities and the fundamental units in which the dimensions of
305
FLUID DYNAMICS
all these quantities can be expressed are length, mass, time and temperature (Q)
and thus there are 94 = 5 non-dimensional numbers.
The simple problems of fluid motion which can be considered are divided into
two classes according as the corresponding Reynolds number is small or large.
In the case of small Reynolds number, viscosity is predominant and the inertia
terms in the equations may be regarded as negligible. The case of large
Reynolds number in which the frictional terms are small and inertia forces are
predominant, was investigated by the German Scientist Ludwig Prandtl in
1904. He made an hypothesis that for fluids with very small viscosity i.e. large
Reynolds number, the flow about a solid boundary can be divided into the
following two regions.
(i) A thin layer in the neighbourhood of the body, known as the boundary
layer, in which the viscous effect may be considered to be confined. The
smaller the viscosity i.e. the larger the Reynolds number, the thinner is this
layer. Its thickness is denoted by . In such layer, the velocity gradient normal
to the wall of the body is very large.
(ii) The region outside this layer where the viscous effect may be considered as
negligible and the fluid is regarded as non-viscous.
In the discussion of unsteady flow over a flat plate, we had obtained that
~ 4 t
i.e. the boundary layer thickness is proportional to the square root of kinematic
viscosity. The thickness is very small compared with a linear dimension L of
the body i.e. << L.
At the stagnation point O, the thickness of the boundary layer is zero and it
increases slowly towards the rear of the wedge. The velocity distribution and
the pattern of streamlines deviate only slightly from those in the potential flow.
We take the x-axis along the wall of the wedge and y-axis perpendicular to it,
so that the flow is two-dimensional in the xy-plane. Within a very thin
boundary layer of thickness , a very large velocity gradient exists i.e. the
velocity u parallel to the wall in the boundary layer increases rapidly from a
value zero at the wall to a value U of the main stream at the edge of the
boundary layer.
u u u 1 p 2u 2u
u v 2 2 (1)
t x y x x y
v v v 1 p 2v 2v
u v 2 2 (2)
t x y y x y
u v
0 (3)
x y
In studying the unsteady flow over a flat plate, we found that the thickness of
the boundary layer is proportional to the square root of the kinematic
viscosity v which is indeed very small. For this reason < < x except near the
stagnation point 0 where the boundary layer begins. In order to compare the
order of magnitude of the individual terms in the above equations, we put them
in non-dimensional form by introducing the non-dimensional notations
307
FLUID DYNAMICS
x y u v t p
x* = , y* , u * , v* t* , p* (4)
l U V l/U p
U u * V v *
0 (5)
l x * y *
Integrating, we get
1 1
U u * V v *
l 0 x *
dy *
0 y *
dy* 0
1
V u *
or
U
l 0 x *
dy*, where (v*) y*1 1 (6)
V
Since the integral in (6) is of the order of unity, the ratio is of order .
U l
Therefore V < < U.
We now obtain the non-dimensional form of (1) using (4) such that
U 2 u * U 2 u * UV u * p p * U 2 u * l 2 2 u *
u* *
l t * l x * y * l x * l 2 x *2 2 y *2
or
u * u * V l u * p p * l 2u * l 2 2u *
u* v* 2 (7)
t * x * U y * U x * R e x *2 2 y *2
1 1
1 1 1 2 1
δ2
lU 1
Reynolds number, Re = = 0()2 as is proportional to v1/2.
R e lU
UV v * UV v * V 2 v *
u* v*
l t * l x * y *
FLUID DYNAMICS 308
p p * V 2 v * V 2 v *
=
y * l 2 x *2 2 y *2
V v * V v * V 2 l v *
or u* 2 v*
U t * U x * U y *
1
2
δ
p l p * Vl 2 v * l 2 2 v *
=
U 2 y * l 2 U 2 x *2 2 y *2
p l p * 1 V 2 v * l 2 2 v *
=
U 2 y * R e U x *2 2 y *2
1
2 1 (8)
δ2
We neglect the terms of the order of and higher as is small. We then revert
back to the dimensional variables to obtain
u u u 1 p 2u
u v 2 (9)
t x y x y
p
0 p = p(x) (10)
y
u v
and 0 (11)
x y
u v 0, y 0
(12)
u U(x, t), y
Since p is independent of y, for given x, p has the same value through the
boundary layer from y = 0 to y = . Thus, in boundary layer theory, there are
only two variable terms u and v instead of three u, v and p in the Navier-Stokes
equations. This is a great simplification in the solution of the differential
equations.
309
FLUID DYNAMICS
Now, U is the velocity outside the boundary layer. The Euler‟s equation in the
main stream (potential flow of non-viscous fluid) is obtained from (9) by
taking v= 0 and
u
v = 0, 0 for y
y
Thus, we get
U U 1 dp
U (13)
t x dx
u u u U U 2u
u v U 2 (14)
t x y t x y
u v
and 0 (15)
x y
Although these equations are obtained for a rectilinear flow but they hold for
curved flow if the curvature of the boundary is small in comparison to the
boundary layer thickness.
The integration of (14) and (15) can be simplified if we can reduce the number
of variables by introducing the stream function .
ψ ψ
where u= , v (16)
y x
ψ ψ
0, y 0
x y
(18)
ψ
U(x, t ), y
y
The exact solution of (17) was given by H. Blasius in 1908, for the case of
steady flow t 0 past a flat plate (U = constant).
FLUID DYNAMICS 310
Let us consider the steady flow of an incompressible viscous fluid past a thin
semi-infinite flat plate which is placed in the direction of a uniform velocity
U. The motion is two-dimensional and can be analysed by using the Prandtl
boundary layer equations. We choose the origin of the co-ordinates at the
leading edge of the plate, x-axis along the direction of the uniformal stream
and y-axis normal to the plate. The Prandtl boundary layer equations, for this
case, are
u u 2u
u v 2 (1)
x y y
u v
0 (2)
x y
where u, v are the velocity components and v is the kinematic viscosity.
y
U
U
u
O x
The boundary conditions are
uv0 when y 0
(3)
u U when y
In this problem, the parameters in which the results are to be obtained, are U,
v, x, y. So, we may take
u
= F(x, y, v, U) = F() (4)
U
Further, according to the exact solution of the unsteady motion of a flat plate,
we have
x
~ t ~ (5)
U
where x is the distance travelled in time t with velocity U. Hence the non-
dimensional distance parameter may be expressed as
311
FLUID DYNAMICS
y y U
= y (6)
δ vx U vx
ψ ψ
= u dy u , v
y x
dy
= U F() dη
dη
x
= U F() d xU f () (7)
U
U
u= xU f ' () = Uf () (8)
y y x
1 U U 1
v = f () vxU f ' () y 3/ 2
x 2 x v 2x
1 U 1 U
v= f () y f ' ()
2 x 2 x
1 U U
= y f ' () f ()
2 x x
1 U
= ( f ' () f ()) (9)
2 x
Also,
u 2
U f ' ' ()
x xy x
1 U 1
U f ' ' (). y
2 v x3/ 2
FLUID DYNAMICS 312
1 U
= η f ' ' (η) (10)
2 x
u U
U (f ' ' (η)) U f ' ' (η) (11)
y y vx
2 u U 2
f ' ' ' (η) (12)
y 2 vx
U f ()
1 U 1 U U
f ' ' (n ) ( f ' () f ())U f ' ' ()
2 x 2 x x
U 2
=v f ' ' ' ()
x
U 2 U2 U2
or η f ' f ' ' ( η f 'f ) f ' ' f ' ' '
2x 2x x
or f f + f f f f = 2f
or 2 f + f f = 0
d 3f d 2f
i.e. 2 f 0 (13)
d3 d 2
Therefore,
Therefore,
313
FLUID DYNAMICS
The third order non-linear differential equation (13) has no closed form
solution, however, Blasius obtained the solution in the form of power series
expansion about = 0.
Let us consider
c 2 2 c3 3
f() = c0 + c1 + + ……. (16)
|2 |3
c3 2 c 4 3
f () = c1 + c2 + η η + ……. (17)
|2 |3
c 4 2 c5 3
f () = c2 + c3 + η η + ……. (18)
|2 |3
c5 2 c 6 3
f () = c3 + c4 + η η + ……. (19)
|2 |3
The constants ci‟s are determined from the boundary conditions (14), (15) and
the differential equation (13). From (14), we get
c0 = c1 = 0
c2 c4
0 = (2c3 + 2c4 + c52 +…..) + (c0 + c1 + 2 +….) (c2 +c3 + 2 ...)
|2 |2
c c c 2
+ c 5 0 4 c1c 3 2 2 +…. = 0
2
|2
c2
i.e. 2c3 + 2c4 + c 5 2 2 .... 0
2
FLUID DYNAMICS 314
c3 = c4 = c6 = c7 = c9 = c10 = 0
c 22 11 3 375 4
c5 = , c8 c2 , c11 c2
|2 4 8
c 2 2 c 22 5 11 3 8 375 4 11
f() = c2 c2 +… (20)
2 2 |5 4 |8 8 | 11
df
1 as n
dη
We write (20) as
f() =
(c1/ 3) 2 1 (c12/ 3) 5 11 (c12/ 3) 8 375 (c12/ 3)11
c12/ 3 2 ...
| 2 2 |5 4 |8 8 | 11
Therefore,
3/ 2
1
c2 = 1/ 3
(22)
lim f ' (c 2 )
The shearing stress 0 on the surface of the plate can be calculated from the
results of the Blasius solution. Thus, we have
315
FLUID DYNAMICS
UC2 0.332 2
= U (23)
vx U R ex
0 0.664
Cf = (24)
1 2 R ex
U
2
The total frictional force F per unit width for one side of the plate of length l is
given by
l
l
0 d x 0.664 U
2
F= (25)
0
U
Equation (25) shows that frictional force is proportional to the 3/2th power of
the free stream velocity U .
lU
Where R el .
x 5.64x
= 5.64 (27)
U Re
x
where the right-hand size signifies the decrease in total flow caused by the
influence of the friction and the left-hand side represents the potential flow that
has been displaced from the wall. Hence the displacement thickness 1 is that
distance by which the external potential field of flow is displaced outwards due
to the decrease in velocity in the boundary layer.
u
i.e. 1 = 1 dy (29)
0 U
u
Using the expressions for and y from (8) and (6) respectively, we find 1
U
for the flow on a flat plate, as
vx
1 = (1 f ' )dη
U 0
Ux
= lim [ f()]
U
vx 1.7208 x
= 1.7208 (30)
U Re x
u u
or 2 = 1 dy (31)
0 U U
Again, using (8) and (6), we obtain 2 for the case of the flow on a flat plate, as
x
2 =
U f (1f ) d
0
x 0.664 x
= 0.664 (32)
U R ex
U u
u (Velocity deficit)
2 1
x
O
FLUID DYNAMICS 318
We have observed that the solution of the steady boundary layer equations is
very difficult. The solution obtained in the previous case is also a very special
case. For engineering problems, it is often acceptable if an approximate
solution can be obtained. One of the most useful methods is the Von Karman-
Pohlhausen method based on the integral theorem. The basic concept of this
method is that the solutions satisfy the differential equations only on the
average, i.e., it is not anticipated that the solution satisfies the boundary layer
equations at every point (x, y) but the momentum integral equation and the
boundary conditions must be satisfied. The momentum integral equation is
obtained by integrating the boundary layer equations with respect to y over the
boundary layer thickness or by the momentum law.
4.1. Momentum Integral Equation for the Boundary Layer (Von Karman Integral
Relation). The Prandtl‟s boundary layer equations are
u u u 1 dp 2u
u v v 2 (1)
t x y ρ dx y
u v
and 0 (2)
x y
Integrating (1) w.r.t. y from y = 0 to y = (x), the outer edge of the boundary layer, we get
u u 1 dp 2u
t 0
u dy 0 u dy 0 v dy 0 dy 0 2 dy (3)
x y dx y
Let us simplify the third term on L.H.S. of (3). We have
u (uv) v
0 v y dy 0 y
dy 0 u
y
dy
v
= 0 d(uv) 0 u y dy
= uv 0 0u v dy
y
v v
= U 0 dy 0 u dy (4)
y y
where u = U at y = .
v u
Replacing by , from the continuity equation, we get
y x
319
FLUID DYNAMICS
u u u
0 v y dy U0 x dy 0 u x dy (5)
y u u 1 dp 2u
t 0 0 x 0 x 0 x 0 dx 0 y 2
u dy u dy U u dy u dy dy dy
u u 1 dp u
t 0
u dy 2 0 u dy U 0
x x
dy 0
dx
dy
y 0
u
2
u 1 dp u u
t 0
u dy 0 x
dy U0
x
dy 0
dx
dy
y y y y0
1 dp
=
dx y y 0
dp 0
= (6)
dx
u
where = = 0 at y = .
y
= 0 at y = 0
Let us further simplify the second and third terms on the L.H.S. of (6). For this we use the
Leibnitz rule according to which
d b( x ) b( x ) f da db
dx a ( x )
f (x, y)dy a ( x )
x
dy f (a, y) f (b, y)
dx dx
u 2 d 2 d
0 x
dy
dx 0
u dy U 2
dx
u d d
and U 0 dy U 0 u dy U 2
x dx dx
Thus, equation (6) reduces to
d 2 d dp 0
t 0
udy
dx 0
u dy U
dx 0
u dy
dx
(7)
This is one form of the Von Karman integral relation and is also called the momentum integral
equation of the boundary layer.
FLUID DYNAMICS 320
4.2. Von Karman Integral Relation by Momentum Law. The Von Karman integral
equation of the boundary layer represents the relation between the overall rate of flux of
momentum across a section of the boundary and the surface forces due to the wall shearing
stress and the pressure gradient. The Von Karman integral equation which we just obtained,
can be derived from the momentum theorem of fluid mechanics.
Let us consider an element of the boundary layer, ABCDA of unit length perpendicular to the
xy-plane as shown in the figure.
1 dp
p 2 dx x x
U
y
C
u d
pδ (pδ)x
p dx
0
O x
x A x D
Let AD = x be the small length of the element in the x-direction and be the thickness of the
boundary layer at a distance x from the leading edge of the plate. We assume that the velocity
of the boundary layer flow at the outer edge of the boundary layer is the same as that of the
potential flow, i.e., u = U at y = .
0
u dy +
x
u dy
0
x
d
dx
u dy
0
x (1)
Since there is no flow across the surface of the plate AD, so by continuity
equation, the rate of mass flow out of the element across BC must be
d
dx
u dy
0
x (2)
U
d
dx
u dy
0
x (4)
where U is the velocity across BC in the x-direction. Total outward flux of momentum
becomes
d 2 d
dx 0 u dy U dx 0 u dy x (5)
t 0 u dy x (6)
The forces acting on the fluid due to the shearing stress at the wall is
u
x = 0 x, in the x-direction (7)
u y0
and due to the difference of pressure along AB and CD is
d 1 dp d dp
p p (p)x p x x x (8)
dx 2 dx dx dx
where we have neglected terms of order x.
Now, according to the momentum law, we have
Rate of change of momentum in x-direction. = Total force in x-direction
d 2 d
t 0 u dy x dx 0 u dy U dx 0 u dy x
do
= x 0 x (9)
dx
Dividing both sides of (9) by x, we get
d 2 d dp 0
t 0
u dy
dx 0
u dy U 0 udy
dx
dx
(10)
which is the required Von Karman integral equation, being the same as obtained by integrating
Prandtl‟s boundary layer.
4.3. Other Forms of the Von Karman Integral Equation. It is often convenient to have the
integral equation in terms of displacement and momentum thicknesses. The momentum
integral equation of the boundary layer is
FLUID DYNAMICS 322
d 2 d dp 0
t 0
u dy
dx 0
u dy U
dx 0
udy
dx
(1)
U U 1 dp
U (2)
t x dx
dp
where U is the velocity of the potential flow, is the pressure gradient, is the density
dx
, is the thickness of the boundary layer and 0 is the shearing stress at the wall. For
a steady flow, we obtain from (1) and (2)
d 2 d dU 0
dx 0
u dy U
dx 0
u dy U
dx
or
d 2 d dU dU
dx 0
u dy 0 U u dy
dx dx 0
udy U
dx 0
dy 0
or
d dU
dx 0
u ( U u ) dy
dx 0
(U u) 0
(3)
u u
1 = 0 1 U dy U1 = 0 1 U dy (4)
and
u u
2 = 0 U 1 U dy U22 = 0 u(U u)dy (5)
d dU
(U 2 2 ) U1 0
dx dx
or
d 2 dU dU 0
U2 2U 2 U1
dx dx dx
or
d 2 1
2 2 1
dU
02 (6)
dx U dx U
323
FLUID DYNAMICS
This is the Von Karman momentum integral equation in terms of displacement and momentum
thicknesses.
4.4. Application of the Momentum Integral Equation to Boundary Layers (Von Karman
Pohlhausen Method). Pohlhausen introduced a fourth degree polynomial for the velocity in
terms of a non-dimensional parameter = y/, 0 1 such that
u
f () a b2 c3 d4 (1)
U
The constants a, b, c d are to be determined from the boundary conditions
2u 1 dp U dU
u = 0, v = 0, , at y = 0 (2)
y 2 dx dx
u 2u
u = U, 0, 2 0, at y = (3)
y y
The first two conditions in (2) and the first condition in (3) are satisfied by all exact solutions
of the boundary layer equations. The second condition in (3) is meant for continuous flow on
the outer boundary of the layer. The third condition in (2) is obtained from Prandtl‟s boundary
layer equation i.e.
u u u 1 dp 2u
u v v 2
t x y dx y
When the flow is steady and u = 0 = v on y = 0, then
2 u 1 dp 1 dp U dU
y 2 dx dx dx
2u
The point where 0 is called a point of inflexion of the velocity profile in the boundary
y 2
layer. From (1), we get
2u 1 d2u U
(2b+6c+12d2) = 0
y 2 δ 2 dη2 δ 2
6d2 + 3c + b = 0
This gives two values of . One of the points is near the wall and other is in the upper region
of the boundary layer. For this reason, the boundary condition.
2u
0 at y = is imposed.
y 2
Let us now use the conditions (2) and (3) in (1) to find out
2u U dU 2bU U dU
= 0, 2
y 2
dx dx
FLUID DYNAMICS 324
2 dU 2 dU
b= ,
2 dx 2 dx
λ
= 1, u = U a c d 1
2
u
= 1, = 0 a +3c +4d = 0 (4)
y
2u
= 1, 0 + 6c + 12d = 0
y 2
Solving (4), we get
a = 2+ ,b , c 2 , d 1 (5)
6 2 2 6
Therefore, the velocity in (1) has the expression
u λ λ λ λ
f (η) 2 η η 2 2 η3 1 η 4
U 6 2 2 6
η η 2 η3 η 4
= 223 + 4 +
6 2 2 6
= F() + G() (6)
η η
G() = (1 3η 3η 2 η3 ) = (1)3
6 6
2 dU
= (7)
dx
The velocity profile expressed in terms of in (6) constitute a one-parameter family of curves
with a dimensionless parameter which depends mainly on the pressure gradient of the flow.
may be written as
δ 2 dU dp δ
=
v dx dx μU / δ
which can be interpreted physically as the ratio of the pressure force to viscous force. This is
known as the shape factor.
λ
f() = (223+4) + (1)3
6
Therefore,
325
FLUID DYNAMICS
df λ
(262 + 43) + [(1)3 3(1)2]
dη 6
λ
= (262 + 43) + (1u) (1)2
6
d 2f λ
and (1 η)12 1η λ
dη 6
2
df λ
when = 0, 2
dη 0 6
df
0 = 12
dη 0
This is taken as lower limit of . The upper limit of can be determined from the condition of
d 2f λ λ
zero curvature of the velocity, i.e. 0 which gives = 1 . It is seen that
dη 2 12 6
for 12, 1.0 and for > 12, < 1.0. Hence for > 12, the point of inflexion occurs
within = 1.0 i.e. the velocity profile in the boundary layer becomes greater than the velocity
in the potential flow. This is not justified physically. Therefore, we take 12. So, the limits
of are 12 12.
u u
2 = U 1 U dy (9)
0
y
Using (6) and = , we have
δ
1 1
0
[1F() G()] d
1
1 2 2 4 (1 ) 3 d
3
=
0 6
1
η4 η5 λ (1 η) 5 λ (1 η) 4
η η
2
=
2 5 6 5 6 4 0
FLUID DYNAMICS 326
3 λ
= (10)
10 120
and
2 1
0
[F()+ G()] [1F()G()]d
37 λ λ2
= (11)
315 945 9072
The shearing stress 0 at the wall is given by
u μU λ
0 = 2 (12)
y y0 δ 6
d 2 1 dU
(2 2 1 ) 02
dx U dx U
U 2
by to find out
U 2 d 2 2 dU 0 2
2 1 2 (13)
dx 2 dx U
To simplify (13), we use the parameters
2
2 2 dU 22 2
z = 2 , K zU' 2 (14)
dx 2
λ 12 δ 2 λ 12 37 λ λ2
(17)
6 δ 6 315 945 9072
=
δ 2 dδ 2 1 dz
Using the values of z, k, f1(k), f2(k) in (13), and noting that , the momentum
v dx 2 dx
integral equation takes the form
U dz
(2+f1(k)) k = f2(k)
2 dx
dz F(k)
or (18)
dx U
where F(k) = 2f2(k) 2 (2+f1(k)) k (19)
Equation (18) is a non-linear differential equation of the first order for z.
dz
At the stagnation point x = 0 and U = 0. At this point cannot be infinite and so F(k) = 0.
dx
This gives the value of at the stagnation point. Thus, we have
Using the values from (15), (16) and (17), it is obtained that initial value 0 of at the
stagnation point is 0 = 7052.
5.1. Physical Approach : The decelerated fluid particles in the boundary layer do not remain
in the thin layer which adheres to the body along the whole wetted length of the wall. In some
cases the thickness of the boundary layer increases considerably in the downstream direction
and the flow in the boundary layer become reversed. The decelerated fluid particles then no
longer remain in the boundary layer but forced outwards, which means that the boundary layer
separates from the wall. Such phenomenon is known as boundary layer separation and the
point at which the boundary layer separates is known as point of separation.
The phenomenon of boundary layer separation is primarily connected with the pressure
distribution in the boundary layer and is very common in the flows about blunt bodies, such as
circular and elliptic cylinders or spheres. The fluid flow in the boundary layer is determined
by the following three factors.
(ii) It is pulled forward by the free stream velocity above the boundary layer
We have already observed that the pressure in the boundary layer is the same as it is outside
the boundary layer. Let us consider a curved surface as shown in the figure
FLUID DYNAMICS 328
dp
dp
0 0
U(x) dx
dx
y
2u
0
y 2
2u
2u 0
Upstream of the highest point
0 y 2
y 2 the stream lines of the outer flow converge, resulting in an
increase of the free stream velocity U(x) and a consequent fall of pressure with
dp
x 0 i.e. favourable pressure gradient. Downstream of the highest point the stream
dx
dp
lines diverge, resulting in a decrease of U(x) and a rise of pressure with x 0 . In the
dx
dp
region with rising pressure , 0 i.e. adverse pressure gradient. along the wall, the
dx
retarded fluid particles with small momentum and energy cannot penetrate too far. Thus, the
forward flow is brought to rest and thereafter a back flow sets in the direction of the pressure
gradient. This causes a boundary layer separation and the point at which the forward flow is
brought to rest is called the point of separation.
5.2. Analytical Approach. In this approach, the separation phenomenon may be explained by
applying the Prandtl‟s boundary layer equations both outside the boundary layer and at the
wall. Outside the boundary layer, the equation is
dU 1 dp
U (1)
dx dx
and at the wall, i.e. at y = 0, we have u = v = 0, the equation is
2 u dp
μ 2 = (2)
y 0 dx
u 2u
It may be noted that at the outer edge of the boundary layer both and 2 tend to zero,
y y
u 2u
from the positive side whereas from the negative side, as at the outer edge the
y y 2
maximum value of u i.e. U should occur and the boundary layer flow merges smoothly with
the potential flow.
Since it is clear from equation (1) that the curvature of the velocity
profiles in the immediate neighbourhood of the wall depends only on the
pressure gradient, we consider the following three cases :
329
FLUID DYNAMICS
dp
(i) = 0 i.e. zero pressure gradient i.e. constant pressure :
dx
2u u
In this case
y 2 0 and hence the velocity gradient y decreases steadily from a
0
positive value at the wall to zero at the outer edge of the boundary layer. The velocity profile
must therefore have a steadily decreasing form (figure 1).
y y
y
u 2u
O u O
y
O y 2
Figure 1
3u 4u
The point of inflexion occurs on the wall since
y 3 0 but y 4 0, which can
0
easily be verified by differentiating the boundary layer equation w.r.t. y and evaluating the
value at y = 0. The fluid particles continue to move forward and therefore separation of
boundary layer does not occur.
dp
(ii) < 0 i.e. favourable pressure gradient :
dx
2u
For this case, from equation (2), we conclude that
y 2 0 and therefore it increases
0
steadily to the value zero at the outer edge (y = ) of the boundary layer. The velocity gradient
u
again decreases steadily from a positive value at the wall to the value zero at the outer
y
edge of the boundary layer. The velocity profile does not have any point of inflexion (figure 2)
and has a form similar to the case of zero pressure gradient. In this case also, the fluid particles
continue to move forward and so there is no boundary layer separation.
y y
y
dp u 2u
(iii) >0 i.e. adverse pressure gradient :
dx u
O O y O
y 2
Figure 2
FLUID DYNAMICS 330
2u
will be a positive quantity. In order to have a positive value of u at
2
In this case
y 2 y 2
0
u
y = 0, the slope of the velocity gradient at y = 0 must be positive. But the boundary
y
u
condition requires = 0 at y = . Therefore, the slope of the velocity gradient must change
y
signs from positive to negative in the boundary layer which results in point of inflexion of the
velocity profile in the boundary layer (fig. 3). The velocity gradient at the wall is much
smaller compared to the case of zero pressure gradient.
y
y y
P.I.
O u O
u O 2u
y y 2
Figure 3.
As the adverse pressure gradient increases further, the velocity profile may become
u
increasingly distorted until the velocity gradient at the wall is zero, as shown in figure
y 0
3. At this point, separation of flow from the wall begins. Further downstream, a back flow in
the direction of the pressure gradient sets in.
It should be noted here that the type of velocity profile shown in figure 3 is naturally unstable
and it frequently happens that the transition to turbulent flow in the boundary layer will take
place before laminar separation can occur. Under such circumstances, the turbulent boundary
layer will be maintained and separation of flow from the wall will be delayed.
Further, the point of separation is defined as the limit between forward and reverse flow in the
layer in immediate neighbourhood of the boundary wall. In other words, the point of
u
separation is the point at which = 0 i.e. 0 = 0.
y 0