Académique Documents
Professionnel Documents
Culture Documents
November 2004
We have seen that, in general, high Reynolds number flow past a slender body such as an airfoil
can be considered as an irrotational outer flow (that can be determined in principle at least
from potential flow theory) and a thin inner flow regime in which viscous effects are important
and lead to the generation of vorticity. This thin inner region is often referred to as a viscous
boundary layer and denoted generically ! (x) . Here we shall consider the inner flow region in
detail and wish to see what simplifications to the equations of motion are possible due to the
thinness of the boundary layer.
We consider a 2D boundary layer next to a solid wall on which the no-slip boundary condition is
to be applied. We shall use a boundary layer coordinate system in which x is along the
surface and y is normal to the surface. A similar analysis can be performed in three-dimensions
but requires an explicit assumption of the lateral scale of the three-dimensional flow. The
following analysis is also applicable to boundary layers on axisymmetric objects provided that
the boundary layer is thin enough that local curvature terms are always negligible (i.e.
! (x) R(x) ! 1 ).
The boundary layer thickness grows with x but is small with respect to the length scale of the
surface ! ( x ) L << 1 . We proceed as follows:
i)
An estimate of the thickness where both inertial and viscous forces balance each other
is (as we have seen before)
"vx &
" 2 vx *
#
!
v
~
,
%$ x
(
"x '
"y 2 ,
+
U0
U0 ,
!U 0
~ 2
,L
)
!~
L
U"
or
!
~
L
1
ReL
G.H.McKinley
!vx !vy
+
=0
!x
!y
We express this in dimensionless form with scaled variables x* = x L , y* = y ! and
u = vx U 0 , v = vy V (where V is, at the moment, unknown):
ii)
If we have picked the characteristic y-component V correctly (so that the scaled partial
derivatives, !u !x * and !v !y * are both 0(1)) then we expect:
"!%
V ~ $ ' U0
# L&
The variations in the y-component of the velocity are thus much smaller than the variations in vx
(at least within the boundary layer). This is analogous to our approach in Lubrication Theory so
far.
iii)
We next consider the magnitude of pressure variations within the boundary layer.
The first order solution to this flow problem at Re >>1 is the inviscid flow given by
Eulers Equation (or Bernoullis equation) and we thus expect that Pc ~ !U 0 2 .
The pressure field obtained from Bernoullis equation gives an excellent description
of the actual pressure field measured experimentally for slender bodies. This should
be expected since introducing viscosity into the problem affects the tangential
shear stresses only [you should be able to show that along a solid surface
!vi !xi = 0 for i = j ].
The full nondimensional equations of motion in the 2D boundary layer are thus:
Scalings:
x* = x L
u = vx U 0
p* = p !U 02
y* = y !
v = vy (!U " L )
!=L
Re = vL U"
x-component:
!u
!u
!p * 1 !2u
!2u
u
+v
="
+
+
!x *
!y *
!x * Re !x* 2 !y* 2
(1)
y-component:
1 " !v
!v %
1 !2 v
1 !2 v
!p *
u
+v
+
+
=(
Re $# !x *
!y * '&
!y * Re2 !x* 2 Re !y* 2
(2)
!u
!v
+
=0
!x * !y *
(3)
continuity:
G.H.McKinley
For high Reynolds number flows these equations can be simplified; we ignore all items
O (1 Re) , O 1 Re2 . The resulting equations when expressed in dimensional terms are:
!vx
!vx
! 2 vx
1 !p
vx
+ vy
="
+$ 2
!x
!y
# !x
!y
0=!
"p
"y
(4)
(5)
!vx !vy
+
=0
!x
!y
(6)
Note the difference between these equations and the lubrication analysis (for quasi-steady,
locally fully-developed flows) we have discussed previously. In the flow direction, eq. (4)
shows that inertial, viscous and pressure forces are all important. In the lubrication scaling only
pressure and viscous forces remain.
Finally a note about pressure fields.
Equation (5) tells us that to leading order there is no variation in the pressure across the
thickness of the boundary layer. The pressure field inside the boundary layer can thus be
equated with the pressure field in the inviscid flow far from the solid surface.
This outer pressure field is said to be imposed on the inner flow and can be determined
by solving the inviscid Euler equations (or by using Bernouilli) a priori (for a given p! , U 0 ).
The boundary conditions for the equations are given by (in dimensionless terms):
u(x*, 0) = 0
No slip
v(x*, 0) = 0
No Penetration
u(x*, !) = U(x*)
Initial Condition
Note that because of the nonlinear terms, such as u x !u x !x etc., these equations are not elliptic
(like the lubrication equations) but rather are PARABOLIC convection-diffusion equations and
thus require an initial condition. Here x plays the role of a time-like variable describing diffusion
in the y -direction as convection in the x -direction increases.
The fundamental concept of the boundary layer equations given by (13) or (46) is that we
introduce a stretching of the small variable (in this case the length scale y ) so that terms in
G.H.McKinley
the equations become of order O(1) . At a position x we know ! ( x ) ~ " x U 0 and we stretch
y by rescaling it (i.e. by dividing by a small parameter ) to become y* = y ! where 0 ! y* < 1 :
Hence y* =
U 0 "1 2
x
y
!
or
y* = Re x ( y x )
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Consider a laminar boundary layer or the laminar sublayer of a turbulent boundary in two-dimensional ow.
The uid is incompressible and has constant viscosity.
Show that, at the wall, the velocity prole is concave upwards in ow with a favorable pressure gradient
(p/x 0). Whereas it is concave downwards for ow with an unfavorable pressure gradient (p/x > 0).
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Porous plate
Suction
Consider a at porous surface with no lengthwise pressure gradient to which suction is applied for the purpose
of sucking o the laminar boundary layer. Let us denote by v0 the downward component of velocity at the
surface of the plate (y = 0). Then it may be shown that for very small values of v0 /U compared with
unity, the laminar boundary layer becomes constant in both thickness and velocity at large distances from
the leading edge, provided that v0 is constant.
For this case of small v0 /U and large distance from the leading edge, nd:
(a) the velocity prole, by relating u/U to v0 y/
(b) the displacement thickness Reynolds number, v0 /.
(c) the skin-friction coecient, 20 /U 2 , in terms of v0 , U , and .
(d) develop an estimate for the distance (in x) at which the boundary layer thickness approaches the
asymptotic constant value found in Parts (a)-(c).
Hint: Start with the dierential equations of the laminar boundary layer.
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Consider the two-dimensional, incompressible, steady ow of a uid of constant viscosity in the entrance
region of a slit of width 2h. The ow enters the tube (x = 0) with the uniform velocity V . At distance x
from the entrance, the boundary layer thickness is , and the core ow has the speed U . The boundary layer
is laminar.
The ultimate objective is to analyze the entrance region, using momentum integral method, to determine
how the boundary layer thickness, the skin friction stress, and the pressure gradient all vary with distance
x. Since it is known that the fully-developed ow has parabolic velocity prole, it is agreed to assume,
with some approximation, that the boundary layer velocity prole in the entrance region is also parabolic,
following the equation
u
y
=2
! "2
y
(9.09a)
Carry your analysis only to the point of obtaining a dierential equation relating the dimensionless boundary
layer thickness /h as dependent variable to the dimensionless length x/h as the independent variable. Any
constants of the problem may of course appear in the dierential equation. Do not attempt to integrate the
latter.
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Fall 2009
Denote by (x) the displacement thickness of the boundary layer and by U(x) the horizontal fluid
velocity at the edge of the displacement thickness. Assume that the horizontal velocity profile
u(x, y) inside the boundary layer at a distance x downstream of the leading edge is given by the
expression
3
" y #
" y #
u(x, y)
= a( x) $
% + b( x ) $
% , 0 < y < ! ( x)
U ( x)
& ! ( x) '
& ! ( x) '
Assume that at the edge of the displacement thickness of the boundary layer the horizontal fluid
velocity is equal to that of the inviscid flow above it.
a) (3 points) Determine a( x) and b( x) by providing appropriate physical justifications.
b) (3 points) Derive an expression relating U (L) with U ! , d and ! (L) .
c) (3 points) Apply the mass conservation principle and derive an expression for the
displacement thickness at x=L, i.e., ! (L) .
d) (3 points) Determine the pressure variation at x =L : p ( L, y ), 0 < y < ! (L) .
e) (4 points) Apply the momentum conservation principle to determine the horizontal force
acting on the plate over its length 0<x<L.
f) (4 points) Assuming that the velocity profile u(x, y) for x>L is given by the same
expression as for x<L, obtain the shear stress (as a function of x) for x>L on the plate.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider a laminar boundary layer or the laminar sublayer of a turbulent boundary in two-dimensional ow.
The uid is incompressible and has constant viscosity.
Show that, at the wall, the velocity prole is concave upwards in ow with a favorable pressure gradient
(p/x 0). Whereas it is concave downwards for ow with an unfavorable pressure gradient (p/x > 0).
c 2010, MIT
Copyright @
Boundary Layers
Solution:
Lets consider laminar boundary layer equation, i.e.,
2u
1 p
u
u
+ 2
=
+v
x
y
y
x
0=
1 p
y
u v
+
=0
x y
(9.03a)
(9.03b)
(9.03c)
u =
(9.03d)
(9.03e)
2u
1 p
= 2
x
y
(9.03f)
where the sign of the Eq. (9.03f) determines the sign of pressure gradient which also determines the sign of
curvature. The separation point is dened as a point where shear stress becomes zero, i.e.,
o =
u
=0
y
(9.03g)
D
Problem Solution by J.Kim, Fall 2009
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Porous plate
Suction
Consider a at porous surface with no lengthwise pressure gradient to which suction is applied for the purpose
of sucking o the laminar boundary layer. Let us denote by v0 the downward component of velocity at the
surface of the plate (y = 0). Then it may be shown that for very small values of v0 /U compared with
unity, the laminar boundary layer becomes constant in both thickness and velocity at large distances from
the leading edge, provided that v0 is constant.
For this case of small v0 /U and large distance from the leading edge, nd:
(a) the velocity prole, by relating u/U to v0 y/
(b) the displacement thickness Reynolds number, v0 /.
(c) the skin-friction coecient, 20 /U 2 , in terms of v0 , U , and .
(d) develop an estimate for the distance (in x) at which the boundary layer thickness approaches the
asymptotic constant value found in Parts (a)-(c).
Hint: Start with the dierential equations of the laminar boundary layer.
c 2010, MIT
Copyright @
Boundary Layers
Solution:
Assumptions:
1.
dp
dx
2. vy = v0 at y = 0 where v0 = constant.
3. For
v0
U
n
1 dp
2 vx
vn
vx
x
vxn
= nn +
+vy
y
y 2
n x
n dx
assumption 3
assume steady
assumption 1
From continuity
vx
x
vy
=0
y
vy = constant = v0
assumption 3
0=
B.C.
2 vx
vx
+ v0
2
y
y
(9.04a)
at y = 0 x
as y x
vx = 0
vx = U
(a) Solve for vx by assuming vx = A exp(By) + C and substitute that into Eq. (9.04a):
v0
B=
0 = AB 2 exp(By) + v0 AB exp(By)
v0 y
vx = A exp
+C
v0 y
vx = U 1 exp
(9.04b)
1
0
vx
U
dy =
exp
0
v0
=1
vx
U
dy =
v0
(9.04c)
c 2010, MIT
Copyright @
Boundary Layers
(c)
u
0 =
y y=0
U
U v0
=
= U v0
2
2
U
v0
Cf =
=
2
U
U
OR 0
Cf =
2v0
U
(9.04d)
(d) We could think of the problem in terms of the Rayleigh problem, then
t ,
(9.04e)
but what is t ? t is the characteristic time for the uid particles to reach the point x, then t =
Then,
x
,
U
x
U .
(9.04f)
which is the growth of the BL for the normal Blasius Solution. Now, when is the characteristic vertical
length equal to ? Lets assume that the characteristic vertical length is , i.e. the BL reaches its
steady state, then from mass conservation,
U
v0
+
= 0,
x
(9.04g)
then, v0 x U . Now, elevating to the second power this expression, and replacing the order of
magnitude of obtained before,
v02
2
x U
,
U x2
(9.04h)
U
,
v02
(9.04i)
simplifying,
x
then, for x
U
v02
Note: Using suction to enhance the B.L adhesion is an idea that has been investigated since 1940, one of the
last research airplanes that incorporated this technology is the NASA experimental version of the F-16XL.
It has laser-cut holes to allow suction and enhance the attachment of the boundary layer.
c 2010, MIT
Copyright @
Boundary Layers
http://grin.hq.nasa.gov/ABSTRACTS/GPN-2000-001935.html
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider the two-dimensional, incompressible, steady ow of a uid of constant viscosity in the entrance
region of a slit of width 2h. The ow enters the tube (x = 0) with the uniform velocity V . At distance x
from the entrance, the boundary layer thickness is , and the core ow has the speed U . The boundary layer
is laminar.
The ultimate objective is to analyze the entrance region, using momentum integral method, to determine
how the boundary layer thickness, the skin friction stress, and the pressure gradient all vary with distance
x. Since it is known that the fully-developed ow has parabolic velocity prole, it is agreed to assume,
with some approximation, that the boundary layer velocity prole in the entrance region is also parabolic,
following the equation
y
u
=2
2
y
(9.09a)
Carry your analysis only to the point of obtaining a dierential equation relating the dimensionless boundary
layer thickness /h as dependent variable to the dimensionless length x/h as the independent variable. Any
constants of the problem may of course appear in the dierential equation. Do not attempt to integrate the
latter.
c 2010, MIT
Copyright @
Boundary Layers
Solution:
Velocity prole in boundary layer assumed parabolic and it is give as
u
y
y
= 2
U
u = U
in
0y
in
(9.09b)
yh
(9.09c)
Lets apply mass conservation to determine U = U (x). At region (1) and (2), the volume ow rates are
respectively
V dy
Vh
at
(1)
(9.09d)
0
h
u dy
2U
U 2
U
y 3 y3
=
=
V =U
y
y
U
=U
3h
dy +
U dy
(9.09e)
+ U (h )
(9.09f)
1
h
3
at
U (x) =
(2)
(9.09g)
V
1
(9.09h)
(x)
3h
Now lets consider Karman momentum integral equation (See Kundu textbook p.362 - 364 for derivation).
d
dU
o
U 2 + U
=
dx
dx
(9.09i)
1
0
u
U
dy ,
=
0
u
u
1
U
U
dy
(9.09j)
d 2
U
dx
u
u
1
U
U
dy + U
dU
dx
1
0
B
u
U
dy =
|{z}
(9.09k)
where the integral range has been replaced from 0 to because 1 u/U is zero outside boundary layer.
Lets calculate each term. If we substitute the given velocity prole into A is, then it is
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
Boundary Layers
d 2
U
dx
A :
y y 2
2
y y 2
12 +
dy
(9.09l)
(x)
(9.09m)
then
Z
1
d
U 2
2 2 1 2 + 2 d
dx
0
2 d
=
U 2
15 dx
(9.09n)
(9.09o)
U
dx
B :
Z
0
dU
1
1 2 + 2 d = = U
dx
3
(9.09p)
Finally, C becomes
o
=
C :
u
y
= U
2 2y
2
=2
y=0
(9.09q)
9U
dU
d
U
+ 2U 2
= 30
dx
dx
(9.09r)
We cannot solve this equation because both U and are function of x but we have only one equation. For
the relation between U and , we can use Equation (9.09g) which was obtained by mass conservation. For
this, we may use the following algebras.
dU
dx
U
dU
dx
V
1
V
1
1
2
3h
(x)
d
dx
(9.09s)
3h
2
2
3
(x)
3h
1
3h
d
dx
(9.09t)
c 2010, MIT
Copyright @
Boundary Layers
9
V2
1
2
3
(x)
3h
1
3h
d
V2
d
V
+ 2
= 30
2
dx
dx
1 (x)
1 (x)
3h
3h
(9.09u)
d
V
dx
1
1 13 h
3 h
1
1
3h
+2
= 30
(9.09v)
,
h
x
h
(9.09w)
Vh
2 + 7/3
2
1 1/3
= 30
dx
(9.09x)
U
1 p
=
x
x
O(1)
(9.09y)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Solution:
The given horizontal velocity prole u(x, y) is given inside the boundary layer as
u(x, y)
= a(x)
U (x)
y
5(x)
+ b(x)
y
5(x)
Firstly, substitute y = 5(x) into the given velocity prole, in which u(x, 5(x)) = U (x).
(1)
1 = a(x) + b(x)
@u
=0
@y
a(x) + 3b(x) = 0
(2)
a(x) =
3
,
2
b(x) =
1
2
(3)
c 2009, MIT
Copyright @
Since inviscid assumption is valid in outer ow, apply the Bernoulli from x = 0 to x = L along the top
streamline.
1 2
1
pa + U1
+ gd = pa + U (L)2 + g5(L)
2
2
( 2
+ 2g(d
) U (L) = U1
5(L))
(4)
)1/2
(5)
c) Determine 5(L)
U1 d
5(L)
u(L, y) dy
0
5(L)
U (L)
0
3
2
(6)
y
5(x)
5
U (L)5(L)
8
8U1 d
5(L) =
5U (L)
1
2
y
5(x)
3 !
dy
(7)
=
)
(8)
(9)
5(L) =
2
5 (U1
8U1 d
+ 2g(d
5(L)))
(10)
1/2
0=
@p
@y
@p
=
@y
(11)
The boundary condition for this ODE is the atmospheric pressure at y = 5(L). Then,
p(L, y) = pa + g (5(L)
y)
(12)
c 2009, MIT
Copyright @
Momentum ows in the x-direction through planes A and B and through the y = 0 plane via shear stresses
acting on the plate. No shear stresses contribute to the x-component of the momentum ux on plane B. No
momentum is owing through the free surface.
So the drag on the plate of the length L is equal to the momentum ux decit in x-direction :
=
=
U 2 d + g
U 2 d + g
Z 5(L)
y) dy
u(L, y)2 dy
0
5(L)2
17
U (L)2 5(L)
2
35
(d
0
d2
2
(L)
(5
y) dy
(13)
(14)
xy =
@u
2 U (x)
|y=0 =
@y
3 5(x)
(15)
We do not know how U (x) and 5(x) evolve along the distance. To obtain this, use mass conservation.
Q = U1 d
5
5
U (L)5(L) = U (x)5(x)
8
8
8 Q
) U (x) =
5 5(x)
=
(16)
(17)
or
xy
xy
2 8
Q
16 Q
=
15 5(x)2
3 5 5(x)2
2 5 U (x)2
5 U (x)2
=
3 8
Q
12 Q
(18)
(19)
Assume that over a very thin layer on the free surface at the edge of the boundary layer the ow is inviscid.
Then Bernoulli applies and
c 2009, MIT
Copyright @
1 2
1
U + gd = U (x)2 + g5(x)
2 1
2
(20)
1 2
1
8gQ
U + gd = U (x)2 +
2 1
2
5U (x)
(21)
This is a cubic equation for U (x) in terms of known quantities. Thus, the shear stress xy can be obtained
by using the solution of the above equation.
c 2009, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A at plate is hinged at one side to the oor, as shown, and held at a small angle 0 (0 1) relative to
the oor. The entire system is submerged in a liquid of density . At t = 0, a vertical force is applied and
adjusted continually so that it produces a constant rate of decrease of the plate angle .
d
= = Const,
dt
(5.18a)
c 2008, MIT
Copyright
Momentum Theorems
Solution:
(a) Recall from problem 3.5 (Shapiro and Sonin) that mass conservation gives:
u(t) =
x
.
2(t)
(5.18b)
x
.
2(0 t)
(5.18c)
(b) Now, we can analyze the wedge using form A of the Conservation of Momentum. Lets consider a
triangular, deforming C.V. encompassing the moving wedge and the uid beneath at all the times, as shown
in gure (2).
FH =
2.25 Advanced Fluid Mechanics
d
dt
VxF luid dV +
CV
CS
(5.18d)
c 2008, MIT
Copyright
Momentum Theorems
FH =
FH
d
dt
d
=
dt
U xtan()dzdx +
L 2
bL,
2
(5.18e)
x
L 2
xtan()bdx +
bL,
2
2
(5.18f)
Expanding the rst term, and simplifying using the small angle approximation, 1 ,
FH
d
=
dt
x
d
xtan()bdx =
2
dt
x
bdx = 0,
2
(5.18g)
Where the derivative of the integral is zero because it does not have any time dependence, all the quantities
inside the integral are constant (only is time dependent, but was cancelled as shown).
Finally, the sum of forces acting on the x direction are zero because outside the C.V. there is only atmospheric
pressure, and we have assumed a massless object compressing the uid. Therefore,
FH =
2 L3
b.
4
(5.18h)
Note that we assumed the velocity to be always parallel to the x-axis in this problem. This is a good
assumption but not completly true, specially near the plate, where we do have a downward component of
the velocity as well. This is partially compensated by having a wall nearly perpendicular to the moving
plate.
D
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Contents
1
3
3
4
4
4
5
5
5
6
6
6
7
7
8
9
9
9
11
11
13
13
14
14 a
15
16
2
4
17
19
1 For
a historical note on control volume analysis in engineering, see Chapter 4 of Walter G. Vincentis
What Engineers Know and How They know It, John Hopkins University Press, 1990.
2 A material volume is the same as a closed system in thermodynamics.
4
intrinsic properties are essentially uniform within it. An observer moving with a particle
(sitting on it, as it were) would see its properties change with time only (Fig. 2).
Fig. 2
v
For a material particle with infinitesimal volume V(t) , density (t), and velocity v , the
four laws have the following familiar forms:
Mass conservation
d
( V ) = 0
dt
(1)
This law asserts that the mass M = V of a material particle remains invariant.
(The prefix indicates quantities that are of infinitesimal size, and the prefix d refers to
changes that occur in the indicated property in time dt.)
Newtons law of (non-relativistic) linear motion
v
v
dv
( V)
= F , or
dt
v
d v
( v V ) = F
dt
(2)
Newtons law states that, relative to an inertial reference frame3, thev product of a
particles mass and acceleration is at every instant equal to the net force F(t) exerted on
it by the rest of the universe, or alternatively, that the rate of change of a particles
momentum (a vector quantity) is equal at every instant to the force applied to the particle
by the rest of the universe. (Actually the law states that the rate of change of momentum
is proportional to the applied force, with the coefficient being universal, but in most
3 An
inertial reference frame is one in which the particle would move at a perceptibly constant velocity if
all the forces acting on it were removed.
5
systems of measurement the universal coefficient is set equal to unity, which determines
the units of force in terms of those of acceleration and time.)
Newtons law applied to angular motion
v
d v
v
(r vvV ) = r F
dt
(3)
This law figures in rotary motion. The rate of change of a particles angular
v
momentum (the quantity in brackets on the left side of (3), r (t) being the particles
position vector) is at every instant equal to the net torque exerted on the particle by the
rest of the universe. This is not a new law, but one that follows from Eq. (2). Equation
v
(3) is obtained by taking the cross product of r (t) and Eq. (2), using Eq. (1), and noting
v
v v v
that dr dt v = v v = 0 . Like the law it is derived from, Eq. (3) is valid only in inertial
reference frames. Actually the law states that the rate of change of momentum is
proportional to the applied force, with the coefficient being universal, but in most
systems of measurement the universal coefficient is set equal to unity, which determines
the units of force in terms of those of acceleration and time.
First law of thermodynamics
d( etV ) = W + Q
(4)
The increase of a material particles total energy in a time interval dt (e t is its total
energy per unit mass, internal plus kinetic plus potential) is equal to the work W done in
the interval dt by forces exerted by the rest of the universe on the material volumes
boundary (that is, not counting work done by volumetric body forces), plus the heat Q
added to the particle at its boundary during this interval. Equation (4) is one part of the
definition of the quantity we call heat.
Second law of thermodynamics
Q
(5)
T
The increase of a particles entropy (s represents the particles entropy per unit mass)
in a time dt is greater than or equal to the heat added to the particle at its boundary during
this interval divided by the absolute (thermodynamic) temperature, T .4 The equality sign
applies in the limit of a reversible process.
d (s V )
According to the Second Law the temperature in Eq. (5) should be that of the reservoir from which the
heat is supplied to the material particle. In this case the heat comes from the material that bounds the
infinitesimal particle, where the temperature differs infinitesimally from the particles own average
temperature T.
d
dt
(r ,t)dV = 0 ,
(6)
MV (t )
v
where (r ,t) is the materials density field, dV=dxdydz represents a volume element
inside the material volume, and MV(t) under the integral sign signifies integration over
the material volume at the instant t.
d
dt
v v
(r ,t)v (r ,t)dV = F
MV
(t) .
(7)
MV (t )
The usual term conservation equation is a bit of a misnomer, since mass is the only one of these
quantities that is actually conserved.
7
understood that Eq. (7) applies only in inertial (non-accelerating) reference frames under
non-relativistic conditions.
Motion (angular momentum)
v
d
v v
v v
r v dV = TMV (t) = ri Fi
dt MV (t )
i
(8)
This equation is obtained by summing the angular momentum law for a material
particle, Eq. (3), over all the particles that comprise a finite material volume. The law
states that the rate of increase of a material volumes angular momentum, expressed as
the integral over
v the volume of the angular momentum per unit volume, is equal to the
vector sum TMV (t) of all torques exerted by the rest of the universe on the material
volume. This form of the law assumes that the torques exerted between two particles
within the volume are equal and opposite, or zero, which is the case except in rare
circumstances. Note again that Eq. (8) is not a new law, but a corollary of Newtons law
of motion and subject to the same restrictions.
First law of thermodynamics
d
dt
e dV = Q
t
MS
(9)
MV (t )
This law is obtained by summing Eq. (4) over all the particles that comprise the
material volume and noting that the particle-to-particle heat transfer and work terms
cancel for all particles inside the material volume when the sum is taken (what comes
from one goes into the other).
The law states that the rate of increase of a material volume's energy ( et is the total
energy per unit massinternal plus kinetic plus gravitational) is equal to the sum of two
source terms which represent interactions with the rest of the universe at the volumes
boundary. The first source term is the net heat flow rate into the material volume across
its bounding surface
Q MS (t) =
v v
q n dA ,
(10)
MS(t)
where
v
q = kT
(11)
8
is the conductive heat flux vector at a point on the material volumes boundary, k isthe
v
materials thermal conductivity, T is its local thermodynamic temperature, n is the
outward-pointing unit vector at the bounding surface, and dA is an elemental area on the
bounding surface. The symbol MS(t) denotes integration over the closed bounding
surface of the material volume at time t. The second source term in (9) is the rate at
which work is done by the rest of the universe on the material volume at its boundary.
This may be evaluated as
WMS (t) =
v v
v dA
(12)
MS(t )
v
v
were is the vector stress exerted on the boundary by the rest of the universe and v is
v
the materials local velocity at dA. The quantity dA is the force exerted by the rest of
the universe on the surface element dA of the control volume.
Equation (9) thus has the form
d
v v
v v
et dV = q ndA + v dA
dt MV (t )
MS(t )
MS(t )
(13)
d
dt
MV (t )
sdV
v v
q n dA
T
MS(t)
(14)
The rate of increase of a material volumes total entropy is greater than or equal to the
sum of all the local heat inflows at the boundary when each contribution is divided by the
local thermodynamic (absolute) temperature at the point on the material volumes surface
where the transfer takes place.
This law provides a bounding value of the rate of entropy increase, but not the actual
value, and is less useful in dynamics than the other laws. It does, however, have some
important uses in dynamics. One can for example discard from the dynamically possible
solutions (those that satisfy mass conservation and the equation of motion) those that are
unrealizable because they violate the Second Law, and one can predict the entropy
change in limiting cases of negligible dissipation, where the equality sign applies.
d
dt
(r ,t)dV
(15)
MV (t )
r
in which a quantity ( r ,t) that represents something per unit volumemass, momentum,
energy, or entropyis first integrated over a material volume and the result then
differentiated with respect to time. When the material is flowing and deforming, the
volumes boundary moves with it and is not known as a function of time until the
problem is solved. It seems, therefore, that one must know the solution before one can
apply these laws to find the solution. Clearly, we need to find a way of applying the
basic laws to systems of our own choice, that is, to control volumes.
A control volume is an arbitrarily defined volume with a closed bounding surface (the
control surface) that separates the universe into two parts: the part contained within the
control volume, and the rest of the universe. The control surface is a mental construct,
transparent to all material motion, and may be static in the chosen reference frame, or
moving and expanding or contracting in any specified manner. The analyst specifies the
v v
velocity vc (r ,t) at all points of the control surface for all time.
We shall show next how the universal laws for a material volume can be rewritten in
terms of an arbitrarily defined control volume. This opens the way to the application of
the integral laws in engineering analysis.
2.2 Rate of change of a volume integral over a control volume
We begin by considering a time derivative like Eq. (15) for a control volume rather
than a material volume. The time rate of change of the integral of some field quantity
v
(r ,t) over an arbitrarily defined control volume CV(t) is by definition
d
dt
dV = lim
CV( t )
t 0
( r ,t + t)dV ( r ,t)dV
CV (t + t )
CV (t)
(16)
10
The first integral on the right hand side is evaluated at the advanced time over the
advanced volume, and the second is evaluated at time t over the volume at time t (Fig. 3).
v
At any point r we can write for small t
v
v
(r ,t + t ) = (r ,t) + t .
t
(17)
v
d
(r ,t)
CV (t)
dV =
dV + lim CV (t +t )
t 0
dt CV( t )
t
t
CV (t )
(18)
where the integrals on the right are evaluated based on the values of /t and at time t.
In the limit t0, the difference between the two volume integrals in the second term
can be evaluated (see Fig. 3) by means of an integral over the material surface at time t:
v
CV (t + t )
CV (t)
(19)
CS(t )
v
v v
Here vc (r ,t) is the velocity of the control surface element d A, n is the outwardlyv v
directed unit normal vector associated with dA, and vc ntdA is the control volume size
increase in time t due to the fact that the surface element dA has moved in that time
interval. The integral on the right side is taken over the entire (closed) bounding surface
CS(t) of the control volume.
11
Substituting Eq. (19) into Eq. (18), we obtain for an arbitrarily chosen control volume
CV(t),
d
dt
v
v v v
(r ,t )
dV =
dV + (r ,t)vc ndA
t
CV( t )
CV (t)
CS(t )
(20)
d
dt
v
v v v
(r ,t)
dV = t dV + (r ,t)v ndA
MV (t )
MV (t )
MS(t )
(21)
Fig. 4
12
We apply Eq. (20) to our CV and Eq. (21) to the MV that coincides with it at time t ,
and note that because the volumes coincide, the integrals on the right-hand side of Eq.
(21) may be evaluated over either the CV instead of the MV. This yields two alternative
equations for the time derivative of an integral over a material volume, expressed in
terms of a CV that coincides with the material at the time involved:
Form A
Form B
d
dt
dV =
(22)
MV (t )
d
r r r
dV + (v vc ) n dA
dt CV (t )
CS (t)
d
dt
dV =
r r
dV + v ndA
t
CV (t)
CS (t )
(23)
MV( t )
Equation (22) is obtained by subtracting Eq. (20) from Eq. (21). Equation (23) is Eq. (21)
with the integrals referred to the CV instead of the MV, the two being coincident. Recall
r
r
that v is the local material velocity, vc is the local control surface velocity at the surface
r
element dA, and n is the outward-pointing unit normal vector associated with dA..
Both forms A and B are valid for arbitrarily moving and deforming control volumes
(i.e. control volumes that may be expanding, translating, accelerating, or whatever), and
for unsteady as well as steady flows. The two forms express exactly the same thing, but
do the bookkeeping in different ways.
Remember that represents something per unit volume. Both forms express the
material-volume time derivative on the left as a sum of two terms that refer to the control
volume that coincides with the material volume at the instant t.
In form A, the first term on the right is the rate of change of the amount of inside
the control volume at time t (the volume integral is evaluated first, then the time
derivative), and the second term is the net rate of outflow of across the control
volume's boundary.
In form B, the first term on the right is the volume integral of the partial time
derivative of over the control volume at time t (the CS is held fixed at its position at
time t while the integration is performed). The second term accounts for the fact that the
material volumes boundary, which appears in the integral on the left, does not maintain
the shape it has at time t, but envelops more volume (and more of the quantity ) when it
v
expands, every point moving with the local material velocity v . The control surface
velocity does not enter at all in form B.
We shall see that Form A is usually more convenient in unsteady applications than
Form B. This is particularly true in cases where t is singular at some surface inside
the control volume (as it is at a moving flame front inside a solid-propellant rocket, for
example, if is the material density distribution in the rocket), in which case it is
difficult to evaluate the volume integral in Form B. The volume integral in From A, on
the other hand, can be calculated straightforwardly and then differentiated with respect to
time.
13
r r
vn = v n = vcos
(24)
is the outward normal component of the materials absolute velocity at the control
r
r
surface, being the angle between v and the outward-pointing normal unit vector n , and
r r r
vrn = (v vc ) n = v n vcn
(25)
is the outward normal component of the material's velocity relative to the control surface,
vcn being the outward normal component of the control surface's velocity.
3.1 Mass conservation
Setting = in Eqs. (22) and (23), we transform Eq. (6) into two alternative forms
for a CV:
Form A
Form B
d
dt
dA = 0
(26A)
dV + vn dA = 0
t
CV (t )
CS (t )
(26B)
dV + v
CV( t )
rn
CS(t )
14
Equation (26A) states the mass conservation principle as follows: The rate of increase
of the mass contained in the CV, plus the net mass flow rate out through the (generally
moving) CS, equals zero at every instant.
Equation (26B) states the same principle in different but equally correct terms: The
rate of increase of the mass contained in the fixed volume defined by the control surface
at time t, plus the net mass outflow rate through the fixed bounding surface of that
volume, equals zero at all times.
3.2 Linear momentum
v
Putting = v in Eqs. (22) and (23) and substituting into (7), we obtain the following
alternative forms for the equation of motion applied to a CV:
Form A
Form B
d
dt
v
vdV +
CV(t )
v
v
v
v
dA
=
F
CV (t)
rn
(27A)
CS (t)
v
v
( v )
v
dV
+
v
v
dA
=
F
CV (t)
t
n
CV (t )
CS (t )
(27B)
v
Here, FCV (t) is the vector sum of all the forces exerted at time t by the rest of the universe
on the control volume, including volumetric forces and stresses exerted on the control
volumes boundaries. For a continuous distribution of surface and body forces,
v
v
FCV (t) = dA +
CS
G
dV
(28)
CV
Equation (27A) states that the rate at which the linear momentum contained in the CV
increases with time, plus the net flow rate of linear momentum out through the control
surface, is equal at every instant to the force exerted by the rest of the universe on the
material within the control surface.
Equation (27B) states it in different terms: The rate of increase of the momentum
contained in a frozen volume identical to the control surface at time t, plus the net mass
outflow rate through the frozen bounding surface of that volume, is equal at time t to
the net force exerted by the rest of the universe on the material in the control volume.
3.3 Angular momentum
v v
Setting = r v in either (22) or (23) and substituting into (8) yields the angular
momentum theorem for a CV in two alternative forms:
15
d
dt
Form A
Form B
v v
( r v)dV +
CV(t )
v
v v
(
v
)v
dA
=
T
rn
CV (t)
(29A)
CS(t )
v
v v
v v
(
)
v
dV
+
(
v)v
dA
=
T
n
CV (t)
t
CV (t )
CS (t )
(29B)
v
v
Here r is the position vector from an arbitrary origin, TCV (t) is the sum of all the torques
(relative to the chosen origin) that the rest of the universe exerts on the control volume,
including those resulting from both surface forces (pressure and shear) and volumetric
body forces (e.g. gravity). An inertial reference frame is presumed. For a continuous
distribution of surface and body forces,
v
v v
TCV (t) = r dA +
CS
v v
r
GdV .
(30)
CV
v
where is thev vector stress exerted on the boundary element dA by the rest of the
universe, and G is the body force exerted by the rest of the universe on unit mass of
material within the volume.
Equation (29A) states the following: The rate at which the angular momentum inside
the control volume increases with time, plus the net rate at which angular momentum
flows out of the control surface, is equal to the net torque exerted by the rest of the
universe on the matter in the control volume (on the boundary as well as on the mass
within). The reader will be able to interpret (29B) based on the comments been made
above with reference to (26B) and (27B).
3.4 First law of thermodynamics (energy equation)
Setting = et in Eqs. (22) and (23) and substituting into (13), we obtain two forms of
the first law for a CV:
Form A
Form B
d
dt
( et )
v v
v v
dV + et vn dA = q ndA + v dA
t
CV (t )
CS (t)
CS (t )
CS (t )
(31B)
CS( t )
et vrn dA =
v v
(31A)
CV( t )
etdV +
v v
q ndA + v dA
CS (t )
CS (t )
Equation (31A) states that the rate at which the total energy contained in the CV
increases with time, plus the net rate at which total energy flows out of the CS, is equal to
the sum of two terms on the right. The first term is the rate at which heat is conducted
into the CV via the control surface. The second is the rate at which the rest of the
16
universe does work on the material volume whose bounding surface coincides with the
CS at the instant in question. The work done at the control surface,
WCS (t) =
r r
v dA ,
c
(32)
CS(t )
depends on the control surface velocity distribution, which is chosen at will by the
analyst and obviously has no place in a universal law.
3.5 Second law of thermodynamics
Form A
d
dt
v v
q n
sdV + svrn dA T dA
CV( t )
CS (t )
CS (t )
(33A)
Form B
v v
( s)
qn
t dV + CS(t ) svndA CS(t ) T dA
CV (t )
(33B)
Equation (33A) states that the rate of increase of the entropy contained in the CV (s is
the entropy per unit mass), plus the net rate of entropy convection out of the control
surface, never exceeds the integral over the control surface of the normal heat influx
divided by the local absolute temperature.
17
Calculate the time derivative of the volume integral that appears on left-hand side of
your integral equation.
Step 7
From steps 4, 5 and 6, substitute into your integral equations.
18
Step 8
If you set out to solve a practical problem using the control volume theorems, you
must write down enough equations to ensure that their number equals the number of
unknowns in the equations. The four integral laws that we have described are totally
general and rigorous, but these laws contain more unknowns than equations. You will
need to draw also on other physical laws, e.g. gravitational theory and/or electromagnetic
field theory to define the external body force field, and various constitutive equations
(e.g. the thermodynamic equations of state, the form of the stress tensor, etc.).
In most applications you will also make simplifying approximations wherever they
are appropriate. Uniform flow over inlet and exit planes is a typical engineering
approximation; integral relations by themselves provide no information about velocity
distributions. If you have reason to believe that the flow may be approximated as
inviscid, you invoke Bernoullis equation6 and when you have obtained your solution
check that it is consistent with that approximation. If you that density varies little, you
write = constant and later check that the this approximation is justified, based on the
fluids thermal equation of state and the predicted pressure excursion in your problem.
Step 9
Solve for the unknowns
Step 10
Check, by suitable order-of-magnitude estimates, that your solution is consistent with
any approximations that you made.
6 Bernoullis
equation is derived from Newtons law of motion, just like the linear momentum theorem. By
invoking Bernoulli, we are not simply writing down the same equation twice. The linear momentum
equation applies generally, to viscous and inviscid flows. By introducing Bernoullis equation we add the
additional constraint that the flow is inviscid.
19
d
( V ) = 0
dt
( V)
v
d v
v
(r vvV ) = r F
dt
d( etV ) = W + Q
d (s V )
v
v
v
dv
d v
= F or
( v V ) = F
dt
dt
Q
T
Material volumes
Mass conservation
d
dt
d
dt
d
dt
dV = 0
MV (t )
v
v
v
dV
=
F
MV (t)
MV (t )
v
v v
v
dV
=
T
MV (t )
MV (t )
d
dt
MV (t )
d
dt
v v
q n dA
sdV T
MV (t )
MS(t)
e dV = Q
t
MS
20
Control volumes
Reynolds theorem
v v
dV =
dV + v ndA
dt MV (t )
CV (t ) t
CS(t )
Mass conservation
Form A
Form B
d
dt
dV + ( v v ) n dA = 0
CV( t )
CS(t )
v v
t dV + v ndA = 0
CV (t )
CS (t )
Linear momentum
Form A
Form B
d
dt
vdV + v v
rn
v
dA = FCV (t)
CV(t )
CS (t)
v
v
( v )
v
dV
+
v
v
dA
=
F
CV (t)
t
n
CV (t )
CS (t )
Angular momentum
Form A
Form B
d
dt
( r v)dV + (r v )v
CV(t )
rn
v
dA = TCV (t)
CV (t )
v
v v
v v
(
)
v
dV
+
(
v)v
dA
=
T
n
CV (t)
t
CV (t )
CS (t )
First law
Form A
Form B
d
dt
CV( t )
etdV +
CS( t )
et vrn dA =
v v
CS (t )
Form B
CS (t )
( e )
v v
v v
t t dV + et vndA = q ndA + v dA
CV (t )
CS (t)
CS (t )
CS (t )
Second law
Form A
v v
q ndA + v dA
v v
q n
sdV + svrn dA T dA
CV( t )
CS (t )
CS (t )
v v
( s)
qn
t dV + CS(t ) svndA CS(t ) T dA
CV (t )
d
dt
21
Definition of velocities
v
v = local fluid velocity in an inertial reference frame
v
vc = local control surface (CS) velocity in same frame
v v v
vr = v vc =fluid velocity relative to local CS
v
vn = outward normal component of v
v
vrn = outward normal component of vr
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Simple Problem
We shall use a very simple example to illustrate the variety of ways in which a control
volume theorem can be applied to a particular application, depending on the choice of
control volume and which of the two alternative forms of the control volume theorem is
used. The exercise provides a few basic insights into the thought processes that are used
in control volume analysis.
Figure 1 depicts something like a cylindrical syringe, or a grease gun, in which a
solid piston of radius R 1 is pushed at a speed U(t) into a fluid-filled cylinder with the
same internal radius, forcing the fluid out through a tube with internal radius R2 and
length L. The piston, cylinder, and tube are inflexible and made of material with density
s; the fluid has density and can be considered incompressible.
Fig. 1: The system and the control volume (broken red line) for Methods 1 and 2. This control volume is
fixed in the inertial reference frame of the cylinder.
Given the aforementioned quantities, what is the average flow speed V(t) of the
fluid at the exit plane? The governing principle is clearly mass conservation, which can
be written for a control volume in two alternative forms,
Form A
Form B
d
dt
dV + v
dA = 0
(1)
dV + vn dA = 0
t
CV (t )
CS (t )
(2)
CV( t )
rn
CS(t )
r r
where vn = v n = vcos is the outward normal velocity component of the fluid at the
v v v
control surface, and vrn = (v vc ) n is the outward normal component of the fluid
velocity relative to the control surface at that point. The answer is simple:
V(t) =
R12 U(t)
.
2
R2
(3)
Let us look at some of the different ways this result can be obtained.
In all the examples that follow, the reference frame is fixed in the solid cylinder.
Method 1
Control volume: As shown in Fig. 1, with all points of the control surface held fixed in
the chosen reference frame. Note that the top of this CS is placed just below the piston
face, that is, we choose to look at the situation at an instant before the pistons face
reaches the fixed elevation of the top of our control surface. All parts of the CS reside in
the fluid, where the density is .
Equation: We shall apply Form A of mass conservation,
d
dt
dV + v
CV( t )
rn
dA = 0
(1)
CS(t )
Analysis: The first term in (1) is zero because the fluid density is invariant and our
control volume has a fixed volume, so that the mass contained within it is invariant. (We
are considering a time just before the pistons face penetrates the top of the CS!) In the
second term, vn = U (t) at the top of the CS where the piston face pushes the fluid down
into the CV through a stationary CS with area R12 . At the exit plane the area is R22 and
the velocity is vn = V(t ) on the average. Elsewhere on the CS, vn = 0 . Equation (1) thus
gives
VR22 + ( U )R12 = 0 ,
(4)
Method 2
Control volume: Same as in Method 1.
Equation: Form B of the mass conservation equation,
dV + vn dA = 0 .
t
CV (t )
CS (t )
(2)
Analysis: In this case the volume integral is zero because the density is invariant with
time. The integral over the CS is exactly the same as in Method 1, and we end up with (3)
again.
Method 3
Control volume: As shown in Fig. 2, i.e. similar to the one in Fig. 1 except that the top
surface moves downward at the piston speed U, keeping just a hair ahead of the piston
face, that is, the CV height h(t) is such that
dh
= U(t) .
dt
(5)
d
dt
dV + v
CV( t )
CS(t )
rn
dA = 0
(1)
Analysis: In this case the first term in (1) is non-zero because the CVs volume changes
as the piston moves down and h decreases. The outflow term in the surface integral is the
same as before, but there is no inflow into the CV from the top because the fluid moves
down at the same speed as the piston. Equation (1) yields
d
R12 h(t) + R22 L] + V(t)R22 = 0
[
dt
(6)
Fig. 2: Control volume for Methods 3 and 4 (broken red line). The top moves down under the piston.
where the quantity in square brackets in (6) is the volume integral of the density within
the CV. We evaluate the time derivative in (6) using (5), and again obtain (3).
Method 4
Control volume: Same as in Method 3. See Fig. 2.
Equation: Form B of the mass conservation theorem,
dV + vn dA = 0 .
t
CV (t )
CS (t )
(2)
Analysis: The first term is zero because the density does not vary with time. The surface
integral involves the normal fluid velocity, not the normal relative velocity. (The control
surface velocity appears nowhere in (2)!) Equation (2) reads
0 + [U (t)R1 + V (t)R2 ] = 0
2
(7)
Method 5
Control volume: This time we deliberately pick a less thoughtful control volumeone
with a fixed upper bounding surface that cuts through the piston at a fixed elevation ho
(Fig. 3).
Equation: Form A of the mass conservation theorem,
d
dt
dV + v
CV( t )
rn
dA = 0
(1)
CS(t )
Fig. 3: Control volume for Methods 5 and 6 (broken red line). The control volume is fixed.
Analysis:
d
R12 h + sR12 (ho h ) + R22 L] + [ sUR12 + VR22 ] = 0
[
dt
(8)
Here h(t) is the elevation of the piston face, the first term in square brackets is the volume
integral of density, and the second term in square brackets is the surface integral of mass
flux.. Evaluating the time derivative using (5), we find that (8) reduces to (3).
Method 6
Control volume: Same as in Method 5.
Equation used: Form B of the mass conservation theorem,
dV + vn dA = 0 .
t
CV (t )
CS (t )
(2)
Analysis: When we now try to apply Form B, we encounter a difficulty: the partial time
derivative of density is zero everywhere except at the piston face, where it is singular. At
any height z inside the cylinder, the density is equal to the fluid density before the
piston arrives, jumps discontinuously to the solid density s at the instant the piston
arrives, and stays at that value thereafter. If for example the piston velocity is constant,
this can be expressed as
z < ho Ut
z = ho Ut
= s
z > ho Ut
(9)
The singularity makes it difficult to evaluate the volume integral of the time derivative of
the density. It is possible to integrate this particular singularity by representing the partial
time derivative of the density with a suitable Dirac delta function. We will not do that
here, for it requires an introduction to generalized functions (see, however, Problem 3.7
for an analogous problem with a solution provided). A better way to proceed with this
CV choice (if one is fixated on it) is to use Form A of the mass conservation theorem,
that is, Method 5.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A long, at plate of breadth L (L being small compared with the length perpendicular to the sketch) is
hinged at the left side to a at wall, and the gap betweeen the plate and wall is lled with an incompressible
liquid of density . If the plate is at a small angle (t) and is depressed at an angular rate
!(t) = -
d
,
dt
obtain an expression for the average liquid speed u(x, t) in the x-direction at station x and time t.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A long, at plate of breadth L (L being small compared with the length perpendicular to the sketch) is
hinged at the left side to a at wall, and the gap betweeen the plate and wall is lled with an incompressible
liquid of density . If the plate is at a small angle (t) and is depressed at an angular rate
(t) =
d
,
dt
obtain an expression for the average liquid speed u(x, t) in the x-direction at station x and time t.
c 2010, MIT
Copyright
Conservation of Mass
Solution:
L
y
u = (u, v)
CV
x
x=0
Conservation of mass:
time rate of change
of mass in the volume
rate of mass
rate of mass
ow in
ow out
(3.05a)
If the plate moves in time, t, uid in the shaded area, must ow out through the control surface (unit
depth into the page).
time rate of change
of mass in the volume
1
1
x(x ttannn)
= x2 ;
e
2
t
2
From (3.05a)
1
x
.
x2 = 0 uav x uav =
2
2
More formally by conservation of mass:
dV =
t V
(u VCV ) n ds .
dx
= 0,
3
dy =
0
= 0,
VCV
x
dV =
(3.05b)
1
d
1
x(x)
dV = x2
2
dt V
2
2
x
=
1
dy = x
=un
1
x
t
x
0
nn
uav
u dy
e
/ / x/2 =
2
2
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
Conservation of Mass
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Water ows from a large reservoir through a very long pipe under constant head h. When the valve is slowly
closed, the head h remains constant, but the volume ow rate is reduced.
a) Neglecting friction and compressibility of the water, demonstrate that the gage pressure just upstream of
the valve at any instant during the closure period is given approximately by
p = gh
Q2
L dQ
A dt
2A2
(4.25a)
Ae (t)
=
A
t
1
1+
2
gh
1
2
12
(4.25b)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Water ows from a large reservoir through a very long pipe under constant head h. When the valve is slowly
closed, the head h remains constant, but the volume ow rate is reduced.
a) Neglecting friction and compressibility of the water, demonstrate that the gage pressure just upstream of
the valve at any instant during the closure period is given approximately by
Q2
L dQ
p = gh
2
A dt
2A
(4.25a)
Ae (t)
=
A
t
1
12 12
L
2
1+
gh
(4.25b)
c 2010, MIT
Copyright
Inviscid Flows
Solution:
g
h
Valve
Q(t)
a) If we assume that the ow is inviscid, irrotational and incompressible, but not steady, we may apply the
unsteady Bernoulli equation along a streamline between stations 1 and 2, which takes the form
s2
s1
v
1
1
ds + p2 + v22 + gz2 p1 + v12 + gz1 = 0
t
2
2
(4.25c)
We dene the gage pressure p = p2 p1 . Additionally, we recognize that uid speed at station 1 is essentially
zero because h remains constant in time since the reservoir area is much greater than the pipe area A. Finally
we note that the speed of the uid in the horizontal pipe is spatially uniform along its length, L, but changes
in time, so we obtain
v
1
L + p + v22 gh = 0
t
2
(4.25d)
The speed at station 2 is related to the volume ow rate by v2 = Q/A, so when we substitute in this result
into Eq. (4.25d) and rearrange this result, we nd
L dQ
Q2
p = gh
A dt
2A2
(4.25e)
b) Consider now a third station, station 3, which is just downstream of the valve where the pressure is once
again atmospheric, such that p3 = p1 . Our objective is to have a volumetric ow rate Q(t) that decreases
linearly in time to zero in a period , therefore
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
Inviscid Flows
t
1
Q(t) = Q0
!
(4.25f)
where Q0 is the initial steady state flow rate before the valve begins to close. This flow rate, Q0 , is determined
by applying the steady Bernoulli equation from station 1 to station 3, and it is
p
2ghA
Q0 =
(4.25g)
p2 p3 =
(4.25h)
Since p2 p3 is equal to the gage pressure, we may replace it with Eq. (4.25e) and rewrite speeds in terms
of volumetric flow rates to obtain
Q2
L dQ
gh
2
2A
A dt
1
Q2
Q2
=
2
A2e
A2
!
(4.25i)
which simplifies to
gh
L dQ
1 Q2
=
A dt
2 A2e
(4.25j)
Now we may substitute for Q in Eq. (4.25i) with (4.25f) and its derivative and (4.25g) to obtain
gh
L
A
1p
2ghA
2
t
2
2
ghA
1
1
A2e
(4.25k)
This expression may be rearranged to obtain our final result in the desired form
Ae (t)
=
A
t
1
!"
1+
2
gh
! 12 # 12
(4.25l)
Returning now to our earlier assumption that the flow between stations 2 and 3 was steady, we consider
the relative magnitude of the unsteady term in the unsteady Bernoulli equation based on our results from
the pseudo-steady assumption, were we to have included it in Eq. (4.25h). The unsteady Bernoulli equation
between stations 2 and 3 is:
c 2010, MIT
Copyright
Inviscid Flows
s3
p2 p3 =
s2
v
1
1
ds + v32 v22
t
2
2
(4.25m)
We must compare the relative magnitudes of the unsteady term with the difference between the square of
the velocities. If we assume that the characeristic velocity in the unsteady term is v3 = Q(t)/Ae (t), then
s3
s2
!
Q(t)
l
Ae (t)
v
d
ds =
t
dt
(4.25n)
Q(t)
d
= 0 for all t
where l is the length of the nozzle. Since Q(t) (1 t/ ) and Ae (t) (1 t/ ), dt
Ae (t)
and its exclusion from Eq. (4.25h) is clearly justified. If, on the other hand, we were to have taken v2 as the
characteristic velocity in the integral, then
s3
s2
v
d
ds =
t
dt
!
!
!
Q(t)
Q0
2g h
l =
l =
l
A
A
(4.25o)
1
1
(v32 v22 ) =
2
2
"
!2
Q0 (1 t )
A(1 t )[1 +
L 2 12 12
( gh ) ]
Q0 (1 t )
A
!2 #
(4.25p)
"
1
L
2
2
(v3 v2 ) = gh 1 +
2
2
gh
! 12
t
1
!2 #
(4.25q)
1
(v32 v22 )t=0 =
2
2gh
L
(4.25r)
Comparing the magnitudes of Eq. (4.25o) and (4.25r), we conclude that provided L l, the exclusion of
the unsteady term from Eq. (4.25h) is justified.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
u(r,t)
r
Gas
Liquid, density
pg
R(t)
u=
R2 dR
r2 dt
(4.27a)
(b) Show that the rate of growth of the bubble is described by the equation
d2 R 3
R 2 +
2
dt
dR
dt
!2
2(
p g - p1
=
R
(4.27b)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
u(r,t)
r
Gas
pg
Liquid, density
R(t)
u=
R2 dR
r2 dt
(4.27a)
(b) Show that the rate of growth of the bubble is described by the equation
d2 R 3
R 2 +
dt
2
dR
dt
2
pg p
=
(4.27b)
c 2010, MIT
Copyright
Inviscid Flows
Solution:
CV
R(t)
CS1
u(r,t)
CS2
(a) If the control volume is chosen correctly, it is possible to determine u using either Form A or B of the
integral mass conservation equation, however, here we will use Form A. Form A is written
d
dt
=0
(u uCS ) ndA
dV +
CV
(4.27c)
CS
We choose a control volume taking the shape of a hollow sphere whose inner control surface, CS1 , has radius
R(t) and whose outer control surface, CS2 , has arbitrary radius r. Furthermore, the internal surface CS1
is selected to move radially outward at exactly the rate of expansion of the bubble dR/dt. Let us rst
evaluate the volume integral term, by noting that the density is constant and the total volume of our CV is
V = 43 (r3 R(t)3 ). Accordingly,
d
dt
dV =
CV
4
d 3
dR
r R(t)3 = 4R2
3 dt
dt
(4.27d)
Next, we evaluate the surface ux integrals. At CV1 we note that the liquid velocity is exactly equal to the
velocity of the gas-liquid interface, dR/dt which is also the speed at which CV1 is moving, hence there is no
relative velocity between the liquid and CV1 so u uCS = 0, and
(u uCS ) ndA
=0
(4.27e)
CS1
Conversely, at CS2 , r may be arbitrary, but it is xed in time so that uCS2 = 0, and provided R < r, then
the liquid velocity at this r-value is u(r, t), so the surface ux integral at CS2 is
= u(r, t)
er er 4r2 = 4r2 u(r, t)
(u uCS ) ndA
(4.27f)
CS2
4R2
2.25 Advanced Fluid Mechanics
dR
+ 4r2 u(r, t) = 0
dt
2
(4.27g)
c 2010, MIT
Copyright
Inviscid Flows
and hence
u(r, t) =
R(t)
pg
R2 dR
r2 dt
(4.27h)
(b) Neglecting gravity, and provided the flow is inviscid and irrotational, we may apply the unsteady Bernoulli
equation along a streamline with station 1 located just on the liquid side of the gas-liquid interface and station
2 located at a great distance from the bubble where p = p and the liquid velocity is approximately zero,
u2 0. The unsteady Bernoulli equation along this streamline is
Z
s2
s1
du
1
ds + p2 = p1 + u21
dt
2
(4.27i)
2
The pressure p1 differs from the pressure in the bubble pg by the Laplace pressure such that p1 = pg R(t)
,
where is the surface tension of the gas-liquid interface. Substituting our result for u in Eq. (4.27h) into
Eq. (4.27i) and setting ds = dr, s1 = R(t) and s2 = r , we obtain
R(t)
!2
!
R2 dR
2 1 dR
dr + p = pg
+
dt
r2 dt
R
2
d
dt
(4.27j)
R(t)
2R
r2
dR
dt
!2
!
!2
R2 d2 R
2 1 dR
+ 2
dr + p = pg
+
r dt2
R
2
dt
(4.27k)
2R
dR
dt
!2
R2 d2 R
r dt2
!r=r
!2
2 1 dR
+ p = pg
+
R
2
dt
(4.27l)
r=R(t)
which reduces to
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
Inviscid Flows
dR
2
dt
!2
d2 R
+R 2
dt
!
+ p
!2
2 1 dR
= pg
+
R
2
dt
(4.27m)
Finally, we rearrange this result to obtain the governing equation for R in its desired form
d2 R 3
R 2 +
dt
2
dR
dt
!2
+
2
pg p
=
R
(4.27n)
(c) Since our governing equation, Eq. (4.27h) is second order in time, we require two initial conditions, such as
the initial bubble radius R0 and interfacial velocity dR/dtt=0 . Furthermore, we require a relationship between
bubble pressure pg and bubble radius R(t), which could be obtained from the ideal gas law, p = RT and
some reasonable assumption about the nature of the bubble expansion (e.g. adiabatic or isothermal).
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A pipe of area A1 carries a gas at density p and velocity V1 . A converging nozzle is mounted at the end of
the pipe, as shown, to increase the gas velocity as it emerges into the atmosphere. The ow in the nozzle is
incompressible.
(a) Use the momentum theorem to derive the x and y components of force, in excess of those required
to support weight, exerted by the nozzle on its support.
(b) What gage pressure will the presence of the nozzle induce at the pipe (where the area is A1 )?) You
may model the velocity at station (1) as being uniform and assume that the velocity is also uniform
at (2). As regards part (a), there is clearly ambiguity in the problem as being stated, since the x
component of the force on the support will depend on the compression force applied to the gasket, as
well as on the uid ow. In answering (a), consider just the ow-induced force which will be exerted
when the compression force on the gasket is zero.
(c) Apart from the assumption that conditions at (2) have attained uniformity, does the result in (a)
depend in any way on the contour of the nozzle between (1) and (2)?
(d) What is the direction of the force (a) if A2 < A1 ? If A2 > A1 ? Explain.
Note: See also problem 4.4.
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A pipe of area A1 carries a gas at density and velocity V1 . A converging nozzle is mounted at the end of
the pipe, as shown, to increase the gas velocity as it emerges into the atmosphere. The ow in the nozzle is
incompressible.
(a) Use the momentum theorem to derive the x and y components of force, in excess of those required
to support weight, exerted by the nozzle on its support.
(b) What gage pressure will the presence of the nozzle induce at the pipe (where the area is A1 )?) You
may model the velocity at station (1) as being uniform and assume that the velocity is also uniform
at (2). As regards part (a), there is clearly ambiguity in the problem as being stated, since the x
component of the force on the support will depend on the compression force applied to the gasket, as
well as on the uid ow. In answering (a), consider just the ow-induced force which will be exerted
when the compression force on the gasket is zero.
(c) Apart from the assumption that conditions at (2) have attained uniformity, does the result in (a)
depend in any way on the contour of the nozzle between (1) and (2)?
(d) What is the direction of the force (a) if A2 < A1 ? If A2 > A1 ? Explain.
Note: See also problem 4.4.
c 2008, MIT
Copyright
Linear Momentum
Solution:
(a) Lets use mass conservation in the CV shown to get more information about the entrance and exit
velocities, V1 and V2 respectively.
A1
(M assCV ) V1 A1 + V2 A2 = 0, V2 = V1 .
t
A2
(5.02a)
Steady=0
Now, in order to perform a momentum conservation analysis, we still need to know the pressures at
the control volume boundaries. Lets use Bernoulli to gure out the pressure inside the pipe 1 , then
1
1
1
A2
P1 + V12 = P2 + V22 , P1 = Patm + V12 ( 12 1).
!"#$ 2 !"#$
2
2
A2
=Patm
(5.02b)
A2
=V12 12
A2
Now, lets use a CV to nd out the extra force needed to hold the nozzle, considering that the system
is steady,
(V1 )A1 (V1 ) + (V2 )A2 (V2 ) = Fx + A1 (
P1
!"#$
A2
=Patm + 12 V12 ( 12
A2
) A1 ( P2 ),
1)
(5.02c)
=Patm
1 Notice that we are assuming that the ow is inviscid to use Bernoulli, and that is possible to draw a streamline between
(1) and (2), thus the ow is assumed not turbulent too.
c 2008, MIT
Copyright
Linear Momentum
then,
1
A2
A2
1
A2
V12 A1 + V22 A2 = Fx + A1 V12 ( 21 1), V12 21 A2 = Fx + A1 V12 ( 21 + 1),
!"#$
2
A2
A2
2
A2
(5.02d)
A2
=V12 12
A2
nally,
1
A2
A1
1
A1
Fx = A1 V12 ( 12 2
1)2 .
+ 1) = A1 V12 (
A2
A2
A2
2
2
(5.02e)
=V12
(5.02f)
A2
1
A2
2
(c) Yes, if the ow exhibits separation, the use of Bernoulli is not valid and the solution is not valid
too.
(d) From the form of the solution, in both cases the force applied goes in the negative direction. The
force can be divided in two parts, one part due to the pressure dierences, and another from the
dierence in momentum uxes. When there is a reduction in the area, the momentum ux dier
ence dominates. On the other hand when there is a expansion, the pressure at the inlet is smaller
than atmospheric and this force dominates. To show this, lets take the terms from the momentum
conservation analysis before simplication
V12 A1 + V12
A21
1
A2
A2 = Fx + A1 V12 ( 12 1),
2
A2
2
A2
1
A2
A1
A1 V12 (1 12 ) + V12 A1
1
A2
2
A2
"#
"#
P ressure
= Fx .
(5.02g)
(5.02h)
M omentumF lux
D
Problem Solution by MC, Fall 2008
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
1
1
Reaction Zone
V1
2
2
;p0
p0,1
V2
p0,2
The sketch shows a liquid emulsion (a nely-divided mixture of two liquids) of mean density 1 entering a
reaction zone of a constant-area reactor with speed V1 . The components of the emulsion react chemically,
and leave the reaction zone as a liquid at the density 2 . Pitot tubes are installed upstream of the reaction
zone. (Pressure inside a pitot tube is stagnation pressure, p0 = p + 12 V 2 ).
It is agreed to assume that the ow is inviscid, steady and one-dimensional, that the original emulsion is
incompressible, and that the liquid leaving the reaction zone is incompressible.
Calculate the value of (p0,1 p0,2 )/( 12 1 V12 ) in terms of the density ratio 2 /1 .
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
1
1
Reaction Zone
V1
2
2
p0
p0,1
V2
p0,2
The sketch shows a liquid emulsion (a nely-divided mixture of two liquids) of mean density 1 entering a
reaction zone of a constant-area reactor with speed V1 . The components of the emulsion react chemically,
and leave the reaction zone as a liquid at the density 2 . Pitot tubes are installed upstream of the reaction
zone. (Pressure inside a pitot tube is stagnation pressure, p0 = p + 12 V 2 ).
It is agreed to assume that the ow is inviscid, steady and one-dimensional, that the original emulsion is
incompressible, and that the liquid leaving the reaction zone is incompressible.
Calculate the value of (p0,1 p0,2 )/( 12 1 V12 ) in terms of the density ratio 2 /1 .
c 2010, MIT
Copyright
Linear Momentum
Solution:
Reaction Zone
y
1
V1
x
CV
CS1
V2
CS2
Taking the reaction zone as our control volume, and noting that the cross-sectional area of the reactor is
constant, we nd from our integral mass conservation relation that
1
V2
=
V1
2
(5.09a)
One might be tempted to apply the Bernoulli equation across the streamline, however, this approach would
be invalid since the density changes in this region and the ow is well mixed so that it may not be possible
to dene a streamline here. Instead, we consider an integral momentum conservation relation to determine
the pressure drop across the reaction zone. Here we consider Form A:
d
dt
=
u(u uCS ) ndA
ud +
CV
(5.09b)
CS
Since the ow is steady in time, the rst term in Eq. (5.09b) is zero and we must balance the momentum
ux and forces acting on the control surfaces. Furthermore, our control volume is not moving or changing
shape in time so uCS = 0. The momentum ux at CS1 is
= 1 V12 Aex
u(u uCS ) ndA
(5.09c)
= 2 V22 A
ex
(5.09d)
CS1
CS2
The forces acting on the control surfaces simply result from the pressure acting upstream and downstream
of the reaction zone so that
F = (p1 p2 )Aex
2.25 Advanced Fluid Mechanics
(5.09e)
c 2010, MIT
Copyright
Linear Momentum
1 V12 + 2 V22 = p1 p2
(5.09f)
Outside the reaction zone, the ow is well-dened such that the Bernoulli equation may be applied along a
streamline, but only between two points that both lie either upstream of the reaction zone or downstream
of it. The relationship between static and dynamic pressure is p0 = p + 12 V 2 , so the relationships between
p1 and p0,1 and between p2 and p0,2 are
1
p0,1 = p1 + 1 V12
2
&
1
p0,2 = p2 + 2 V22
2
(5.09g)
(5.09h)
or alternatively
p0,1 p0,2 =
1
2 V22 1 V12
2
(5.09i)
p0,1 p0,2
2
=
1
2
1
2 1 V1
V2
V1
(5.09j)
(5.09k)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
PB = PA + p
PA
(1)
area A2
(2)
(a)
(b)
jet: Vj , Aj
pump
The device connected between compartments A and B is a simplied version of a jet pump. A jet pump, or
ejector, is a simplied device which uses a small, very high-speed jet with relatively low volume ow rate to
move uid at much larger volume ow rates against a pressure dierential p (see the gure).
The pump in the gure consists of a contoured inlet section leading to a pipe segment of constant area A2 .
A small, high velocity jet of speed Vj and area Aj injects uid, drawn from compartment A, at the entrance
plane (1) of the pipe segment. Between (1) and (2), the jet (the primary stream) and the secondary uid
ow which is drawn in from compartment A via the contoured inlet section mix in a viscous, turbulent
fashion and eventually, at station (2), emerge as an essentially uniform-velocity stream.
We shall assume that the ows are incompressible, that the ow from compartment A to station (1) is
inviscid, and that, although viscous forces dominate the mixing process between (1) and (2), the shear force
exerted on the walls between those stations is small compared with A2 . The pump operates in steady
state.
Neglect gravitational eects.
(a) Derive an expression for p as a function of the total volume ow rate Q from compartment A to
compartment B. The given quantities are Aj , A2 , , and Vj . You may assume Aj A2 to simplify
your expression.
(b) Sketch the relationship p vs. Q (the pump curve) for positive p and Q. Indicate the value of
Q when p = 0 (the short-circuit volume ow rate). Show that for Aj A2 , the latter is large
compared with the volume ow rate Vj Aj of the jet.
(c) Sketch the pressure distribution along the line ab for the case when p = 0 and for a case when
p > 0.
(d) Is your formulation in (a) valid when Q = 0, i.e. when the total ow rate for A to B is zero? Explain.
What is the minimum value for Q which your formulation is valid?
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
PB = PA + p
PA
(1)
area A2
(2)
(a)
(b)
jet: Vj , Aj
pump
The device connected between compartments A and B is a simplied version of a jet pump. A jet pump, or
ejector, is a simplied device which uses a small, very high-speed jet with relatively low volume ow rate to
move uid at much larger volume ow rates against a pressure dierential p (see the gure).
The pump in the gure consists of a contoured inlet section leading to a pipe segment of constant area A2 .
A small, high velocity jet of speed Vj and area Aj injects uid, drawn from compartment A, at the entrance
plane (1) of the pipe segment. Between (1) and (2), the jet (the primary stream) and the secondary uid
ow which is drawn in from compartment A via the contoured inlet section mix in a viscous, turbulent
fashion and eventually, at station (2), emerge as an essentially uniform-velocity stream.
We shall assume that the ows are incompressible, that the ow from compartment A to station (1) is
inviscid, and that, although viscous forces dominate the mixing process between (1) and (2), the shear force
exerted on the walls between those stations is small compared with A2 . The pump operates in steady
state.
Neglect gravitational eects.
(a) Derive an expression for p as a function of the total volume ow rate Q from compartment A to
compartment B. The given quantities are Aj , A2 , , and Vj . You may assume Aj A2 to simplify
your expression.
(b) Sketch the relationship p vs. Q (the pump curve) for positive p and Q. Indicate the value of
Q when p = 0 (the short-circuit volume ow rate). Show that for Aj A2 , the latter is large
compared with the volume ow rate Vj Aj of the jet.
(c) Sketch the pressure distribution along the line ab for the case when p = 0 and for a case when
p > 0.
(d) Is your formulation in (a) valid when Q = 0, i.e. when the total ow rate for A to B is zero? Explain.
What is the minimum value for Q which your formulation is valid?
c 2010, MIT
Copyright
Linear Momentum
Solution:
(1)
jet
inow
outow
(2)
vn
Area
Pressure
ex
ex
ex
vj
ex
v1
ex
v2
ex
vj
v1
v2
Aj
(A2 Aj )
A2
P1
P1
PB
CV
0
0
VV
visc n
dA +
g
VdV + Fext
V
CV
V
0
0
v dV +
v(v vc ) n
dA =
CS
n dA +
p
CS
CS
Substituting the values from the table into the above equation gives
vj
ex (vj )Aj + v1
ex (v1 )(A2 Aj ) + (v2
ex )v2 A2
jet
outow
inow
= P1 (
ex )Aj P1 (
ex )(A2 Aj ) PB (
ex )(A2 )
To solve for P1 in terms of PA apply Bernoullis from compartment A to section (1):
1
1 V
PA + v
VA 2 = P1 + v1 2
2
2
V
1
P1 = PA v1 2
2
1
v2 2 A2 vj 2 Aj v1 2 (A2 Aj ) = (PA PB )A2 v1 2 A2
2
A2
Aj
2
v2 2 A2 vj 2 Aj v1 2
(5.10a)
CV
dV +
t
(v vc ) n
dA = 0
CS
vj Aj v1 (A2 Aj ) + v2 A2 = 0
v1 =
v2 A2 vj Aj
A2 Aj
Q vj Aj
A2
Q2
+ vj 2 Aj +
Aj
A2
A2 A j
2
Q2
Aj
(Q vj Aj )2
+
p = 2 + vj 2
A2
A2
2A2 2
pA2 =
Aj2
Aj
Qvj Aj
1 Q2
p = 2 + vj 2
+ vj 2
2
2 A2
A2
A2
2A2 2
Note the last term is negligible since Aj 2 /A2 2 Aj /A2 .
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
Linear Momentum
v
A
)
(Q
Q0 2
j
0
j j
+
0 = 2 + vj 2
p0
A2
A2
2A2 2
Q0 =
2vj 2 Aj A2
Aj
A2
Q0
Q
D
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
p1
p2
a
stre
secondary flow
nes
mli
entrainment
2R
Two large compartments are separated by a wall in which there is a small nozzle, or hole, of radius R. The
pressure p1 in the left-hand compartment far from the nozzle is greater than the pressure p2 in the right-hand
compartment, and a steady volume ow Q takes place from (1) to (2).
The ow through the nozzle is an incompressible, high Reynolds number ow, typical of the ones rather
loosely termed inviscid. However, as in all such ows, viscous forces are responsible for the phenomenon
of ow separation which gives rise to a profound dierence between the inow and outow regions of the
nozzle ow eld.
In compartment (1), on the inow side, the ow is directed approximately radially inward toward the nozzle
entrance until one gets close to the nozzle, and is essentially inviscid.
In compartment (2), however, the ow separates from the boundaries and emerges from the nozzle as a
horizontal jet.
Inside compartment (2) viscous forces slow the jet and also cause it to become turbulent and to mix with the
surrounding uid, some of which is dragged along with the jet as the latter penetrates into the compartment
and gradually slows down. The process whereby the jet drags some of the ambient uid along with it is called
entrainment. and gives rise in compartment (2) to a secondary bulk ow which is directed approximately
radially inward toward the jet as sketched. The velocities associated with this secondary ow are relatively
small, however, and the pressure in compartment (2) can be modeled as being approximately uniform.
(a) Consider a disc-shaped portion of the wall extending a radial distance r from the nozzle centerline.
Using a control volume whose left side is a hemisphere of radius r, show that as r/R , the
x-component of external force required to hold this portion of the wall in place is given by
F = (p1 p2 )r2 +
2.25 Advanced Fluid Mechanics
Q2
R2
c 2010, MIT
Copyright
Linear Momentum
(b) Consider the integrals of (i) mass ux and (ii) x-direction momentum ux across a plane at station x
to the right of the nozzle exit. Do these integrals grow, decrease, or remain constant as x increases?
How do they compare with their values at the nozzle exit plane?
Gravity is to be neglected in this problem.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
p1
p2
lines
m
strea
secondary
flow
entrainment
2R
Two large compartments are separated by a wall in which there is a small nozzle, or hole, of radius R. The
pressure p1 in the left-hand compartment far from the nozzle is greater than the pressure p2 in the right-hand
compartment, and a steady volume ow Q takes place from (1) to (2).
The ow through the nozzle is an incompressible, high Reynolds number ow, typical of the ones rather
loosely termed inviscid. However, as in all such ows, viscous forces are responsible for the phenomenon
of ow separation which gives rise to a profound dierence between the inow and outow regions of the
nozzle ow eld.
In compartment (1), on the inow side, the ow is directed approximately radially inward toward the nozzle
entrance until one gets close to the nozzle, and is essentially inviscid.
In compartment (2), however, the ow separates from the boundaries and emerges from the nozzle as a
horizontal jet.
Inside compartment (2) viscous forces slow the jet and also cause it to become turbulent and to mix with the
surrounding uid, some of which is dragged along with the jet as the latter penetrates into the compartment
and gradually slows down. The process whereby the jet drags some of the ambient uid along with it is called
entrainment. and gives rise in compartment (2) to a secondary bulk ow which is directed approximately
radially inward toward the jet as sketched. The velocities associated with this secondary ow are relatively
small, however, and the pressure in compartment (2) can be modeled as being approximately uniform.
(a) Consider a disc-shaped portion of the wall extending a radial distance r from the nozzle centerline.
Using a control volume whose left side is a hemisphere of radius r, show that as r/R , the
x-component of external force required to hold this portion of the wall in place is given by
F = (p1 p2 )r2 +
Q2
R2
c 2010, MIT
Copyright
Linear Momentum
(b) Consider the integrals of (i) mass ux and (ii) x-direction momentum ux across a plane at station x
to the right of the nozzle exit. Do these integrals grow, decrease, or remain constant as x increases?
How do they compare with their values at the nozzle exit plane?
Gravity is to be neglected in this problem.
c 2010, MIT
Copyright
Linear Momentum
Solution:
hemispherical
volume
p1
1
1
p1 + v12 = p2 + v22
2
2
v2
2(p1 p2 )
=
p2
2R
v1
Fx
r sin
(5.13a)
rd
Fx
As the uid moves from First we make a table of the relevant parameters
(1)
(2)
v2
jet
wall
vc
(v
ex ) = vx
Pressure
er
ex
ex
vr
er
v2
ex
0
0
0
0
vr cos
v2
0
p1
p2
p2
%
d
%
v%
dV +
v(v %
v%
dA
ex
F
ex =
c) n
dt%%
CS
% CV
p1 r2 p2 r2 + Fx =
(1)
vx
(2)
dA
/2
cos sin d + v2 2 R2
0
= vr 2 r2 + v2 2 R2
To nd vr , apply mass conservation
d
dt
dA = 0
(v %
v%
c) n
dV +
CV
CS
/2
vr =
v2 R2
2r2
c 2010, MIT
Copyright
Linear Momentum
Fx = (p1 p2 )r
v2 R2
2r2
r2 + v2 2 R2
= (p1 p2 )r2 + v2 2 R2 1
R2
4r2
(5.13b)
As r/R
Fx = (p1 p2 )r2 + v2 2 R2 = (p1 p2 )r2 +
Q2
R2
If we substitute our previous relation for v2 [Eq. (5.13a)] into Eq. (5.13b)
Fx = (p1 p2 )r2 1 + 2
R
r
1
2
R
r
Note that we have applied conservation of momentum on the uid, and as a result, this is the force of
the wall on the uid. The force of the uid on the wall is equal and opposite Fx .
p2
secondary ow
2R
vx
D
v2
x
As x increases, the secondary ow into this control volume increases and thus, the mass ux through
the control volume increases. However, since the x-component of the the secondary ow velocity is
small (if we choose the diameter of our control volume D to be large compared to diameter of the ow),
the x-momentum ux is relatively constant with increasing x. Although the x-momentum is constant,
it does diuse radially and we must take a control volume with larger D as we increase x.
Far away from the wall (x ) we expect the uid velocity to go to zero. If we applied Bernoullis
equation from (1) a point far to the left of the wall to (2) a point far to the right of the wall, we would
expect the pressures to be equal since the velocities are roughly zero.
p
p1
p1
1
2
2 v2
p2
x
However, we cannot apply Bernoulli between these two points, because viscosity is important far from
the wall and is the reason that uid is entrained. Far to the right of the wall, the pressure does not
increase to p1 and energy is burned by viscous dissipation. The is an example of how uid will ow
smoothly go down a pressure gradient (favorable pressure gradient), but not up a pressure gradient
(adverse pressure gradient).
D
Problem Solution by Tony Yu, Fall 2006
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A at plate is hinged at one side to the oor, as shown, and held at a small angle e0 (e0 1) relative to
the oor. The entire system is submerged in a liquid of density . At t = 0, a vertical force is applied and
adjusted continually so that it produces a constant rate of decrease of the plate angle e.
de
= = Const,
dt
(5.18a)
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
V (t)
g
M (t)
After its second booster has been red, a space vehicle nds itself outside the earths atmosphere, moving
vertically upward at a speed V0 against gravity g. Its total mass at that point is M0 . At t = 0, the vehicles
third stage is turned on and the rocket burns propellant at a mass rate mr kg/s, ejecting gas from the exit
plane (area Ae ) at speed Ve relative to the rocket.
Show that if the gravitational acceleration remains essentially constant at the vehicle during the rocket ring,
the velocity V (t) of the vehicle after time t will be given by
V (t) V0 = Ve ln
M0
g[M0 M (t)]
m
r
M (t)
where M (t) is the mass of the system at time t. Assume that although the pressure of the gas at the rocket
exit plane is Pe (the rocket exhaust is supersonic, and hence the pressure at the exit is not balanced with
the zero pressure of space), the eect of the nite exit plane pressure on the thrust is negligible.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
V (t)
g
M (t)
After its second booster has been red, a space vehicle nds itself outside the earths atmosphere, moving
vertically upward at a speed V0 against gravity g. Its total mass at that point is M0 . At t = 0, the vehicles
third stage is turned on and the rocket burns propellant at a mass rate mr kg/s, ejecting gas from the exit
plane (area Ae ) at speed Ve relative to the rocket.
Show that if the gravitational acceleration remains essentially constant at the vehicle during the rocket ring,
the velocity V (t) of the vehicle after time t will be given by
V (t) V0 = Ve ln
g[M0 M (t)]
M0
M (t)
m
r
where M (t) is the mass of the system at time t. Assume that although the pressure of the gas at the rocket
exit plane is Pe (the rocket exhaust is supersonic, and hence the pressure at the exit is not balanced with
the zero pressure of space), the eect of the nite exit plane pressure on the thrust is negligible.
c 2010, MIT
Copyright
Linear Momentum
Solution:
V (t)
Given:
V (t = 0) = V0
M (t = 0) = M0
Ae (area over which gas exits)
Ve (relative velocity of gas leaving booster)
m
r (rate of mass leaving)
ez
M (t)
CV
Unknown: V (t)
CS, Ae
Since the mass ow rate of gas is constant, the mass of the rocket can be expressed as:
M (t) = M0 m
rt
(5.20a)
By Mass Conservation,
d
dt
(v vc ) n
dA = 0
dV +
CV
CS
d
M + Ve Ae = 0
dt
Combine the above equation with Eq. (5.20a) to give
dM
r
= Ve Ae = m
dt
(5.20b)
Ve 2 Ae = M (t)g M (t)v(t)
"
m
r Ve
dv
m
r
=
Ve g
dt
M
t
m
r
dV =
Ve g dt
M
0
where t =
M0 M (t)
m
r
V (t)
V0
(5.20c)
M (t)
M0
n
m
r Ve dM
n
g
M n
m
n
r
t
0
M0
gt
M (t)
M0 M (t)
M0
g
= Ve ln
M (t)
m
r
dt = Ve ln
c 2010, MIT
Copyright
Linear Momentum
V (t) V0 = Ve ln
M0
M0 M (t)
g
M (t)
m
r
D
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
The sketch shows a lawn sprinkler with two horizontal arms of radial length R, at the termination of which are
nozzles (exit Area A2 ) pointing in a direction which is at an angle relative to the tangent of a circumferential
line, as shown. The sprinkler is free to rotate, but the bearing on which it is mounted exerts s torque k
in the direction opposing the rotation, being the angular rate of rotation. A constant volume ow rate Q
passes through the sprinkler, the ow being incompressible at density .
(a) Find an expression for the steady-state angular velocity of sprinkler in terms of the given quan
tities R, A2 , , Q, , and k.
(b) In the steady state, what is the velocity vector of the uid emerging from the nozzles, as seen by
an observer in the non-rotating reference frame? What is the uid velocity at the nozzle vent if the
bearing is frictionless (k = 0)?
(c) If the pipe area at station 1 near the bearing is A1 , and the ow from that point to the nozzles is
inviscid, what gage pressure is required at station 1 to sustain the ow rate in this steady state?
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
The sketch shows a lawn sprinkler with two horizontal arms of radial length R, at the termination of which are
nozzles (exit Area A2 ) pointing in a direction which is at an angle relative to the tangent of a circumferential
line, as shown. The sprinkler is free to rotate, but the bearing on which it is mounted exerts a torque k
in the direction opposing the rotation, being the angular rate of rotation. A constant volume ow rate Q
passes through the sprinkler, the ow being incompressible at density .
(a) Find an expression for the steady-state angular velocity of sprinkler in terms of the given quan
tities R, A2 , , Q, , and k.
(b) In the steady state, what is the velocity vector of the uid emerging from the nozzles, as seen by
an observer in the non-rotating reference frame? What is the uid velocity at the nozzle vent if the
bearing is frictionless (k = 0)?
(c) If the pipe area at station 1 near the bearing is A1 , and the ow from that point to the nozzles is
inviscid, what gage pressure is required at station 1 to sustain the ow rate in this steady state?
c 2010, MIT
Copyright
Angular Momentum
Solution:
(a)
Approaching the problem from the laboratory reference frame and taking our CV as uid plus sprinkler,
we have
t
\
CV
r V dV ol +
dA = k.
r V (V V CS ) n
(5.29a)
CS
steady,=0
The angular momentum theorem is applied here in a direction opposite to . At the exit of each
Q
horizontal arm, we have a tangential velocity V = 2A
cos R, then
2
2
Q
Q
cos R R
= k,
2
2A2
(5.29b)
Q cos
.
k
2A2 (R + QR
)
(5.29c)
then
=
Approaching from a rotating reference frame, and taking the same CV, we have
t
\
But now, Vr =
Q
2A2
r V dV
dA = k + TF ict ,
r V (V V CS ) n
VC
(5.29d)
CS
steady,=0
Q
Q
cos R
= k +
2
2A2
Q
2.
2rvr dV.
(5.29e)
c 2010, MIT
Copyright
Angular Momentum
Q2 R cos
= k + 2Q
2A2
rdr = k + QR2 .
(5.29f)
Q cos
.
k
2A2 (R + QR
)
(5.29g)
(b) We have the velocity vector emerging from the sprinkler as:
Q
Q
Ve =
sin er + R
cos e .
2A2
2A2
(5.29h)
If the bearing is frictionless, the angular momentum equation (5.29b) would give R
Hence the velocity is
Q
V =
sin er .
2A2
Q
2A2
cos = 0.
(5.29i)
(c) We have to apply Bernoulli from station (1) to nozzle exit. But for that, we should move to noninertial rotating reference frame, since only then can we identify a stationary point in space on the
nozzle exit! Applying Bernoulli from station 1 to nozzle exit in the rotating frame, we have 1
1 2
1
1
pa + Ve,rel
2 R2 = p1 + V12 ,
(5.29j)
2
2
2
2
2
= AQ2 , and the additional 12 2 R2 term
where in the rotating reference frame, we simply have Ve,rel
arises from the pressure rise associated with the centrifugal acceleration of the fluid. (If you find this
term puzzling, think about the result in the case when the sprinkler is rotating, but the inlet at station 1
2
2
has been blocked so that there is no radial flow in the pipe. It results from p
r = V /r = r .)
Hence, gauge pressure p1g is
p1g =
Q
2A2
2
2 R2
Q
A1
2
.
(5.29k)
Note that if the area A1 is quite less than the area at the T-juntion, then it will almost be like a
jet problem where there is a significant internal viscous dissipation. In that case, we cannot apply
Bernoulli across the T-junction. We are not given this area ratio in this problem and hence nothing
can be said here.
In case the two areas are comparable, our result in part (c) is valid.
that we have neglected the radius of the inlet tube, and in the moving reference frame, the water velocity at the inlet
is practically the same as in the non rotating reference frame.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
V0
v
The steady sink ow in the sketch is set up by injecting water tangentially through a narrow channel near
the periphery and letting it drain through a hole at the center. The vessel has a radius R. At the point of
injection, the water has a velocity V and depth h0 ; the width of the injection channel, b, is small compared
with R.
In what follows, we consider the region of the ow act too close to the drain, and assume that everywhere
in the region (i) the ow is essentially incompressible and inviscid, (ii) the radial velocity component |vr | is
small compared with the circumferential velocity component v , and at the periphery.
(a) Show, by applying the angular momentum theorem to a control volume comprising the water between
r = r and r = R, that
v = V R/r
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
V0
v
The steady sink ow in the sketch is set up by injecting water tangentially through a narrow channel near
the periphery and letting it drain through a hole at the center. The vessel has a radius R. At the point of
injection, the water has a velocity V and depth h0 ; the width of the injection channel, b, is small compared
with R.
In what follows, we consider the region of the ow act too close to the drain, and assume that everywhere
in the region (i) the ow is essentially incompressible and inviscid, (ii) the radial velocity component |vr | is
small compared with the circumferential velocity component v , and at the periphery.
(a) Show, by applying the angular momentum theorem to a control volume comprising the water between
r = r and r = R, that
v = V R/r
c 2010, MIT
Copyright
Angular Momentum
Solution:
Assume: incompressible, inviscid ow, vr v
V
1
CS
vc
(v vc ) n
1
2
V
e
e vr
er
v
er
0
0
V
vr
r vr
R
T=
d
dA
[r v] dV + [r v](v vc ) n
dt
CV
CS
"
.,
(5.32a)
steady
CS(1) :
er
rv = R
0
CS(2) :
rv =
e
0
V
er
r
vr
ez
0 = RV
ez
0
e
0
v
ez
0 = rv
ez
0
V dA +rv
CS,(1)
"
.,
Qin
vr dA = 0
CS,(2)
"
.,
Qout
RV = rv
v =
RV
(5.32b)
Substitute in V = rv /R:
v bh0 jr
= vr 2jrh(r)
R
vr v
Finally, if
b
R
b
R
1,
vr v
(5.32c)
* Recall 4.21 and how much HARDER it was to solve using Bernoulli !!
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright
Angular Momentum
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
At t = 0, a circular tank of radius R contains water at rest, with a depth h. Between 0 < t < , a water
hose is sprayed onto the surface of the water in the tank at a volume ow rate Q and an exit velocity Vj .
The jet impacts tangentially on the water at a radius Rj , with an angle relative t the horizontal.
After the time , the hose is turned o. Eventually, because of friction within the water, all the water in
tank will end up rotating like a solid body.
Derive an expression for the nal angular rate of rotation of the water, assuming shear forces between the
water and the walls of the tank are negligible.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
At t = 0, a circular tank of radius R contains water at rest, with a depth h. Between 0 < t < , a water
hose is sprayed onto the surface of the water in the tank at a volume ow rate Q and an exit velocity Vj .
The jet impacts tangentially on the water at a radius Rj , with an angle relative t the horizontal.
After the time , the hose is turned o. Eventually, because of friction within the water, all the water in
tank will end up rotating like a solid body.
Derive an expression for the nal angular rate of rotation of the water, assuming shear forces between the
water and the walls of the tank are negligible.
c 2010, MIT
Copyright
Angular Momentum
Solution:
dA = 0,
(V V CS ) n
dV ol +
CV
(5.33a)
CS
then
R2
dh
Q = 0,
dt
(5.33b)
Q
.
R2
(5.33c)
(5.33d)
dA dt = 0,
r V (V V CS ) n
(5.33e)
CS
r V dV ol +
dA = 0,
r V (V V CS ) n
r V dV ol +
CV
CV
CS
R
(5.33f)
where hnew is the increment in the tank water level due to the added hose water.
Q
Also, note that writing dV = 2rhdr is only an
From (5.33c), we have, hnew = R
2 for any t < .
approximation, since the free surface is not straight but rather curved! (recall the calculation of this surface
equation during the uid statics lecture).
As the water hose is sprayed, its radial component adds angular momentum in the CV (nally stabilizing
as rigid body rotation). If rigid is the time, after the hose stops, for the uid to stabilize into rigid body
rotation, we can integrate the right hand side of (5.33e) from 0 to ( + rigid ), and the left hand side form 0
to R 1 , thus
R4
h+
2
+rigid ,
hnew
Q
= R
2
+rigid
(5.33g)
0,0
, t>
Now, noting that Q = 0 for < t < + rigid , and = at t = + rigid , we have
1 Note
that since the uid has stabilized, is nor a function of the radius anymore.
c 2010, MIT
Copyright
Angular Momentum
R4
Q
h+
= Rj Vj cos Q,
2
R2
2Rj Vj cos Q
.
Q
R4 h + R
2
(5.33h)
(5.33i)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25
Shapiro 7.14
V elocity [ms1]
Unsteady
Laminar Flow
T ime [s]
Figure 1: Laminar Flow (Steady vs. Unsteady)
V elocity [ms1]
Unsteady
Turbulent Flow
V
V = V +V
V
V = V +V
3
2
3
T ime [s]
(1)
Any variable can be resolved into a mean value V plus a uctuating value V i where by denition:
V
1
T
t0 +T
V dt
(2)
t0
MIT 2012,B.K.
2.25
Shapiro 7.14
where T is large compared to the relevant period of the uctuations. The mean values V can
also change slowly with time, which is referred as unsteady turbulent ow. What in Shapiro
7.14 is mentioned as steady and fully developed in the mean describes the fact that:
Vz
Vz
=0 ,
=0
t
z
(3)
The important fact is that when you try to deduce the governing equations of motion you
should carry these uctuation terms, V i , all along. Finally you will end up getting an equation
very similar to Navier-Stokes which is called the Reynolds equation (there is still an ongoing
debate on wether the idea of time-averaging in the turbulent ow is the best approach but
those discussions are way beyond the scope of this course).
The Reynolds equation will have the following form:
DV
+
vi vi =
Dt
xj i j
P +
V + g
(4)
For the simple pipe ow, in the turbulent case (unlike the laminar case), even though the ow
is steady and fully developed in mean i.e. DV
Dt = 0 the left hand side will not turn into zero.
The equation (when simplied) turns into the following:
vii vji =
xj
P +
(5)
which shows that both and are important parameters in the problem and we can not get
rid of density as we did in the laminar case.
MIT 2012,B.K.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A metal ball falls at steady speed in a large tank containing a viscous liquid. The ball falls so slowly that
it is known that the inertia forces may be ignored in the equation of motion compared with the viscous forces.
(a) Perform a dimensional analysis of this problem, with the aim of relating the speed of fall V , to the
diameter of the ball D, the mass density of the ball b , the mass density of the liquid l , and any other
variables which play a role. Note that the eective weight of the ball is proportional to (b l )g.
(b) Suppose that an iron ball (sp. gr.=7.9, D=0.3 cm) falls through a certain viscous liquid (sp. gr. =
1.5) at a certain steady-state speed. What would be the diameter of an aluminum ball (sp. gr. = 2.7)
which would fall through the same liquid at the same speed assuming inertial forces are negligible in
both ows?
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A metal ball falls at steady speed in a large tank containing a viscous liquid. The ball falls so slowly that
it is known that the inertia forces may be ignored in the equation of motion compared with the viscous forces.
(a) Perform a dimensional analysis of this problem, with the aim of relating the speed of fall V , to the
diameter of the ball D, the mass density of the ball b , the mass density of the liquid l , and any other
variables which play a role. Note that the eective weight of the ball is proportional to (b l )g.
(b) Suppose that an iron ball (sp. gr.=7.9, D=0.3 cm) falls through a certain viscous liquid (sp. gr. =
1.5) at a certain steady-state speed. What would be the diameter of an aluminum ball (sp. gr. = 2.7)
which would fall through the same liquid at the same speed assuming inertial forces are negligible in
both ows?
c 2010, MIT
Copyright
Dimensional Analysis
Solution:
= FD
(7.03a)
Wef f = (b l )g
3
2
l
[ML
[ML
V
T
[LT
[L ]
FD
[ML1 T2 ]
Thus we have
n = 5 variables
k = 3 primary variables
j = 2 dimensionless group
For our primary variables, we choose (1) a uid property: l , (2) a ow parameter: V , and (3) a
geometric parameter: D. Therefore, the rst dimensionless group is
= f1 (l , V, D)
or
1 = K1 al V b Dc
K1
=
Re
l V D
(7.03b)
FD
l V 2 D 2
(7.03c)
FD
1
2
2 V A
c 2010, MIT
Copyright
Dimensional Analysis
Di 3
4
3
2
2 D2
i
(a l )g
=
4
3
Da 3
2
V 2 Da2
(7.03d)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A strong explosion (like an atomic bomb) causes a spherically symmetric shock wave to move through the
air radially out from the origin. As the shock sweeps by, it causes a sudden rise in the pressure and sets the
initially static air into radially outward motion.
It can be argued from strong shock wave theory that if the undisturbed atmosphere is homogeneous at a
density a , the velocity vs of the shock, as well as the pressure ps and the wind speed just behind the shock
wave, should depend only on the density a , the total distance rs of the shock wave from the origin, and the
total energy E released by the explosion.
(a) Show that:
E
vs = const.
a
1
2
32
rs
ps = const. E rs3
(7.09a)
(7.09b)
(b) Obtain an expression for the shocks radial position as a function of time (the expression may involve
one unknown dimensionless constant). Show how the strengths of two dierent bomb explosions, as
measured by their energy releases, can be compared based on lm information about their shock wave
positions as a function of time.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A strong explosion (like an atomic bomb) causes a spherically symmetric shock wave to move through the
air radially out from the origin. As the shock sweeps by, it causes a sudden rise in the pressure and sets the
initially static air into radially outward motion.
It can be argued from strong shock wave theory that if the undisturbed atmosphere is homogeneous at a
density a , the velocity vs of the shock, as well as the pressure ps and the wind speed just behind the shock
wave, should depend only on the density a , the total distance rs of the shock wave from the origin, and the
total energy E released by the explosion.
1
2
32
rs
ps = const. E rs3
(7.09a)
(7.09b)
(b) Obtain an expression for the shocks radial position as a function of time (the expression may involve
one unknown dimensionless constant). Show how the strengths of two dierent bomb explosions, as
measured by their energy releases, can be compared based on lm information about their shock wave
positions as a function of time.
c 2010, MIT
Copyright
Dimensional Analysis
Solution:
(7.09c)
a
b
Equivalently, this result can be written in terms of LT 1 = M L2 T 1 M L3 Lc . Solving for
each exponent, we find that a = 0.5, b = 0.5 and c = 1.5, and accordingly,
E
vs = const.
a
! 12
23
rs
(7.09d)
Also, we wish to find ps = g(E, a , rs ), where we have assumed, for the moment, that ps can depend
on a . Again,
ps = const. E a ba rsc
(7.09e)
a
b
This result gives M L1 T 2 = M L2 T 1 M L3 Lc . For this result to be valid, a = 1, b = 0 and
c = 3 and so
ps = const. E rs3
(7.09f)
! 12
32
rs
(7.09g)
! 12
32
rs
(7.09h)
! 12
t
(7.09i)
and hence E r5 at any instant t. The above expression indicates that any instant t, the energy of
the explosion is proportional to the 5th power of the radial extent rs of the corresponding shock wave.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider an incompressible ow through a series of geometrically similar machines such as fans, pumps,
hydraulic turbines, etc. If Q denotes volume ow, rotational speed, D impeller diameter, uid viscosity,
and uid density,
(a) show that dynamic similarity requires that Q/D3 and Q/D be xed.
(b) Show that if Q/D3 and Q/D are xed in a series of tests, then P/ 2 D2 must remain constant,
where P is the change in head across the machine, expressed in pressure units.
(c) Find the form of the relation between the work output per unit mass of uid W , and the the given
variables, in a series of tests where Q/D3 and Q/D are xed.
c 2011, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider an incompressible ow through a series of geometrically similar machines such as fans, pumps,
hydraulic turbines, etc. If Q denotes volume ow, rotational speed, D impeller diameter, uid viscosity,
and uid density,
(a) show that dynamic similarity requires that Q/D3 and Q/D be xed.
(b) Show that if Q/D3 and Q/D are xed in a series of tests, then P/ 2 D2 must remain constant,
where P is the change in head across the machine, expressed in pressure units.
(c) Find the form of the relation between the work output per unit mass of uid W , and the the given
variables, in a series of tests where Q/D3 and Q/D are xed.
c 2011, MIT
Copyright
Dimensional Analysis
Solution:
(a) The variables in the problem are related through f (Q, , D, , , P ) = 0 where
Q [L3 T1 ] : Volume ow rate
[T1 ] : Rotational Speed
D [L] : Impeller diameter
[ML1 T1 ] : Fluid viscosity
[ML3 ] : Fluid density
P [ML1 T1 ] : Change in head across machine
As our primary variables, we pick for the uid, for the ow and D for the geometry. We have
n = 6 variables
r = 3 primary dimensions
j = 6 3 = 3 dimensionless groups
Now,
1 =
Q
a b D c
(7.12a)
Q
D3
(7.12b)
2 =
D2
(7.12c)
Similarly, we nd
Let
1
2
Q
D2
=
D3
Q
=
D
=
(7.12d)
(7.12e)
(7.12f)
(7.12g)
(7.12h)
P
2 D2
2
(7.12i)
c 2011, MIT
Copyright
Dimensional Analysis
= f(
,
)
D3 D
2 D2
Q
Now if 1 = D
3 and =
equation (7.12k) implies
Q
D
(7.12j)
(7.12k)
(7.12l)
w = QP
(7.12m)
(7.12n)
w
= KD2 3
D3
(7.12o)
c 2011, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A uid of density and viscosity ows through a long pipe of diameter D at the volume ow rate Q.
(a) Demonstrate that if the ow is laminar (i.e., totally steady) and fully developed (the velocity prole
and the pressure gradient no longer change with downstream distance x), the pressure gradient in the
direction of ow must have the form
dp
Q
= K 4
dx
D
where K is constant.
(b) Demonstrate that if the ow is turbulent (i.e., unsteady) but steady and fully developed in the mean,
the mean pressure gradient in the direction of ow must have the form
dp
=
dx
Q2
D5
Q
D
c 2012, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A uid of density and viscosity ows through a long pipe of diameter D at the volume ow rate Q.
(a) Demonstrate that if the ow is laminar (i.e., totally steady) and fully developed (the velocity prole
and the pressure gradient no longer change with downstream distance x), the pressure gradient in the
direction of ow must have the form
dp
Q
= K 4
dx
D
where K is constant.
(b) Demonstrate that if the ow is turbulent (i.e., unsteady) but steady and fully developed in the mean,
the mean pressure gradient in the direction of ow must have the form
dp
=
dx
Q2
D5
Q
D
c 2010, MIT
Copyright
Dimensional Analysis
Solution:
Q
dp
dx
2 2
[ML
Q
]
[L T
[L]
[ML
[ML
T1 ]
k = 3 primary variables
j = 1 dimensionless group
Our primary variables must be , Q, and D (since there are no other variables)
dp a b c
Q D
dx
[ ] = [M0 L0 T0 ] = [ML2 T2 ][ML1 T1 ]a [L3 T1 ]b [L]c
=
M0 = M1 Ma
2
2 1 3 c
T =T
L =L
a = 1
T T
L L
c=4
dp D4
= constant
dx Q
dp
Keep in mind that dx
and Q have opposite signs (i.e.,
dene constant K to ensure that K 0.
b = 1
dp
dx
dp
Q
= K 4
dx
D
For our primary variables, we choose (1) a uid property: , (2) a ow parameter: Q, and (3) a
geometric parameter: D. Follow the same procedure as in part (a) to nd
dp D5
dx Q2
D
2 =
Q
1 =
c 2010, MIT
Copyright
Dimensional Analysis
Since 1 = (2 ),
dp
Q2
= 5
dx
D
D
Q
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A researcher is concerned with the mechanics of sh propulsion. To determine how the thrust force generated
by a sh of a given geometry depends on sh size (L = sh length) and on the frequency of oscillation of
the tail (f , in cycles/sec), she builds a mechanical model, having this geometry, of length L = 1 m. She
then mounts this model in a xed position deep within a large tank containing stagnant water at room
temperature, and measures the thrust force F , over a large range of frequency of tail oscillation. She nds
that her data can be described by the empirical equation
F =
0.49 104 f 3
Newtons
1 + 0.74 103 f
(7.18a)
where f is in cycles/sec.
(a) Suppose we want to infer, from these results, the thrust generated by sh of other sizes held in still
water having dierent temperature (i.e. dierent density, viscosity). What relation must be satised
between the frequency, size, and uid condition of the real sh and of the model experiments
(b) From the empirical equation given above for the thrust of a 1 m model in room temperature water,
develop a formula for the thrust of a sh of any given size and tail frequency, held in water at any
given density and viscosity.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A researcher is concerned with the mechanics of sh propulsion. To determine how the thrust force generated
by a sh of a given geometry depends on sh size (L = sh length) and on the frequency of oscillation of
the tail (f , in cycles/sec), she builds a mechanical model, having this geometry, of length L = 1 m. She
then mounts this model in a xed position deep within a large tank containing stagnant water at room
temperature, and measures the thrust force F , over a large range of frequency of tail oscillation. She nds
that her data can be described by the empirical equation
F =
0.49 104 f 3
Newtons
1 + 0.74 103 f
(7.18a)
where f is in cycles/sec.
(a) Suppose we want to infer, from these results, the thrust generated by sh of other sizes held in still
water having dierent temperature (i.e. dierent density, viscosity). What relation must be satised
between the frequency, size, and uid condition of the real sh and of the model experiments
(b) From the empirical equation given above for the thrust of a 1 m model in room temperature water,
develop a formula for the thrust of a sh of any given size and tail frequency, held in water at any
given density and viscosity.
c 2010, MIT
Copyright
Dimensional Analysis
Solution:
We want an expression for the thrust force F in terms of all other variables, such that
F = f (L, , , f )
L
2
[MLT
[L]
[ML
[ML
f
T
[T1 ]
Thus we have
n = 5 variables
k = 3 primary variables
j = 2 dimensionless groups
For our primary variables, we choose (1) a uid property: , (2) a ow parameter: f , and (3) a geometric
parameter: L.
Follow the same procedure as in Problem 7.14(a) to nd 1 , and 2 :
F
L4 f 2
1
=
2 =
Re
L2 f
F
=
L2 f
L4 f 2
1 =
Re = Reynolds number =
Inertia
Viscosity
(a) In order to ensure that the real sh and experiments are dynamically similar, the Reynolds number
must remain constant
=
2
L f experiments
L2 f real
(b) We know that the thrust force can be expressed as:
F
=
L4 f 2
L2 f
1
Re
F =
1
Re
L4 f 2
(7.18b)
This says that the thrust force is some unkown function of the Reynolds number times f 2 . However,
the dependence of Eq. (7.18a) on f is a bit more complicated. Also, note that the coecients in the
given emipirical equation must have dimensions for the right-hand side to have the units of force:
units = [Ns3 ]
F =
(0.49 104 ) f 3
1 + (0.74 103 ) f
units = [s]
Since Re f , we will try to replicate Eq. (7.18a) with Eq (7.18b) by substituting Re (with an unknown
coecient) whenever we need an extra f :
F =
C1 (2 L6 /)f 3
C1 L4 f 2 Re
=
1 + C2 (L2 /)f
1 + C2 Re
(7.18c)
c 2010, MIT
Copyright
Dimensional Analysis
kg
m3
= 103 Pa s = 103
kg
ms
kg2
6
3 kg
3
m6 )(1 m )/(10
ms )f
kg
kg
C2 (103 m3 )(1 m2 )/(103 ms )f
C1 (106
1+
C1 (109 kg m s)f 3
1 + C2 (106 s)f
C1 = 0.49 105
C1 = 0.74 103
Now substitute the coecients above into Eq. (7.18c) such that
F =
L2 f /
= 1 =
(0.49 105 ) Re
1 + (0.74 103 ) Re
Recall, that the coecients in front of Re are dimensionless such that both sides of the above equation
are dimensionless. This new equation is more general than the empirical equation given because
kg
changes in , , and L are taken into account (the original equation was only valid for = 103 m
3,
3
= 10 Pa s, L = 1 m).
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Similarity
The principle of similarity underlies the entire subject of dimensional analysis. There are three
necessary conditions for complete similarity between a model and a prototype.
Geometric similarity: the model must be the same shape as the prototype, but may be
scaled by some constant factor.
Kinematic similarity: the velocity at any point in the model flow must be proportional by
a constant scale factor to the velocity at the homologous point in the prototype flow.
(That is, the flow streamlines must have the same shape.)
Dynamic similarity: all forces in the model flow must scale by a constant factor to the
corresponding forces in the prototype flow. In other words, the relative importance of
different types of forces (e.g., viscous and inertial forces) must be the same for the model
and prototype. This requires that the model and prototype have the same dimensionless
parameters (e.g., the same Reynolds number), although they may (and usually do) have
different dimensional variables. Mathematically, for all p pi groups that can be defined
for two different flow situations, dynamic similarity requires that
k ,model k ,prototype , k 1... p.
Thus, geometric and kinematic similarity are necessary but insufficient conditions for dynamic
similarity. That is, it is possible to have geometric and kinematic similarity, but not dynamic
similarity.
Further reading
engel, Y.A. and Cimbala, J.M. Fluid Mechanics: Fundamentals and Applications. Boston:
McGraw Hill, 2010, pp. 291-292.
Panton, R. Incompressible Flow. Wiley, 2013, p. 170.
Sonin, A.A. The Physical Basis of Dimensional Analysis.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
66
NATURE
NO.
2368,
VOL.
95J
[MARCH
18, 1915
NATURE
M=1!:F(~),
g
grra 2
1z=v!d'/(v!vd),
where f is arbitrary.
As a last example let us consider, somewhat in
detail, Boussinesq's problem of the steady passage of heat from a good conductor immersed in
a stream of fluid moving (at a distance from the
solid) with velocity v. The fluid is treated as
incompressible and for the present as inviscid,
while the solid has always the same shape and
presentation to the stream.
In these circumstances the total heat (h) passing in unit time is
a function of the linear dimension of the solid
(a), the temperature-difference (0), the streamvelocity (v), the capacity for heat of the fluid
per unit volume (e), and the conductivity (K).
The density of the fluid clearly does not enter into
the question. We have now to consider the
" dimensions" of the various symbols.
.v .
. . . ().
.... e.
. . . . . K,.
. . . . . h ..
(Length)!,
(Length)! (Time)-!,
(Temperature)!,
(Heat)! (Length)-3 (Temp.)-!,
(Heat)l (Length)-l (Temp.)-l (Time)-l,
(Heat)! (Time)-l.
Hence if we assume
we have
NO.
heat
I =u+v
temperature o=y-u-v,
length
o=X+2-3zt-V,
time
- 1= - Z - V ;
2368,
VOL.
95J
h=K,a()
(a~er
where F denotes an arbitrary function. Experiment shows that F varies but little and that within
somewhat wide limits may be taken to be 3'8.
Within these limits Tate's law that M varies as
a holds good.
In the A':olian harp, if we may put out of
account the compressibility and the viscosity of the
air, the pitch (n) is a function of the velocity of
the wind (v) and the diameter (d) of the wire.
It then follows from similarity that the pitch is
directly as v and inversely as d, as was found
experimentally by Strouhal. If we include viscosity (v), the form is
Those of a are
so that
h=Ka().(~~)zt:r
Here z and ware both undetermined, and the
conclusion is that
ev,
h=.a().F l(ave
-K-, -;j'
68
NATURE
PERISCOPES.
HILE the periscope of the submarine is
developing in the direction of greater
optical perfection and elaboration, there has been
a return to the simplest and earliest types of
periscope for use in land warfare. Some of these
trench periscopes recall the polemoscope, described
by Helvelius in the seventeenth century for
military purposes; this polemoscope in its simplest form consisted of two mirrors with their
reflecting surfaces parallel to each other, and
FIG. I.
2368,
VOL.
9SJ
u(y)
y
Q = Q/W
Q =
H
2
(Utop
Ubottom ) (signed)
w,top =
(Utop Ubottom )
Umax = dP
dx
H2
8
Uavg = 23 Umax
Q = dP
dx
w =
>0
u(y )
H3
12
H
dP
2 ( dx )
dP
dx
x
u(y) =
U H 2 ) dP y
2 ( dx )( H (1
y
H ))
Umax =
Uavg =
>0
U R2 ) dP
4 ( dx )
U R2 )U
8
r
dP
dx
u(r)
2R = D
R
Q = dP
dx 8
w (2RL) = R2 L( dP
dx )
(w in the x direction)
f=
P
L
1
U 2 D
2
u(r) =
16
ReD
U R2 )
4
r 2
( dP
dx )(1 ( R ) )
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Contents
Fig. 1: Locally fully developed flow (left) and fully developed flow (right)
(1)
v
2 v 2v
v
v
P
,
+
u
+
v
=
+
+
t
x
y
y
x 2 y 2
(2)
(3)
and the appropriate boundary and initial conditions. In (1) and (2)
P = p + gz
(4)
not the third Cartesian coordinate that goes with x and y). The term gz in
(4) accounts for the gravitational body force.
The simultaneous presence of the nonlinear inertial terms on the left and
the second order viscous terms on the right makes it difficult to solve the
Navier-Stokes equation (1)-(2) in the general case. Under certain
circumstances, however, all the inertial terms on the left hand sides of (1) and
(2), while not zero, are small enough compared with the viscous term to be
neglected, and the y-derivative in the viscous term dominates over the xderivative. Under these conditions (1) and (2) simplify to
P
2u
0
+ 2
x
y
0
P
y
(5)
(6)
These equations are similar in form to the equations for a truly fully
developed flow. The velocity profile at a station x in this diverging and
possibly unsteady flow will thus be identical to the profile in a fully
developed flow with the same height, the same boundary conditions at y=0
and y=h, and the same pressure gradient. The flow can be said to be locally
fully developed, that is, having at every station x essentially the same velocity
profile as a fully developed flow with the same cross-sectional geometry and
boundary conditions. For example, if the flow is bounded by solid, immobile
walls such that u=0 at y=0 and y=h, the local solution is the familiar
parabolic one
u
h 2 y
y P
.
1
2
h h x
(7)
u
2u
<< 2
t
y
(8)
u
2u
<< 2
x
y
(9)
u
2u
<< 2
y
y
(10)
2u
2u
<<
x 2
y 2
(11)
where we have implied, but not indicated, that absolute magnitudes are
involved in the inequalities. These criteria can be expressed in more useful
form by estimating the orders of magnitude of the various derivatives in
terms of specified quantities. Let U be a characteristic, or typical, streamwise
velocity such that
u U
(12)
where the symbol in this case stands for order of magnitude, by which we
mean an estimate that is correct to within a factor of 3, say, implying that we
know the decade in which the quantity's numerical value lies on a logarithmic
scale. U might be the average flow speed at the channel's entrance at a
particular time, say. Similarly, let h be a characteristic value of the transverse
distance h in the problem, e.g. the value at the channel's entrance. Since the
dependence on y is expected to be parabolic when (1) serves as a good
approximation, we estimate, to order of magnitude, that
u U
.
y h
(13)
x L
(14)
.
t
(15)
x 2 L2
(16)
and (13) and the expected (near-) parabolic variation of u with y imply that
2u U
.
y 2 h 2
(17)
u
dy .
x
0
(18)
U
h .
L
(19)
Indicating only the orders of magnitude of all terms except the pressure
gradients, we can now write (1) and (2) as
U U 2 U 2
U U
p
+
+
+ 2 + 2
L h
x
L
L
U U 2 U 2 h
U U h
p
+
+
+ 2 + 2
L h L
y
L
L L
(20)
(21)
(22)
h2
<< 1
(23)
Uh h
<< 1 .
(24)
The same criteria also ensure that (2) reduces to (6). This becomes apparent
when we think of (20) and (21) as providing order-of-magnitude estimates
for the respective pressure gradients on their right-hand sides. A comparison
of (21) and (20) shows that
P h P
.
y L x
(25)
what is small enough, but it can be done only on a case by case basis after a
particular solution has been obtained. Equations (22)-(24) serve as an
adequate estimate, however, and are conveniently expressed in general terms.
They suffice if the estimates come out truly small so that there is little doubt
of the assumptions being satisfied.
pviscous =
12UL
h2
(26)
The pressure drop at the entrance and across the transition region between
the two segments depends on whether the Reynolds number is small or
large. If the Reynolds number is small, we estimate it from (26) with Lh,
assuming the transition occurs in a distance of at most a few h; if it is large,
we claim it cannot exceed the maximum Bernoulli drop. Thus,
Fig. 2: Pressure distribution under (a) inertia-dominated, (b) intermediate and (c) highly
viscous (locally fully developed) conditions.
10
12Uh
ptransitions
h 2
ptransitions
U 2
2
if Re =
Uh
<1
(27)
if Re =
Vh
>1
(28)
if
Uh
<1
(29)
ptransitions 1 Uh h
12 L
pviscous
if
Vh
>1
(30)
(31)
Uh h
<< 1
L
(32)
h2
<< 1
(33)
Note that the requirement (31) for constant pressure across area changes is
more severe than (22), which refers to the flow inside the channel sections
only.
11
(34)
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25
Contents
1.
2.
3.
4.
5.
Surface stress . 2
The stress tensor 3
Symmetry of the stress tensor 7
Equation of motion in terms of the stress tensor 9
Stress tensor for Newtonian fluids . 12
The shear stresses and ordinary viscosity . 12
The normal stresses .. 13
General form of the stress tensor; the second viscosity 18
6. The Navier-Stokes equation 21
7. Boundary conditions .. 23
Appendix A: The Navier-Stokes and mass conservation equations
in cylindrical coordinates, for incompressible flow .24
Appendix B: Properties of selected fluids .. 26
Surface Stress
Quantities like density, velocity, and pressure are defined by a value at every point in
!
!
the fluid at every time t. The density !(r ,t) and pressure p( r ,t) are scalar fields. They
! !
have a numerical value at every point in space at any instant in time. The velocity v (r,t)
is a vector field; it is defined by a direction as well as a magnitude at every point.
The surface stress is a more complicated type of quantity. One cannot talk of the
stress at a point without first defining the particular surface through that point on which
!
the stress acts. A small fluid surface element centered at the point r is defined by its area
A (the prefix ! indicates a very small but finite quantity) and by its outward unit normal
!
!
vector n . The stress exerted by the fluid on the side toward which n points on the surface
element is defined as
!
"F
!
! = lim
"A#0 "A
(1)
!
where !F is the force exerted on the surface by the fluid on that side (only one side is
involved). In the limit A 0 the stress is independent of the magnitude of the area, but
!
will in general depend on the orientation of the surface element, which is specified by n .
In other words,
! ! ! !
! = ! ( x ,t, n ) .
(2)
!
!
The fact that ! depends on n as well as x, y, z and t appears at first sight to
complicate matters considerably. One apparently has to deal with a quantity that depends
!
on six independent variables (x, y, z, t, and the two that specify the orientation n ) rather
3
!
than four. Fortunately, nature comes to our rescue. We find that because ! is a stress, it
!
must depend on n in a relatively simple way.
We have seen that, in the absence of shear forces, Newton's law requires that the
surface stress have the particularly simple form
!
!
! = " pn (no shear forces)
(3)
!
where p, the magnitude of the normal compressive stress, is a function of r and t only.
This is Pascal's principle, which states that in the absence of shear forces, at any point in
the fluid, the stress is always normal to the surface on which it acts, and its magnitude is
independent of the surface orientation. In the absence of shear stresses, therefore, the
stress on any surface, anywhere in the fluid, can be expressed in terms of a single scalar
!
field p( r,t) provided there are no shear forces. This gives rise to the relatively simple
form of the equation of motion for inviscid flow.
When shear forces are present, as they always are in practice except when the fluid is
totally static in some reference frame, Newton's law imposes a somewhat more
!
!
complicated constraint on the relationship between ! and n . We shall see that the stress
on any surface anywhere in the fluid can in general be specified in terms of six scalar
functions of x, y, z, and t. These six are the independent components of a quantity called
the stress tensor.
The first and simplest thing that Newton's law implies about the surface stress is that,
!
at a given point, the stress on a surface element with an orientation n must be equal in
magnitude, but opposite in direction, to that on a surface element with an opposite
!
orientation ! n , that is,
! ! !
! !
!
! ( r,t," n ) = "! (r ,t, n)
(4)
!
This result can be obtained by considering a thin, disc-shaped fluid particle at r , as
shown in Fig. 2, with very small area A and thickness h. One side of the disc has an
!
!
orientation n and the other ! n . The equation of motion for this fluid particle reads
!"h"A
!
!
Dv ! !
! !
= # ( n )"A + # ($ n )"A + !"h"AG
Dt
(5)
!
where G is the body force per unit mass. When we let h approach zero, so that the two
faces of the disc are brought toward coincidence in space, the inertial term on the left and
the body force term on the right become arbitrarily small compared with the two surface
force terms, and (4) follows immediately.
Newton's law also implies that the stress has a more profound attribute, which leads
to the concept of the stress tensor. The stress at a given point depends on the orientation
of the surface element on which it acts. Let us take as "reference stresses," at a given
!
point r and instant t, the values of the stresses that are exerted on a surface oriented in
the positive x-direction, a surface oriented in the positive y-direction, and a surface
oriented in the positive z-direction (Fig. 3). We can write these three reference stresses,
which of course are vectors, in terms of their components:
!
!
!
! !
! (i ) = " xx i + " yx j + " zxk
!
!
!
! !
! ( j ) = " xy i + " yy j + " zy k
!
!
!
! !
! ( k ) = " xz i + " yz j + " zzk
(6)
Thus, ! xx , ! yx and ! zx represent the x, y, and z components of the stress acting on the
surface whose outward normal is oriented in the positive x-direction, etc. (Fig. 3). The
first subscript on ! ij identifies the direction of the stress, and the second indicates the
outward normal of the surface on which it acts. In (6) the ! ij ' s are of course functions of
position x, y, z, and time t, and the reference stresses themselves also depend on x, y, z,
and t; we have simply not indicated this dependence.
We shall now show, again by using Newton's law, that the stress on a surface having
!
!
! !
any! orientation! n at the point r can be expressed in terms of the reference stresses ! ( i ) ,
!
!
! ( j ) , and ! ( k ) or, more specifically, in terms of their nine components ! xx , ! yx ,..., ! zz .
Consider a fluid particle which at time t has the shape of a small tetrahedron centered
!
at x, y, z. One of its four faces has an area A and an arbitrary outward normal n , as
shown in Fig. 4, and the other three faces have outward normals in the negative x, y and z
directions, respectively. The areas of the three orthogonal faces are related to A by
(7)
where Ax represents the area of the surface whose outward normal is in the negative x!
!
direction, ! nx is the angle between n and the x-axis and nx is the x-component of n , and
so on.
6
Consider what Newton's law tells us about the forces acting on the tetrahedron as we
!
let it shrink in size toward the point r around which it is centered. Since the ratio of the
mass of the tetrahedron to the area of any one of its faces is proportional to the length of
any one of the sides, both the mass times acceleration and the body force become
arbitrarily small compared with the surface force as the tetrahedron is shrunk to a point
(c.f. (5) and the paragraph that follows it). Hence, in the limit as the tetrahedron is shrunk
to a point, the surface forces on the four faces must balance, that is,
! !
! !
! !
! ( n)"A + ! ( j )"Ax + ! (k )"Az = 0 .
(8)
!
Now we know from (4) that the stress on a! surface pointing in the ! i direction is the
negative of the stress on a surface in the + i direction, etc. Using this result and (7) for
the areas, (8) becomes
! !
! !
! !
! !
! ( n) = ! (i )nx + ! ( j )ny + ! ( k )nz .
(9)
Alternatively, if we use (6) to write the reference stresses in terms of their components,
! !
we obtain the components of ! ( n) as
!
! x (n ) = " xx nx + " xy ny + " xz nz
!
! y (n ) = " yx n x + " yy ny + " yz nz
(10)
!
! z ( n) = " zxnx + " zyny + " zznz .
!
!
! !
Thus the stress ! ( n) acting at x ,t on a surface with any arbitrary orientation n can be
expressed in terms of the nine reference stress components
xx xy xz
yx
yy
yz
xz
zy
zz .
These nine quantities, each of which depends on position and time, are the stress tensor
components. Once the stress tensor components are known at a given point, one can
compute the surface stress acting on any surface drawn through that point by determining
!
the components of the outward unit normal n of the surface involved, and using (10).
Equation (10) can be written more succinctly in conventional tensor notation, where i
and j can represent x, y, or z and where it is understood that any term which contains the
same index twice actually represents the sum of all such terms with all possible values of
the repeated index (for example, ii xx + yy + zz). In this notation (10) reads
! !
!
! i ( r,t, n ) = " ij (r ,t)n j .
(11)
The importance of the stress tensor concept in continuum theory is this: It allows us
to describe the state of stress in a continuum in terms of quantities that depend on
position and time, but not on the orientation of the surface on which the stress acts.
Admittedly, nine such quantities are needed (actually only six are independent, as we
shall see shortly). Still, it is far easier to deal with them than with a single quantity which,
at any given position and time, has a doubly infinite set of values corresponding to
!
different surface orientations n .
Physically, the stress tensor represents the nine components of the three reference
!
stresses at the point r and time t in question. The reference stresses are by custom chosen
as the stresses on the three surface elements that have outward normals in the direction of
the positive axes of the coordinate system being used. Thus in our Cartesian coordinates,
the reference stresses are the stresses on the surfaces pointing in the positive x, y, and z
directions, and the stress tensor is made up of the nine components of these three stresses,
! ij being the i-component of the stress on the surface whose normal points in the jdirection. In a cylindrical coordinate system, the stress tensor would be comprised of the
components of the stresses acting on the three surfaces having outward normals in the
positive r, and z directions.
Why are the quantities ! ij "tensor components," and not just an arbitrary bunch of
nine scalar quantities? The answer lies in the special way the values of these nine
quantities transform when one changes one's reference frame from one coordinate system
to another. Equation (10) tells us that when a coordinate change is made, the three sums
! ij nj must transform as components of a vector. A set of nine quantities ! ij that transform
in this manner is by definition a tensor of second rank. (A tensor of first rank is a vector,
whose three components transform so that the magnitude and direction of the vector
remain invariant; a tensor of zeroth rank is a scalar, a single quantity whose magnitude
remains invariant with coordinate changes.)
8
at essentially zero angular distortion), since in the limit x, y, z 0, a finite angular
distortion would require infinite shear in a viscous fluid. If the cube has an angular
velocity !z in the z-direction, say, and rotates like a solid body, we can derive from
Newton's law
written in angular momentum form for a material volume, that at any given instant its
angular velocity increases according to
Iz
d! z
= Tz ,
dt
(12)
where
+#x 2 +#y 2 +#z 2
$ $ $ !(x
Iz =
+ y 2 )dxdydx
![("x) 2 + ("y) 2 ]
12
"x"y"z
(13)
is the moment of inertia of the cube and Tz is the net torque acting on the cube, relative to
an axis running through the center of the cube parallel to the z-axis. Equation (13) is
obtained by writing the cubes angular velocity as v! = !(t)r , where r 2 = x 2 + y 2 , x and
y being the Cartesian coordinates fixed in the rotating cube.
The torque in (12)! is obtained by considering the stresses acting on the cube (Fig.
!
5). On the face with n = i , for example, there is by definition a stress ! xx in the positive
!
!
x-direction and a stress ! yx in the positive y-direction. On the face with n = !i , the
corresponding stresses have the same magnitudes but opposite directions [see (10) or
9
(4)]. The net torque about an axis through the cube's center, parallel to the z- axis, is
caused by the shear forces (the pressure forces act through the cubes center) and by any
volumetric torque exerted by the external body force field. A body force field like gravity
acts through the cube's center of mass and exerts no torque about that point. Let us
assume for the sake of generality, however, that the external body force may exert a
!
torque t per unit volume at the particle's location. The net torque in the z-direction
around the particle's center would then be
Tz = 2
!x
!x
" yx !y!z # 2 " xy!y !z + t z!x !y!z
2
2
= (! yx " ! xy + tz )#x#y#z
(14)
! yx " ! xy + t z =
# d$z
(%x) 2 + (%y) 2 ]
[
12 dt
(15)
! yx = ! xy " tz ,
or, more generally, the result that the off-diagonal stress tensor components must satisfy
! ji = ! ij + t k ,
(16)
where i, j, k form a right-hand triad (e.g. in Cartesian coordinates they are in the order x,
y, z, or y, z, x, or z, x, y).
Volumetric body torque can exist in magnetic fluids, for example (e.g. see R. E.
Rosensweig, Ferrohydrodynamics, 1985, Chapter 8). In what follows we shall assume
that volumetric body torque is absent, in which case (16) shows that the off-diagonal or
shear terms in the stress tensor are symmetric,
! ji = ! ij (i ! j) .
(17)
This means that three of the nine components of the stress tensor can be derived from the
remaining ones; that is, the stress tensor has only six independent components.
10
! ij +
"! ij #x
"x 2
! ij "
#! ij $x
#x 2
(18)
Since xyz is the particle's volume, we identify the quantity within the brackets as the
net x-component of surface force per unit volume at a point in a fluid. The expressions
for the y and z components are similar, except that the first subscript x is replaced by y
and z, respectively.
11
The equation of motion can now be written down directly for the cubical fluid particle
in Fig. 6. The x-component of the equation states that the mass times the acceleration
equals the net surface force plus the body force acting on the particle:
!"x"y "z
Dvx %' #$ xx #$ xy #$ xz (
=
+
+
"x "y "z + !"x"y "zGx
Dt & #x
#y
#z )
Here, D/Dt represents the substantial derivative, which is defined elsewhere, and Gx is
the x-component of the external body force per unit mass. This yields
(20a)
(20b)
(20c)
Dvi "# ij
=
+ !Gi
Dt "x j
(20)
12
where a summation over j=x, y, and z is implied. Equation (20) states that at a given point
and time, the mass per unit volume times the acceleration in the i-direction (the left-hand
term) equals the the net surface force per unit volume in the i-direction (the first term on
the right) plus the body force per unit volume in the i-direction (the second term on the
right). The equation applies quite generally to any continuous distribution of matter,
whether fluid or solid, and is not based on any assumption other than that the continuum
hypothesis applies.1 Eq. (20) is, however, incomplete as it stands. To complete it, one
must specify the stress tensor components and the body force components, just as one
must define the forces acting on a solid particle before one can derive its motion. The
specification
of the body force is straightforward. In a gravitational field, for example, the
!
force G per unit mass is well known and is of the same form for all substances. The
form of the stress tensor is different for different classes of materials.
There remains the task of specifying the relationship between the stress tensor
components and the flow or deformation field. The simplest model of a solid continuum
is the well-known elastic one, where stresses and strains are linearly related. The defining
attribute of a simple fluid, however, is that it keeps deforming, or straining, as long as any
shear stress, no matter how small, is applied to it. Obviously, no unique relation can exist
between the shear stresses and the shear strains if strain can increase indefinitely at
constant shear. It is observed, however, that a fluid tends to resist the rate of deformation:
the higher the applied shear stress, the faster the rate of shear deformation. In many fluids
the relation between stress and rate of strain in a fluid particle is linear under normal
conditions.
The Newtonian model of fluid response is based on three assumptions:
(a)
(b)
(c)
A Newtonian fluid is the simplest type of viscous fluid, just like an elastic solid
(where stresses are proportional to strains) is the simplest type of deformable solid.
1In
static solid deformations, the acceleration term is absent and the gravitational loads induced by the
weight of the structure itself are often negligible compared with externally applied forces. In such cases the
equation of motion reduces to the simple statement that the net surface stress per unit volume is zero at
every point in the medium.
13
The shear stresses and the ordinary viscosity
To implement the Newtonian assumptions we consider first a typical shear term in the
tensor, e.g ! xy . Fig. 7 depicts the deformation of a fluid particle as it moves between time
t and time t+dt. In this interval the shear stress ! xy produces in the fluid particle an
incremental angular strain d! xy
d! xy
$& "v x
' $& "vy
'
#ydt
#xdt
% "y
( % "x
(
=
+
.
#y
#x
The rate of angular (or shear) strain in the fluid particle as seen by an observer sitting on
it is therefore
D! xy "vx "v y
=
+
.
Dt
"y "x
(21)
! xy =
D" xy
$ #v #v y '
= & x +
.
% #y #x (
Dt
(22a)
! xz =
D" xz
$ #v x #v z &
=
+
% #z #x '
Dt
(22b)
14
! yz =
D" yz
$ #v #v '
= & y + z
% #z #y (
Dt
(22c)
or in general,
# "v "v &
! ij = % i + j (
$ "x j "xi '
(i j) .
(22)
The coefficient of proportionality is the same in all three shear stresses because a
Newtonian fluid is isotropic.
The normal stresses
Next consider a typical normal stress, that is, one of the stress tensor's diagonal terms,
say ! xx . The derivation of such a term's form is not as simple as that of the shear terms,
but can nevertheless be done in fairly physical terms by noting that linear and shear
deformations generally occur hand in hand. The trick is to find how the linear stresses
and deformations are related to the shear stresses and deformations.
Consider a small fluid particle which at time t is a small cube with sides of length h
parallel to the x, y and z axes. We will again be considering the limit of a particle "at a
point", that is, the limit h ! 0 . At time t , its corner A is at (x, y, z). Between t and t+dt,
it moves and deforms as in Fig. 8. The sides AB and AD will in general rotate by
unequal amounts. This will result in a shear deformation of the particle. The shear
deformation will cause one of the diagonals AC and BD to expand and the other to
contract, that is, it will give rise to linear deformations in the x' and y' directions which
are rotated 45o relative to the x and y axes.
Now, we know the relationship between the shear stress and the rate of angular strain
of the particle in the (x, y) frame. If we can connect the shear stresses in this frame and
the stresses in the rotated (x', y') frame, and the shear strain rates in the (x, y) frame and
the strain rates on the (x', y') frame, we will arrive at a relation between the stresses and
the strain rates in the (x', y') frame. Since the reference frames are arbitrary, the
relationship between stresses and rates of strain for the (x', y') frame must be general in
form.
15
We start by considering the forces acting on one half of the fluid particle in Fig. 8: the
triangular fluid particle ABD as shown in Fig. 9. Since we are considering the limit
!h " 0 , where the ratio of volume to area vanishes, the equation of motion for the
particle will reduce to the statement that the surface forces must be in balance. Figure 9
shows the surface forces on particle ABD, expressed in terms of the stress tensor
components in the original and the rotated reference frames. A force balance in the x'direction requires that
! =
" xx
" xx + " yy
+ " yx .
2
(23)
Similarly, a force balance in the y'-direction on the triangular particle ACD requires that
" yy! =
" xx + " yy
# " yx .
2
(24)
(25)
Using the relation (22a) between the shear stress and the rate of strain, this becomes
D$ xy
Dt
(26)
which relates the diagonal stress tensor terms in the (x', y') frame to the angular strain rate
in the (x, y) frame.
16
To close the loop we must relate the angular strain rate in the (x, y) frame to the strain
rates in the (x', y') frame. Figure 10 shows the deformations of the triangular particles
ABD and ACD between t and t+dt. The deformations a, b, c, and d in the figure are
related to the incremental linear strains d! and angular strains d! in the (x, y) frame by
c
h
a
d! y =
h
b+ d
.
d! xy =
h
d! x =
(27)
Here, d! x is the linear strain (increase in length divided by length) of the particle in the
x-direction, d! y is its linear strain in the y-direction, and d! xy is the angular strain in the
x-y plane.
17
The linear strain in the x' direction can be computed in terms of these quantities from
the fractional stretching of the diagonal AC, which is oriented in the x' direction.
Recalling that ACD is an isoscoles triangle at time t, and that the deformations between t
and t+dt are infinitesimally small, we obtain
a+ d b+c
+
d(AC)
2
2 = 1 # c + a + b + d & = 1 (d! + d! + d) ) .
=
d! x " =
(
%
x
y
xy
(AC)
2$h h
h ' 2
h 2
(28)
The linear strain in the y' direction is obtained similarly from the fractional stretching of
the diagonal BD of the triangular particle ABD as
d! y" =
d(BD) 1
= (d! x + d! y # d$ xy ) .
(BD) 2
(29)
The sum of the last two equations shows that the difference of the linear strains in the x'
and y' directions is equal to the angular strain in the (x, y) plane:
d! x " # d! y" = d$ xy .
(30)
The differentials refer to changes following the fluid particle. The rates of strain
following the fluid motion are therefore related by
D! x " D! y " D$ xy
.
#
=
Dt
Dt
Dt
(31)
If we now eliminate the reference to the (x, y) frame by using (26), we obtain
D$ (
% D$
! x "x " # ! y"y " = 2 ' x " # y " .
& Dt
Dt )
(32)
The linear strain rates can be evaluated in terms of the velocity gradients by referring to
Fig. 11. Between t and t+dt, the linear strain suffered by the fluid particle's side parallel
to the x' axis is
#vx "
$xdt #v
#
x
"
d! x " =
= x " dt
$x
#x "
so that
18
(33)
A similar equation is obtained for the linear strain rate in the y' direction. Using these
relations in (32), we now obtain
% $v
$v (
! x "x " # ! y"y " = 2 ' x " # y " .
& $x " $y " )
(34)
(35)
! x "x " =
+ 2
(36)
Since the coordinate system (x', y') is arbitrary, this relationship must apply in any
coordinate system. We thus have our final result:
! xx = " pm + 2
where the quantity
#vx 2
!
" $ % v
#x 3
(37)
19
pm = !
("
xx
+ " yy + " zz )
3
=!
" ii
3
(38)
where
(39)
!ij = 1 if i=j
= 0 if ij
is the Kronecker delta. Note that (39) represents any single component of the tensor, and
no sum is implied in this equation when one writes down the general form of the diagonal
terms by setting j = i.
The mechanism whereby stress is exerted by one fluid region against another is
actually a molecular one. An individual molecule in a fluid executes a random thermal
motion, bouncing against other molecules, which is superposed on the mean drift motion
associated with flow. Normal stress on a surface arises from average momentum transfer
by the fluid molecules executing their random thermal motion, each molecule imparting
2
Note added by G.H.McKinley; in order to avoid confusion, the notation for the second viscosity in the
following section has been modified from AAS original notes to match that of Kundu (3rd Edition), the
second viscosity (or bulk viscosity is thus given by ! =
2
3
+".
20
an impulse as it collides with the surface and rebounds. Normal stress is exerted even in a
static, non-deforming fluid. Shear stress arises when there is a mean velocity gradient in
the direction transverse to the flow. Molecules which move by random thermal motion
transverse to the flow from a higher mean velocity region toward a lower mean velocity
region carry more streamwise momentum than those moving in the opposite direction,
and the net transfer of the streamwise molecular momentum manifests itself as a shear
stress on the macroscopic level at which we view the fluid.
The molecular theory of the shear viscosity coefficient is quite different for gases and
liquids. In gases the molecules are sparsely distributed and spend most of their time in
free flight rather than in collisions with each other. In liquids, on the other hand, the
molecules spend most of their time in the short-range force fields of their neighbors (see
for example J. O. Hirschfelder, C. F. Curtiss and R. B. Bird, Molecular Theory of Liquids
Gases and Liquids). The shear viscosity is mainly a function of temperature for both
gases and liquids, the dependence on pressure being relatively weak. There is, however,
one big difference between gases and liquids: the viscosity of gases increases with
temperature, while the viscosity of liquids decreases, usually at a rate much faster than
the increase in gases. The viscosity of air, for example, increases by 20% when
temperature increases from 18oC to 100oC. The viscosity of water, on the other hand,
decreases by almost a factor of four over the same temperature range.
Equation (39) contains only a single empirical coefficient, the shear or ordinary
coefficient of viscosity . A second coefficient is, however, introduced in our quest for a
complete set of flow equations when we invoke the fluid's equation of state and are
forced to ask how the "thermodynamic" pressure which appears in that equation is related
to the mechanical pressure pm . The equation of state expresses the fluid's density as a
function of temperature and pressure under equilibrium conditions. The "thermodynamic"
pressure which appears in that equation is therefore the hypothetical pressure that would
exist if the fluid were in static equilibrium at the local density and temperature.
Arguments derived from statistical thermodynamics suggest that this equilibrium
pressure may differ from the mechanical pressure when the fluid is composed of complex
molecules with internal degrees of freedom, and that the difference should depend on the
rate at which the fluid density or pressure is changing with time. The quantity that
provides the simplest measure of rate of density change is the divergence of the velocity
!
vector, ! " v , which represents the rate of change of fluid volume per unit volume as seen
by an observer moving with the fluid. It is customary to assume a simple linear
relationship which may be thought of as being in the spirit of the original Newtonian
postulates, but in fact rests on much more tenuous experimental grounds:
!
pm = p ! "# $ v .
(40)
Here, is an empirical coefficient which has the same dimension as the shear viscosity ,
and is called the expansion viscosity (Batchelor, An Introduction to Fluid Dynamics;
alternative names are "second coefficient of viscosity" and "bulk viscosity").
Thermodynamic second-law arguments show that must be positive. This implies that
21
the thermodynamic pressure tends to be higher than the mechanical pressure when the
!
mechanical pressure is decreasing (volume increasing, ! " v > 0), and lower than the
!
mechanical pressure when the pressure is increasing (volume decreasing, ! " v < 0 ). In
other words, the thermodynamic pressure always tends to "lag behind the mechanical
pressure when a change is occurring. The difference depends, however, on both the rate
!
of expansion ( ! " v ) and the molecular composition of the fluid (via : see below).
Written in terms of the thermodynamic pressure p, the Newtonian stress tensor reads
$ 3v 3v j '
,
!/
$2
'
! ij = " . p + & " # ) * + v 1 2 ij + & i +
).
%3
(
0
% 3 x j 3 xi (
(41)
!
The term ! " v is associated with the dilation of the fluid particles. The physical
!
interpretation of ! " v is that it represents the rate of change of a fluid particle's volume
recorded by an observer sitting on the particle, divided by the particle's instantaneous
volume.
It can be shown rigorously that = 0 for dilute monatomic gases. For water is about
three times larger than , and for complex liquids like benzene it can be over 100 times
!
larger. Nevertheless, the effect on the flow of the term which involves ! " v and the
expansion viscosity is usually very small even in compressible flows, except in very
special and difficult-to-achieve circumstances. Only when density changes are induced
either over extremely small distances (e.g. in the interior of shock waves, where they
occur over a molecular scale) or over very short time scales (e.g. in high-intensity
!
ultrasound) will the term involving ! " v be large enough to have a noticeable effect on
the equation of motion. Indeed, attempts to study the expansion viscosity are hampered
by the difficulty of devising experiments where its effect is significant enough to be
accurately measured. For most flows, therefore, including most compressible flows
where the fluid's density is changing, we can approximate the stress tensor by
% $v $v (
! ij = " p# ij + ' i + j *
& $x j $xi )
or
! xx = " p + 2
#vx
#x
! yy = " p + 2
#v y
#y
#vz
#z
#% "v x "v y &
! xy = ! yx =
+
$ "y "x '
! zz = " p + 2
(42)
(43)
22
! xz = ! zx =
# "v x "vz %
+
$ "z
"x &
The Navier-Stokes equation is the equation which results when the Newtonian stress
tensor, (41), is inserted into the general equation of motion, (20):
!
(44)
Dvi
#
="
# xi
Dt
!0 #
%2
(
/ p + '& 3 " $ *) + , v 2 + # x
.
1
j
- % #v j #v ( 0
+ i * 2 + !Gi
/ '
#
x
/. & i # x j ) 21
23
!
!
Dv
!
!
(
%1
!
= "#p + ' + $ * #(# + v) + # 2 v + !G
&3
)
Dt
(45)
where
!2 =
"
"
"
2 +
2 +
"x
"y
"z2
(46)
!
is a scalar operator, operating in (45) on the vector v , just like D/Dt on the left side is the
!
well-known scalar operator that operates on v .
For incompressible flows with constant viscosity,
!v j
!
= "# v = 0 ,
!x j
(47)
# 2 vi
Dvi
#p
="
+
+ !Gi ,
Dt
#xi
#x j #x j
(48)
!
!
Dv
!
!
= "#p + #2 v + ! G .
Dt
(49)
As mentioned above, (48) or (49) are in many cases a very good approximation even
when the flow is compressible. Written out fully in Cartesian coordinates, (48) reads
# "v
# " 2v " 2v " 2v &
"v
"v
"v &
"p
! % x + vx x + v y x + vz x = ( + % 2x + 2x + 2x + !Gx
$ "t
$ "x
"x
"y
"z '
"x
"y
"z '
(50a)
(50b)
#% "vz
#% " 2 v z " 2 v z " 2 vz &
"vz
"vz
"v z &
"p
!
+ vx
+ vy
+ vz
=( +
+ 2 + 2 + !Gz
$ "t
$ "x 2
"x
"y
"z '
"z
"y
"z '
(50c)
24
by one fluid particle against another. Later, Barr de Saint Venant (in 1843) and George
Gabriel Stokes (in 1845) derived the equation starting with the linear stress vs. rate-ofstrain argument.
Boundary conditions
A particular flow problem may in principle be solved by integrating the NavierStokes equation, together with the mass conservation equation plus whatever other
equations are required to form a complete set, with the boundary conditions appropriate
to the particular problem at hand. A solution yields the velocity components and pressure
at the boundaries, from which one obtains the stress tensor components via equation (42)
[or (43)] and the stress vector from (11).
In the absence of surface tension, the boundary conditions consistent with the
continuum hypothesis are that (a) the velocity components and (b) the stress tensor
components must be everywhere continuous, including across phase interfaces like the
boundaries between the fluid and a solid and between two immiscible fluids. That this
must be so can be proved by applying mass conservation and the equation of motion to a
small disc-shaped control volume at a point in space, similar to the disc depicted in Fig.
1, and considering the limit where the thickness of the disc go to zero. The proof for the
continuity of ! ij is essentially the same as the one for equation (4), with the requirement
!
that the equation of motion must be satisfied at every point for any orientation n of the
surface.
Surface tension gives rise to a discontinuity in the normal stress at the interface
between two immiscible fluids.
25
Appendix A
The Navier-Stokes equation, its stress tensor, and the mass conservation equation in
cylindrical coordinates (r, ! , z), for incompressible flow
%' "v r
"v r v# "v r v#2
"v (
!
+ vr
+
$ + vz r * =
& "t
"r
r "# r
"z )
"p
( 1 " # "vr % vr 1 " 2 vr 2 "v' " 2 vr +
!
+ *
r
+ 2 - + .Gr
! 2+ 2
2 ! 2
"r
r "'
r "'
"z ,
) r " r $ "r & r
v "v v v
$ "v#
"v
"v
+ vr # + # # + r # + vz # & =
% "t
"r
r "#
r
"z '
1 "p
1 " 2 v# 2 "vr " 2 v# +
( 1 " $ "v# & v#
!
+ *
r
+ 2 - + .G#
! 2+ 2
2 + 2
r "#
r "#
r "#
"z ,
) r "r % " r ' r
$ "vz
"v v "v
"v
+ vr z + # z + v z z & =
% "t
"r
r "#
"z '
"p
( 1 " # "vz % 1 " 2 vz " 2 v z +
+ .Gz
!
+ *
r
+ 2
2 +
"z
"z 2 -,
) r "r $ "r & r "'
(A.1)
(A.2)
(A.3)
26
Stress tensor components (cylindrical coordinates)
! rr = " p + 2
#v r
#r
! "" = # p + 2
% 1 $v" vr '
+
& r $" r (
! zz = " p + 2
#vz
#z
(A.4)
1 #vr +
( # $v
! r" = ! "r = * r % " &' +
) #r r
r #" -,
! "z = ! z" =
! rz = ! rz =
# "vr "vz %
+
$ "z
"r &
1
!" 1 !
( r"vr ) + ! ("v# ) + ! ( "vz ) = 0
+
r !#
!z
!t r !r
(A.5)
27
Appendix B
Density
Viscosity
Thermal
conductivity
Isothermal
compressibility
k
(W/m K)
Coefficient
of thermal
expansion
(1)
T
(m2/N)
Specific heat
at constant
pressure
cp
(J/kgK)
(kg/m3)
(kg/m s)
Helium
0.164*
1.92x10-5
0.150
3.41x10-3 *
1.00x10-5 *
5.21x103 *
Air
1.19*
1.98x10-5
0.0262
3.41x10-3 *
1.00x10-5 *
1.00x103 *
Water
1.00x103
1.00x10-3
0.597
1.8x10-4
4.6x10-10
4.18x103
Glycerin
(C3H803)
1.26x103
1.49
0.286
5.0x10-4
3.7x10-10
2.39x103
Mercury
1.36x104
1.55x10-3
8.69
1.82x10-4
0.40x10-10
1.39x102
The compressibility ! and the coefficient of thermal expansion ! are defined by the
equation
d!
= "dp # $dT
!
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
V0 = R +
3 cD
4 cL
Mg
0V0 A0
(5)
where A0 is the exit area of a jet. Equation (5) reduces to a quadratic equation for V0, which can be
solved straightforwardly. Putting the equation in the form (5) is useful since it shows at a glance
that the torque due to aerodynamic drag becomes negligible in the limit
3 cD
4 cL
Mg
<< R
0V0 A0
(6)
PART D
The power required for hovering flight is the torque exerted on the two rotor blades Eq.
(4)multiplied by O .
NOTE:
The above solution accounts for the dynamic effects of the jet exit velocity but neglects the effects
of the inflow into the jet engines. This is a reasonable approximation: the hot exhaust gases are
much less dense than the cool inlet air. Mass conservation implies that, with inlet and exit areas of
the same order of magnitude, both the velocity and the momentum flux will be much higher at the
exit than in the cool gas at the inlet.
P L = f ( , vo , R1, 1, 2 , R, )
(1)
Hence n = 8; r = 3; (n r) = 5 Pi groups. Note that only (any) two of the length scales can enter
as they are constrained by R1 = R - 8 . Gravity does not enter since there is no density contrast
to drive density waves at interface ( lP = P1 - P2 = 0 ). Pick as primary variables the average oil
2
R1
velocity vo = Qo 7 R12 (flow), core radius R1 (geometry) and density p . Obtain following Pi
groups:
P/L
vo2 R1
R
, 2 , ,
2
vo R1 vo R1 R1 vo R1
(2)
where the dimensionless groups may be recognized as, respectively, a friction factor, Reynolds
numbers for the inner and outer fluids, a geometric ratio and the Weber number (inertia/
capillarity) We = p vo2 R1 ( . Alternatively you may have picked the inner fluid viscosity as the
third primary variable (characterizing the flow), in which case you obtain:
MP / L
1vo R12
( P vo R1 2 R ( )
, , ,
1 1 R1 1vo
(3)
Where the last group is inverse of capillary number. Note that Ca = We Re . In general, the way
this problem would be written by inspection would be in terms of a friction factor as a
function of Reynolds number and other characteristic ratios:
f =
( Re1, 2
1 , R1 R,We)
(4)
Under flowing conditions, surface tension is the stabilizing force which resists the formation of
curved interfaces (i.e. this would lead to extra interfacial area which is energetically
unfavorable). Hence we require
We = vo2 R1
<< 1
(5)
b) The appropriate boundary condition for the shear stress on the interface r = R1 assuming the
interface is cylindrical can be written.
rz r= R
1
rz r= R +
1
or 1
vz
r
= 2
r= R1
vz
r
(6)
r= R1+
!
. This increase is
R1
negligible if the capillary pressure correction is small compared to the hydrostatic change from top
to bottom that we are also neglecting; i.e. ( R1 << p g(2R1 ) or ( (2 p gR12 ) << 1 . Even if this term
dp
dp
MP
is included, it does not change the conclusion that 2 = 1 = _
.
dz
dz
L
The capillary pressure increase across the interface gives p1 (z) = p2 (z) +
c) Assuming that the steady-state velocity field is v = [0, 0, vz (r)D z r and substituting into the
z-component of Navier-Stokes equations gives
0=-
dp 1 d ( dvz )
+
r
0 r R1
dz r dr
dr
(7a,b)
dp d ( dvz )
0=- + 2
r
R1 r R
dz
r dr
dr
Using these boundary conditions together with the condition that there is a finite interfacial
velocity given by vi = vz (r = R11 ) = vz (r = R1+ ) gives the following expressions for the fullydeveloped velocity field vz (r) :
vz(1) =
2
2
2
2
R1 r
+
R
R1
4 2 L
4 1 L
2
2
=
R1 r
+ vi
4 1 L
vz(2) =
1
R
2
4 2
L
r2
for 0
R1
(8a)
R.
(8b)
v z(2) (r = R1 ) = vz(1) (r = R1 ) = vi =
R1
(9)
R
vi
w =
1 ( P X 2
R - R12 ]
4 2 L
oil
2R ( P L )
water
R
R1
d) The flow in the pipeline is typically started impulsively by imposing a sudden increase in the
pressure gradient along the pipe,
2
2
2
and the flow takes a period of time
water
center of the pipe moves as a
Plug flow
plug until viscous effects diffuse
oil
r
near middle
in from the no-slip boundary.
2
z
t
t
t
1 R1
1
1
Engineering estimates of the total time taken for the flow field in each domain to reach steady
state are thus:
2
2
(R
R1 )
R12
R12
=
1
(10)
typically 2 << 1 whereas 8 , R1 might be only a factor of two or three different so the first
inequality dominates. The cross-over for these inequalities comes when 8 2 R12 > 2 1 .
e) The volume flow rate of each component is given by:
R
P
1 r
r
R12 r 2
+
R
2 R12
dr
2
L
0 1
2
P
2 2
P
4
R1
R1
R
R12
+
8 1
L
4 2
L
R1
Qo =
2 rvz(1)dr =
= R12 vi +
(11)
P
4
R1
8 1 L
The last line in eq.(11) shows that the oil flow rate consists of the usual Poiseuille result (second
term) PLUS an additional contribution from the core with a finite interfacial velocity.
Qw =
R 2
1
rvz(2)dr =
R
2
8 2
L
R12
(12)
Using these results, the total power dissipated by the pumping operation per unit length is:
2
( " !P % * 1 2
" !P %
!
WL = $
Qo + Qw ) = $
R ) R12
(
+
'
'
# L &
8 # L & , 2
2 2 2
1
R1 R ) R12 + R14 .
2
1
/
f) Differentiating the expression for Qo with respect to the core radius R1 gives:
0 =
Qo
2R
2 R1
R1 4 2
L
4R13
+
4R13
8 1
L
(13)
assuming the viscosities 1, 2 and density p are all held constant and that there is a fixed value
of the imposed pressure gradient MP L . Solving gives:
R1* =
[ 2 - 2
1 ]
1/2
(14)
for the optimal value of the core radius that maximizes the volume flow rate of oil through the
pipe. In the limit 2 1 << 1 this value approaches R1* " R 2 . A smaller core radius (2nd term
in eq. 11) cuts down on the amount of oil transported; a larger core means a smaller outer shell
and an increase in the dissipation within the water phase.
5
vi =
R
P
w
=
2
2
L
2
(15)
vi
water
oil
Alternately: note that when the gap is very small then the flow in the 0
R1 R
outer (water) phase becomes almost a Couette flow. To see this
linearize equation 8b using a new variable that starts at the wall ( y = R - r ) to get r 2 " R 2 - 2Ry
and thus
1 ( MP )
2Ry
4 2 L
dvz
dv
R ( MP )
= - 2 z =
dr
dy
2 L
and a maximum velocity at the edge of the Couette shell layer (i.e. at position y = ! ) given by
which again looks like a Couette flow introduced in class with T = 2
vz(2) (y = ) "
R ( MP )
=
Tw
2 2 L
2
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Final Examination
P, 2
P, 1
Qo
water
oil
R1
Figure 1: geometry
of a lubricated
pipeline
z
R
Use dimensional analysis to determine an appropriate dimensionless form for expressing the
dP MP
as a function of the other
fully-developed pressure drop per unit length in the pipe =
dz
L
relevant parameters in the problem. Use the average oil velocity vo = Qoil 7 R12 and core
radius R1 as two of your primary variables together with as many other parameters as you need.
Which dimensionless group is important in determining whether waves will develop. Based on
your physical understanding of interfacial processes, express an appropriate inequality on the
range for this dimensionless parameter in order for waves not to form.
b) Assuming that your criterion above is satisfied so that the flow in the pipe remains a perfect
smooth core-annular flow as shown in the sketch, write down the appropriate boundary
Final Examination
condition for the shear stress on the interface r = R1 . Furthermore, provide a criterion under
which the change in pressure across the interface is negligible.
c) Use these boundary conditions to find expressions for the fully-developed velocity field
vz (r) that are valid in the core domain 0 r R1 and the shell R1 r R. On a single large
graph (at least 0.5 page in size), sketch the velocity profile and the shear stress profile across
the entire pipe (i.e. for the region 0 r R).
d) The flow in the pipeline is typically started impulsively by imposing a sudden increase in the
pressure gradient along the pipe, and the flow takes a period of time to become fully
developed. Draw a large diagram and sketch the shape of the velocity field vz (r,t) as a
function of time. Provide an engineering estimate of the total time taken for the flow field to
reach steady state.
e) Find expressions for the volume flow rate of oil Qo and for the volume flow rate of water
Qw through the pipeline as a function of the imposed pressure MP and the other physical
parameters defined in the figure.
f) The results of your analysis can be used to optimize the lubricated pipeline operation. For
example; consider the viscosities 1, 2 and density p to all be held constant. Show that at
any fixed value of the imposed pressure gradient MP L there is an optimal value of the core
radius (denoted R1* ) that maximizes the volume flow rate of oil through the pipe. Derive an
expression for R1* and explain (very briefly) why this occurs.
NOT REQUIRED FOR EXAM (extra credit):
g) If the outer layer of fluid becomes very thin ( 8 << R ) and of very low (but still non-zero!)
viscosity ( 2 << 1 ) then a number of simplifying approximations can be made in the
governing equations. Show that in this limit the inner fluid can have a large velocity very
close to the wall which makes it appear to slip at the wall (i.e. at r = R1 z R ) with a slip
velocity vw that is proportional to the wall shear stress. Find the coefficient of
proportionality.
Navier Stokes equation in {r,e, z} coordinates:
p
[ dvr
dt
+ vr
[ d [ 1 d(rvr ) J 1 d2 vr d2 vr 2 dve .
dvr ve dvr ve2
dv J
dp
+ p gr
+
+ vz r = - +
+ 2
+ 2 - 2
dr
r de
r
dz
dr
dz
r de
r de 2
dr r dr
[ 1 d [ dvz J 1 d2 vz d2 vz .
dvz ve dvz
dvz J
dp
[ dvz
p
+ vr
+
+ vz
r
+ 2 2 + 2 + p gz
=- +
dt
dr
r de
dz
dr
dz
dz
r de
r dr
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2R1
liquid of
density and
2R2
viscosity
d/dt
(6.05a)
where is the shear stress and d/dt is the rate of change of the angle between two uid lines which at time
t are mutually perpendicular, the rate of change being measured by an observer sitting on the center of mass
of the uid particle.
(a) Show that in terms of streamline coordinates,
= (dV /dn V /R)
(6.05b)
where V is the resultant velocity, R is the radius of curvature of the streamline, and n is the outwardgoing normal to the streamline.
2.25 Advanced Fluid Mechanics
c 2012, MIT
Copyright
Viscous Flows
(b) A long, stationary tube of radius R1 is located concentrically inside of a hollow tube of inside radius
R2 , and the latter is rotated at constant angular speed . The annulus contains uid of viscosity .
Assuming laminar ow, and neglecting end eects, demonstrate that
P
4
=
2
2 R22
(R2/R1) 1
(6.05c)
where P is the power required to turn unit length of the hollow tube.
(c) Find the special form of (b) as R2 /R1 1, in terms of the gap width h = R2 R1 and the radius
R.
c 2012, MIT
Copyright
Viscous Flows
Solution:
(a) Consider two uid lines perpendicular to each other at time t for a particle at position P and select
one of the lines to be parallel to the velocity vector/streamline ds. The other line will be consequently
parallel to dn. Now track the particle till it reaches point P ' at time t + dt. Figure 2 shows the
mentioned geometry and depicts the possible deformations as the uid particle travels on a streamline.
from the denition we have the following:
S ( n + dn )
at time t : LBPA
S ( n)
at time t + dt : LDPC
A
C
LBPA = LBPA = 90
LBPD =
LAPC = LPOP =
Figure 2: Deformation for a particle traveling from P at time t to P I at time t + dt along a streamline.
d/dt
/dt /dt
(6.05d)
(6.05e)
(6.05h)
so we will have:
/dt =
V
R
(6.05i)
(6.05j)
Note: if you feel that the geometry relationships are hard to visualize try to solve this problem assuming
that your coordinate system is locally cylindrical with center O and your motion (locally) has only Vs
which is V so Vr = 0. Using the relationships for strain rate in cylindrical coordinates (you can nd
them in Kundu) you will get something exactly similar to the geometry proof.
2.25 Advanced Fluid Mechanics
c 2012, MIT
Copyright
Viscous Flows
(b) Consider the -component in the Navier Stokes equations for cylindrical coordinates:
1 1 P
V r V
V
+
+ (V . ) V +
=
r
t
r
in which:
V +
2 Vr
V
2
r2
r
+
+ Vz
r
r
z
1 2
r
+ 2 2+ 2
z
r
r
(6.05k)
(V . ) = Vr
(6.05l)
1
r r
(6.05m)
and
2
Now if you consider that / = 0 due to axisymmetry in the problem and also note that Vr = Vz = 0
then (k) simplies to
1
V
V
0=
r
2
(6.05n)
r r
r
r
thus:
r2
V
2 V
+r
V = 0
r
r2
(6.05o)
r2
dV
rV
dr
=0
r2
dV
rV
dr
= const.
(6.05p)
const
.
r3
(6.05q)
V
r
c1
c1
V
= 2 + c 2 V =
+ c2 r .
r
r
r
(6.05r)
in which c1 and c2 are two integration constants which need to be determined from boundary conditions.
Boundary conditions for this steady problem are:
at r = R1 : V = 0
(6.05s)
at r = R2 : V = R2
Satisfying the boundary conditions one will get the following values for c1 and c2 :
R2 R22
c1 = R21R
2
2
1
c2 =
R22
R22 R12
(6.05t)
R2 R12
(R22 R12 ) r
R2
(6.05u)
Note that at the limit of R1 0 the velocity eld becomes exactly similar to solid body rotation i.e.,
V = r.
2R12 R22
(R22 R12 ) r2
(6.05v)
c 2012, MIT
Copyright
Viscous Flows
thus at r = R2 :
=
2R12
(R22 R12 )
(6.05w)
(6.05x)
(6.05y)
(c) Now in the limit where R2 /R1 1 we have R2 = R1 + h R2 /R1 = 1 + h/R in which
R = R1 . If we plug this in the relationship for stress (v) we can easily show that at the limit of
R2 /R1 1 R1 R2 R we will have::
=[
2R22
R 2R22
R
] 1
2
2
2h
r
h
(1 + h/R) 1 r2
(6.05z)
Notice that (z) shows that at the limit of R2 /R1 1 the shear stress is almost constant in the gap and
its values is very close to what we had in simple plane Couette ow. This fact plus the benets of circular
and long geometries are the main ideas for making rheometers out of similar geometries
(Taylor-Couette
(
)
geometries). Plugging the value from (z) in the relationship for P leads to P/ 2 R2 = 2R/h
D
c 2012, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Consider a steady, fully developed laminar ow in an annulus with inside radius R2 and outside radius R1 .
(a) Find a relation between the pressure gradient
2
and R
R1 .
dp
dx ,
(b) Fin the limiting form of the relation for a very thin annulus by expressing it in terms of R1 and
h
h
R1 , where h = R1 R2 , and taking the limit R1 0. Compare with the formula for fully developed
laminar ow between parallel at plates separated by a distance h.
2
(c) In the opposite limit R
R1 0, does the relation of (a) reduce to the formula for Hagen-Poiseuille
ow in a circular pipe of radius R1 ? Discuss your answer.
c 2008, MIT
Copyright
Viscous Flows
Solution:
(6.04aa)
where the rst term is approximately a constant across the space between the cylinders (long cylinder
approximation), then
(
)
1
vx
0 = K +
r
,
(6.04ab)
r r
r
then, integrating,
Krdr =
(
)
vx
r
dr,
r
r
r2
vx
+ C1 = r
,
2
r
r
C1
vx
+
=
.
r
2
r
(6.04ac)
(K
r
C1
+
)dr =
2
r
vx
)dr,
r
r2
+ C1 ln(r) + C2 = vx ,
4
(6.04ad)
vx (R1 ) = 0, vx (R2 ) = 0,
(6.04ae)
R2
R12
+ C1 ln(R1 ) + C2 = 0, OR K 2 + C1 ln(R2 ) + C2 = 0.
4
4
(6.04af)
K 2
R1
= 0,
(R R22 ) + C1 ln
R2
4 1
(6.04ag)
then,
C1 =
K
2
4 (R1
ln
R22 )
R1
R2
(6.04ah)
K
4 (1
2 )
,
ln
C1 = R12
K
2
4 (
1)
,
ln
(6.04ai)
C1 ln(R2 ).
(6.04aj)
where, = R2 /R1 .
C2 = K
2.25 Advanced Fluid Mechanics
R12
C1 ln(R1 ),
4
C2 = K
2
R22
4
c 2008, MIT
Copyright
Viscous Flows
Upon substitution of C1 ,
C2 = K
R12
+
4
K
2
4 (R1
ln
R22 )
R1
R2
ln(R1 ),
C2 = K
R22
4
K
4
(R12 R22 )
ln
R1
R2
ln(R2 ),
(6.04ak)
simplifying,
)
(R12 R22 )
ln(R
)
,
1
1
ln R
R2
)
(
K
(R2 R2 )
C2 =
R22 + 1 R1 2 ln(R2 ) .
4
ln R2
C2 =
K
4
R12 +
(6.04al)
(6.04am)
[
( )]
1 dp 2
R12 R22
r
r R22
ln
R2
4 dx
ln(R1 /R2 )
(6.04an)
R1
R2
R1
vx rdrd =
R2
R1
R2
)
r2
+ C1 ln(r) + C2 rdrd = 2
4
)
r2
K + C1 ln(r) + C2 rdrd,
4
R1
R2
(6.04ao)
)
r3
+ C1 r ln(r) + C2 r dr
4
(6.04ap)
After integration,
2
R1
R2
)
(
[
])
r3
Kr4
r2
r2
1
+ C2 + C1
ln(r)
+ C1 r ln(r) + C2 r dr = 2
4
16
2
2
2
R1
(6.04aq)
R2
then, nally,
Q = 2
[
]
[
])
R2
(R2 R22 )
1
R2
1
K(R14 R24 )
+ C2 1
+ C1 1 ln(R1 )
C1 2 ln(R2 )
.
16
2
2
2
2
2
(6.04ar)
R4
R2 R2
R2
(R22 R12 )2
R14
+ 2 1 2 (R12 R22 ) 1 +
4
4
2
2
4 ln(R1 /R2 )
(6.04as)
C2 =
KR12
4
1+
)
1 2
ln(R1 ) ,
ln
C2 =
KR12
4
1+
2 1
ln
)
ln(R1 ) .
(6.04at)
KR14
2
(2 1)2
4
1
ln
(2 1)
2
(6.04au)
c 2008, MIT
Copyright
Viscous Flows
c 2008, MIT
Copyright
Viscous Flows
Problem 6.04b
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Solution:
Now, factorizing taking into account that R1 = R2 + H, then R2 = R1 H,
Q=
K
8
1
1
2(R1 H)2 R12 + (R1 H)4 ln( R1RH
) + (R1 H)4 R14 ln( R1RH
) + R14
1
ln( R1RH
)
(6.04ba)
R1
R1
4
4
4
2(R1 F R1 )2 R12 + (R1 F R1 )4 ln( R1 F
R1 ) + (R1 F R1 ) R1 ln( R1 F R1 ) + R1
R1
ln( R1 F
R1 )
)
(6.04bb)
Now, taking the limit as F 0, but keeping the higher order terms,
Q=
2 4
R KF 3 ,
12 1
(6.04bc)
Q=
1 dP
R1 H 3 ,
6 dx
H3
12
dP
dx
(2R1 ),
(6.04bd)
(6.04be)
which corresponds to the solution of a pressure driven ow between two plates separated by a distance
H, over a length equal to the average circumference of the annulus.
NOTE: SEE PLOTS OF THE SOLUTIONS USING THE ATTACHED MATLAB FILES, PLAY
WITH THE SOLUTIONS TILL THE LIMITS MAKE SENSE TO YOU.
c 2008, MIT
Copyright
Viscous Flows
c 2008, MIT
Copyright
Viscous Flows
Problem 6.04c
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Solution:
(
)
R13 dP
lim Q =
(2R1 ),
0
16 dx
(6.04ca)
(6.04cb)
(6.04cc)
(6.04cd)
which is the solution for Poiseuille ow for a simple tube. You may have guessed that the solution did
not converge to this value, i.e. the velocity prole had a hole in the center, but this is wrong. The
solution converges to the simple tube ow because as the inner cylinder becomes smaller, the area that
it uses to transmit vorticity decreases, and as the area decreases, its inuence decreases too (Think of
a small string (hot wire) inside the tube for measuring ow, and think how small are the disturbances
that it creates in the ow).
To further verify that the solution makes physical sense, lets look at the product r viscous to show
that the viscous force per unit length decreases as r 0. Using the velocity prole, the viscous stress
can be obtained,
)
(
dvx
R12 R22 1
= K 2r
,
dr
ln(R1 /R2 ) r
(6.04ce)
&
(
)
dvx &&
R12 R22
2
=
K
2R
,
2
ln(R1 /R2 )
dr &R2
lim R2
R2 0
&
dvx &&
= lim K(2R22 ) = 0,
dr &R2 R2 0
(6.04cf)
(6.04cg)
NOTE: SEE PLOTS OF THE SOLUTIONS USING THE ATTACHED MATLAB FILES, PLAY
WITH THE SOLUTIONS TILL THE LIMITS MAKE SENSE TO YOU.
c 2008, MIT
Copyright
Viscous Flows
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider a steady, fully developed laminar ow in an annulus with inside radius R2 and outside radius R1 .
(a) Find a relation between the pressure gradient
2
and R
R1 .
dp
dx
,
(b) Find the limiting form of the relation for a very thin annulus by expressing it in terms of R1 and
h
h
R1 , where h = R1 R2 , and taking the limit R1 0. Compare with the formula for fully developed
laminar ow between parallel at plates separated by a distance h.
2
(c) In the opposite limit R
R1 0, does the relation of (a) reduce to the formula for Hagen-Poiseuille
ow in a circular pipe of radius R1 ? Discuss your answer.
c 2011, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Consider a gas bubble of xed mass and radius R(t) which is expanding or contracting in an innite sea of
incompressible liquid. The speed of the interface is dR/dt. The local Eulerian coordinate in the liquid is r.
Let pR , p, and p be, respectively the pressure at r = R (on the liquid side of the interface), at r = r, and
at r = .
(a) Determine the viscous contribution to the normal stress rr in the liquid.
(b) Show that the dimensionless overpressure, (pR p )/(dR/dt)2 , is independent of whether the uid
is viscous or inviscid.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider a gas bubble of xed mass and radius R(t) which is expanding or contracting in an innite sea of
incompressible liquid. The speed of the interface is dR/dt. The local Eulerian coordinate in the liquid is r.
Let pR , p, and p be, respectively the pressure at r = R (on the liquid side of the interface), at r = r, and
at r = .
(a) Determine the viscous contribution to the normal stress rr in the liquid.
(b) Show that the dimensionless overpressure, (pR p )/(dR/dt)2 , is independent of whether the uid
is viscous or inviscid.
c 2010, MIT
Copyright
Solution:
rr
rr
(a) First, we must determine the velocity eld in the liquid at any point in time. We choose a control
volume taking the shape of a hollow sphere with inner control surface at radius R(t), which moves
outward at exactly the rate of expansion of the bubble dR/dt, and outer surface at an arbitrary radius
r, such that its volume is V = 43 (r3 R(t)3 ). Using Form A of the integral mass conservation equation,
d
dt
(u uCS ) ndA
= 0
dV +
CV
(8.13a)
CS
R2 dR
r2 dt
(8.13b)
Using Eq. (8.13b) we can determine the average rate of strain from the following equations
ur
rr = 2
r
(8.13c)
and
ur
1 u
+
r
r
(8.13d)
ur
u cot
1 u
+
+
r sin
r
r
(8.13e)
= 2
and
= 2
R2 dR
r3 dt
(8.13f)
c 2010, MIT
Copyright
and
= 2
R2 dR
r3 dt
(8.13g)
= 2
R2 dR
r3 dt
(8.13h)
and
For a Newtonian fluid, ij = ij , where is the dynamic viscosity. Accordingly the normal stresses
for this flow are
rr = 4
R2 dR
r3 dt
(8.13i)
and
= 2
R2 dR
r3 dt
(8.13j)
= 2
R2 dR
r3 dt
(8.13k)
and
(b) The complete equation of motion in the radial direction for spherical coordinates is
!
u2 + u2
ur
ur
u ur
u ur
+ ur
+
+
(8.13l)
t
r
r
r sin
r
"
#
1 2
1
1 r
+
p
= 2 (r rr ) +
(r sin ) +
+ gr (8.13m)
r r
r sin
r sin
r
r
Neglecting gravity and retaining only the non-zero terms, we have
ur
ur
+ ur
t
r
"
#
1 2
+
p
= 2 (r rr ) +
r r
r
r
(8.13n)
Substituting Eq. (8.13i), (8.13j) and (8.13k) into Eq. (8.13n) to evaluate the net contribution of viscous
stresses acting on a fluid element, we obtain
ur
ur
+ ur
t
r
"
=
4
2
r r
R2 dR
r dt
#
R2 dR
p
4 4
r dt
r
(8.13o)
When we differentiate this term, we find that the net contribution of viscous stresses acting radially
is exactly zero and hence there is no net dissipation associated with this flow. Consequently, the
governing equation is
ur
ur
+ ur
t
r
!
+
p
=0
r
(8.13p)
c 2010, MIT
Copyright
Integrating Eq. (8.13p) along dr, which is indeed the streamline coordinate, we obtain the unsteady
Bernoulli equation
r=
r=R
ur 0
dr +
t
ur, =0
ur = dR
dt
u0r du0r +
dp0 = 0
(8.13q)
pR
which is
Z
r=
r=R
R 2 d2 R 0 1
dR
dr
r02 dt2
2
dt
!2
+ p pR = 0
(8.13r)
which gives
d2 R 1
dR
R 2
dr
2
dt
!2
+ p pR = 0
(8.13s)
2R ddtR
pR p
2
2 = 2 1
1
dR
dR
2 dt
dt
(8.13t)
So we have shown that the dimensionless overpressure is indeed independent of whether the fluid is
viscous or inviscid. Note that the dimensionless overpressure can be positive or negative depending on
the rate of change of the surface velocity of the gas bubble.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
"p
0=!
"y
Form A:
" !vx !v y %
$# !x + !y '& = 0
Form B:
h
!h !Q"
+
= 0 where Q! = " vx! dy
0
!t
!x
2
!h $ " gh sin # ' !h " g cos # ! * 3 !h +&
) =
,h
/
!t %
3 !x + !x .
( !x
0!
dt
cv
Incompressible Flow
!"v = 0
"p
+ # gy
"y
Inviscid fluid Re ! !
(Euler Equation)
Dv
!
= "#p + !g
Dt
! =0
! =0
Hydrostatics
!p = "(g # a)
Across streamlines
(outward pointing normal n)
!p " v 2
=
R
!n
Unsteady Bernoulli Equation along a streamline
2
# !v " dr +
%
$ !t
1
! = $ " # dA = !
$ v # ds
L = !V"
D=0
Young-Laplace Equation
# 1
1 &
!p = " % + (
$ R1 R2 '
Kutta-Zhukowski Theorem
Gareth H. McKinley
December 2006
!"
=#
% dV + ( ( &b)v ' n dA
CS
$ !t
#2 v
#p
+ 2x + $ gx
#x
#y
dBsys
The Equations of
Fluid Mechanics
d
" !dV + "CS (#b) ( v $ vc ) % n dA
dt cv
Continuity
1 D!
= " (# $ v )
! Dt
0!"
dt
dBsys
# dp
1 2
1 2
+ ( (2 & (1 ) = 0
2 v 2 & 2 v1 + %
$ '
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25 Quiz 1
Fall 2013
Fall 2013
QUIZ 1
OF EQUAL WEIGHT
You may use the course textbook (Kundu or Panton), a binder containing your class notes,
recitation problems and ONE page of handwritten notes summarizing the key equations.
2.25 Quiz 1
Fall 2013
2.25 Quiz 1
Fall 2013
is constant everywhere
throughout the flow field, not only along a streamline. Consider a Cartesian coordinate
system and let gravity act in the negative y direction.
A liquid flows steadily along an open channel of rectangular cross section having a uniform
width (ro ri). Viscous and compressibility effects are negligible. At station (1), where the flow
enters a circular bend, the depth of the flow is uniformly equal to h1 and the velocity magnitude
is uniformly equal to v1. We assume that the streamlines are straight and parallel upstream of the
bend and that in the bend the streamlines are approximately circular arcs having a common
center at the origin, 0. Note that the depth h in the bend will depend on r because the free
surface is exposed to the uniform pressure of the atmosphere.
(c) [2] Show that the velocity in the bend is independent of both and z.
(d) [2] Argue that the Bernoulli constant is the same for all streamlines. Extend this result to
show that the velocity in the bend is given by
(1)
where K is a constant.
2.25 Quiz 1
Fall 2013
(e) [2] Show that the shape of the liquids free surface in the bend is given by
(2)
(f) [2] Show, by applying an appropriate control volume theorem, that the constant K in (1)
must satisfy
(3)
Finally, consider steady rigid-body rotation of a viscous liquid at an angular velocity :
(4)
The fluid is in a circular bucket and the fluid height in the center (r = 0) is H. The free surface of
the liquid is exposed to the uniform pressure of the atmosphere.
(g) [1] Explain why Bernoullis equation is applicable to this case, given that the liquid has a
nonzero viscosity.
(h) [2] Show that the Bernoulli constant Bsb for this case does not depend on z. Show that Bsb
does depend on r and determine its value for a given r. Explain why Bsb varies in this
case.
2.25 Quiz 1
Fall 2013
We will represent the free surface height as h(x) and the velocity by the function u(x). The fluid
has constant density and we will treat the problem as one-dimensional. You can assume that
viscous stresses along the control surfaces of the volume shown above are negligibly small, and
neglect the density of air.
PART I:
a) [2 points] consider a streamline drawn (line AB in the figure) just above the smooth flat
lower surface of the channel. How is the static pressure in the fluid along this line related
to the height of the river? How does the static pressure vary along the line DEA?
b) [4 points] Using the control volume shown in the sketch develop two expressions that
relate the velocity and height of the stream at station 1 and the velocity and height of the
stream at station 2. Developing a table of relevant quantities along each face of the
control volume ABCDEA is highly recommended!
c) [2 points] Combine your expressions from (a) and (b) together to show that the speed of
the river can be simply evaluated from simple measurements of the river height (e.g.
using marked yardsticks attached to the channel floor):
gh2
u1 =
(1)
( h1 + h2 )
2h1
Part II: A deeper question to answer is why is the water moving so fast locally to begin with. To
answer this we must consider the topography of the river bed that is upstream of station 1, as
shown in the drawing below. We denote the height of the fluid stream above the river bed as h(x)
and the height of the riverbed by b(x):
d) [1 point] Consider a slice of river dx and show that conservation of mass can be written in
the form:
dh(x)
du(x)
+ h(x)
u(x)
= 0 (2)
dx
dx
2.25 Quiz 1
Fall 2013
e) [2 points] As the riverbed elevation rises (or the flow goes over a bump such as a
submerged tree, (or house!) etc.) so that b(x) increases in the flow direction as shown
what should happen to the height of the free surface h(x)? If viscous effects are
negligible in this upstream region, use the Bernoulli equation along a streamline at the
free surface to obtain an additional differential expression involving U(x), h(x) and b(x).
f) [2 points] Combine your equations from (d) and (e) to show that the velocity of the river,
the height of the surface, and the underlying topography are all inter-related by the
differential equation:
1 du(x)
db(x)
u(x)2 gh(x) + g
=0
u(x) dx
dx
(3)
g) [2 points] Finally, consider the specific point along the flow direction xm at which the
riverbed height b(xm ) locally reaches a maximum, so that db dx x = 0 as shown in the
m
figure above. Describe very briefly in words happens to the velocity (and sketch the free
surface) just beyond this bump in the river for the two different cases given by:
i) u(xm )2 gh(xm ) < 0 ?
and
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25 Quiz 2
Fall 2013
Fall 2013
QUIZ 2
OF EQUAL WEIGHT
You may use the course textbook (Kundu or Panton), a binder containing your class notes,
recitation problems and TWO pages of handwritten notes summarizing the key equations.
2.25 Quiz 2
Fall 2013
2.25 Quiz 2
Fall 2013
It is important, but not always straightforward, to be able to measure the EOF-driven flow rate in
microchannels. One apparatus to do this is depicted in the figure above. A small, cylindrical
capillary tube with diameter Dc and length Lc connects two wells. The wells and capillary tube
are filled with an aqueous solution having density and viscosity . An electric field pointing
from right to left drives EOF in the same direction at a uniform velocity veo. The right well is
open to the atmosphere and the left well is connected to a needle having diameter Dn and length
Ln. The needle feeds into a microsyringe with a smaller diameter Dm, which is open to the
atmosphere. Lm is the length of the microsyringe that is filled with fluid. The entire system is
kept horizontal to eliminate the effects of gravity.
As fluid accumulates in the left well, over time a pressure difference builds up between
the two wells that drives fluid from left to right. This back-flow has an average velocity given
by
At the same time, fluid is driven upward into the needle and microsyringe, where the
average velocities are
respectively.
In the following, neglect the effects of surface tension (i.e. assume the meniscus at the
air/liquid interface in the microsyringe is perfectly horizontal) and assume steady, fully
developed, incompressible flow everywhere in the capillary tube.
(a) [2 points] Show that the velocities in the capillary tube, needle, and microsyringe are
related by
(1)
(b) [3 points] Determine the velocity profile in the capillary tube at a point near the middle of
the tube (far from either end), in terms of an unknown pressure gradient and
Clearly
3
2.25 Quiz 2
Fall 2013
list your assumptions and boundary conditions. Sketch the velocity profile for two
different magnitudes of the pressure gradient (you dont need to specify the values just
one stronger, one weaker).
(c) [2 points] Neglecting entrance effects at the junction between any two pieces of the
device, show that
(2)
The utility of this experimental setup is that it enables one to indirectly measure the electroosmotic velocity by directly measuring the average meniscus velocity
and using
mathematical manipulations of (1) and (2) to convert this result to veo.
is relatively easy to
measure by fixing a camera on the microsyringe and tracking the position of the meniscus over
time.
(d) [1 point] Show that the needle and microsyringe can be lumped together as a single tube
with average fluid velocity
and diameter Dm if we assume its effective length is
(e) [3 points] Show that the electro-osmotic velocity is related to the average meniscus
velocity and the average pressure-driven back-flow velocity by
(3)
(4)
(f) [2 points] Determine the maximum pressure difference that the capillary can support, in
terms of the electro-osmotic flow velocity veo.
(g) [2 points] How does the meniscus height scale with time at short times? That is, find n in
the expression
Assume here that the situation is quasi-steady, i.e. that
everything you have derived above for the steady case is still valid. There is no need to
use the unsteady momentum or the unsteady Bernoulli equation.
2.25 Quiz 2
Fall 2013
Motor
Motor &
Rotating Be
ll
Spraypainted part
Bell Cup
2.25 Quiz 2
Fall 2013
the density and surface tension of the paint as two of your primary variables.
b) [2 points] Write down the conservation of mass in spherical polar coordinates and show
that a velocity vector v = [vr (r, ), v (r, ), r sin 0 ] consisting of a two-dimensional
axisymmetric velocity profile in the film, combined with a constant angular velocity
throughout the film (at each value of r) v = r sin 0 is able to satisfy the continuity
equation. Make a scaling argument to determine if v is small or large compared to vr ?
c) [4 points] Now consider the radial component of the conservation of momentum equation
for the paint film on the rapidly-spinning conical applicator shown in the figure above.
Because the layer is thin (so that the value of doesn't vary much), it can be argued that
it is valid to simplify the angular dependence of the radial velocity component, so that
sin sin 0 everywhere. Show that the radial component of the equation of motion can
be simplified to the following form for a very viscous thin film of paint:
1 2 vr 2 r 2
sin 0 0
+
r 2 2
(2.1)
Give three appropriate constraints that identify when this approximation is valid.
d) [1 point] Give appropriate boundary conditions at the free surface between the paint and
the air and at the rigid wall between the cup and the paint.
[Hint: To do this, it is easiest to sit in the rapidly and steady rotating frame of the cup
with angular velocity v r sin 0 = ; you can then focus just on the radial
(outwards) velocity towards the edge of the bell].
e) [2 points] Find an expression for the radial velocity profile in the fluid film.
[Hint: you may find it useful to define a new coordinate as shown in the figure with
y = r sin ( 0 ) and also recall that the function sin x x + O(x 3 ) for x << 1.
f) [2 points] Integrate your expression for the velocity field through each small annular slice
of area dA = 2 (r sin 0 )dy to show that the constant volume flow rate Q of paint and the
thickness h(r) of the paint film are related by the expression
1/3
3Q
(2.2)
h(r) =
2 2
3
2 r sin 0
g) [1 point] In the laboratory reference frame, fluid streamlines and particle pathlines actually
follow spiral trajectories as the paint flows outwards of the conical bell cup, because of the
combined radial and azimuthal (i.e. rotational) flow components. Use the definition of a
streamline to show that the expression for the spiral trajectory of a material point rs ( )
(such as a small bubble or particle) that is on the paint surface is given by:
drs
= Ars 1/3
(2.3)
d
and give an expression for the constant A.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Fall 2013
2rj
g
uj
, , ,u j
h(r)
u
r1
r=R
Figure 1: Viscous Savart Sheet. Image courtesy: Villermaux et. al. [1]. This kind of
geometry was rst studied by Felix Savart in 1833.
(a):
We know that the following independent properties are important for determining the value of
R: radius of the disk (r1 ), radius of the jet (rj ), viscosity (), density (), surface tension (),
velocity of the jet (uj ), and gravity (g), so:
R = f (r1 , rj , , , , uj , g)
thus, n = 8 and knowing that there are three dimensions involved in these parameters ([M ],[L],
and [T ] r = 3) we can conclude that there are 5 dimensionless groups. We select r1 ,vj , and
as the repeating parameters and using Buckingham-Pi theorem following dimensionless groups
will be identied:
1 = R/r1
2 = rj /r1
3 = uj /
4 = gr1 /u2j
5 = u2j r1 /
So the relationship for R can be written in the following dimensionless form:
rj uj gr1 u2j r1
R/r1 = f
,
,
,
r1 u2j
(1)
One can easily see that 3 is the Capillary number (comparing the jet velocity with visco
capillary velocity scale) and 4 is the inverse of Froude number (comparing the gravity and
inertia). The other dimensionless number found in this problem, (5 ) is indeed the Weber
Fall 2013
number comparing the inertia stresses with capillary pressure. Thus, in order to ignore the
eects of gravity and viscosity we have to satisfy the following1:
gr1
<< 1
u2j
Ca
for ignoring viscosity: Re
<< 1
We
for ignoring gravity: F r1
2 Satisfying
r1
(2)
(3)
(b):
From conservation of mass written for the dashed control volume in Figure 2 we will have (note
that since viscous losses are ignored in this part, by Bernoulli, we can say that velocity at the
edge of the disk is equal to the jet velocity i.e. along a streamline the velocity is unchanged.):
(4)
2rj
proper
s
uj
, , ,u j
h(r)
u
r1
r=R
Figure 2: Simplied geometry with the selected control volume in dashed green line.
For calculating the thrust force on the disc we need to apply the conservation of linear momen
tum for y-component, lets assume that the disc is exerting F ey to the control volume:
Momentum. Cons.: F ey = vj2 (rj2 )ey
1
Another approach to do the dimensional analysis is picking , , r1 as the repeating parameters and that
will lead to R/r1 = g(rj /r1 , Oh = / r1 , Bo = gr12 /, W e = vj2 r1 /) and the criteria for ignoring gravity
1
and viscosity will respectively be F r
= Bo/W e << 1 and Oh << 1; the physical reason that F r1 << 1
means that gravity is negligible is the fact that inertia is ghting against gravity to keep the sheet at rather
than capillarity so the F r1 << 1 is a physically more meaningful criteria for ignoring gravity eects compared
to Bo << 1.
2
Stating the argument for neglecting the viscous eects by picking the Oh number or the Ca number is
considered as the right answer but the most physical one is the Re << 1
Fall 2013
thus an equal and opposite thrust force will be exerted from the liquid to the disc, i.e.
FT ey = vj2 (rj2 )ey .
(c):
For nding a relationship for the value of R we have to consider a slice of the rim as our control
volume (Figure 3) . The x-component of linear momentum enters the control volume at 1 but
vanishes as the drops have zero/negligible velocity in the x-direction. This must happen from
the force acting on the uid in the control volume thus the conservation of linear momentum
for the selected C.V. will be:
W h(R)U 2 = 2W
substituting U = uj and h(R) = rj2 /2R from conservation of mass:
u2j rj2 /(2R)
u2j rj
R
= 2
=
4
rj
(5)
u2j r1
R
=
4
r1
rj
r1
which matches well with our predictions from dimensional analysis (equation 3).
F = 2 R
2
2 Rh(R)U
F = 2 R
r~R
Figure 3: A slice of the sheet rim picked as our control volume (with h << R) and length taken
to be W = Rd.
(d):
Using the boundary layer scaling for diusion of momentum away from the plate we have the
following:
t r1 /uj
and using the fact that h1 (inviscid) = rj2 /2r1 we will have:
=
h1 (inviscid)
4r13
uj rj4
! 1/2
2
Rej
r1
rj
3/2
(6)
Fall 2013
=f
!
u2j
rj
, Oh
,We
r1
(/r1 )
r1
3/2
r1
rj
If viscous eects are important then the momentum of the incoming jet will get dissipated in
the boundary layer and thus the liquid sheet emerging from the jet will be slower and thicker.
(e):
At the end of the disc there will be loss of momentum compared to the initial ux from the
impinging jet due to the losses in the boundary layer. Through the concept of momentum
thickness we know that the loss in the radial momentum at the edge of the disk should be:
(7)
where h(r)inviscid = rj2 /2r is the variation in the sheet thickness observed in the inviscid case.
U
h(r)
(r)
r = r1
Fall 2013
(f):
ur
1 u
+
= 2
r
r
knowing that by axisymmetry the rst term on the right hand side is zero (there is no variation
in the direction):
U
= 2
r
now we can compare the scale of inertial terms with the viscous terms in the liquid sheet:
Ur
inertia
(1/2)U 2
=
viscous
2U/r
4
if this ratio is larger than unity then viscous terms in the sheet are dominated by inertia terms.
(g):
We can repeat the same argument we had in part (c) but only this time we need to use new
(8)
rj
r1
2
(1 )
(9)
Oh1/2
1
W e1/4
r1
rj
3/2 !
(10)
which is exactly in accordance to what we expected from dimensional analysis in the absence
(h):
By putting crit = R and using the result in equation (8) one can get:
R=
u2j rj2 (1 )
2 u4j rj2 (1 )3
10
40 2
10
=
4
2
a /
a uj2 (1 )2
a u2j (1 )2
40/
(11)
in which a / and W ecrit u2j rj /. In the inviscid case for water in the air (a / =
1.2/1000) we get W ecrit 900. As the eects of viscosity get larger and larger ( = 0) the
critical Weber number shifts to higher values (Equation 11) and the rim radius at a given value
of the Weber number is reduced, from equation (8). Interestingly because the dependence of
equation (11) on is stronger than equation (8) this means the viscous Savart sheet actually
extends a little bit further out radially before it starts apping (Figure 5).
5
Fall 2013
References
[1] E. Villermaux, V. Pistre, and H. Lhuissier. The viscous Savart sheet. Journal of Fluid
Mechanics, 730:607625, August 2013.
L
Dimensions of w :
T
L
Dimensions of U :
T
Dimensions of a 2 and z 2: L2
L2 L
L2
T T
2 1
1
2
y/a
0.5
0
-0.5
-1
-1.5
-2
-4
-3
-2
-1
x/a
Figure 1: A holiday-themed plot of velocity potential contours (green) and streamlines (red) for potential flow
around a flat plate of length 2a and negligible thickness. Lines of constant potential and streamlines are
everywhere perpendicular, except at the stagnation point at the origin. Flow is from left to right. Note: the
vertical streamlines at x = 0 are not physical; they are artifacts of discontinuities in the stream function at
x = 0.
U x iy
U z
dw
vx ivy
dz
z 2 a2
x 2ixy y 2 a 2
(1.1)
To find vx along the centerline, set y = 0 and isolate the real part of dw / dz:
vx y 0
U x
x2 a2
(1.2)
vx y 0
U x
x2 a2
U x
x2 a2
x0
(1.3)
x0
Similarly, the y-direction slip velocity on the plate is determined by setting x = 0 and looking at
the imaginary part of dw / dz:
vy x 0
U y
a2 y2
(1.4)
2
2
a y
v y x 0, a y a
U y ,
a2 y2
(1.5)
Figure 2 shows the signs of the velocity components in each of the four quadrants.
II
vx > 0, vy > 0
I
vx > 0, vy < 0
III
vx > 0, vy < 0
IV
vx > 0, vy > 0
ABCD
v dr
v dr v dr v dr v dr
AB
BC
CD
(1.6)
DA
where v is the velocity vector and dr is a differential line segment along ABCD. The integrals
along BC and DA vanish because the plate is infinitely thin. The integrals along AB and CD
reduce to
U ydy
a2 y2
v dr v dr
AB
CD
2U
ydy
a2 y2
U y dy
a2 y2
(1.7)
This integral may be evaluated by, for example, making the substitution u a 2 y 2 , du 2 ydy
to obtain
2U a.
(1.8)
Curve II
One can also integrate along a rectangle that includes the entire half-domain y > 0. This
rectangle extends to infinity in the +x, +y, and x directions and coincides with the x-axis. Since
the fluid contains no vorticity flux (it is a potential flow, which by definition is irrotational), all
of the vorticity within curve II is contained in the top half of the plate. Thus, the answer for
circulation should be the same as for curve I.
Figure 4: The desired circulation may also be computed by integrating around the entire top half of the domain. The
result for circulation is the same because the flow around the plate is irrotational, and so the entire vorticity flux is
bound into the top half of the plate.
In this case, the y-components of velocity along the left and right segments of the curve are zero
(since v Ue x as distance from the plate approaches infinity). The circulation becomes
U dx U
x
dx U
1 dx
2
2
x a
x a
(1.9)
x
2U
1 dx 2U lim x 2 a 2 x 2U a.
2
2
0
L
0
x a
(1.10)
Since the flow is symmetric about the x-axis, the circulation around the bottom half of the plate
is equal and opposite to (1.8). Thus the net circulation around the entire plate is zero.
10
bottom top 2U a.
(1.11)
The vorticity may be computed from Stokes Theorem. Consider the bottom vortex, which has
positive circulation:
bottom 2U a z dA z R 2 z ,bottom z ,top
A
2U a
.
R2
(1.12)
z
2
Ua
r.
R2
(1.13)
Ua
.
R
(1.14)
Evaluating at r = R,
v r R
One can also arrive at this answer by assuming the flow field outside the vortex to be that of an
irrotational vortex v / 2 r , setting r = R, and using the result from part (e):
v r R
2U a U a
.
2 R 2 R R
(1.15)
The pressure is obtained from applying the Bernoulli equation between a point at infinity (where
the fluid is stationary) and the edge of the vortex. Recall that the Bernoulli equation may be
applied between any two points since the flow outside the vortices is irrotational.
Since there are no body forces to consider, the Bernoulli equation reads
11
1
2
0
2
fluid is stagnant
at infinity in the
laboratory reference
frame
1
p r R v2 r R
2
(1.16)
Note that we are considering each vortex to be isolated from the other, i.e. the velocity field
induced by the other vortex is ignored in formulating (1.16); we will justify this assumption later.
Solving for the desired pressure,
2
1 U a
p r R p .
2 R
(1.17)
U a
, r R.
2 r
r
(1.18)
Each vortex will induce the other one to move. Each vortex core moves at a speed given by
evaluating (1.18) at r = H:
v r H
U a
.
H
(1.19)
In the laboratory reference frame, the vortices propagate to the left. Thus,
V
Ua
from right to left.
H
(1.20)
Equation (1.20) justifies the assumption made in part (f) that the velocity field induced by the
other vortex may be neglected when computing the tangential velocity at the edge of the vortex
(r = R). From part (f), the tangential velocity at the edge of the vortex scales as 1/R. Equation
(1.20) shows that the contribution to the tangential velocity due to the other vortex will scale as
1/H. Since H
12
v r R .
v
1
rv .
t
r r r
(1.21)
Axisymmetric / 0
Unidirectional vr vz 0
v r , t 0
.
2 r
(1.22)
Here 2U a is the circulation around the bottom vortex found in part (e).
Boundary conditions (valid for t > 0): (1) The velocity is zero in the center of the vortex (similar
to solid-body rotation), and (2) very far away the velocity field looks like an irrotational vortex
(because the fluid very far away from the vortex core has not yet felt the influence of viscosity):
v r 0, t 0, t 0
v r , t
, t 0
2 r
(1.23)
Following the problem statement, introduce the dimensionless velocity and similarity variable
v
r2
F
,
.
/ 2 r
4 t
(1.24)
Using the chain rule, we can transform the partial derivatives with respect to t and r in (1.21) to
ordinary derivatives with respect to :
d r 2 d
,
t
t d 4 t 2 d
d
r d
.
r
r d 2 t d
13
The left and right-hand sides of equation (1.21) are then rewritten as
v
r 2 dF
,
t 2 r 4 t d
1
1 dF
r d 2F
rv
r r r
r 2 2 t d
4 t 2 t d 2
8 t d 8 t d 2
d 2 F dF
0.
d 2 d
(1.25)
(1.26)
F 0 0,
F 1.
(1.27)
The solution can be found by several different methods. Here, we write the characteristic
equation for the ODE (1.26) as
s 2 s s s 1 0 s 0, 1.
F 1 e ,
(1.28)
r 2
Ua
v r , t
1 exp
.
r
4 t
(1.29)
This result implies that the propagation speed V of the vortices decays with time, according to
V t
H 2
Ua
1
exp
.
H
4 t
14
(1.30)
z r , t
r2 U a
r2
U a
1
1 v
rv r exp exp 2 ,
r r
r 2 t
4 t 2 t
0
(1.31)
The vorticity has a normal distribution centered at the vortex core with standard deviation .
Note that the peak value (at r = 0) changes with time. It is convenient to define the length scale
2 t.
Even without knowing the solution to (h), one can still make an argument based on previous
problems we have considered throughout the course that deal with viscous diffusion of vorticity
(e.g., Rayleigh problem, boundary layers, etc.). From a similar argument one could conclude
that ~ t .
Part (j) (3 points)
The vorticity distributions will begin to overlap when their characteristic widths are
approximately equal to H/2. This happens at a characteristic time tc, where
H
H2
tc tc ~
.
2
4
(1.32)
The vortex propagation speed scales with the value found in part (g):
V~
Ua
.
H
(1.33)
Then the distance x that the vortices travel before they decay away is roughly the product of V
and tc:
x Vtc
U a H 2 U H a
H 4 4
ReH
a ~ ReH .
(1.34)
A Reynolds number clearly emerges from (1.34). This is not surprising, as the Reynolds number
should govern the distance that the vortices propagate before they are spread out by viscosity.
Thus, in this case the Reynolds number can be interpreted as a dimensionless propagation
distance.
15
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
u
u
p
v ,
y
x
x
v
v
p
v g.
y
y
x
v u
v u
0 .
x y
x y
ez
u
v
1 p
dx v dx
dx,
x
x
x
u
v
1 p
u dy v dy
dy gdy.
y
y
y
u
u
u v
v
1 p
p
u dx dy v dx dy dx dy gdy.
y x
y
x
y
x
du
dv
dp
Since the terms in parentheses are the total differentials of u, v, and p along dr , this equation
becomes
udu vdv
dp
gdy,
or
u2
v 2 dp
d d
gdy 0.
2
2
Integrating indefinitely, and assuming the flow is incompressible, we obtain Bernoullis
equation,
u 2 v2 p
gy B,
2
where B is a constant. Since we have made no assumptions about the direction of the
displacement vector dr , we conclude that the constant B must be the same for any dr , that is, it
must be the same everywhere in the flow.
Elegant solution:
Using the vector identity v v v v / 2 v v , one can rewrite the general vector
Euler equation in the form
v2 p
v
gy v v v .
2
Here v is the velocity magnitude. In this case the right-hand side is identically zero because
the flow is irrotational. If we assume a steady flow and integrate the resulting equation, we are
left with
v
2
gy B
This equation applies everywhere in the flow field; therefore, the Bernoulli constant is the same
everywhere in an arbitrary inviscid, irrotational (potential) flow.
Part c (show that velocity is independent of and z)
There are several ways to do both parts. Below we show the most straightforward way for each.
Independent of : Consider the differential statement of the conservation of mass for this
steady, incompressible flow in cylindrical coordinates:
v
1
1 v v
rvr z 0.
r r
r z
0
0
Since it was given that the streamlines in the bend are circular arcs, this implies that vr vz 0.
Thus, the first and third terms vanish, and we are left with v / 0. Clearly, this implies that
velocity in the bend is independent of . This part is also relatively straightforward to show with
the control-volume statement of mass conservation.
Independent of z (actually shows independent of as well): Consider Eulers equation in z:
p gz 0.
z
We can substitute this result into Bernoullis equation normal to a streamline in the bend (which
is really Eulers equation in the r direction):
v2
p gz patm gh r .
r
r
r
Simplifying,
v2 rg
dh
.
dr
since the fluid height at the inlet is equal to h1 everywhere. Since all streamlines in the bend
originate at the inlet, and since B is constant along a streamline by definition, the value of B for
any streamline is equal to Binlet:
v2
v12
p
gz
gh1 Binlet .
atm
2
2
bend
Thus, the Bernoulli constant is the same for all streamlines in this flow.
There are many ways to determine the velocity profile. One possible way is to take the radial
derivative of the Bernoulli equation along a streamline to obtain:
B
v2
v
v2
p
gz
gz
v
r
2
r
r 2
r
r
where B / r 0 because B is constant everywhere in the flow. We also know from Bernoullis
equation in the radial direction that
v2
.
p gz
r
r
Thus we can set
v2
r
v
.
r
dv dr
0 d ln v d ln r 0.
v
r
Integrating,
ln v ln r C,
v r eC K ,
where K is another constant.
Consider a streamline along the free surface extending from the inlet to somewhere in the bend.
Bernoullis equation along this streamline is
B Binlet patm gh1
v12
2
patm gh r
v2
2
Equation (24) is valid for any streamline (i.e., at any value of r or z) because B is the same for all
streamlines. Simplifying,
gh1
v12
2
gh r
v2
2
gh r
K / r
2
Solving for h,
v12
K2
h r h1
.
2 g 2 gr 2
Part f (derive a condition on K from a control-volume theorem)
Consider a control volume that surrounds the liquid in the bend, from the inlet to some arbitrary
cross section in the bend. Applying form A of the control-volume statement of conservation of
mass,
d
dV CS v vc ndA 0.
dt CV
0, steady
Noting that the control surface is not moving ( vc 0 ), the flux integral over the inlet is
v ndA v1h1 ro ri
Inlet
ro
Outlet
v ndA v r h r dr.
ri
ro
ri
v12
K
K2
h1
dr
r
2 g 2 gr 2
v 2 ro dr K 3
Kh1 1 1
ri r
2
gh
2g
ro
ri
dr
r3
v2 r K 3 1 1
Kh1 1 1 ln o
2 2
2 gh1 ri 4 g ri ro
v 2 r K 3 ri 2
v1h1 ro ri Kh1 1 1 ln o
1 2 .
2
2 gh1 ri 4 gri ro
Part g (explain why Bernoulli is valid for rigid-body rotation)
Although the fluid is viscous, there is no viscous stress in steady rigid-body rotation since there
is no relative motion between fluid elements, by definition. Thus, there are no viscous losses to
consider. Since the flow is also incompressible, steady, and without energy input, Bernoullis
equation may be applied along a streamline.
If one wishes to make a more rigorous argument, one can look at the velocity gradient tensor,
v. All components of this tensor vanish except for two:
vr
r
v
v
r
1 vr v
r r 0
.
1 v vr 0
r r
0 0
1
T
v v
.
0
0
2
As expected, fluid elements experience no deformation, only pure rotation. Thus, if we wish to
calculate the viscous stress tensor, e.g. 2 e for a Newtonian fluid, we see that there is no
viscous stress in the flow (at steady state) regardless of how viscous the fluid is. Thus, there are
no viscous losses, and Bernoullis equation is applicable.
Part h (show that the Bernoulli constant does not depend on z; show that it does depend on r and
determine its value for a given r; explain why it varies in this case)
The Bernoulli equation for this case is
p
v2
2
gz p
r 2 2
2
gz Bsb const.
p gz 0 p gz patm gh r .
z
Here h(r) is the concave, parabolic shape for the free surface that we derived in class for rigidbody rotation. Substituting this relation into the Bernoulli equation,
Bsb patm gh r
r 2 2
2
This shows both that the Bernoulli constant does not depend on z and that it does depend on r.
This equation is an acceptable final answer for the Bernoulli constant, but one can also substitute
the height profile from the class notes (not required for credit) to obtain
Bsb patm
r 2 2
2
r 2 2
2 2
g H
patm gH r .
2g
Since the flow is rotational, the Bernoulli constant is not the same everywhere in the flow field,
and varies across streamlines as indicated above.
Interpretation
Recall that the Bernoulli constant B is roughly a measure of the total energy of a fluid element.
As one looks outward in r, the fluid elements are moving faster and thus have greater kinetic
energy. Thus it makes sense that the total energy of fluid elements increases as r increases, i.e.
that B varies with r.
Later in the class, we will show that B is actually constant along both streamlines and vortex
lines. In this flow, the vortex lines point in the z direction, so it makes sense that the Bernoulli
constant does not depend on z.
Fall 2013
u1
h1
u2
h2
P Pa
gh2
gh1
xA
xB
Fall 2013
y
yD
u1
yE
h1
yA
gh1
P Pa
u1
P1 (y)
h1
P2 (y) h2
gh2
gh1
u2
Figure 3: A schematic of the selected control volume for the hydraulic jump problem.
Table 1 summarizes all the important parameters acting on dierent control surfaces for the
selected control volume:
Fall 2013
Table 1: Bookkeeping of all the related properties at dierent control surfaces in the control
volume.
n
vc
(v vc ).n
P Pa
AD
ex
u1 ex
u1
h1 W
gy
AB
ey
?ex
gh(x)
BC
ex
u2 ex
u2
h2 W
gy
CD
ey
Now we can start by writing the conservation rules using the RTT. It is important to notice
that due to the turbulent mixing happening in the region of the hydraulic jump, energy will
not be conserved and thus either applying the conservation of energy or the Bernoulli equation
will not be the right approach. If we write the conservation of mass for the selected control
volume then we will have:
C.O.Mass: 0 =
d
dt
dV +
c.v.
c.s.
(v vc ).ndA
Knowing that the problem is steady state and using the tabulated quantities, conservation of
mass can be simplied to:
u1 h1 = u2 h2 u1 h1 = u2 h2
(1)
The conservation of linear momentum in the x direction can also be written in the RTT form:
C.O.Momentum:
1
W
Fx =
d
dt
c.v.
vx dV +
c.s.
vx (v vc ).ndA
The net of external forces acting in the x-direction on the control volume neglecting the wall
shear eect is a result of pressure forces acting on the AD and BC control surfaces:
1
W
Fx =
AD
(P Pa )dy
BC
(P Pa )dy =
h1
0
gydy
h2
gydy = g
0
h21 h22
2
2
The right hand side of the RTT for the conservation of linear momentum can also be simplied
to (knowing that the problem is steady and using the tabulated identities):
R.H.S. of RTT for C.O. Momentum= u22 h2 u21 h1
thus the conservation of linear momentum implies that:
g
h21 h22
2
2
= u22 h2 u21 h1
10
g 2
h h22 = u22 h2 u12 h1
2 1
(2)
Fall 2013
using the result from conservation of mass (equation (1)) one can eliminate u2 from equation
(2) to give:
g 2
h1 h22 = h1 u21 (h1 /h2 1) u1 =
2
gh2
(h1 + h2 )
2h1
(3)
For the selected control volume (Figure 4) one can easily write the conservation of mass using
du
dh
x)(h(x) +
x)
dx
dx
which after ignoring the second order terms such (x2 ) it can be rewritten as:
dh
du
x u(x)
+ h(x)
dx
dx
=0 u
dh
du
+h
=0
dx
dx
(4)
Another way to reach the same result is to say that since the ow is incompressible then the
volumetric ow rate should remain unchanged thus d(uh)/dx = 0 which will lead to the same
result we just derived in equation (4).
h(x0 )
h(x0 + x)
u(x0 )
u(x0 + x)
Figure 4: An arbitrary control volume selected to derive the conservation of mass in the
dierential form.
11
Fall 2013
tential energy from nowhere and is thus unphysical1 . For further explanations see the appendix.
h(x)
y
b(x)
b(xm )
h(x) is decreasing
y
b(x)
b(xm )
b(x)
b(xm )
h(x) = const.
Figure 5: (a) Flow approaching the bump (b) The case for low speed ows in which F r1 < 1
(c) The non-physical case in which h remains constant.
Since the incoming ow is smooth and has not undergone tany mixing and also viscous
eects are negligible we can think of writing Bernoulli equation on a streamline very close to
the water surface:
[
]
1
d
1
2
2
Pa + u(x) + g(h(x) + b(x)) = const. = 0
Pa + u(x) + g(h(x) + b(x)) = const.
2
dx
2
which can be simplied to:
u(x)
du
dh
db
+g
+g
=0
dx
dx
dx
(5)
dh
du
dh
h(x) du(x)
+h
=0
=
dx
dx
dx
u(x) dx
(6)
The mentioned argument holds for cases in which the incoming kinetic energy of the ow before the bump
is small compared to the potential energy of the uid (i.e. u21 < gh1 or F r1 < 1). If the initial ow has a high
kinetic energy compared to its potential energy (i.e. u21 > gh1 or F r1 > 1) then it is possible to see a dierent
solution in which h(x) does increase as the liquid goes over the bump. Later in the solution we will see that
combining the conservation of mass and Bernoulli equation it can be shown that: dh/dx = db/dx(F r 2 1)1
and consequently the rise or decrease in h(x) depends on the Froude number of the entering ow (F r1 ).
12
Fall 2013
plugging the result from (6) into (5) will give the following result:
]
1 du(x) [
db(x)
u(x2 ) gh(x) + g
=0
u(x) dx
dx
(7)
(g): As the ow reaches xm , the slope db/dx becomes zero. Slightly after the bump (x+
m ) we
+
will have a negative value for db/dx (i.e. db/dx
[ < 0 at x
m ) ]and thus the sign of the quantity
du/dx depends on the value of the quantity u(x)2 gh(x) . Using equation (7) one can see
that it means that at x
+
m we will have the following:
]
1 du [ 2
u gh |x+
> 0
m
u(x) dx
2
+
thus depending on the sign of u(x
+
m ) gh(xm ) there will be two situations:
(i):
1 du
[ 2
]
+
u gh
+ < 0 at x
m : u(x) dx
< 0
xm
which means that the ow will decelerate after the bump and with decrease in velocity the
height will increase due to the conservation of mass, equation (6) as shown in gure 6(a).
(ii):
[
1 du
]
+
u2 gh
+ > 0 at x
m :
u(x) dx
> 0
xm
which means that the ow will accelerate after the bump and with the increase in velocity the
height will decrease due to the conservation of mass (Figure 6(b)). This will lead to a state that
is called a super-critical ow (F r > 1) and slightly after the bump the viscous eects become
important since the speed is increasing and the height is decreasing (remember that v/h)
and the ow ultimately reaches a point at which it has no more kinetic energy to continue its
acceleration into the super-critical zone. As a result, hydraulic jump will occur with a lot of
turbulent energy dissipation and the ow returns into a sub-critical stage (F r < 1). Figure
6(b) shows a sketch of this ow.
13
Fall 2013
h(x) is decreasing
y
b(xm )
b(x)
x
b(xm )
Figure 6: (a) Physical image for the (i) case. (b) Physical image for the (ii) case.
1 du
1 d(u2 (x))
d
du(x)
du h(x)
1 d(u2 (x))
g
=
gQ 2
=
+ gQ
2
dx
dx
dx
dx u(x)
2
dx
u (x) dx
1
u(x)
1
u(x)
+g
db(x)
=0
dx
(8)
Integrating equation (8) from point 1 to point 2 (far after the bump at which point the surface
becomes at again, b(x) = 0) will give the following:
1
1
1
1
1
1
+ gb(x) = const. u21 + gQ
= u22 + gQ
(9)
u(x)2 + gQ
2
u2
2
u(x)
2
u1
Equation (9) is nothing but a reformulation of Bernoulli equation or conservation of energy
between any two arbitrary points on the water surface. Looking at equation (9) it is easy to
see that one possible solution is u2 = u1 which is the case (i) studied in the previous part. Also
it is worthy to mention that there is another root which satises the following:
1
1
1
1
1
1
1
1
1
= u12 + (u2 u1 +u22 ) = u22 + (u2 u1 +u12 ) = u22 +gQ
gQ = u1 u2 (u1 +u2 ) u21 +gQ
2
2
2
2
2
u2
2
2
u1
14
Fall 2013
This second solution or root happens in the case in which the ow is sub-critical before the
bump, becomes critical at the bump point and then becomes super-critical after the bump.
Using the denition of the Froude number (F r u2 /gh) one can manipulate the derived result
in equation (9) in the following way (using the hint that F r = u3 /gQ):
gQ 1 2 gQ 1 2
1
1
+ u1 =
+ u2 ) u2 (gQ + u13 ) = u1 (gQ + u23
)
2
u2
2
2
2
u1
1
u1
1
1
1
u2
(1 + F r2 ) (gQF r2 )1/3 (1 + F r1 ) = (gQF r1 )1/3 (1 + F r2 )
(1 + F r1 ) =
gQ
2
gQ
2
2
2
u1 u2 (
1/3
2F r2
1/3
1/3
1/3
1/3
+ F r2 F r1 = 2F r1 + F r1 F r2 2(F r2
(
)
1/3
1/3
1/3
1/3
2 = F r1 F r2
F r1 + F r2
1/3
1/3
1/3
2/3
F r1 ) = F r1 F r2 (F r2
2/3
F r1
)
(10)
15
Fall 2013
Appendix:
The ow that has just been studied in this problem is a well known subject in the hydraulic
literature. It is worthwhile to mention a few words about dierent physical aspects of these
ows which are generally named open channel ows in hydraulics. One way to study these
problems is to introduce two identities called the specic energy head (E) and momentum
of the ow (M ):
1 2
1 q2
+h=
u +h
2
2g
2 gh
q2
h2
M
+
gh
2
(11)
knowing that q uh one can easily see that E is the total specic energy and has similari
ties to Bernoulli constant whereas M is the linear momentum of the ow and when dened
in this way will have dimensions of length squared. At a constant ow rate (q) one can plot
the relationship between h and E (Figure 7) or h and M (Figure 8) using the relationships in
equation (11). One can easily see that in both equations dE/dh and dM/dh will be zero at
a critical height hc = (q 2 /g)1/3 . It is also possible to show that at h = hc the local Froude
number which is dened as F r u2 /gh becomes equal to 1 and in fact another way to dene
the Froude number is to dene it as the ratio of local height (h) over the critical height (hc )
(i.e. F r h/hc ). Based on this we can categorize open channel ows in three dierent types:
Sub-critical ow in which the Froude number is lower than one and the ow is dominated
by gravity rather than inertia. This is a characteristic for low speed, deep river ows.
Critical ow in which the Froude number is one and the inertia and gravity are equally
important.
Super-critical ow in which the Froude number is higher than one and inertia dominates
over the gravity.
Looking to Figures (7) and (8) it is worthwhile to notice that at a constant ow rate for any
given specic energy (E) or momentum (M ) value there are two possible heights: one in the
sub-critical zone and the other one in the super-critical zone. Also it is noteworthy that for
specic energies lower than the value at Ec at hc there is no possible physical solution.
One benet of these diagrams is enabling us to nd solutions for hydraulics problems just by
visual inspection of the curves. For example in the bump problem we know that the Bernoulli
equation between points on the free surface is equivalent to:
1 2
1
1
1
1
E2 = E1 b(x)
u1 + gh1 = u22 + gh2 + gb(x) g(E1 ) = g(E2 + b(x))
hc
hc
2
2
hc
(12)
What equation (12) is showing is the fact that knowing the value of E1 we can easily nd the
solution for E2 and h2 by following the constant q curve and subtracting the b(x) from the
16
Fall 2013
6
5
h/h c
Fr < 1
3
2
1
0
0
Fr > 1
Fr = 1
E/h c
Figure 7: The specic energy (E)-height (h) diagram for a ow with constant ow rate. In
which hc is dened as (q 2 /g)1/3 .
6
5
h/h c
Fr < 1
3
2
1
0
0
Fr > 1
Fr = 1
M/h 2c
Figure 8: The momentum of the ow (M )-height (h) diagram for a ow with constant ow
rate. In which hc is dened as (q 2 /g)1/3 .
17
Fall 2013
6
b(x)/hc
h/h c
3
2
1
0
0
E/h c
Figure 9: Solution to the bump problem using the constant q curve in the specic energyheight diagram. The green points show the sub-critical solution while the red points show the
super-critical solution.
Using the visual solution from Figure 9 one can easily detect that there are two possible
branches of solution for the ow over a bump based on the initial Froude number. The subcritical solution (F r < 1) predicts that as the ow goes over the bump the value of h will
decrease and the ow will speed up while the super-critical solution (F r > 1) predicts the
opposite. If the ow starts from sub-critical branch in the upstream as it reaches the bump
it is possible to become very close to the critical point in the specic energy-height diagram
(Figure 10(b)) but right after the bump with decrease in b(x) it will return to the original point
and thus all through the process the ow will remain sub-critical. The solution for the change
in height of the ow over the bump for this case is shown in Figure 10(a).
18
Fall 2013
[h(x) + b(x)]/h c
(x xm )/h c
6
5
h/h c
4
3
2
1
0
0
E/h c
Figure 10: Flow over the bump. Sub-critical solution for the entire ow.
If the bump is high enough then it is possible that the ow initially is sub-critical but as it passes
over the bump (where x = xm and db/dx = 0) it becomes critical and with a small perturbation
after (x = xm) the solution will follow the super-critical branch (Figure 11 (b)). This means that
before the bump h(x) will decrease as it reaches the bump and after xm this decrease will con
tinue since the ow has become super-critical (Figure 11). In most hydraulics labs the transition
from sub-critical to critical and super-critical ow is demonstrated in a water tank experiment of
the ow over the bump (link on youtube: http://www.youtube.com/watch?v=cRnIsqSTX7Q).
19
Fall 2013
[h(x) + b(x)]/h c
(x xm )/h c
6
5
b(xm )/hc = 1
h/h c
4
3
2
1
0
0
E/h c
Figure 11: Flow over the bump. Sub-critical solution for the ow before the bump, critical at
xm and super-critical after the bump.
20
Fall 2013
[h(x) + b(x)]/h c
(x xm )/h c
6
5
h/h c
4
3
2
1
0
0
E/h c
Figure 12: Flow over the bump. Super-critical solution for the entire ow.
It is also important to notice that although the application of specic energy-height diagrams is
quite useful in the case of the ow over the bump, special caution should be taken in interpreting
the diagram. As we saw in this problem the hydraulic jump does not conserve the total energy
(but in fact dissipates a lot). The conserved identity is the linear momentum. Thus a good
approach would be to nd the point on the super-critical branch before the jump and nd a
corresponding point on the sub-critical branch with an equal value of momentum and see where
this new point sits on the energy diagram (as shown in Figure 13). It is noticeable that the
jump will lead to a considerable drop of the energy head due to viscous dissipation involved in
the jump region.
21
Fall 2013
E/hc
h/h c
4
3
1
0
0
E/h c
6
5
h/h c
4
3
h = h2
2
1
0
0
h = h1
2
4
M/h 2c
Figure 13: (a) Blue curve showing h/hc as a function of E/hc for constant ow rate. (b) Red
curve showing h/hc as a function of M/h2c for constant ow rate. The ow starts at point
1 on each curve and after the hydraulic jump transitions to point 2 for downstream of the
jump.
22
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Oil Spills may occur in ports where oil tankers are loaded. The density of oil, is less than that of water
w , and the two uids are immiscible, so that when a spill occurs the oil simply spreads out in a layer on
top of the water. To contain any possible spills, a semi-circular oil boom is deployed at a radius R around
the dock where the loading takes place.
The boom is a barrier which oats on the water, its bottom submerged and its top a bit above the water
surface, as shown. This barrier prevents the oil from spreading past it, at least if the spill is not too great
(see part b).
Suppose a volume V of oil is spilled inside the boom. After sucient time has elapsed for the situation to
reach static conditions, calculate, in terms of 0 , w , R, V and g,
(a) the depth h1 of the bottom surface and the elevation h2 of the top surface of the contained oil
relative to the water surface outside the boom;
(b) the components of force parallel to and transverse to the dock exerted by one of the moored boom
ends on the dock.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
The Swiss scientist Auguste Picard developed a navigable diving vessel, the bathyscape, to investigate the
ocean at great depths (http://en.wikipedia.org/wiki/Bathyscaphe). In 1960, his son Jacques, accompanied
by Lt. Don Walsh of the U.S. Navy, reached a depth of 10, 916 m in the Pacics Mariana Trench.
Suppose that the ocean is at constant temperature, has a density of 1030 kg/m3 at sea level, and is charac
terized by a constant isothermal bulk compressibility
1 @
= 4.6 10-10 m2 /N.
(1.10a)
T
@p T
Compute the pressure at a depth of 11 km,
(a) assuming the density is constant at the sea level value, and
(b) taking the waters compressibility into account.
For part (b), derive an expression for the pressure as a function of depth below the surface, considering the
sea level density 0 and pressure p0 , as well as T , as given quantities.
Courtesy of the U.S. Naval History Center. Photograph in the public domain.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
The Swiss scientist Auguste Picard developed a navigable diving vessel, the bathyscape, to investigate the
ocean at great depths (http://en.wikipedia.org/wiki/Bathyscaphe). In 1960, his son Jacques, accompanied
by Lt. Don Walsh of the U.S. Navy, reached a depth of 10, 916 m in the Pacics Mariana Trench.
Suppose that the ocean is at constant temperature, has a density of 1030 kg/m3 at sea level, and is charac
terized by a constant isothermal bulk compressibility
T
(1.10a)
Courtesy of the U.S. Naval History Center. Photograph in the public domain.
c 2010, MIT
Copyright
Fluid Statics
Solution:
(a) Assuming the density is constant at the sea level value and the pressure at sea level is p0 = 1.01105 Pa,
we nd that
p = p0 gh
(1.10b)
5
1
d
T dp =
p0
T (p p0 ) = ln ln 0 = ln
= 0 eT (pp0 )
(1.10c)
At this point, you may be inclined to substitute this expression for into the pressure equation
[Eq. (1.10b)] used in part (a). However, we note that this pressure distribution assumes a constant
density (see Kundu & Cohen [K&C] pp.11). Instead, we use the more general form of the pressure
gradient [Eq. (1.8) in K&C] and substitute Eq. (1.10c) to give
dp
= g
dz
= 0 eT (pp0 ) g.
Again, we separate variables and integrate:
p
T (pp0 )
p0
dp =
0 gdz
0
T (pp0 )
1 = 0 gh
p = p0
ln(1 + T 0 gh)
T
= 1.01 105 Pa
(1.10d)
1
ln 1 + (4.6 1010 )(1030)(9.8)(11, 000)
4.6 1010 m2 /N
= 114 MPa
(1.10e)
As a check on our pressure equation [Eq. (1.10d)], take the limit as x = T 0 gh is small. Note that
ln(1 + x) x for small values of x. Thus,
1
ln(1 + T 0 gh)
T
1
p0
T 0 gh
T
= p0 0 gh
p = p0
c 2010, MIT
Copyright
Fluid Statics
Note, the equation above is the the same as the incompressible pressure equation [Eq. (1.10b)] in part
(a).
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Accelerometer
It is a proposed to use the type of system shown in the sketch as an accelerometer for measuring the horizontal
acceleration, ax , and to obtain ax from the formula
ax =
h
g,
b
(1.13a)
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Accelerometer
It is a proposed to use the type of system shown in the sketch as an accelerometer for measuring the horizontal
acceleration, ax , and to obtain ax from the formula
ax =
h
g,
b
(1.13a)
c 2010, MIT
Copyright
Fluid Statics
Solution:
(a) The uid in the column behaves as if it were part of a larger container as long as it remains
connected to the large reservoir (except for the neglected eects of surface tension). First, lets nd
the lines of constant pressure in the uid. From hydrostatics,
p
= g,
z
p
= ax .
x
(1.13b)
(1.13c)
(1.13d)
p = gz + f0 (x),
(1.13e)
p = ax x + f1 (z).
(1.13f)
(1.13g)
(1.13h)
(1.13i)
h
g.
b
(1.13j)
The density is not involved because even when the weight of the water column depends on it, the
pressure driving the uid up, also depends linearly on it, and the eect of the density cancels out.
(b) If we draw lines of constant pressure, well notice that the perpendicular distance from the surface
acts as an eective depth distance. Then, b amplies the magnitude of the read length, h, (thats
why we have to divide h by it when getting the acceleration). Then, the longer b is, the better accuracy
well have when reading the measurement. If b is too short, then the measurement error will grow,
which sets a lower limits to the minimum required size of the device. Also, if too small surface tension
and the eect of the meniscus wiil degrade the measurement.
c 2010, MIT
Copyright
Fluid Statics
Also, getting a number for rapid changing accelerations will prove to be challenging, since the height
will change constantly. Besides that, any other form of acceleration will produce a considerable error
(like rotation).
Improvements: There is always room for improvements
(i) Measure the height dierence between the front liquid level (not at the center as it is done
now) and the meniscus in the tube.
(ii) Reduce the volume of liquid in the container, most of it is not needed (not much because
this volume stores the liquid that is driven into the tube when acceleration occurs, and acts as a
hydraulic capacitor).
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
(a) Demonstrate that when a cylindrical can of liquid rotates like a solid body about its vertical axis
with uniform angular velocity, , the free surface is a parabolic of revolution.
(b) Demonstrate that the pressure dierence between any two points in the uid is given by
p2 p1 = g(z2 z1 ) + 2 (r22 r12 )/2,
(1.14a)
c 2012, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
(a) Demonstrate that when a cylindrical can of liquid rotates like a solid body about its vertical axis
with uniform angular velocity, , the free surface is a parabolic of revolution.
(b) Demonstrate that the pressure dierence between any two points in the uid is given by
p2 p1 = g(z2 z1 ) + 2 (r22 r12 )/2,
(1.14a)
c 2008, MIT
Copyright
Fluid Statics
Solution:
ps
rs rs
ps
zs zs
V2
= 2 r,
r
(1.14b)
p
= 2 r,
r
(1.14c)
p
= g.
z
(1.14d)
= 0, or 2 rs rs gz s = 0, then
2 rs rs = gz s ,
2 rs rs =
gz s , zs =
2 rs2
+ Const,
2g
(1.14e)
p
z ,
we notice that g =
p(z, r) = 2
f
z ,
(1.14f)
r2
gz + Const,
2
(1.14g)
r22
r2
1 ) g(z2 z1 ),
2
2
(1.14h)
(c) No practical dierence, the surface just would be cut by two planes instead of a cylinder (the eects
of surface tension would be dierent, but this is unimportant as long as the surface tension contribution
is small).
D
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Fall 2013
x1
12
+ a
21
y, 2
+ b
x2
12
21
x, 1
Figure 1: Taylor series expansion for the shear stresses acting on a material element of size
x1 , x2 .
Knowing that the normal stresses acting on each plane will not lead to any net torque around
axis x3 passing through point O one can calculate the net exerted torque on the element (MO )
by accounting for all the shear stresses acting on the element:
L
MO =(12 + a )(x2 x3 )(x1 /2) + (12 a )(x2 x3 )(x1 /2)
(21 + b )(x1 x3 )(x2 /2) (21 b )(x1 x3 )(x2 /2)
1
Fall 2013
MO = (12 21 )x1 x2 x3
(1)
I = x1 x2 x3 (x21 + x22 )
(2)
12
Combining (1) and (2) will result in:
3 =
12 12 21
x21 + x22
It is easy to see that if one shrinks the element to a very small volume (i.e. x1 and x2 0)
the rotational acceleration of the element (3 ) will diverge to innity unless the shear stress
dierence also tends to zero at least as fast as x2i 0 (thus 12 21 = 0). Since innite
rotational acceleration is not physically possible the stress tensor should be symmetric, ij =
ji .1
The mentioned proof is true in the absence of magneto-hydrodynamic forces or other non-conservative body
forces.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
1
Equation of Motion in Streamline Coordinates
Ain A. Sonin, MIT, 2004
Updated by Thomas Ober and Gareth McKinley, Oct. 2010
2.25 Advanced Fluid Mechanics
Eulers equation expresses the relationship between the velocity and the pressure fields in
inviscid flow. Written in terms of streamline coordinates, this equation gives information
about not only about the pressure-velocity relationship along a streamline (Bernoullis
equation), but also about how these quantities are related as one moves in the direction
transverse to the streamlines. The transverse relationship is often overlooked in
textbooks, but is every bit as important for understanding many important flow
phenomena, a good example being how lift is generated on wings.
A streamline is a line drawn at a given instant in time so that its tangent is at every
point in the direction of the local fluid velocity (Fig. 1). Streamlines indicate local flow
direction, not speed, which usually varies along a streamline. In steady flow the
streamline pattern remains fixed with time; in unsteady flow the streamline pattern may
change from instant to instant.
In what follows, we simplify the exposition by considering only steady, inviscid flows
with a conservative body forces (of which gravity is an example). A conservative force
per unit mass G is one that may be expressed as the gradient of a time-invariant scalar
function,
G = -\U(r) ,
and the steady-state Euler equation reduces to
(1)
1
V VV = - Vp - VU(r) .
P
(2)
A uniform gravitational force per unit mass g pointing in the negative z direction is
represented by the potential
U = gz .
(3)
A streamline coordinate system is not chosen arbitrarily, but follows from the
velocity field (which, we note, is not known priori). Associated uniquely with any point
r and time t in a flow field are (Fig. 2): the streamline that passes through the point
(streamlines cannot cross), the streamlines local radius of curvature R and center of
curvature, and the following triad of orthogonal unit vectors:
! d ! d !d
V = is + in
+i
ds
dn
dl
(4)
V = isV
(5)
and
where V is the magnitude of the velocity vector V . From (4) and (5),
!
d
V V =V
ds
and thus
1
f (s,n,l )
is defined by
Vf dr df (s, n, l ) =
dr = is ds + in dn + il dl
(6)
!f
!f
!f
ds +
dn +
dl . Equation
!s
!n
!l
(4) follows
(V
)V = V
(V is ) = is
V2
i
+V2 s .
s 2
s
(7)
The unit vector in the last term of (7) changes orientation as one moves along the
streamline. The change dis in is from s to s+ds is obtained with the construction shown
in Fig. 2 as
!
!
! ds
dis = -in d8 = -in
R
(8)
d is
i
=_ n
ds
R
(9)
d (V 2 J
V2
- in
ds 2
R
(10)
The first term on the right is the convective acceleration in the direction of the velocity,
and the second is the centripetal acceleration, toward the center of curvature.
The pressure gradient in streamline coordinates is
p = is
p
p
p
+ in
+ il
s
n
l
(11)
4
Using (10) and (11) in (2), we obtain the equation of motion in streamline coordinates for
steady, inviscid flow as
s-direction:
n-direction:
d (V 2 l
1 dp dU
=ds 2
P d s ds
(12)
V2
1 dp dU
=R
p dn dn
(13)
1 dp dU
P dl dl
(14)
0=-
l-direction:
(
s
1
2
V2 =
1 p
s
n-direction:
V2
1 dp
dz
=-g
R
P dn
dn
l-direction:
0=-
1 dp
dz
-g
P dl
dl
z
s
(15)
(16)
(17)
For constant-density flow in a uniform gravitational field, the equations simplify further
to
s-direction:
d
LV 2
( p + Lgz +
J=0
ds
2
(18)
n-direction:
d
pV 2
( p + pgz) = R
dn
(19)
l-direction:
(p+
gz) = 0
(20)
The s-direction equation (18) states Bernoullis theorem: the total pressurethe sum
p + pgz + pV 2 2 of the static, gravitational, and dynamic pressuresremains invariant
along a streamline.
The n-direction equation (19) states that when there is flow and the streamlines curve,
the sum p + pgz (which is constant in when the fluid is static) increases in the ndirection, that is, as one moves away from the local center of curvature.
5
The l-direction equation (20) states that p + pgz remains constant for small steps in
the binormal direction, that is, the pressure distribution is quasi-hydrostatic distribution in
the l-direction.
EXAMPLE
Consider the simple case of 2D, inviscid air flow over a smooth hill (Fig. 3). Far
upstream of the hill the incident velocity is uniform at V= . The hill deflects the air around
it, and a uniform flow is again established far downstream. Far upstream, above, and
downstream of the hill, the pressure is constant at p= and the streamlines are straight (the
hill does not perturb the flow at infinity). We shall assume that gravitational effects are
negligible (the medium is air and the hills elevation is modest) and the free streams
Mach number is small, so that and the density can be taken as constant. Based on the
available equations, what can we say about the pressure and velocity distributions over
the hillwhere is the velocity higher than V= , for example, and where lower?
To answer this question accurately we need to know the shapes of the streamlines
throughout the flow fieldor, at least, in the region that is perturbed by the hill. We
dont have this information, so we proceed by drawing a rough estimate of the streamline
pattern, as shown in Fig. 3. The difference between the pressure at infinity and at the top
of the hill, point (3), can be estimated by integrating equation (19) along the vertical path
from (3) to ( 0 ). Since this path follows the local n-direction, R>0 everywhere along it.
Neglecting the gravitational term, (19) gives
dp pV 2
=
dn
R
from which we see that
(21)
p3 =
V 2 dn
>0
R
(22)
Thus p3 < p= , and according to Bernoullis equation (18), it follows that V3 > V= . Using
similar arguments, we conclude that p1 = p= and V1 = V= , and p2 > p= and V2 < V= , etc.
In principle, if R(n) and V(n) can be established or estimated, the integral in (21) can
be evaluated. For example if we find that the flow perturbation caused by the hill is
negligible at elevations greater than some multiple of the height of the hilltop, we might
write for the path from (3) to ( 0 )
n
(23)
Rhill e H
where Rhill is the streamlines radius of curvature in the vicinity of the hilltop, n is
measured from the top of the hill upward, and H = {h is some multiple of the actual
height h of the hilltop, the coefficient being an empirical number. From Bernoullis
equation (18) we also have that
p+
V2
V2
,
=p +
2
2
(24)
Substituting for R and V into (21) from (23) and (24), respectively, we integrate (23) and
obtain
V 2 R hill
p3 =
e
2
2H
(25)
For a low hill such that 2H<<Rhill, the exponential term can be expanded and (25)
simplified to
V 2H
Rhill
p3
(26)
The velocity at point (3) now follows from (24) and (25) as
V3 = V e
H
R hill
(27)
V 1+
H
Rhill
(28)
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
atmosphere, pa
A1
A(x)
A2
nozzle
A nozzle with exit area A2 is mounted at the end of a pipe of area A1 , as shown. The nozzle converges
gradually, and we assum that the ow in it is (i) approximately uniform over any particular station x, (ii)
incompressible, and (iii) inviscid. Gravitational eects are, furthermore, taken as negligible. The volume
ow rate in the nozzle is given as Q and the ambient pressure is pa .
(a) Derive an expression for the gage pressure at a station where the area is A(x).
(b) Show, by integrating the x-component of the pressure force on the nozzles interior walls, that the net
x-component of force on the nozzle due to the ow is independent of the specic nozzle contour and is
given by
2
(A1 A2 )
F = Q2
2A1 A2 2
(c) The expression in (b) predicts that F is in the positive x-direction regardless of whether the nozzle is
converging (A2 < A1 ) or diverging (A2 > A1 ). Explain.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A1
A(x)
A2
nozzle
A nozzle with exit area A2 is mounted at the end of a pipe of area A1 , as shown. The nozzle converges
gradually, and we assum that the ow in it is (i) approximately uniform over any particular station x, (ii)
incompressible, and (iii) inviscid. Gravitational eects are, furthermore, taken as negligible. The volume
ow rate in the nozzle is given as Q and the ambient pressure is pa .
(a) Derive an expression for the gage pressure at a station where the area is A(x).
(b) Show, by integrating the x-component of the pressure force on the nozzles interior walls, that the net
x-component of force on the nozzle due to the ow is independent of the specic nozzle contour and is
given by
2
(A1 A2 )
F = Q2
2A1 A2 2
(c) The expression in (b) predicts that F is in the positive x-direction regardless of whether the nozzle is
converging (A2 < A1 ) or diverging (A2 > A1 ). Explain.
c 2010, MIT
Copyright @
Inviscid Flows
Solution:
Given: Q, A1 , A2 are constants.
(a) By mass conservation,
(mass in) = v1 A1 = v(x)A(x) = (mass out)
Since there is no change in the mass inside the CV:
v1 A1 = Q = v(x)A(x)
v(x) =
Q
A(x)
A(x)
A2
x
1
A1
Note that all the assumption required for Bernoulli have been satised:
(a) inviscid
(b) along a streamline
(c) steady
(d) constant density
(e) no work/energy input or loss
1
1
2
p(x) + v(x) = pa + v2 2
2
2
station 1
station 2
Therefore,
pg (x) = p(x) pa =
1
v2 2 v 2
2
pg (x) =
1 2
Q
2
1
1
2
A2 2
A(x)
(4.04a)
(b) Integrate the pressure along the nozzle to obtain the x-component of pressure force
z1
Q
A(x)
A1
Fx =
dFx =
=
A1
projected vertical
= A1 A
area of the
inner wall
A2
pg d (A1 A) =
A2
A1
pg dA
c 2010, MIT
Copyright @
Inviscid Flows
We can reverse the integration limits to get rid of the minus sign in front and substitute Eq. (4.04a):
Fx =
A1
A2
2
= Q
pg dA
A1
A2
1
1
2 A2
A2
dA
Fx =
Q2 (A1 A2 )
2
A1 A2 2
(4.04b)
(c)
For (A2 < A1 ), pg (x) is positive. Since the pressure is
greater on the inside than the outside, the net pressure
force acts on the inner wall.
A1
For (A2 > A1 ), pg (x) is negative. Thus, the net pressure force acts on the outer wall, still pointing to the
right.
A1
A2
A2
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
(b) Determine the integrated forms of these equations for an incompressible uid.
(c) Derive the appropriate equations for unsteady frictionless, compressible ow, in a stream tube of crosssectional area which depends on both space and time.
c 2013, MIT
Copyright
Inviscid Flows
Solution:
A+dA
+d
V+dV
CS3
CS2
CV
ds
A
p+dp
dz
gdV
p
CS1
(a) Here we consider an arbitrary control volume, CV , sitting along a streamline of length ds. For steady
ow, we may write the integral mass conservation equation as
u ndA
=0
(4.05a)
CS
To evaluate this integral we must decompose it into three integrals for the three sub-control surfaces of this
volume. For CS1 located at the upstream portion of the CV, the integral is
u ndA
= V A
(4.05b)
CS1
CS2
d
d+V Ad+V ddA+
d AddV
d+ddV
d
dd
ddA
undA
u ndA
=0
(4.05d)
CS3
Combining Eq. (4.05b), (4.05c) and (4.05d) into Eq. (4.05a) we obtain
2.25 Advanced Fluid Mechanics
c 2013, MIT
Copyright
Inviscid Flows
V A + V A + V dA + AdV + V Ad = 0
d dA dV
+
+
=0
A
V
(4.05e)
Fs,CS2 =(p+dp)(A+dA)
Fs,CS3 12 (2p+dp)dA
Fs,CS1 = pA
Z
u u n
dA =
(4.05f)
CS
To calculate the left hand side of Eq. (4.05f), we calculate the momentum ux across CS1
u u n
dA = V 2 A
(4.05g)
CS1
u u n
dA = ( + d)(V + dV )2 (A + dA) V 2 A + 2V AdV + V 2 Ad + V 2 dA
(4.05h)
CS2
c 2013, MIT
Copyright
Inviscid Flows
F s = Fgravity,s + Fpressure,s
The gravitational force is
Fgravity,s = ()dg sin
where the angled brackets indicate the average value. Setting () =
and sin = dz
ds , we obtain
Fgravity,s =
1
2
+(+d) and =
1
1
2 + d 2A + dA gdz = Agdz (dA + dA)gdz
4
2
1
2
A+(A+dA) ds
1
ddAgdz
ddd
4d
(4.05i)
where we neglect all terms higher than rst order. The force arising from the pressure acting on the control
volume is
Fpressure,s = pA (p + dp)(A + dA) + (p)ACS3 sin
where we set (p) =
equation we have
1
2
p + (p + dp) and ACS3 sin = dA. Having made these substitutions into the above
1
1 d
d
2p + dp dA = dpA d
dpdA
2
2
(4.05j)
AV dV + V 2 A
dV
d dA
+
+
V
= Agdz dpA
(4.05k)
AV dV = Agdz dpA
(4.05l)
dp + V dV + gdz = 0
(4.05m)
c 2013, MIT
Copyright
Inviscid Flows
(b) When we integrate Eq. (4.05e) from station 1 to station 2 on the streamline, we have
d
+
A2
A1
dA
+
A
V2
V1
dV
=0
V
(4.05n)
ln
2
1
!
+ ln
A2
A1
V2
V1
+ ln
!
= ln
2 V2 A2
1 A1 V1
!
=0
(4.05o)
1 A1 V1 = 2 V2 A2
(4.05p)
Again, when we integrate Eq. (4.05m) from station 1 to station 2 on the streamline, we have
p2
V2
dp +
z2
V dV +
p1
V1
gdz = 0
(4.05q)
z1
1
p2 p1 + V22 V12 + g(z2 z1 ) = 0
2
(4.05r)
(c) For unsteady, frictionless, compressible ow, the integral mass conservation equation is
CV
d +
t
u ndA
=0
(4.05s)
CS
The surface integrals in Eq. (4.05b), (4.05c) and (4.05d) remain valid, and the time varying volume integral
is
CV
d =
Ads
t
t
(4.05t)
since in the limit ds 0, dA 0 and thus volume can be written as Ads. Combining Eq. (4.05b), (4.05c),
(4.05d) and (4.05t) into Eq. (4.05s) and dividing by AV we obtain
1
d dA dV
ds +
+
+
=0
V t
A
V
2.25 Advanced Fluid Mechanics
(4.05u)
c 2013, MIT
Copyright
Inviscid Flows
u
d +
t
X
u u n
dA =
F
(4.05v)
CS
The surface integrals and forces in Eq. (4.05g), (4.05h), (4.05i) and (4.05j) remain valid and the time
dependent integral term is
Z
CV
u
d
V Ads
d =
t
dt
(4.05w)
again, since in the limit ds 0, dA 0 and thus volume is written as Ads. Combining Eq. (4.05g), (4.05h),
(4.05i), (4.05j) and (4.05w) into Eq. (4.05v) we obtain
2
2
2
2
t V Ads V A + V A + 2V AdV + V Ad + V dA = Agdz dpA
Eliminating terms, expanding the time derivative, dividing by A, and rearranging the result, we have
d dA
V
1
dV
t ds + V dV + V 2 V t ds + V + + A
!
= gdz dp
Substituting Eq. (4.05u) into the result above , rearranging and dividing by we have
V
dp
ds +
+ V dV + gdz = 0
t
(4.05x)
Integrating Eq. (4.05x) from station 1 to station 2 on the streamline, we obtain the unsteady Bernoulli
equation
s2
s1
V
ds +
t
s2
s1
dp
1 2
ds +
V2 V12 + g(z2 z1 ) = 0
(4.05y)
(4.05z)
Problem Solution by Thomas Ober (2010), updated by Shabnam Raayai, Fall 2013
2.25 Advanced Fluid Mechanics
c 2013, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
T a , a
d
h
dHh
1
2
H
Ta
Ta + T
Consider a furnace of height H with a tall cylindrical smoke stack of diameter d (d H) and height h
(h H). Air, an ideal gas (P = RT ), enters the furnace at atmospheric density and temperature and
at local atmospheric pressure. Between stations 1 and 2, heat is added at constant pressure and the air
temperature is raised by an amount T . Thereafter, heat addition is negligible and the air rises through
the stack at a sensibly constant density.
(a) On the assumption that viscous eects are negligible, derive an expression for the steady mass ow
rate of air drawn by a stack of given height, h, in terms of the temperature rise in the furnace.
(b) If the chimney were capped o at the top, what would be the pressure dierntial across the cap,
assuming that T would not be altered by the ow stoppage?
Note: The height h of the stack is small compared with the length RTa /g over which the atmosphere density
falls by 1/e (see Problem 1.8). Hence, gravitational density changes can be neglected.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
T a , a
d
h
dHh
2
H
Ta
Ta + T
Consider a furnace of height H with a tall cylindrical smoke stack of diameter d (d H) and height h
(h H). Air, an ideal gas (P = RT ), enters the furnace at atmospheric density and temperature and
at local atmospheric pressure. Between stations 1 and 2, heat is added at constant pressure and the air
temperature is raised by an amount T . Thereafter, heat addition is negligible and the air rises through
the stack at a sensibly constant density.
(a) On the assumption that viscous eects are negligible, derive an expression for the steady mass ow
rate of air drawn by a stack of given height, h, in terms of the temperature rise in the furnace.
(b) If the chimney were capped o at the top, what would be the pressure dierntial across the cap,
assuming that T would not be altered by the ow stoppage?
Note: The height h of the stack is small compared with the length RTa /g over which the atmosphere density
falls by 1/e (see Problem 1.8). Hence, gravitational density changes can be neglected.
c 2010, MIT
Copyright
Inviscid Flows
Solution:
(a) First consider the eect of heat addition on the air density. Since the pressure at stations 1 and 2 is
Pa , the heat is added at constant pressure such that
Pa = 1 RT1 = 2 RT2
a Ta = 2 (Ta + T )
a
2 =
1 + T /Ta
3
d
h
2
H
Ta Ta + T
As stated, the density is reasonably constant (i.e. 2 = 3 ) and viscous eects are negligible so we can
apply Bernoullis equation along the streamline shown above.
1
1
P2 + 2 v2 2 + 2 gh2 = P3 + 2 v3 2 + 2 gh3
2
2
If we assume that the streamlines at the exit of the smoke stack at station 3 are parallel, then we
also set P3 equal to the local ambient pressure at the top of the stack. Provided the air outside of
the furnace and stack is isothermal at Ta and has roughly constant density a , then we can relate the
pressure at station 3, P3 , to the ambient pressure, Pa , at ground level by stations 1 and 2 using our
knowledge of hydrostatic pressure, such that Pa = P2 = P3 + a gh. Hence the pressure at the top of
the smoke stack, P3 , is below the pressure at ground level, Pa = P1 = P2 . Accordingly,
1
2 (v3 2 v2 2 ) = (P2 P3 ) + 2 g(h2 h3 )
2
2
v3 2 v2 2 = (P
P
3 + a gh
3 ) 2gh
2
a
v3 2 v2 2 = 2
1 gh
2
Conservation of mass from stations 2 to 3 tells us that
v2
d
H
v3
v3 =
a
1 gh
2
v3 =
2gh
T
Ta
(b) If the chimney were capped, there would be no ow and we can apply our knowledge of uid statics.
c 2010, MIT
Copyright
Inviscid Flows
h
W
1 + T /Ta
a
P3 = Pa
gh
1 + T /Ta
The ambient pressure above the cap is calculated from our knowledge of hydrostatic pressure as before,
Pa,cap = Pa a gh. Hence the pressure dierential Pcap = P3 Pa,cap across the cap is
Pcap =
a ghT
Ta + T
D
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
An incompressible, inviscid liquid ows with speed V vertically downward from the nozzle of the radius R.
The liquid density is high compared with that of the ambient air. The surface tension between the liquid
and the air is .
(a) Obtain an expression which relates the local radius r of the liquid stream to the distance x from the
nozzle.
(b) Show that for suciently large x,
V2
rR
2gx
1
4
(4.11a)
(c) (Optional-Since you need to use dimensional analysis for this part and it is not covered yet in the class)
Write down all the criteria which must be satised for this expression to be a good approximation.
State each criterion as x must be very large compared with y, where y is some combination of the
given quantities V , R, g, and .
c 2012, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
An incompressible, inviscid liquid ows with speed V vertically downward from the nozzle of the radius R.
The liquid density is high compared with that of the ambient air. The surface tension between the liquid
and the air is .
(a) Obtain an expression which relates the local radius r of the liquid stream to the distance x from the
nozzle.
(b) Show that for suciently large x,
V2
rR
2gx
) 1
4
(4.11a)
(c) Write down all the criteria which must be satised for this expression to be a good approximation.
State each criterion as x must be very large compared with y, where y is some combination of the
given quantities V , R, g, and .
c 2008, MIT
Copyright
Inviscid Flows
Solution:
(4.11b)
2 1 2
1
2
( V + gx) = V22 ,
2
2
(4.11c)
simplifying,
then,
Also, from mass conservation
V 2 + 2gx = V22 , V2 =
R2 V = r2 V2 ,
2gx
V2
( )2
r
R
V2
2gx
) 21
r
R
)2
r
R
)2
V + 2gx
2
V
.
V2
(4.11d)
(4.11e)
(4.11f)
,
r
R
V 2 + 2gx.
V2
1,
2gx
V2
2gx
) 41
OR
(4.11g)
V 2
1,
2gx
(4.11h)
then,
x
V2
2g .
(4.11i)
c 2008, MIT
Copyright
Inviscid Flows
= r 1,
1.
gx
rgx
Px
(4.11j)
Now, lets get an estimate of the order of magnitude of r(x) from (b),
rR
)
1
1
RV 2
V 2 4
, r
1 ,
2gx
(gx) 4
(4.11k)
(gx) 4
1
2
RV gx
1, x
4
RV 2
g 4
(4.11l)
nally,
x%
4
3
4
4
2
R 3
g 3
V
3
(4.11m)
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A soap bubble (surface tension ) is attached to a narrow glass tube of the dimensions shown. The initial
radius of the bubble is R0 . At t = 0, the end of the tube is abruptly opened.
a) Obtain a solution for R(t), assuming that the ow is : (i) incompressible ans (ii) inviscid, that (iii) grav
itational eects are negligible, and that (iv) the temporal acceleration term in Eulers equation is negligible
(we are referring to the term involving the partial derviative of the velocity with time).
b) Derive a criterion for when assumption (iv) is satised.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A soap bubble (surface tension ) is attached to a narrow glass tube of the dimensions shown. The initial
radius of the bubble is R0 . At t = 0, the end of the tube is abruptly opened.
a) Obtain a solution for R(t), assuming that the ow is : (i) incompressible ans (ii) inviscid, that (iii) grav
itational eects are negligible, and that (iv) the temporal acceleration term in Eulers equation is negligible
(we are referring to the term involving the partial derviative of the velocity with time).
b) Derive a criterion for when assumption (iv) is satised.
c 2010, MIT
Copyright
Inviscid Flows
Solution:
(a) To obtain an expression for R, we will write mass conservation across the moving control volume
outlined in the gure above then since the ow is inviscid, incompressible and steady, we can write
Bernoullis equation (conservation of momentum) across a streamline from point 1 to point 2.
Mass conservation
Using form A for mass conservation on the moving control volume drawn above, we have
d(
CV (t)
dV )
dV
dt
(v vc ).n dA = 0
dt
(4.12a)
CS(t)
CV (t)
dV )
=
dt
dV
dR
= 4R2
dt
dt
(4.12b)
(4.12c)
Bernoullis equation
Since the ow is inviscid, incompressible and steady, we can use steady Bernoulli across a stream
line going from point 1 to point 2.
1
p1 = pa + V22
2
(4.12d)
4
R(t)
(4.12e)
This equation results from the fact that for a very thin soap bubble in air, there are two air-soap
interfaces: one on the inside of the bubble and the other on the outside, each having the same
radius of curvature, R. Hence the total Laplace pressure within the bubble is twice what it would
be for a bubble having a single interface (e.g. air bubble in water), i.e. 2 2/R. Combining
Eq. 4.12d and Eq. 4.12e, we get
c 2010, MIT
Copyright
Inviscid Flows
V2 = 2
2
R
(4.12f)
dR
+
dt
2
A=0
(4.12g)
2
A=0
(4.12h)
2R 2
4 d(R 2 )
+
7 dt
R(t) = (R02
7
4
2
2
At) 7
(4.12i)
(b) For the unsteady term in Eulers equation (or Bernoulli) to be negligible, we need
V
V
.
V
x
t
(4.12j)
Let l be a characteristic length scale in the x direction, a characteristic time scale and
R0
characteristic velocity.
From Eq. 4.12i, we see that a characteristic time scale for this process is =
R03
A
R0
.
R0
(4.12k)
Which gives
lA R03
(4.12l)
The volume of the pipe needs to be negligible compared to the volume of the bubble for this process
to be considered pseudo-steady.
c 2010, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
from pump
pa
mass
M
ground
clearance: h
bellows
slot width: w
settling chamber
p0
pb
A circular hovering platform of radius R is to support a mass M (its own mass plus a load). A thin, sheet-like
jet (width w) is directed downward at the platforms periphery, as shown. The jet is fed from a settling
chamber which is maintained at a pressure p0 by an external pump. The system is to hover at an elevation
h which is large compared to the width w of the sheet-like jet, but small compared with R.
When the jet is turned on, the pressure under the platform builds up and the platform rises until a steady
state is reached. It is this steady state that we are concerned with.
(a) Describe the physical mechanism which allows the pressure pb under the platform be higher than the
atmospheric pressure pa , in steady state, and thus to support a weight
M g = (pb pa ) R2
(b) Given the system weight M g, the platform radius R, the jet width w, and the air density , derive
approximate expressions for (i) the volume ow rate Q of air required and (ii) the gage pressure
p0 required in the settling chamber, in order to maintain a ground clearance h. You may assume
incompressible, inviscid ow in the peripheral jet, and make physical approximations consistent with
the jet being thin compare with h (w h) and the gage pressure pb pa below the platform being
very small compared with the gage pressure p0 pa in the settling chamber.
c 2012, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
from pump
pa
mass
M
ground
clearance: h
bellows
slot width: w
settling chamber
p0
pb
A circular hovering platform of radius R is to support a mass M (its own mass plus a load). A thin, sheet-like
jet (width w) is directed downward at the platforms periphery, as shown. The jet is fed from a settling
chamber which is maintained at a pressure p0 by an external pump. The system is to hover at an elevation
h which is large compared to the width w of the sheet-like jet, but small compared with R.
When the jet is turned on, the pressure under the platform builds up and the platform rises until a steady
state is reached. It is this steady state that we are concerned with.
(a) Describe the physical mechanism which allows the pressure pb under the platform be higher than the
atmospheric pressure pa , in steady state, and thus to support a weight
M g = (pb pa ) R2
(b) Given the system weight M g, the platform radius R, the jet width w, and the air density , derive
approximate expressions for (i) the volume ow rate Q of air required and (ii) the gage pressure
p0 required in the settling chamber, in order to maintain a ground clearance h. You may assume
incompressible, inviscid ow in the peripheral jet, and make physical approximations consistent with
the jet being thin compare with h (w h) and the gage pressure pb pa below the platform being
very small compared with the gage pressure p0 pa in the settling chamber.
c 2008, MIT
Copyright
Inviscid Flows
Solution:
Given: M g, R, w, ; Unknown: Q, p0 ; Assume: incompressibility, inviscid, R h w
pa
pb
Fz = 0 = M g pa (R2 ) + pb (R2 )
p b pa =
Mg
R2
(4.19a)
pa
pb
out h
in
n
s
The normal component of Eulers equation states that
v2
1 p
z
=
g
r
n
n
v2
(p + gz) =
n
r
in
out
in
out
v2
v2
dn w
r
h
pb pa =
v2
w
h
(4.19b)
v2
Mg
w=
R2
h
v=
M gh
R 2 w
(4.19c)
Since Q = v Area,
Q = (2R)w
jet
area
M gh
R 2 w
Q=2
M gwh
(4.19d)
2
1 The
c 2008, MIT
Copyright
Inviscid Flows
1 2
v
2
po =
1
2
M gh
R2 w
(4.19e)
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Fall 2013
vs
P
vs
+ vs
=
+ gs
t
s
s
vs
t
s
1
1
which can be simplied to:
1
vs
1
ds + (P + vs2 + gz)2 (P + vs2 + gz)1 = 0
t
2
2
(1)
This is not a very useful result in general since vs /t can change dramatically from one point
to another; to use this in practice we need to be able to draw streamline shapes at each instant
in time. It works especially for simple cases such as impulsively started conned ows where
streamlines have the same shape at each instant and we are interested in time required to start
the ow.
Example: Flow out of a long pipe connected to a large reservoir (steady and
transient starting stages)
A1
h
A2
a
b
2
Figure 1: Discharge of water from a long pipe connected to a large uid reservoir with cross
section area A1 >> A2 . The problem approaches steady state when the valve has been open
for a long time but is transient in the starting stage.
Consider the ow in the discharge of water through a long pipe connected to a big reservoir. If
the area of the tank is much larger than the pipe cross section area (i.e. A1 >> A2 ) then the
1
Fall 2013
solution for steady state case, in which the discharge valve has been open for a while, can be
easily done by writing Bernoulli between points 1 and 2 :
J
1
1
Pa + (0)2 + gh = Pa + v22 + g(0) v2 = 2gh
2
2
where v1 A1 = v2 A2 v1 c 0 because A2 << A1 .
This result was known to Torricelli in the 1630. Now consider the analysis for a more general
case which includes the starting time. Although the velocity is changing with time in the pipe
during the transient stage one can easily conclude that conservation of mass says that velocity
has to be constant at any instant along the length of the pipe and it just changes with time.
Applying unsteady Bernoulli equation, as described in equation (1) will lead to:
2
vs
1
1
ds + (Pa + (v2 )2 + g(0)) (Pa + (0)2 + gh) = 0
(2)
t
2
2
1
Calculating an exact value for the rst term on the left hand side is not an easy job but it is
possible to break it into several terms:
vs
ds =
t
vs
ds +
t
vs
ds +
t
vs
ds
t
If the reservoir area is much larger than the pipe area then it the integral from 1 to a is
negligible compared to the integral along the pipe length ( b to 2 ) because vs in the tank is
small, also knowing that the entry region is small compared to the length of the pipe we can
easily neglect the integral from a to b compared to the corresponding integral over the pipe
length. Thus the following estimate is an acceptable approximation for the unsteady term in
the Bernoulli equation:
2
vs
ds c
t
vs
ds =
t
v2
v2
ds =
L
t
t
(3)
1
2L
dv2
L + v22 = gh
dv2 = dt
dt
2
2gh v22 /2
integrating from t = 0 and knowing that v2 = 0 at t = 0 gives the following integral:
v2
0
dv2
=
2gh v22 /2
t
0
dt
1
2L
2 2gh
v2
1
1
+
2gh v2
2gh + v2
2gh
t
L
v2 =
dv2 =
J
2ghtanh(t/ )
dt
2L
(4)
The described relationship for the transient velocity and time is plotted in Figure 2.
Fall 2013
v 2 / 2gh
1
0.8
0.6
0.4
0.2
0
0
t/
Figure 2: Velocity in the pipe as a function of time. The characteristic time constant is
L( 2gh)1
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Contents
Fig. 1: Locally fully developed flow (left) and fully developed flow (right)
(1)
v
2 v 2v
v
v
P
,
+
u
+
v
=
+
+
t
x
y
y
x 2 y 2
(2)
(3)
and the appropriate boundary and initial conditions. In (1) and (2)
P = p + gz
(4)
not the third Cartesian coordinate that goes with x and y). The term gz in
(4) accounts for the gravitational body force.
The simultaneous presence of the nonlinear inertial terms on the left and
the second order viscous terms on the right makes it difficult to solve the
Navier-Stokes equation (1)-(2) in the general case. Under certain
circumstances, however, all the inertial terms on the left hand sides of (1) and
(2), while not zero, are small enough compared with the viscous term to be
neglected, and the y-derivative in the viscous term dominates over the xderivative. Under these conditions (1) and (2) simplify to
P
2u
0
+ 2
x
y
0
P
y
(5)
(6)
These equations are similar in form to the equations for a truly fully
developed flow. The velocity profile at a station x in this diverging and
possibly unsteady flow will thus be identical to the profile in a fully
developed flow with the same height, the same boundary conditions at y=0
and y=h, and the same pressure gradient. The flow can be said to be locally
fully developed, that is, having at every station x essentially the same velocity
profile as a fully developed flow with the same cross-sectional geometry and
boundary conditions. For example, if the flow is bounded by solid, immobile
walls such that u=0 at y=0 and y=h, the local solution is the familiar
parabolic one
u
h 2 y
y P
.
1
2
h h x
(7)
u
2u
<< 2
t
y
(8)
u
2u
<< 2
x
y
(9)
u
2u
<< 2
y
y
(10)
2u
2u
<<
x 2
y 2
(11)
where we have implied, but not indicated, that absolute magnitudes are
involved in the inequalities. These criteria can be expressed in more useful
form by estimating the orders of magnitude of the various derivatives in
terms of specified quantities. Let U be a characteristic, or typical, streamwise
velocity such that
u U
(12)
where the symbol in this case stands for order of magnitude, by which we
mean an estimate that is correct to within a factor of 3, say, implying that we
know the decade in which the quantity's numerical value lies on a logarithmic
scale. U might be the average flow speed at the channel's entrance at a
particular time, say. Similarly, let h be a characteristic value of the transverse
distance h in the problem, e.g. the value at the channel's entrance. Since the
dependence on y is expected to be parabolic when (1) serves as a good
approximation, we estimate, to order of magnitude, that
u U
.
y h
(13)
x L
(14)
.
t
(15)
x 2 L2
(16)
and (13) and the expected (near-) parabolic variation of u with y imply that
2u U
.
y 2 h 2
(17)
u
dy .
x
0
(18)
U
h .
L
(19)
Indicating only the orders of magnitude of all terms except the pressure
gradients, we can now write (1) and (2) as
U U 2 U 2
U U
p
+
+
+ 2 + 2
L h
L
L
x
U U 2 U 2 h
U U h
p
+
+
+ 2 + 2
L h L
L
L L
y
(20)
(21)
(22)
h2
<< 1
(23)
Uh h
<< 1 .
(24)
The same criteria also ensure that (2) reduces to (6). This becomes apparent
when we think of (20) and (21) as providing order-of-magnitude estimates
for the respective pressure gradients on their right-hand sides. A comparison
of (21) and (20) shows that
P h P
.
y L x
(25)
what is small enough, but it can be done only on a case by case basis after a
particular solution has been obtained. Equations (22)-(24) serve as an
adequate estimate, however, and are conveniently expressed in general terms.
They suffice if the estimates come out truly small so that there is little doubt
of the assumptions being satisfied.
pviscous =
12UL
h2
(26)
The pressure drop at the entrance and across the transition region between
the two segments depends on whether the Reynolds number is small or
large. If the Reynolds number is small, we estimate it from (26) with Lh,
assuming the transition occurs in a distance of at most a few h; if it is large,
we claim it cannot exceed the maximum Bernoulli drop. Thus,
Fig. 2: Pressure distribution under (a) inertia-dominated, (b) intermediate and (c) highly
viscous (locally fully developed) conditions.
10
12Uh
ptransitions
h 2
ptransitions
U 2
2
if Re =
Uh
<1
(27)
if Re =
Vh
>1
(28)
if
Uh
<1
(29)
ptransitions 1 Uh h
12 L
pviscous
if
Vh
>1
(30)
(31)
Uh h
<< 1
L
(32)
h2
<< 1
(33)
Note that the requirement (31) for constant pressure across area changes is
more severe than (22), which refers to the flow inside the channel sections
only.
11
(34)
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
h
h
that
1 , Re L 1 , we
know that the
velocity
profile
in the
gap
is:
L
L
2
y
y h(x,t)2 dp y
+
vx (x, y,t) = U 1
2 dx h(x,t) h(x,t)
h(x,t)
Couette
Poiseuille
(wall driven) (pressure driven)
(1)
Integrating this expression gives an expression for the local flow rate q at any given slice:
q(x,t)
h(x,t )
vx dy =
h(x,t)3 p 1
+ Uh(x,t)
12 x 2
(2)
**Note that this may not be constant along the channel because of the squeezing flow
induced by vertical motion of the channel boundary at h(x,t).
2. We combine with this with an integrated analysis of the flow in the vertical direction, and
the use of what is commonly called the kinematic boundary condition at the upper plate.
This links the displacement in the boundary motion to the local
fluid
flow rate
Essentially
we apply conservation of mass to the thin strip of fluid h ( x,t ) dx
a).
We start with
vx vy
+
= 0 , integrate
in y to
obtain
x
y
h( x )
h( x )
vx
dx +
x
0
(? ) + Vy
y= h ( x,t )
y= 0
vy
y
dy = 0 (3a)
=0
(3b)
a x dy = x
a f dy
f
b
y=b
db
f
dx
y= a
da
dx
vx
d h
dh
dy =
vx dy vx
x
dx 0
dx
y= h
d(0)
U
y= 0
dx
(4)
The first term
on the right-hand
side is nothing
but the
variation
in the
flow rate
of fluid
along
the gap (which again
we re-emphasize might not be constant because of the
squeezing
induced
by
the
top plate
etc).
The terms
in { } again have to be evaluated
based
on the
actual boundary
conditions
of the
specific
problem at the top and bottom
plates.
Combining eq.(3) with 2(b) results in the expression:
dq dh
U
dx dx
y= h
+V
y= h
=0
(5)
[Alternately you can apply a control volume analysis directly to a thin slice of the form
shown
in the
sketch
above
to
arrive
at exactly the same final result].
b) To simplify this result, we can recognize that the equation for the location of the free
surface is simply the solution of the equation: y h(x,t) = 0 . Since material points on the
surface
stay
on the
surface as it moves and deforms, we can
take the convected derivative of
this expression
so that
D
( y h(x,t)) = 0
Dt
h
vx
t
y=h
dh
+ vy
dx
y=h
=0
(6)
This is commonly known as the
kinematic boundary
condition and it is of exactly the same
form
as the
expression
derived in eq. (5). Combining eq. (5) and (6)
thus
results
in the
following
much simpler expression:
dq h
+
=0
dx t
(7)
3) We can now combine this with the result obtained
in eq. (2) to
obtain
the following
nonlinear coupled PDE:
1 3 p
h
h
+ 12
h
= 6U
x
x
t
x
(8)
This
is referred to
as
the general form
of Reynolds Lubrication
Equation.
If the shape
of the
boundary and its displacement are specified (i.e. h(x,t) is given) then
it
gives a second
order differential equation
that can
be
solved
for the
pressure
field
p(x,t) (subject to
appropriate boundary
conditions on
the pressure field at the edge of the fluid film).
Alternatively it can be combined with a global force balance on a moving block or plate or
other object that has a thin film
of viscous fluid underneath it to develop
a differential
equation for the motion of the block. It is thus the starting
point for many different
research
topics
as
well as
homework or qualifying problems!
2
1
or
1 or
1
L
L
V
L
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Oil is conned in a 10 [cm] diameter cylinder by a piston with a clearance of 0.0002 [cm]. The piston is 5
[cm] long, and the oil has a viscosity coecient of 0.05 [kg/ms] and a density of 920 [kg/m3 ].
A total weight of 100 [kg] is applied to the piston. Estimate the leakage rate of oil past the piston, in
liters/day. Justify any approximations you use in arriving at your estimate.
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
air, Pa
water, w > o
w o
oil; o , o
oil layer
An oil barge has developed a ne crack in its side, running a length L perpendicular to the sketch. Oil leaks
out of the crack and runs up the side of the barge (inclined at an angle ) in a very thin layer, as sketched.
Assume that the ow in the oil layer is highly viscous, that the oil is less dense than the water (0 < w ),
and that it is much more viscous than water (0 w ).
(a) If the oil layer is found to have a thickness b, what is the oil volume ow rate Q out through the slit?
(b) Describe qualitatively how the eld diers when the viscosity of the water is not negligible compared
with the oil viscosity.
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
air, pa
g
h(x, t)
viscous liquid
A rigid plane surface is inclined at an angle relative to the horizontal and wetted by a thin layer of highly
viscous liquid which begins to ow down the incline.
(a) Show that if the ow is two-dimensional and in the inertia-free limit, and if the angle of the inclination
is not too small, the local thickness h(x, t) of the liquid layer obeys the equation
h
h
+c
=0
t
x
where
c=
gh2
sin
(b) Demonstrate that the result of (a) implies that in a region where h decreases in the ow direction,
the angle of the free surface relative to the inclined plane will steepen as the uid ows down the
incline, while in a region where h increases in the ow direction, the reverse is true. Does this explain
something about what happens to slow-drying paint when it is applied to an inclined surface?
(c) Considering the result of (b) above, do you think that the steady-state solutions of the previous
problems would ever apply in practice? Discuss.
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A at plate of breath L and length much greater than its breadth is attached to a plane oor by a hinge.
The hinge has a radius R as shown. The plate is initially at a small angle 0 relative to the oor, and the
region between it and the oor is lled with a viscous liquid. Starting at t = 0, the plate is forced toward
the oor at a constant angular rate d
dt = .
Obtain an expression for the pressure distribution p(x, t) under the plate in the limit of highly viscous
(inertia-free) ow. The given quantities are L, R, , , , , and the atmospheric pressure pa outside
the plate.
Derive an expression for the vertically force Fytip (t) which must be applied at the right-hand tip of the
plate to make it close down at the specic constant angular rate.
Write down the criteria which must be satised for your solutions to apply
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
The sketch shows a circular bearing pad which rests on a at base through the intermediary of a lm of
viscous liquid of instantaneous thickness h(t). The load W causes the pad to sink slowly at the speed S, and
this squeezes the liquid out from under the pad. Assume that h D, that the viscosity is very high, and
that the speed S is very small.
(a) Making approximations (state them precisely) consistent with theses assumptions, show that the
settling speed is
32 W h3
S=
.
(6.21a)
3 D4
(b) An apparatus with two very at plates of 0.3 m diameter carries a load of 100 kg on a lm 0.003
2
gm
cm thick. If the liquid is a heavy oil with a kinematic viscosity of 10 cm
s and a density of 0.93 cm3 ,
estimate the speed S.
(c) If the load W is constant, and the gap width is h0 at time zero, show that the width h varies with
time accordingly to
12
h
64 W h20
= 1+
t
.
(6.21b)
3 D4
h0
(d) Calculate, for the initial conditions of part (b), the time (in hours) required for the gap width to
be decreased to half its initial value.
(e) Suppose now that the initial thickness is h0 , and that a constant upward force F pulls the disk
away from the base. Show that the disk will be pulled away in a time
t =
3 D4
.
64 h20 F
(6.21c)
NB When h0 is very small, the time t is very large. This is the basis for the phenomenon of viscous
adhesion, e.g., adhesives such as Scotch tape, or the apparent adhesion of accurately-ground metal surfaces.
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
An innitely long, cylindrical container of radius R rotates at the angular speed S. It contains water which
is also in solid-body rotation with angular speed S. At time t = 0, the container suddenly stops rotating,
and the contained water gradually comes to rest.
In all that follows, the possible eects of turbulence and other instabilities are to be considered absent.
(a) Sketch curves of Ve versus r, showing how the circumferential velocity varies with radius for several
successive times, t > 0.
(b) What is the order of magnitude of the time, tR , up to which the Rayleighs solution for impulsive
start of a at plate would describe the motion near the wall?
(c) Suppose that S = 33 - 1/3 [rpm], R = 10[cm], and that the uid is water at 20 degrees celcius.
Make a very rough estimate of the time, in seconds required for most of the motion to disappear.
c 2008, MIT
Copyright
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider Stokes First Problem: impulsive start of a at plate beneath a semi-innite layer of initially
quiescent incompressible uid. The governing equations (presuming parallel ow no instabilities) for
u(y, t) are:
2u
,
y 2
u
t
u(y = 0, t)
U ,
(2)
u(y , t)
0,
(3)
u(y, t = 0)
0.
(4)
0<y<,
(1)
u
y
(5)
.
y=0
Here is the density and is the dynamic viscosity. The shear stress at the wall will be of the form
W = CU 1 2 3 t4 ,
(6)
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
(1)
where the phase is independent of t and C is a non-dimensional constant. Find the exponents 1 , 2 , 3
and 4 by dimensional analysis.
Hint: (one approach): See Hint for Stokes First Problem; make good use of the steady-periodic form
of the solution.
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Oil is conned in a 10 [cm] diameter cylinder by a piston with a clearance of 0.0002 [cm]. The piston is 5
[cm] long, and the oil has a viscosity coecient of 0.05 [kg/ms] and a density of 920 [kg/m3 ].
A total weight of 100 [kg] is applied to the piston. Estimate the leakage rate of oil past the piston, in
liters/day. Justify any approximations you use in arriving at your estimate.
c 2010, MIT
Copyright @
Viscous Flows
Solution:
Lets calculate the Reynolds number for this problem,
Re =
V H
920 U 0.00002
=
= 0.368U,
0.05
(6.01a)
then, even assuming a velocity of 0.1[m/s], the Reynolds number is small, and the lubrication approximation
1
(Low Reynolds number) can be used as an initial assumption 2 . Also, well assume a pressure driven ow
because, from the ow geometry,
Upiston =
Qoil
UOil,avg 2RH
H
=
= 2UOil,avg ,
2
R
R
Apiston
(6.01b)
0=
1 p 1
vz
+
r
,
z
r r
r
(6.01c)
where the rst term, pressure drop rate, is approximately a constant across the space between the cylinders
(long cylinder/small gap approximation), then
1
vz
0 = K +
r
.
r r
r
(6.01d)
Furthermore, since the gap is small, 1 h/R, it is possible to approximate this problem using a local
cartesian coordinate system, (see problem 6.4, Sonin and Shapiro)
0 = K +
(vz )2
.
y 2
(6.01e)
then, integrating,
vz = K
y2
+ C1 y + C2 .
2
(6.01f)
vz (y = 0) = 0,
vz (y = H) = 0,
(6.01g)
(6.01h)
1 The lubrication approximation, Re h < 1 is really less restrictive, the Reynolds number can be large, as long as this
hL
h
combination of parameters gives a small number still, but since L
< 1 having the Reynolds number small is more than enough.
2 Well verify this later.
c 2010, MIT
Copyright @
Viscous Flows
where H = R2 R1 = 0.002[cm], and the origin of the reference coordinate system, y = 0, is located at
r = R1 . Now, using the B.C., the constants can be obtained,
C2 = 0,
H
C1 = K ,
2
(6.01i)
(6.01j)
then,
vz = K
Hy
y2
.
2
2
(6.01k)
then, the resulting velocity prole is a Poiseuille Flow (notice that this could have been inferred since the
beginning of the problem). Now, to obtain the ux, lets integrate this expression,
vz = 2R
2R
0
K
y2
Hy
dy.
2
2
(6.01l)
After integration,
Q = RK
y3
Hy 2
3
2
(6.01m)
0
then, nally,
Q=
R
RP 3
KH 3 =
H .
6
6L
(6.01n)
(6.01o)
In this case the weight of the piston creates almost all the total pressure and viscous stresses can be neglected.
To justify this, lets take a look at the viscous stresses for a second. The viscous force upwards is supplied
c 2010, MIT
Copyright @
Viscous Flows
by the pressure force exerted on the liquid just beneath and entering the gap between the piston and the
cylinder, and cant be higher than that (think about Momentum Conservation in a CV that encompasses
the oil around the piston), then,
Fviscous P Agap 2RHP,
(6.01p)
on the other hand, the force coming from the pressure beneath the piston is
FP iston
F ace
= P R2 ,
(6.01q)
then, the viscous force is much more smaller than the pressure force beneath the piston and almost completely
balances by itself the piston weight 3 . Then,
W + P R2 = 0.
(6.01r)
W
mg
100 9.8[N ]
39.2e4
+ Pa =
+ Pa =
+ Pa =
[N/m2 ] + Pa
(0.05)2 [m2 ]
Area
(r)2
(6.01s)
Here, the change in pressure due to gravity was neglected because the distance is small to be important,
more precisely, the total pressure change due to gravity is
P = gH = 920 9.8 0.05[N/m2 ] = 450[P a],
(6.01t)
(6.01u)
3 The acceleration is small because we have assumed a slow process as part of the low Reynolds assumption, to be checked
later.
c 2010, MIT
Copyright @
Viscous Flows
Q=
(0.05[m])(39.2e4[P a])
((2e(5))[m])3 = 1.04e(8)[m3 /s],
6(0.05[m])(0.05[kg/(ms)])
(6.01v)
(6.01w)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
air, Pa
water, w > o
w o
oil; o , o
oil layer
An oil barge has developed a ne crack in its side, running a length L perpendicular to the sketch. Oil leaks
out of the crack and runs up the side of the barge (inclined at an angle ) in a very thin layer, as sketched.
Assume that the ow in the oil layer is highly viscous, that the oil is less dense than the water (0 < w ),
and that it is much more viscous than water (0 w ).
(a) If the oil layer is found to have a thickness b, what is the oil volume ow rate Q out through the slit?
(b) Describe qualitatively how the eld diers when the viscosity of the water is not negligible compared
with the oil viscosity.
c 2010, MIT
Copyright @
Viscous Flows
Solution:
(a) Since we are told that the oil layer is both thin and viscous, we assume fully developed ow. This, along
with assumption that the ow is steady, allows us to ignore all inertial components of the x-momentum
equation (Navier-Stokes).
p
2 vx
2 vx
+
+ o gx
0=
+ o
2
x
x
y 2
d
Also because of our thin-lm approximation, we can neglect variations in the x-direction ( dx
0).
Now we have that
0=
p
+ o
x
d2 vx
dy 2
+ o gx
p = f (y)
h(x)
y
dp
= w g cos
dx
w
gx = g cos
y
equation:
0 = w g cos + o
d2 vx
o g cos
dy 2
d2 vx
w o
=
g cos
2
dy
o
w o
g cos y 2 + C1 y + C2
2o
To solve for C1 and C2 we must apply boundary conditions. At y = 0, we know that vx = 0 such
that C2 = 0. At the interface between oil and water (y = b) we know that the shear stress must be
continuous such that
dvx
dvx
o
= w
dy y=b+
dy y=b
But since were told that w o , we can say that
dvx
dx
y=b
dvx
dy y=b
w o
g cos b + C1 = 0
o
0. Therefore,
C1 =
w o
g cos b
o
w o
o
2
by
y2
2
c 2010, MIT
Copyright @
Viscous Flows
Q=
0
vx dy = g cos
w o
o
b3
3
w , w
o , o
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25
Corrected Solution
Shapiro 6.16
air, pa
g
h(x, t)
viscous liquid
A rigid plane surface is inclined at an angle relative to the horizontal and wetted by a
thin layer of highly viscous liquid which begins to ow down the incline.
1. Show that if the ow is two-dimensional and in the inertia-free limit, and if the angle of
the inclination is not too small, the local thickness h(x, t) of the liquid layer obeys the
equation
h
h
+c
=0
t
x
where
gh2
c=
sin
2. Demonstrate that the result of (a) implies that in a region where h decreases in the ow
direction, the angle of the free surface relative to the inclined plane will steepen as the
uid ows down the incline, while in a region where h increases in the ow direction, the
reverse is true. Does this explain something about what happens to slow-drying paint
when it is applied to an inclined surface?
3. Considering the result of (b) above, do you think that the steady-state solutions of the
previous problems would ever apply in practice? Discuss.
2.25
Corrected Solution
Shapiro 6.16
1. Assumptions:
ReH H
L 1
two dimensional ow
THIN layer
H
L
characteristic height
characteristic length
tU
L
p
p =
P
vx
U
vy
vy =
V
x =
vx =
t =
U vx
V vy
+
=0
L y
L x
Since dimensionless variables are assumed to be of the same order (O(1)),
U
V
L
H
V =
H
U
L
where
H
1
L
x-direction:
U2
L
vx
vx
vx
+
v
+
v
x
y
t
x
y
Divide through by
U H 2
L
P p
+ g sin +
L x
U 2 vx
U 2 vx
+
2
L2 x
H 2 y 2
U
:
H2
vx
vx
vx
+
v
+
v
x
y
t
x
y
PH 2 p gH 2 sin
H
+
+
U L x
U
L
ReH H
=small
L
2 vx 2 vx
+
x 2 y 2
small
y-direction:
vy
vy
vy
+
v
+
v
x
y
t
x
y
U2 H
L L
Divide through by
U H 2
L
P p
H
g cos +
H y
L
U 2 vy
U 2 vy
+
L2 x 2 H 2 y 2
U
:
H2
vy
vy
vy
+
v
+
v
x
y
t
x
y
PH p gH 2 cos
H
U y
U
L
small
=small
ReH ( H
L )
2 vy H 2 vy
+
L y 2
x 2 |{z}
small
2.25
Corrected Solution
Shapiro 6.16
Corrected part starts from here: Note that it is generally better and also more precise
not to do scaling for pressure before writing the simplied Navier-Stokes equation
in both x and y components (long and short scales in the problem...for a problem
in cylindrical coordinates it may turn into r and z for example). One should do
the scaling for pressure after simplifying the NSE and based on the order of the left
terms. This is especially very important in the cases that we have a free surface at
one end of the small scale coordinate (y coordinate in this problem). A common
mistake is to use the scaling for pressure that we had in the lubrication problem for
bearings. Whenever there is a free surface at one end of the smaller scale which its
height is a function of time, i.e. h(x, t) it is a good idea to start from the simplied
N.S.E for the small scale component (here y component):
y-direction:
3 2
vy H 2 vy
vy
vy
v
U H 2
PH p gH 2 cos
H
y
+
v
+
v
=
+
+
x
y
L
t x
y
U y
U
L
L y 2
x 2 |{z}
| {z }
| {z }
small
2
small
ReH ( H
=small
L)
p
h
h
= gcos
gcos << gsin
x
x
L
Now if we look at the simplied x-component of NSE in the dimensional form and
use the fact that p/x << gsin then we will have: (we are assuming here that
is not vanishingly small)
0 = g sin +
2 vx
y 2
MIT 2012, Corrected by B.K.
2.25
Corrected Solution
Shapiro 6.16
Please note that here the gravity is running the ow and the viscous terms are
resisting against it. In free surface lubrication problems usually there is a way to
determine the pressure gradient term p/x and then usually you end up nding
that the main terms driving the ow is gravity, or maybe centripetal acceleration in
the rotational frames and the resisting terms are of viscous nature like the general
case.
The rest of solution will be as it was before(please also check my explanations for
conservation of mass written few lines below): Going back to the dimensional form,
0 = g sin +
2 vx
y 2
(1)
dy
2
y
2 vx
g sin
=
y + C1
2
y
vx
y
= 0 at y = h (free surface),
gh sin
2
g sin
vx
=
(y h)
y 2
C1 =
g sin
1
(hy y 2 )
(2)
(5) Use mass conservation to obtain a single evolution equation for h(x, t).
This is a very strong tool when you have a free surface problem in which h is a
function of time as well as position i.e. h(x, t) to obtain an evolution equation for
h(x, t) Consider the following control volume in the limit of x 0:
d
dt
dV +
CV
lim
x0
v
=0
(v
{
c ) ndA
{
CS
d
p dt
(hx)
h +
p
A
x
A
Rh
0
vx dy
+
t
x
=0
Z
vx dy
0
h Q
+
=0
t
x
(3)
Q
Vx (y).
The above equation can also be derived by combining the kinematic boundary condition,
h
h
t + vx x = vy |y=h(x) , with conservation of mass.
4
2.25
Corrected Solution
Shapiro 6.16
Note that this form is true only for Cartesian Coordinates. It is easy to derive
a similar equation for cylindrical coordinates and I leave it as an excercise for you.
Your nal answer in cylindrical coordinates will be:
h 1 (rQ)
+
=0
t
r r
(6) Combine Eq. (2) with Eq. (3):
Z h
h
g sin
1
+
(hy y 2 )dy = 0
t
x 0
2
3
h g sin h
+
=0
t
x 3
h
g sin 2 h
h
+
=0
t
x
|
{z
}
(4)
Eq. (4) is a nonlinear wave equation with a solution of the form h = f(xct), where
c is the wave speed.
t
x
h
Thus, where h is decreasing locally ( h
x < 0), h increases in time ( t > 0).
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
A at plate of breath L and length much greater than its breadth is attached to a plane oor by a hinge.
The hinge has a radius R as shown. The plate is initially at a small angle 0 relative to the oor, and the
region between it and the oor is lled with a viscous liquid. Starting at t = 0, the plate is forced toward
the oor at a constant angular rate d
dt = .
Obtain an expression for the pressure distribution p(x, t) under the plate in the limit of highly viscous
(inertia-free) ow. The given quantities are L, R, , , , , and the atmospheric pressure pa outside
the plate.
Derive an expression for the vertically force Fytip (t) which must be applied at the right-hand tip of the
plate to make it close down at the specic constant angular rate.
Write down the criteria which must be satised for your solutions to apply
c 2010, MIT
Copyright @
Viscous Flows
Solution:
c) First, in order to realize an order of magnitude analysis of the terms in the N-S equations to show the
validity of the lubrication approximation, lets gather more information about the order of magnitude
of the velocities. First, lets use the mass conservation equation (by now this process should be quite
familiar to you),
u v
= 0,
(6.20a)
+
x y
and then dimensional analysis,
Uavg
L
Vavg
L
= 0,
(6.20b)
simplifying,
Vavg = Uavg .
(6.20c)
Now, lets make an order of magnitude analysis. From the problem statement,
1,
R L,
(6.20d)
(6.20e)
Vavg Uavg .
(6.20f)
then,
Now, using mass conservation again, assuming an incompressible liquid and using a triangular C.V.
whose top surface moves with the hinged plate, and has a xed base length (remember Shapiro 3.5),
V ol
+
t
u(x, t) ndA = 0,
(6.20g)
CS
then, the change in the C.V. volume must be equal to the ow coming out of the C.V..
Now using the C.V. geometry,
V ol
x
=
x = Uavg (x, t)x =
t
2
then,
Uavg (x, t) =
x
.
2
u(x, t) ndA,
(6.20h)
CS
(6.20i)
c 2010, MIT
Copyright @
Viscous Flows
Now, we have enough information to make an order of magnitude analysis. For the N-S equation in
the x direction,
u
u
u
p
2u
2u
+ u
+ v
=
+
+
.
y
x
x2
t_
y 2
._a
. _ax_
._a_
. _a _
._a_
2
2
2
2 2
( x
( x
( x
( x
( x
2 ) /x
2 )/x
2 )/( )
2 ) /x
2 )/(x )
(6.20j)
2x
23 x
x direction
(6.20k)
y direction
(6.20l)
then, the viscous stresses in the y direction are larger than those in the x direction. As a consequence,
the right hand side terms (not considering the pressure) scale as 2
3 x . On the other hand, the left
hand side terms scale as
2 x
2 2 .
23 x
2 x
,
22
(6.20m)
then,
x2
1,
(6.20n)
L2
1,
(6.20o)
since this could be the least viscous point of the system. Notice that this non-dimensional number is
a modied Re number, which can be easily visualized reordering,
(L/)(L)
1.
(6.20p)
dP
dx
h3
,
12
(6.20q)
then,
dP
6x2
6
=
= 3 ,
(6.20r)
x
dx
h3
then, integrating to obtain an equation for pressure (assuming atmospheric pressure outside of the
wedge),
Z L
Z L
dP
6
dx =
dx,
(6.20s)
3
dx
x
x x
nally,
p(x, t) patm =
6
ln
3
L
.
x
(6.20t)
+ patm .
(6.20u)
6
(0 t)3
ln
L
x
In this equation, the pressure blows up as the distance from the hinge is reduced, but is integrable
and thus produces a nite net force upwards on the plate.
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
Viscous Flows
b) For 1, taking moments around the hinge, and assuming the hinge to be thin,
Fytip (t)L =
0
3 L2
.
2 3
(6.20v)
=
=
=
(6.20w)
(6.20x)
(6.20y)
So,
L
3
Fytip (t) = 32 L
3 = 2 (0 t)3 ,
(6.20z)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
The sketch shows a circular bearing pad which rests on a at base through the intermediary of a lm of
viscous liquid of instantaneous thickness h(t). The load W causes the pad to sink slowly at the speed S, and
this squeezes the liquid out from under the pad. Assume that h D, that the viscosity is very high, and
that the speed S is very small.
(a) Making approximations (state them precisely) consistent with theses assumptions, show that the
settling speed is
32 W h3
S=
.
(6.21a)
3 D4
(b) An apparatus with two very at plates of 0.3 m diameter carries a load of 100 kg on a lm 0.003
2
gm
cm thick. If the liquid is a heavy oil with a kinematic viscosity of 10 cm
s and a density of 0.93 cm3 ,
estimate the speed S.
(c) If the load W is constant, and the gap width is h0 at time zero, show that the width h varies with
time accordingly to
12
h
64 W h20
= 1+
t
.
(6.21b)
3 D4
h0
(d) Calculate, for the initial conditions of part (b), the time (in hours) required for the gap width to
be decreased to half its initial value.
(e) Suppose now that the initial thickness is h0 , and that a constant upward force F pulls the disk
away from the base. Show that the disk will be pulled away in a time
t =
3 D4
.
64 h20 F
(6.21c)
NB When h0 is very small, the time t is very large. This is the basis for the phenomenon of viscous
adhesion, e.g., adhesives such as Scotch tape, or the apparent adhesion of accurately-ground metal surfaces.
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
Viscous Flows
Solution:
(a) In terms of the ow geometry, the problem is similar to 6.3; the dierence here being that the
ow is unsteady. First, lets use the continuity equation to get some information about the order of
magnitude of the velocities, 1
1
uz
ur
uz
(rur ) +
= 0,
.
r r
z
r
z
(6.21d)
Now, lets go to the NS equations. Writing the NS equations in cylindrical gap between the bearing
pad and ground,
ur
2 u
u
ur u ur
ur
u2
p
1
ur
1 2 ur 2 ur
+ ur
+
+ uz
= +
r
+ 2
+
2
2
r2
t
r
r r
r
r 2
raa
r aar
rr aa
r zaa r
rz
aa
raa
rr aa
r
aa
} r aa
I
II
=0
III
IV
=0
VI
=0
(6.21e)
where we relabeled the terms to further make an order of magnitude estimation, also notice that due to
(6.21f)
so we can neglect the term IV compared with term V . Also, notice that V I IV , then V I can be
neglected when compared to V . Now, lets compare II to V , then
II
vr D h2
1,
V
D2
(6.21g)
then II can also be neglected when compared to V . Now, when comparing III to V , we rst notice
that II III since
uz /z
II
ur /r
1,
uz /z
III
uz /z
(6.21h)
where we have used the information obtained from mass conservation. Hence, III vanishes when
compared to V .
Now, when comparing I to V , we have
I
h2
,
V
(6.21i)
where is the time scale involved in the process. The source of unsteadyness is the pad settling down,
which renders ur and other ow variables time-dependent. Hence,
h
Sh
,
S
(6.21j)
Since it is given that S is very small and also h is small, we can safely assume that I can be neglected
compared to V . Hence we have the NS equation as
1 By now, you should be familiar with this method of obtaining extra information that can be quite useful when comparing
terms in the NS equations.
2 Well compare the terms with term V because, since the gap is small, this is likely to be the largest derivative.
c 2010, MIT
Copyright @
Viscous Flows
p
2 ur
+ 2 = 0.
r
z
(6.21k)
Q(r) = 2r
0
1
2
p
(zh z 2 )
r
h3 r
ur (z)dz =
6
h3 r
=
6
(6.21l)
dp
dr
(6.21m)
Note that Q is a function of r, since the settling down of the pad drives greater and greater ow rates
as r increases! This can be veried by applying mass conservation in a cylindrical CV, as shown. The
height of this CV changes as h(t).
Since
dh
dt
(V V cs ) ndA = 0,
dV +
CV
(6.21n)
CV
dh
= 2rhur = Q(r).
(r2 h) + 2rhur = 0, r2
t
dt
(6.21o)
= S, we have
Q(r) = r2 S.
(6.21p)
Now, we can nd the value of the pressure gradient using the ow, then
dp
6Q
6r2 S
6rS
= 3
=
=
,
3
h r
h
dr
h r
dp =
6Sr
3S
dr, p(r) = 3
h
h3
D2
r2 ,
4
(6.21q)
(6.21r)
D
2
D2
3SD4
2
r 2rdr =
,
4
32h3
3
(6.21s)
c 2010, MIT
Copyright @
Viscous Flows
(6.21t)
32 100 27 1015
= 1.2 109 [m/s]
3 0.93 81 104
(6.21u)
S=
32W h3
dh
=
dt
3D4
h0
dh'
=
h'3
32W
dt' ,
3D4
(6.21v)
h0
32W t'
=
3D4
1
1
64W t
1
1
64W h20
2 =
2 = 2 1+
t
2
4
h
h0
3D
h
h0
3D4
(6.21w)
which gives
12
h
64W h20
= 1+
t
.
3D4
h0
(6.21x)
(d) Using 2h = h0 , and plugging in the variables, we get the required time as 10.4 hours.
(e) Since S ' =
dh
dt ,
instead of dh
dt , we have
1
1
64F t
2 =
.
2
h
3D4
h0
(6.21y)
t =
3D4
.
64h20 F
(6.21z)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
An innitely long, cylindrical container of radius R rotates at the angular speed . It contains water which
is also in solid-body rotation with angular speed . At time t = 0, the container suddenly stops rotating,
and the contained water gradually comes to rest.
In all that follows, the possible eects of turbulence and other instabilities are to be considered absent.
(a) Sketch curves of V versus r, showing how the circumferential velocity varies with radius for several
successive times, t > 0.
(b) What is the order of magnitude of the time, tR , up to which the Rayleighs solution for impulsive
start of a at plate would describe the motion near the wall?
(c) Suppose that = 33 1/3 [rpm], R = 10[cm], and that the uid is water at 20 degrees celcius.
Make a very rough estimate of the time, in seconds required for most of the motion to disappear.
c 2008, MIT
Copyright @
Solution:
(a)
(b) As soon as the cylinder is stopped, there is diusion of momentum from the wall towards the center.
The present problem in cylindrical geometry diers from Rayleighs problem in that the velocity away
from the wall (at r = 0) is zero at all times. In Rayleighs problem the velocity boundary condition
2
away from the wall is U . By tR = R , he diusion scale, the impact of the wall has diused towards
2
the center, where the BC, as we discussed, is dierent from Rayleighs problem. Hence, by tR = R ,
there is already a back diusion of this BC towards the wall and the near-wall region has started
feeling this eect. For Rayleighs solution to be valid near the wall, we have
tR =
R2
.
(8.02a)
2
(c) For a very rough estimate, we again observe that the diusion time scale, tL = L . We know that
the Rayleighs solution involves the error function. Assuming a similar variation here for a very rough
estimate, we can say
(( )
tR
V
erf
.
(8.02b)
r
t
From the behavior of the error function, we know that all the motion will decay within t 10tR (rough
order of magnitude). Hence, motion-decay time is of the order of 105 seconds.
D
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Consider Stokes First Problem: impulsive start of a at plate beneath a semi-innite layer of initially
quiescent incompressible uid. The governing equations (presuming parallel ow no instabilities) for
u(y, t) are:
2u
,
y 2
u
t
u(y = 0, t)
U ,
(2)
u(y , t)
0,
(3)
u(y, t = 0)
0.
(4)
0<y<,
(1)
u
y
(5)
.
y=0
Here is the density and is the dynamic viscosity. The shear stress at the wall will be of the form
W = CU 1 2 3 t4 ,
(6)
c 2008, MIT
Copyright @
Viscous Flow
Solution:
where u
is dimensionless; y has units of length, L; y has units of length, L; t has units of time, T , and is
given in L2 T 1 . Then, there are three remaining variables and two remaining dimensions; therefore there is
one more dimensional group.
So, 1 = u
(or any multiple), and 2 = y 2 t2 . Now, choosing 2 , 2 such that,
2 + 2
0,
(7)
1 + 2
0,
(8)
2 = 2
2 = 1 ,
(9)
Then,
2 = t/y 2
(10)
1 = f cn0 (02 )
(11)
or
where 02 can be replaced with any function of 02 ; choose (convention) 2 = (02 )1/2 = y/ t
1 = f cn (2 ) ,
(12)
u
= f cn (y/ t) = fI (y/ t)
U
(13)
u
U
|y=o = fI' (0) ,
y
t
(14)
or
W =
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright @
Viscous Flow
=y
such that
u( )
= E,
U
(15)
then,
fI ( ) = E
t
(16)
=C ,
t
(17)
=C
Then, the eect of the wall penetrates as
(18)
t.
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
(1)
where the phase is independent of t and C is a non-dimensional constant. Find the exponents 1 , 2 , 3
and 4 by dimensional analysis.
Hint: (one approach): See Hint for Stokes First Problem; make good use of the steady-periodic form
of the solution.
c 2008, MIT
Copyright @
Viscous Flow
Solution:
Lets start by non dimesionalizing the equations. Now write u =
2 u
u
=
t
y 2
u
U;
thus divided by (U ).
0 < y < ,
(2)
u (y = 0, t) = cos(t),
(3)
u (y , t) 0,
(4)
u (y, t = 0) = 0,
(5)
u = f (y, t, , ),
(6)
and hence,
(7)
2 = ty 2 2 ,
(8)
2 = t,
(9)
a3 = y 3 3 ,
(10)
3 = 1,
(11)
then,
a3 =
,
y2
(12)
3 =
(13)
or,
(14)
u
y
= f ( , t),
U
(15)
or,
c 2008, MIT
Copyright @
Viscous Flow
u
U
u
y
= A !
U
where,
A(0) = 1
y
cos t + !
(16)
(0) = 0.
and
(17)
Furthermore,
W =
u
|y=0
y
(18)
1
1
W = U A1 (0) ! ) cos(t + (0)) A(0) ! ) sin(t + (0))1 (0)
1
W = U !
(19)
(20)
A1 (0)2 + 1 (0)
0.5
cos(t + ( 0) ).
(21)
'
(0)
1
1
where, 0 = arctan A
' (0) ; and A (0) and (0) are universal constants.
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
In what follows, we simplify the exposition by considering only steady, inviscid flows
with a conservative
body forces (of which gravity is an example). A conservative force
per unit mass G is one that may be expressed as the gradient of a time-invariant scalar
function,
G = U(r ) ,
(1)
V V = p U(r) .
(2)
A uniform gravitational force per unit mass g pointing in the negative z direction is
represented by the potential
U = gz .
(3)
A streamline coordinate system is not chosen arbitrarily, but follows from the
velocity field (which, we note, is not known priori). Associated uniquely with any point
r and time t in a flow field are (Fig. 2): the streamline that passes through the point
(streamlines cannot cross), the streamlines local radius of curvature R and center of
curvature, and the following triad of orthogonal unit vectors:
= is + in
+i
s
n
l
(4)
V = isV
(5)
and
where V is the magnitude of the velocity vector V . From (4) and (5),
V =V
s
1
f
f
f
f dr df (s, n, l ) =
ds +
dn +
dl . Equation (4) follows
s
n
l
f (s,n,l )
(6)
is defined by
and thus
V 2
2 is
.
(V )V = V (Vis ) = is + V
s
s 2
s
(7)
The unit vector in the last term of (7) changes orientation as one moves along the
streamline. The change dis in is from s to s+ds is obtained with the construction shown
in Fig. 2 as
ds
dis = in d = in
R
(8)
is
i
= n
s
R
(9)
V 2 V 2
(V )V = is in
R
s 2
(10)
The first term on the right is the convective acceleration in the direction of the velocity,
and the second is the centripetal acceleration, toward the center of curvature.
The pressure gradient in streamline coordinates is
p p p
p = is + in
+ il
s
n
l
(11)
Using (10) and (11) in (2), we obtain the equation of motion in streamline coordinates for
steady, inviscid flow as
s-direction:
n-direction:
1 p U
V 2
=
s s
s 2
(12)
V2
1 p U
=
R
n n
(13)
1 p U
l l
(14)
0=
l-direction:
n-direction:
V2
1 p
z
=
g
n
R
n
l-direction:
0=
1
2
V 2 =
1 p
z
g
s
s
1 p
z
g
l
l
(15)
(16)
(17)
For constant-density flow in a uniform gravitational field, the equations simplify further
to
s-direction:
V 2
p
+
gz
+
=0
s
2
(18)
n-direction:
V 2
( p + gz) = R
n
(19)
l-direction:
( p + gz) = 0
l
(20)
The s-direction equation (18) states Bernoullis theorem: the total pressurethe sum
p + gz + V 2 2 of the static, gravitational, and dynamic pressuresremains invariant
along a streamline.
The n-direction equation (19) states that when there is flow and the streamlines curve,
the sum p + gz (which is constant in when the fluid is static) increases in the ndirection, that is, as one moves away from the local center of curvature.
The l-direction equation (20) states that p + gz remains constant for small steps in
the binormal direction, that is, the pressure distribution is quasi-hydrostatic distribution in
the l-direction.
EXAMPLE
Consider the simple case of 2D, inviscid air flow over a smooth hill (Fig. 3). Far
upstream of the hill the incident velocity is uniform at V . The hill deflects the air around
it, and a uniform flow is again established far downstream. Far upstream, above, and
downstream of the hill, the pressure is constant at p and the streamlines are straight (the
hill does not perturb the flow at infinity). We shall assume that gravitational effects are
negligible (the medium is air and the hills elevation is modest) and the free streams
Mach number is small, so that and the density can be taken as constant. Based on the
available equations, what can we say about the pressure and velocity distributions over
the hillwhere is the velocity higher than V , for example, and where lower?
To answer this question accurately we need to know the shapes of the streamlines
throughout the flow fieldor, at least, in the region that is perturbed by the hill. We
dont have this information, so we proceed by drawing a rough estimate of the streamline
pattern, as shown in Fig. 3. The difference between the pressure at infinity and at the top
of the hill, point (3), can be estimated by integrating equation (19) along the vertical path
from (3) to ( ). Since this path follows the local n-direction, R>0 everywhere along it.
Neglecting the gravitational term, (19) gives
p V 2
=
n
R
from which we see that
(21)
p p3 =
V 2 dn
>0
R
(22)
Thus p3 < p , and according to Bernoullis equation (18), it follows that V3 > V . Using
similar arguments, we conclude that p1 = p and V1 = V , and p2 > p and V2 < V , etc.
In principle, if R(n) and V(n) can be established or estimated, the integral in (21) can
be evaluated. For example if we find that the flow perturbation caused by the hill is
negligible at elevations greater than some multiple of the height of the hilltop, we might
write for the path from (3) to ( )
R Rhill e
n
H
(23)
where Rhill is the streamlines radius of curvature in the vicinity of the hilltop, n is
measured from the top of the hill upward, and H = h is some multiple of the actual
height h of the hilltop, the coefficient being an empirical number. From Bernoullis
equation (18) we also have that
p+
V 2
V 2
= p + ,
2
2
(24)
Substituting for R and V into (21) from (23) and (24), respectively, we integrate (23) and
obtain
2H
V2 R hill
p p3 =
e 1
2
(25)
For a low hill such that 2H<<Rhill, the exponential term can be expanded and (25)
simplified to
V2 H
p p3
Rhill
(26)
The velocity at point (3) now follows from (24) and (25) as
V3 = Ve
H
R hill
(27)
V3 V 1+
Rhill
(28)
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
vx =
y
vy =
x
vz = 0
vr = 1r
v =
r
vz = 0
vr = 1r
z
v = 0
vz = 1r
r
1
vr = r2 sin
1
v = r sin
r
v = 0
vx =
vy =
vz =
vr =
v =
1
r
vz =
vr =
v =
1
r
v =
1
r sin
W (z) = (U iV )z
= Ux + V y
vx = U
= V x + U y
vy = V
r!
!
W (z) =
Q
2
ln(z z0 )
Q
2
ln r'
Q '
2
free vortex
z0
r!
!
W (z) =
=
i
2
vr =
Q 1
2 r'
v = 0
ln(z z0 )
'
2
vr = 0
= 2
ln r'
v =
1
2 r'
forced vortex
z0
r!
!
W (z) =
= Kr2
vr = 0
'2
v = Kr'
doublet (x-orientation)
r!
!
z0
W (z) =
=
c
zz0
c cos '
r'
= c sin
r'
vr = c cos
r '2
'
v = c sin
r'2
'
'
doublet (y-orientation)
r!
!
z0
W (z) =
ic
zz0
c sin '
r'
vr = c sin
r '2
c cos '
r'
v =
c cos '
r'2
z0
U
vr = U cos ' 1
W (z) = + i
R3
2r '2
'
R3
r'3
v = U sin ' 1 +
R3
r'
R3
2r '3
v = 0
shear flow
A
vx = 2Ay '
W (z) = @
z0
y
x
=@
vy = 0
= Ay '2
vz = 0
W (z) = 12 A(z z0 )2
z0
shown
for
A>0
y
x
vx = Ax'
= 12 A(x'2 y '2 )
vy = Ay '
= Ax' y '
vz = 0
Notes:
z = x + iy
z0 = x0 + iy0
0 < 2 =
W (z) = + i
r' = (x x0 )2 + (y y0 )2
' = tan1
yy0
xx0
vx = vr cos v sin
vr = vx cos + vy sin
vy = vr sin + v cos
v = vx sin + vy cos
1
2
dW
dz
= vx ivy
dW
dz
= (vr iv )ei
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
Dv ext +
pn dA
v
F = F
= + g g dV +
CV
(t)
CS(t)
+ B =v
Dt
t
1
Dv
Dv
2
2
p1
) + pg(z
=B-= 0 + g
(p2 =
+ g
2 z1 ) + v2 v1
2
Dt
Dt
Dvp = (g a)
v
Cauchy
V 2
Dv
= -p + g
+ B = (p
v
+ gz) =
p 2
= +pg2 vMomentum
Dt
t
)
n
R
dp
1( 2
Dt
Equation
ds +
+
v2 v12 + g (z2 z1 ) = 0
B = 0
2
1 t
1 v
Dv
+ -B =
(p + gz) = 0
p v
p
= p + g
INVISCID
VISCOUS
t
Dv
d
p=
Dt
Dv = _ +
g(g a)
v(v vc ) n dA
v dV
F=
-B+= 0
=
pI
E 2 Dt = E 2 + g
dt CV (t)
CS(t)
v Dt
dp 1 2
p
ds +
+
v2 v12 + g (z2 z1 ) = 0
Dv
pa)
Generalized
-p = (gBernoulli
t
2
1
1
+ g Eulers
Dv = _p
Streamline
v(v n) dA
F=
(v) dV +
E
E p + g Equation
2
2 t
Dt =
(
)
v
dp
1
Coordinates
CV
(t)
CS(t)
Dtd -OR ds +
+
v22 v12 p+ g (z2 z1 ) = 0
v(v vc ) n dA
v dV +
F=
p
Dv
t
2
1
1 Dv =
v dt CV (t)
momentum equations
Dv
= + g
v(v n) dA
F=
(v) dV +
Dv
Dt
)
t
Dv
=
p
p+
+ 1g
g ( 2
Dt =
CS(t)
2
(p
p
)
+
g(z
z
)
+
= CV
(t)
+ gE
Bernoulli
Dt
2
1
2
1
1
Dv
E
Important
2g 2
Dt
=
p
+
v
Dv
vDt+ B = v 2
limits
+g
pn dA
+
Dv F = Fext + Dt = g dV
V
t + B = v
CS(t)
= p + g CV (t)
t(p + gz) =
v
Dt
n + BB =
R
=0
Dv
1g 2
B=
0v
2
t
=
p
+
Control
v (p2Rigid
p1body
) + g(z z1 ) + v2 v1
Steady
+ Bmotion
= v Dt 2
2
(p
+B(g
gz)
Volume
p
=
0 =a)
p=
=
(g
a)0
t
f
Irrotational
2
(hydrostatics)
v
2
2
B ==vV
2 v
B = 0 t(p +
a)
+ gz)
v ds + 2 dp
dp p+=1
1 (g
vv122 +
vv222
g (z2 z1 ) = 0
n
R
ds
+
+
2
1 + g (z2 z1 ) = 0
j1 2
j1 2 t
2
t
2
p = (g a)
B=0
1 v
1 dp
1 2
+
+
v2 v12 + g (z2 z1 ) = 0
L
d
(p + gz) = 0 1
L t ds
d
2
1 F=
v(v
p = (g a)
FORM A
v(v
v
vcc )) n
n dA
dA
v dV
dV +
+j
F = dt j 1 v
CS(t)
CV
(t)
dt
L
CS(t)
d CV (t)
= pI
v(v vc ) n dA
v dV +
F=
E
L
dt CV (t)
CS(t)
dp
11
dp
v(v n) dA
F=j
(v) dV + j
FORM B
+
gz
BB = 2 vv vv +
++gz
t
L
CV
(t)
CS(t)
v(v n) dA
F=
(v) dV +
CV (t) t
CS(t)
j
j where
L
g dV +
pn dA
F = Fext +
CV (t)
Hosoi (2006)
CS(t)
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
m
+1
aU
1/2
(c) Show that the streamline passing through the stagnation points is given by = 0. Verify that the line
= 0 represents a closed oval-shaped body, whose maximum width h is given by the solution of the
equation
U h
h = a cot
.
m
The body generated by the superposition of a uniform stream and a source-sink pair is called a Rankine
body. It becomes a circular cylinder as the sourcesink pair approach each other.
c 2011, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
33 U 2
64 ag
c 2011, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
c 2011, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
2.25
Potential Flow
Consider the ow past a spinning cylinder. In a real uid, the angular motion would act to
impart a net circulation to the ow through the action of the uid viscosity. This circulation
(denoted by the constant ) may be incorporated articially into an irrotational ow model
by adding an irrotational vortex potential = /(2) to the velocity potential deduced for
potential ow over a cylinder. So the total potential will be of this form:
=U
r+
a2
r
cos /(2)
MIT 2012,B.K.
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
m
+1
aU
1/2
(c) Show that the streamline passing through the stagnation points is given by = 0. Verify that the line
= 0 represents a closed oval-shaped body, whose maximum width h is given by the solution of the
equation
U h
h = a cot
.
m
The body generated by the superposition of a uniform stream and a source-sink pair is called a Rankine
body. It becomes a circular cylinder as the sourcesink pair approach each other.
c 2010, MIT
Copyright @
Potential Flow
Solution:
(a)
W (z) = Wuniform ow + Wsource + Wsink
where W = + i, is the potential function, and the stream function.
Recap from Lecture: W satises the Laplace equation which is linear. Therefore, one can superimpose
its solutions as above.
Wuniform ow = U (x + iy)
m
m
m
ln rei =
ln r + i
Wsource =
2
2
2
m
m
m
Wsink =
ln rei =
ln r + i
2
2
2
y0
m
m
m
ln r + i
ln rei =
Wsource =
2
2
2
m
m
m
Wsink =
ln rei =
ln r + i
2
2
2
Wtotal = U x +
arbitrary
point (x, y)
location of
source/sink
x0
x
_
(x x0 )2 + (y y0 )2
y y0
= arctan
x x0
r=
m
m
(x + a)2 + y 2
ln
+i U y +
arctan
2
2
2
4 (x a) + y
y
x+a
m
arctan
2
y
xa
2
+ y2
m
(x
a)
= U +
x
4 (x + a)2 + y 2
2(x + a)
(x + a)2 + y 2
2(x a)
2
2
+ y2
(x
a)
[(x a)2 + y 2 ]
m
x+a
xa
(x a)2 + y 2
2 (x + a)2 + y 2
2
m
+ y2
2y
(x + a)2 + y 2
(x
a)
vy =
=
2y
2
2
2
2
2
y
4 (x + a) + y
(x
a) + y
[(x a)2 + y 2 ]
vx = U +
vy =
my
1
1
(x a)2 + y 2
2 (x + a)2 + y 2
,
y
vy =
.
x
c 2010, MIT
Copyright @
Potential Flow
m
x+a
xa
=0
vx (x, y = 0) = U +
(x a)2
2 (x + a)2
m
1
1
=0
U +
2 x + a x a
ma
=0
OR (after some algebra. . . ) : x2 a2
U
m
aU
m
y
m
arctan
arctan
2
x+a
2
m
2ay
arctan
= U y
2
x2 + y 2 a2
(x, y)
y
x a
= U y +
1
a
(1 2 )
m
arctan
2
2ay
x2 + y 2 a2
=0
(1)
2ah
h2 a2
2
h a2
2U h
= arccot
m
2ah
U h =
m
arctan
2
h 1
h
a
a 2
h
= 2a cot
2U h
m
(2)
+ bx + c = 0, x =
b2 4ac
2a
c 2010, MIT
Copyright @
Potential Flow
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
33 U 2
64 ag
c 2010, MIT
Copyright @
Potential Flow
Solution:
Due to high speed ow at the top of the sphere, we expect a low pressure at the top of the sphere. This
pressure results in a lift force on the hemsiphere.
Given the velocity eld, the pressure distribution at the surface of the sphere can be found using Bernoulli:
p() pa =
1
(U 2 v(r, )2 )
2
We can then integrate the pressure at the surface of the hemisphere to nd the lift force. The ow around
this hemisphere is the same as that for a sphere because of symmetry about the plate. Thus, streamlines for
this ow can be solved by combining the streamlines for a uniform ow and a doublet.
from Kundu & Cohen pp.192
hemisphere = sphere = uniform + doublet =
1 2 2
m
U r sin
sin2
2
r
1
r2 sin
1
v =
r sin r
vr =
1 2 m
2
vr = 2
Ur
sin
2
r
r sin
(
1
1 2 m
m)
= 2
Ur
2 sin cos = U 2 3 cos
r
r
r sin 2
Similarly for v :
[
J
1
1 2 m
Ur
sin2
r sin r
2
r
(
1 (
m) 2
m)
=
U r + 2 sin = U + 3 sin
r
r sin
r
(1)
v =
(2)
Now we must solve for the doublet strength m. We know there is a stagnation point at r = a and =
(and also for = 0) such that our velocities are zero:
(
m)
vr r=a,= = 0 = U 2 3
a
1 3
m = Ua
2
Now substitute this into Eqs (1) and (2)
[
( a )3 J
vr = U 1
cos
r
[
J
1 ( a )3
sin
v = U 1 +
2 r
(3)
c 2010, MIT
Copyright @
Potential Flow
At the surface of the hemisphere r = a, such that vr = 0 (no ux through the sphere). Thus
3
v(r, )r=a = v (a, ) = U sin
2
Since the pressure is only a function of , we can solve for the lift force by integrating the pressure over the
area of the hemisphere projected on the x-y plane, Ap :
side view
Fp =
Ap
=
Ap
p pa dAp
1
3
U 2 U sin
2
2
ad
z
2
a
x
dAp
1
= U 2
(2a2 sin2 )d
2
0
9
U 2
sin2 (2a2 sin2 )d
8
0
ad sin
top view
y
2a sin
Therefore,
[
J
[
J
1
1
9
sin3 cos 3
Fp =2 U 2 a2 ( sin cos ) a2 U 2
+ ( sin cos )
4
4
8
2
2
0
0
1 2 2 27
11
2 2
2 2
= a U a U = a U
2
32
32
The sign of this force tells us the pressure has a lifting eect (a positive pressure on an upward facing surface
2 2
pushes downward). Thus FL = Fp = 11
32 a U . The weight of the hemisphere is given by
2
W = h gV = h g a3
3
which acts downward. There is also a buoyancy force given by
2
FB = gV = g a3
3
which acts upward. To keep the hemisphere on the plate we need the downward acting force W to be greater
than or equal to the upward acting forces, Fp + FB :
W F p + FB
2
2
11
h g a3 g a3 + a2 U 2
3
3
32
33 U 2
h 1 +
64 ag
D
Problem Solution by Tony Yu, Fall 2006
2.25 Advanced Fluid Mechanics
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
c 2010, MIT
Copyright @
Potential Flow
Solution:
Here we use the method of images and note that in the complex plane the two point sources are located at
z = ia, where i2 = 1. Now the complex potential for this ow is given by
w=
m
m
m
ln(z ia) +
ln(z + ia)
ln a2
2
2
2
(1)
where we have included the term ln a2 to make the arguments of the logarithms dimensionless (recall that
ln z ln a = ln az ). The inclusion of this term, however, does not aect the velocity eld, since it is eliminated
when we take the gradient of w. The above result in Eq.(1) is similar, but not identical to Eq. (6.50) in
Kundu. Why? The equation expands to
w=
m
m
ln a2
ln(x2 y 2 + a2 + i2xy)
2
2
(2)
Since the logarithm of a complex quantity, = ||ei , is ln = ln || + i, for the imaginary part of Eq.(2),
we have
2xy
x2 y 2 +a2
m
tan1
2
2xy
x2 y 2 + a2
(3)
)
.
ux =
m
x(x2 + y 2 + a2 )
=
y
a4 + 2a2 x2 2a2 y 2 + x4 + 2x2 y 2 + y 4
(4)
and
uy =
m
y(x2 + y 2 a2 )
=
4
2
2
x
a + 2a x 2a2 y 2 + x4 + 2x2 y 2 + y 4
(5)
To determine the velocity along the horizontal plane, we solve for ux and uy at y = 0. Accordingly,
ux,p =
m x(x2 + a2 )
m
x
=
4
2
2
4
2
x + a2
a + 2a x + x
(6)
whereas
uy,p = 0
(7)
c 2010, MIT
Copyright @
Potential Flow
Up
1 xa
ux,p a
=
m
1 + xa22
(8)
0.2
Up
0.1
0.1
0.2
15
10
0
x/a
10
15
2
The pressure along the plane is given by the Bernoulli equation P = Pp 12 ux,p
, where P is taken to
be given and is the pressure very far away from the plane where the velocity is essentially zero, but it also
happens to equal the pressure at the stagnation point. So the pressure at the plane is
x2
1
1 m2
Pp = P u2x,p = P 2 2
2
2 (x + a2 )2
(9)
2 a2
x2
a2
1+
x2
a2
(10)
c 2010, MIT
Copyright @
Potential Flow
1 m2
(P p P )/ 2
2 a
0.005
0.01
0.015
0.02
0.025
0.03
15
10
0
x/a
10
15
The net upward (i.e. positive y-direction) force per unit depth, F ' , on the plane is simply the integral of the
gage pressure acting on it. Here we assume that the pressure on the bottom side of the plane is everywhere
P .
F' =
P n
dx =
(Pp P )dx =
1 m2
x2
2 2
dx
2 (x + a2 )2
(11)
x
a
2
2(a + x2 )
1 m2
4a 2 2
(12)
F' =
1 m2
4 a
(13)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
=U r+
cos
r
2
(a) Calculate an expression for the resulting velocity eld.
(b) By examining the location of stagnation points in the ow, deduce the dependence of the form of
c 2013, MIT
Copyright @
Potential Flow
Magnus Eect
Solution:
(a) For a cylinder in uniform ow we have:
w1 (z) = U
a2
z+
z
(1)
Now for simulating the ow around a cylinder which is spinning in uniform ow, we add a clockwise line
vortex of circulation :
w2 (z) =
i
z
ln
2
a
(2)
w(z) = w1 + w2 = U
a2
z+
z
+
i
z
ln
2
a
(3)
w(z) = U
a2
re + ei
r
i
a2
r+
r
i
r
+
ln
+ i = U
2
a
a2
cos + i sin r
r
+
i
r
ln
2
a
2
(4)
c 2013, MIT
Copyright @
Potential Flow
Magnus Eect
a2
(z) = U r +
cos
r
2
r
2
a
(z) = U sin r
+
ln
r
2
a
(5)
(6)
(7)
1
a2
=
= U 1 2 cos
r
r
r
2
1
=
= U 1 + 2 sin
u =
2r
r
r
r
ur =
(8)
(9)
Checking the boundary conditions, at the boundaries of the cylinder (r = a) there should be no ux.
1
ur (r = a) = U
a2
a2
cos = 0
(10)
Since we are in potential ow, we should not care about no-slip as the potential ow does not satisfy the
no-slip boundary condition. The other boundary condition is that at r velocity goes back to U .
u (r ) = U sin
ur (r ) = U cos
|u(r )| =
ur 2 (r ) + u 2 (r ) = U
(11)
(12)
(13)
(b) In order to nd the stagnation point from equations (8) and (9):
ur = 0 r = a or cos() = 0
u = 0 and r = a :
U (2) sin
= 0 sin =
2a
4aU
(14)
(15)
(16)
So, if
4aU
1
1 = sin
4aU
2 = sin
4aU
(17)
(18)
c 2013, MIT
Copyright @
Potential Flow
If
4aU
Magnus Eect
= 1, there there will be only one stagnation point on the cylinder at:
=
(19)
And if 4aU
< 1, then there is no stagnation point on the cylinder, but we can use that fact that at cos = 0
the radial velocity is zero (This cannot be on the cylinder,6 r 6= a). Therefore:
cos = 0 = , =
2
2
a2
= 0 (Impossible f or > 1)
= u = U 1 + 2
2r
2
r
a2
=0
= u = U 1 + 2
2r
2
r
p
1
r=
+ 2 (4aU )2
4U
(20)
(21)
(22)
(23)
Equation (22) has 2 roots, but since one root lies inside the cylinder (r < a), it is discarded and only the root
given in equation (23) is acceptable. A sketch for the ow proles of the 3 cases mentioned is shown in gure
3.
Figure 3: Dierent regimes of ow past a circular cylinder with circulation. Adapted from Fluid Mechanics
4th ed.,P. K. Kundu and I. M. Cohen, Academic Press, 2008
(c) To nd the drag and lift forces, we need to nd the pressure rst; for any point on the cylinder we can
use Bernoulli:
1
P + U 2 = P (r = a, ) + u 2
2
"
2 #
1
P (r = a, ) = P + U 2 2U sin
2
2a
(24)
(25)
Equation (25) is symmetric about the y-axis ( = 2 , 2 ), therefore there is no force component in the x
direction and therefore, there is no drag force on the cylinder.
"
2 #!
Z 2
Z 2
1
2
FD =
P (r = a, ) cos ad = a
P + U 2U sin +
cos d = 0
(26)
2
2a
0
0
(d)To nd lift, again using the pressure from equation (25):
2.25 Advanced Fluid Mechanics
c 2013, MIT
Copyright @
Potential Flow
Magnus Eect
2
L=
P (r = a, ) sin ad = a
P + U 2U sin +
sin d
2
2a
0
0
Z 2
Z
U
U 2 1 cos2
d = U
L = a
sin2 d =
a
0
2
0
(27)
(28)
L = U
(29)
Equation (29) is true for irrotational ow around any 2D object and not only cylinders. This equations is
known as the Kutta-Zhukhousky lift theorem.
D
c 2013, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
ys (x)
x
A liquid of density and surface tension has been spilled on a horizontal plate so that it forms a very large
puddle whose depth (in the central parts) is h. Consider the region near the edge of the puddle, which can
be viewed to good approximation as two-dimensional. If the contact angle is , derive an expression for the
shape of the liquid surface ys (x).
Assume for simplicity that is small, so that the radius of curvature of the surface is large compared with
h and can be approximated by
1
R= 2
d ys
dx2
ans:
ys = h 1 exp
h = tan
g/x
/g
c 2012, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A container is being lled with liquid of density . A small, sharp-edged hole of radius R penetrates the
containers bottom. The surface tension between the liquid and the ambient air is , and the contact angle
for the air/liquid/container combination is (measured from the wall through the liquid to the interface).
(a) Find the critical liquid depth hc at which liquid rst begins to ow through the hole in the bottom.
Assume that R h. (Hint Is the expression dierent depending on whether is greater or smaller
than /2?)
(b) Evaluate hc for the case when the liquid is water at 20 C, R = 0.1 mm, = 0.07 N/m, and = 120 .
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
Suppose we model one of these hairs as a long cylinder of radius R 1 made of completely non-wetting material
(contact angle 180 degrees), and assume that it is set down on the water with its axis parallel to the surface,
as sketched. The surface tension is 0.07[N/m] (water/hair).
(a) Show that, as the cylinder, or hair, is brought into contact with the water and then depressed into
it, the lift force exerted on it by surface tension rst increases, then reaches a maximum at a certain
depression and nally decreases as the cylinder is depressed further. What is the maximum value of
the surface-tension-induced lift force per unit cylinder length?
(b) What is the criterion for the gravitational eects to have a negligible eect on the (maximum) total
lift force? Is it likely that this criterion is satised for the pond skaters tarsi?
(c) If a pond skater weights 0.05 grams (note that this is only a guess, not a gure based on observations
of the real insect) 2 , what minimum local length of hair must it have on its feet to keep it on top of the
water?
(d) What is the shape of the water surface around the leg when the force is maximum? (contact angle,
between the water and the leg, 180 degrees 3 ) Idealize the leg as a perfect cylinder (no hairs) and as
before assume the case that produces maximum force. (Use R' for the leg radius.)
1 Roughly
c 2008, MIT
Copyright @
Surface Tension
Preamble:
If we make a diagram of the leg, we realize that this is a highly idealized problem; the position of the hairs
relative to the leg and the presence of hairs next to each other will change the force that dierent hairs
provide to the leg. Also, if we look at an electron microscope photograph of the hairs, we realize that their
shape is not circular and not constant. Nevertheless,we are not interested in an exact solution, but we want
an estimate of the order of magnitude of the forces involved, and the geometric restrictions. Then, well
consider a single cylindrical hair isolated on the water surface as shown in the problem statement, lets just
keep in mind that we are just calculating approximate values for something as complex as the leg structure
shown in the image.
Scanning electron microscope images of a leg showing numerous oriented spindly microsetae (b) and the ne
nanoscale grooved structures on a seta (c). Scale bars: b, 20 m; c, 200 nm. Taken from Water-repellent
legs of water striders by Xuefeng Gao, and Lei Jiang, Nature 432, 36 (4 November 2004).
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
air
air
plate
plate
liquid
R
A drop of liquid of volume V is squeezed between two parallel smooth plates until the liquid thickness h
is very small compared with the liquids radial extent R. The liquid/plate/air contact angle , and the
liquid/air surface tension is c. Gravitational eects are negligible.
(a) Derive an expression for the downward force F required to hold the plates in position. Express F in
terms V , , c, and R.
(b) If = radians (a perfectly nonwetting situation) and T = 0.07 N/m, say (representing a clean airwater interface), what downward force is required to press a 3 mm3 drop of liquid into a thin disc or
radius R = 2 cm?
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
ys (x)
x
A liquid of density and surface tension has been spilled on a horizontal plate so that it forms a very large
puddle whose depth (in the central parts) is h. Consider the region near the edge of the puddle, which can
be viewed to good approximation as two-dimensional. If the contact angle is , derive an expression for the
shape of the liquid surface ys (x).
Assume for simplicity that is small, so that the radius of curvature of the surface is large compared with
h and can be approximated by
1
R= 2
d ys
dx2
ans:
ys = h 1 exp
h = tan
g/x
/g
c 2010, MIT
Copyright @
Surface Tension
Solution:
h Rpuddle
Po
Pi
ys (x)
x
Given:
Rpuddle
, ,
1
since is small.
d ys
dx2
Radius of curvature, R1
Unknown: ys , h
Find Po Pi :
1
1
+ ll
R1 lR2
Po Pi =
at
Pa
Pa
ys
Pi
Pj
Pj = Pa + g(h y)
R1 =
dx2
1
d 2 ys
dx2
Po Pi =
d2 y s
dx2
d y
g(h ys ) = dx
2
ys = 0
ys = h
at x = 0
as x
(2.02a)
Solve Eq. (2.02a) by assuming the form of the solution to be: ys = AeBx + C
d 2 ys
= AB 2 eBx
dx2
C=h
B=
g
B=
AB 2 eBx
g
(so that ys = h at x )
ys = A exp
g Bx g
g
Ae C = h
g
x +h
c 2010, MIT
Copyright @
Surface Tension
A = h
g
x
ys = h 1 exp
ys = 0 = A + h
Find h by letting
dys
dx
= tan at x = 0:
g
g
dys
=h
exp
x
dx
dys
g
=h
= tan
dx x=0
h=
(2.02b)
tan
g
Alternatively, starting from Young-Laplace, (assuming that is small) such that curvature is equal to
2
d2 ys
dx2
P = ddxy2s ,
d 2 ys
,
(2.02c)
dx2
besides, Pi (x, y) = Pi (x, ys ) + g(ys y), from hydrostatics. Then, combining the information from hydro
statics and Young Laplace,
d2 y s
(2.02d)
Pi (x, y) = 2 + Po + g(ys y).
dx
Now, since its hard to work with so many functions, lets use one more property from hydrostatics, dP
dx = 0,
at any y. Then, dierentiating all the expression,
Pi (x, ys (x)) Po =
0 =
g ,
and setting Lc =
dys
d 3 ys
+ g
,
dx
dx3
(2.02e)
0 = L2c ddxy3s
dys
dx ,
(2.02f)
dys
dx ,
d2 G
G = 0,
dx2
0,
(2.02g)
tan ,
(2.02h)
x .
as
(2.02i)
so
G = C1 e Lc +
x
C2 e Lc
(2.02j)
then,
ys = C1q e Lc + C3 .
x
dys (0)
dx
= tan ,
C3
Lc
(2.02k)
x
Lc
).
x
= tan , ys = Lc tan (1 e Lc ).
x
ys (x) = Lc tan (1 e Lc ),
(2.02l)
from which we notice that Lc measures the extent of the eect of interface tension on the surface prole.
c 2010, MIT
Copyright @
Surface Tension
Note that we can deduce h (for all x) by a force balance as done in class:
gh2
+ cos = ,
2
then, as 0
h2 =
2(1 cos )
g
2(1 cos )
1
2
2(1 (1 2 + ...))
.
=
g
g
g
(2.02m)
(2.02n)
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
A container is being lled with liquid of density . A small, sharp-edged hole of radius R penetrates the
containers bottom. The surface tension between the liquid and the ambient air is , and the contact angle
for the air/liquid/container combination is (measured from the wall through the liquid to the interface).
(a) Find the critical liquid depth hc at which liquid rst begins to ow through the hole in the bottom.
Assume that R h. (Hint Is the expression dierent depending on whether is greater or smaller
than /2?)
(b) Evaluate hc for the case when the liquid is water at 20 C, R = 0.1 mm, = 0.07 N/m, and = 120 .
c 2008, MIT
Copyright @
Surface Tension
Solution:
First, lets assume that the Bond number is small, then surface tension eects dominate. Also, it is very
important to understand the dierence between the case > /2 and < /2.
For >
2:
First lets analyze the case when the drop moves downwards through the hole starting at the container. Lets
calculate the pressure inside the drop, no matter from where we approach the pressure has to be the same,
as long as the drop is at rest (hydrostatics). From the bottom of the drop,
Pi Po =
From the gure:
R2 = R1 =
1
R1
1
R2
"
R
R
=
sin( /2)
cos()
Hole of radius R
>
Pi
2R
/2
R1
Po
gh + Pa Pa = (
h=
2
)
R/ cos
( cos )
gR
Note that we could also have performed a force balance and obtain the same result. The force going down is
simply the hydrostatic pressure force, R2 (gh + Pa ), and the forces going up are the atmospheric pressure
c 2008, MIT
Copyright @
Surface Tension
force acting on the bottom of the drop, and the surface tension, R2 Pa 2R cos . Now, performing a
force balance,
R2 Pa 2R cos = R2 (gh + Pa ), 2 cos = R(gh), h =
2 cos
.
gR
(2.05a)
Now, when the drop reaches the bottom, any angle is allowed, then the factor cos is not a constrain, and
the drop can resist higher hydrostatic pressures. The height reaches its maximum when cos = 1, then,
hmax =
(2.05b)
2
gR ,
After this value, if the drop grows, the force to hold it up reduces and dropping starts. For water,
hmax =
2(70[N/m] 103 )
= 0.14[m].
kg
4
10[ m
s ]1000 m3 1[m] 10
(2.05c)
c 2008, MIT
Copyright @
Surface Tension
For <
2:
Hole of radius R
< /2
2R
Pi
R1
Po
Now, lets use the pressure argument again. From hydrostatics: Pi =! Pa + gh.
" Also, Po = Pa , now
1
1
calculating the pressure inside the drop from the bottom, Po Pi = R1 + R2 From the gure: R1 =
R2 =
R'
sin
!
P!
gh + !
Pa!
a = (
h=
2
R1 / sin
2
(sin )
gR1
Here too, we could also have performed a force balance and obtain the same result. The force going down is
simply the hydrostatic pressure force, R12 (gh + Pa ), and the forces going up are the atmospheric pressure
force acting on the bottom of the drop, and the surface tension, R12 Pa + 2R1 sin . Now, performing a
force balance,
R12 Pa + 2R1 sin = R12 (gh + Pa ), 2 sin = R1 (gh), h =
2 sin
.
gR1
(2.05d)
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
This problem is from Advanced Fluid Mechanics Problems by A.H. Shapiro and A.A. Sonin
The water strider, or pond skater, is a slender insect, about 12 cm long, that runs or skates over the surfaces
of ponds and streams. It stays easily on the surface because its feet (tarsi) are equipped with numerous ne,
non-wetting hairs.
Suppose we model one of these hairs as a long cylinder of radius R made of completely non-wetting material
(contact angle 180 degrees), and assume that it is set down on the water with its axis parallel to the surface,
as sketched. The surface tension is 0.07[N/m].
(a) Show that, as the cylinder, or hair, is brought into contact with the water and then depressed into
it, the lift force exerted on it by surface tension rst increases, then reaches a maximum at a certain
depression and nally decreases as the cylinder is depressed further. What is the maximum value of
the surface-tension-induced lift force per unit cylinder length?
(b) What is the criterion for the gravitational eects to have a negligible eect on the (maximum) total
lift force? Is it likely that this criterion is satised for the pond skaters tarsi?
(c) If a pond skater weights 0.05 grams (note that this is only a guess, not a gure based on observations
of the real insect) 1 , what minimum local length of hair must it have on its feet to keep it on top of the
water?
(d) What is the shape of the water surface around the leg when the contact angle is 180 degrees
between the water and the leg? 2 . Idealize the leg as a perfect cylinder (no hairs) and as before assume
maximum surface tension force.
1 Actually,
accordingly to the Wikipedia the weight of these insects is 0.01 grams, not a bad guess.
to the Wikipedia the eective angle for the legs (due to the properties of the hairs) is roughly 170 degrees,
superhydrophobic.
2 Accordingly
c 2008, MIT
Copyright @
Surface Tension
Preamble:
If we make a diagram of the leg, we realize that this is a highly idealized problem; the position of the hairs
relative to the leg and the presence of hairs next to each other will change the force that dierent hairs
provide to the leg. Also, if we look at an electron microscope photograph of the hairs, we realize that their
shape is not circular and not constant. Nevertheless,we are not interested in an exact solution, but we want
an estimate of the order of magnitude of the forces involved, and the geometric restrictions. Then, well
consider a single cylindrical hair isolated on the water surface as shown in the problem statement, lets just
keep in mind that we are just calculating approximate values for something as complex as the leg structure
shown in the image.
Scanning electron microscope images of a leg showing numerous oriented spindly microsetae (b) and the ne
nanoscale grooved structures on a seta (c). Scale bars: b, 20 m; c, 200 nm. Taken from Water-repellent
legs of water striders by Xuefeng Gao, and Lei Jiang, Nature 432, 36 (4 November 2004).
c 2008, MIT
Copyright @
Surface Tension
Solution:
(a)
If we use the diagram provided in the exercise, we notice that the force per unit length going upwards
due to surface tension is
F; y = 2 sin y,
(2.06a)
which is maximum for = 2 . Then, the upward surface tension force increases till the angle reaches
max
= 2.
(2.06b)
Notice that the diagram shows the force on the hair caused by surface tension, and not the force on
the water surface.
(b) The total upwards vertical force is the sum of the force due to surface tension plus the force resulting
from pressure on the bottom of the hair. The pressure force can be calculated easily if we consider it
as a buoyancy force, then
Fp; = g(h2R sin ) + g
R(1cos )
r(z)dz ,
(2.06c)
The rst term represents the contribution of the water volume displaced by the rectangular area shown
in the gure, and the second term represents the contribution of the area between the rectangle and
the bottom of the cylinder. Then the order of magnitude of this force is
Fp; 2ghR + R2 g,
(2.06d)
1,
2
(2.06e)
Now, for the rst term, lets estimate the height of the water. Using a control volume that encompass
the liquid to the left of the hair and extending till the water surface is horizontal, lets perform a force
balance in the x direction,
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright @
Surface Tension
gh2
+ cos = 0,
2
simplifying,
h=
(2.06f)
(1 cos ),
g
(2.06g)
.
g
(2.06h)
then,
R2 g
#0.5
R2 g
1,
#0.5
(2.06i)
1.
(2.06j)
1, or
Bo 1 , where Bo is the Bond number.
(2.5 10
[m])
[ '#
]
kg
1000 m3 (9.8[ sm2 ])
N
0.07[ m
]
= 9 109 ,
(2.06k)
and can be in eect neglected. Now, if we consider the leg as a whole, the Bond number for a radius
of 200 microns, is 6400 times larger, but still small.
(c) To calculate the minimum hair length, then lets equate the weight force and the surface tension
force,
2lT ot = m g,
then,
lT ot =
(5 105 [kg])(9.9[ m
mg
s ])
=
= 3.6 103 [m],
kg
2
2(0.07[ s2 )]
(2.06l)
hair
then this divided by 6 legs, lper
leg = 0.6mm, which is more than feasible, since these creatures have
many hairs per leg, and each leg is a couple of mm long. (Look again at the electron microscope gure.)
(d) Since we are neglecting the surface irregularities of the leg, our solution will apply only at the water
surface far from the leg, but will give us a good idea of the shape far away. Now, lets draw a similar
diagram for the leg as the one that we made for the hair, and assume ; = .
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright @
Surface Tension
2.
Now, lets obtain an estimate for the maximum depression depth for the water surface. Making again
a force balance for the liquid to the left of the leg,
; =
gh2
,
2
(2.06m)
h =
(2.06n)
then,
2 ;
,
g
If we approach the leg from a point with null curvature, the pressure at this point inside the water is
just P (h) = Pa . Then if we move inside the liquid, the pressure just below this point inside the liquid
is Pa + g(h y) (the origin was set at the minimum height of the surface), but this is the same
pressure at any position at the same height since the liquid is in static equilibrium; then, the pressure
inside the liquid is just determined by its vertical coordinate. Now, making a force balance in the y
direction for a dierential element with a not null curvature, 3
3 Figure
c 2008, MIT
Copyright @
Surface Tension
d 2 hs
+ gh = ghs ,
dx2
(2.06o)
2
now, to make it an equation in terms of one variable, lets add the term ; ddxh2 which is equal to zero,
then,
2
; d
(h hs )
dx2
g(h hs ) = 0,
or
L2c
where Lc =
d2 (h hs )
(h hs ) = 0,
dx2
(2.06p)
h hs 0
h hs = h
x ,
as
x = 0,
when
(2.06q)
(2.06r)
(2.06s)
which gives,
x
(h hs ) = h e Lc ,
simplifying,
x
hs = h (1 e Lc ),
(2.06t)
c 2008, MIT
Copyright @
Surface Tension
N et V ertical F orce =
but,
sin =
d
(sin ),
dx
h;
1
(1 + h;2 ) 2
(2.06u)
(2.06v)
and
d
dx
sin =
(1+h2 ) 2
(1+h2 ) 2
h; h;; ,
simplifying,
sin =
d
dx
then,
h (1+h2 )
3
(1+h2 ) 2
h2 h
(1+h2 ) 2
h;;
d
sin =
3 ,
dx
(1 + h;2 ) 2
(ii) The solution does not satisfy the contact angle condition,
(2.06w)
dhs
dx
Exact Solution:
dene =
x
Lc ,
z=
hs ;
Lc ,
d
d ,
then,
z ;;
(1 + (z ; )2 ) 2
+ z = z ,
(2.06x)
where z = 2. Now, lets multiply by z ; and integrate (after playing with the equation, youll notice
this is the integration factor required),
1
(1 + (z ; )2 )
1
2
z2
z z = Const,
2
(2.06y)
1
(1 (1
z )2 )2
2
# 12
1 ,
(2.06z)
where z(0) = 0. We can integrate numerically to obtain zexact . Note that for = 0, z ; is innite,
1
but z 2 4 as 0, and hence we can start the integration at min with initial condition
1
z(min ) = 2 4 min . In the next gure, we plot zexact and zapprox = 2(1 e ).
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright @
Surface Tension
Final Comment: As the micro technologies progress (microuidics, microfabrication and micromechanics),
the role of surface tension becomes more and more important. Each day new inventions use surface tension.
Below, two of such inventions are shown, the rst Mechanical Water Strider, made at MIT, and the Robotic
Water Strider made in Carnegie Mellon. The mechanical striders are capable of moving slowly on the surface
of water using its many hydrophobic-coated legs, but since they do not have leg hairs, their lift capacity is
smaller.
Water Stride (Top) and First Mechanical Water Strider (Bottom). Courtesy of John Bush, MIT, and Brian
Chan, MIT.
2.25 Advanced Fluid Mechanics
c 2008, MIT
Copyright @
Surface Tension
Water Stride Vortices. As the water strider moves it leaves a trail of vortices that help it move. Before this
observation, the method of propulsion of these insects was a mystery. This particular photo made the
Nature cover some time ago, and even if it is not directly related with the problem, I couldnt resist to
incorporate it to the problem. http://www-math.mit.edu/~dhu/Striderweb/striderweb.html. Courtesy of
John Bush MIT.
D
c 2008, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
air
air
plate
plate
liquid
R
A drop of liquid of volume V is squeezed between two parallel smooth plates until the liquid thickness h
is very small compared with the liquids radial extent R. The liquid/plate/air contact angle , and the
liquid/air surface tension is . Gravitational eects are negligible.
(a) Derive an expression for the downward force F required to hold the plates in position. Express F in
terms V , , , and R.
(b) If = radians (a perfectly nonwetting situation) and T = 0.07 N/m, say (representing a clean airwater interface), what downward force is required to press a 3 mm3 drop of liquid into a thin disc or
radius R = 2 cm?
c 2010, MIT
Copyright @
Surface Tension
Solution:
Given:
liquid volume, V
liquid/plate/air contact angle,
surface tension,
R1
h
2
R2
R1
In this specic problem, (Assuming that the Bo number is small and therefore we can neglect gravity eects
on the shape of the free surface)
R2 = R
1
2 cos
(since R h)
P
=
+
o
i
h
R
h
R1 =
(Bo
1)
2 cos
2 cos
Po Pi = P =
h
F = FP
= P (R2 )
=
2R2 cos
h
where h represents the gap between the plate and the solid surface. The above equation says: the smaller
the gap, the greater the force. This eect is known as capillary stiction.
(a) Since h = V /R2 ,
F =
1 Notice
2 2 R4 cos
V
(2.07a)
that the direction of the applied force was taken as going downwards
c 2010, MIT
Copyright @
Surface Tension
Liquid is nonwetting
(b) Plug in given values:
= ,
= T = 0.07 N/m,
F =
V = 3 mm3 = 3 109 m3 ,
R = 2 cm = 0.02 m
][ ][
]
2(0.07) 2 (0.02)4 cos [
Nm1 m4 m3
9
3 10
= 74 N
F
=
perfectly nonwetting
c 2010, MIT
Copyright @
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.