44 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

When this equation is divided by L W Ax, and the limit taken as Ax approaches zero, we
get

The first term on the left side is exactly the definition of the derivative of 4,: with respect
to x. Therefore Eq. 2.2-7 becomes

At this point we have to write out explicitly what the components +,, and 4,: are, mak-
+
ing use of the definition of in Eqs. 1.7-1 to 3 and the expressions for rxzand T,, in Ap-
pendix B.1. This ensures that we do not miss out on any of the forms of momentum
transport. Hence we get

In accordance with the postulates that v, = v,(x), v, = 0, v, = 0, and p = p(x), we see that
(i) since v, = 0, the pup, term in Eq. 2.2-9a is zero; (ii) since v, = v,(x), the term
-2,u(dv,/dz) in Eq. 2.2-9b is zero; (iii) since v, = v,(x), the term pv,v, is the same at z = 0
and z = L; and (iv) since p = p(x), the contribution p is the same at z = 0 and z = L. Hence
T, depends only on x, and Eq. 2.2-8 simplifies to

1%
I

= pg cos p
I

This is the differential equation for the momentum flux T,,. It may be integrated to give

The constant of integration may be-evaluated by using the boundary condition at the
gas-liquid interface (see 52.1):
B.C. 1: atx=O, r,,=O (2.2-12)
Substitution of this boundary condition into Eq. 2.2-11 shows that C, = 0. Therefore the
momentum-flux distribution is

as shown in Fig. 2.2-3.

Next we substitute Newton's law of viscosity

into the left side of Eq. 2.2-13 to obtain

which is the differential equation for the velocity distribution. It can be integrated to
give
 52.2 Flow of a Falling Film 45

Momentum

Fig. 2.2-3 Final results for the falling film problem,

showing the momentum-flux distribution and the
velocity distribution. The shell of thickness Ax, over
\ which the momentum balance was made, is also shown.

The constant of integration is evaluated by using the no-slip boundary condition at the
solid surface:
B.C. 2 at x = 6, v, =0 (2.2-17)
Substitution of this boundary condition into Eq. 2.2-16 shows that C2 = (pg cos P / 2 4 a 2 .
Consequently, the velocity distribution is

I I

This parabolic velocity distribution is shown in Fig. 2.2-3. It is consistent with the postu-
lates made initially and must therefore be a possible solution. Other solutions might be
possible, and experiments are normally required to tell whether other flow patterns can
actually arise. We return to this point after Eq. 2.2-23.
Once the velocity distribution is known, a number of quantities can be calculated:
(i) The maximum velocity vZ,,,, is clearly the velocity at x = 0; that is,

(ii) The average velocity (v,) over a cross section of the film is obtained as follows:
46 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

The double integral in the denominator of the first line is the cross-sectional area of the
film. The double integral in the numerator is the volume flow rate through a differential
element of the cross section, v,dx dy, integrated over the entire cross section.
(iii) The mass rate of flow w is obtained from the average velocity or by integration of
the velocity distribution

w= IO8
low pv,dxdy = pWS(v,) =
p2g ws3cos p
3~
(2.2-21)

(iv) The film thickness S may be given in terms of the average velocity or the mass
rate of flow as follows:

(v) The force per unit area in the z direction on a surface element perpendicular
to the x direction is +T,, evaluated at x = 6. This is the force exerted by the fluid (re-
gion of lesser x) on the wall (region of greater x). The z-component of the force F of the
fluid on the solid surface is obtained by integrating the shear stress over the fluid-solid
interface:

This is the z-component of the weight of the fluid in the entire film-as we would have
expected.
Experimental observations of falling films show that there are actually three "flow
regimes," and that these may be classified according to the Reynolds number,' Re, for the
flow. For falling films the Reynolds number is defined by Re = 4S(vz)p/p.The three flow
regime are then:
laminar flow with negligible rippling Re < 20
laminar flow with pronounced rippling 20 < Re < 1500
turbulent flow Re > 1500
The analysis we have given above is valid only for the first regime, since the analysis
was restricted by the postulates made at the outset. Ripples appear on the surface of the
fluid at all Reynolds numbers. For Reynolds numbers less than about 20, the ripples are
very long and grow rather slowly as they travel down the surface of the liquid; as a re-
sult the formulas derived above are useful up to about Re = 20 for plates of moderate
length. Above that value of Re, the ripple growth increases very rapidly, although the
flow remains laminar. At about Re = 1500 the flow becomes irregular and chaotic, and
the flow is said to be t ~ r b u l e n t .At
~ , ~this point it is not clear why the value of the

'This dimensionless group is named for Osbome ~ e ~ n b l (1842-19121,

ds professor of engineering at
the University of Manchester. He studied the laminar-turbulent transition, turbulent heat transfer, and
theory of lubrication. We shall see in the next chapter that the Reynolds number is the ratio of the inertial
forces to the viscous forces.
G. D. Fulford, Adv. Chem. Engr., 5,151-236 (1964); S. Whitaker, Ind. Eng. Chem. Fund., 3,132-142
(1964);V . G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, Englewood Cliffs, N.J. (1962),s135.
H.-C. Chang, Ann. Rev. Fluid Mech., 26,103-136 (1994); S.-H. Hwang and H.-C. Chang, Phys. Fluids,
30,1259-1268 (1987).
 s2.2 Flow of a Falling Film 47

Reynolds number should be used to delineate the flow regimes. We shall have more to
say about this in g3.7.
This discussion illustrates a very important point: theoretical analysis of flow sys-
tems is limited by the postulates that are made in setting u p the problem. It is absolutely
necessary to do experiments in order to establish the flow regimes so as to know when
instabilities (spontaneous oscillations) occur and when the flow becomes turbulent.
Some information about the onset of instability and the demarcation of the flow regimes
can be obtained by theoretical analysis, but this is an extraordinarily difficult subject.
This is a result of the inherent nonlinear nature of the governing equations of fluid dy-
namics, as will be explained in Chapter 3. Suffice it to say at this point that experiments
play a very important role in the field of fluid dynamics.

An oil has a kinematic viscosity of 2 X m2/s and a density of 0.8 X 10%g/m3. If we want
to have a falling film of thickness of 2.5 mm on a vertical wall, what should the mass rate of
CalCulation of Film flow the liquid be?
Velocity
SOLUTION
According to Eq. 2.2-21, the mass rate of flow in kg/s is

To get the mass rate of flow one then needs to insert a value for the width of the wall in
meters. This is the desired result provided that the flow is laminar and nonrippling. To
determine the flow regime we calculate the Reynolds number, making use of Eqs. 2.2-21
and 24

This Reynolds number is sufficiently low that rippling will not be pronounced, and therefore
the expression for the mass rate of flow in Eq. 2.2-24 is reasonable.

Rework the falling film problem for a position-dependent viscosity p = which arises
when the film is nonisothermal, as in the condensation of a vapor on a wall. Here pois the vis-
Falling Film with cosity at the surface of the film and a is a constant that describes how rapidly p decreases as x
Variable Viscosity increases. Such a variation could arise in the flow of a condensate down a wall with a linear
temperature gradient through the film.

SOLUTION The development proceeds as before up to Eq. 2.2-13. Then substituting Newton's law with
variable viscosity into Eq. 2.2-13 gives

This equation can be integrated, and using the boundary conditions in Eq. 2.2-17 enables us to
evaluate the integration constant. The velocity profile is then

As a check we evaluate the velocity distribution for the constant-viscosity problem (that is,
when a is zero). However, setting a = 0 gives GO - in the two expressions within parentheses.
48 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

This difficultycan be overcome if we expand the two exponentials in Taylor series (see §C.2),
as follows:

-
- pgs2cos p .-[(-+-a+
Po 0-0
1 1
2 3
. . a ) - ( L~-II',+
282 383 .. I).
which is in agreement with Eq. 2.2-18.
From Eq. 2.2-27 it may be shown that the average velocity is
pgs2cos p
(vz> =
Po [.(A -$+ 4) 21
- -

The reader may verify that this result simplifies to Eq. 2.2-20 when a goes to zero.

s2.3 FLOW THROUGH A CIRCULAR TUBE

The flow of fluids in circular tubes is encountered frequently in physics, chemistry, biol-
ogy, and engineering. The laminar flow of fluids in circular tubes may be analyzed by
means of the momentum balance described in 52.1. The only new feature introduced
here is the use of cylindrical coordinates, which are the natural coordinates for describ-
ing positions in a pipe of circular cross section.
We consider then the steady-state, laminar flow of a fluid of constant density p and
viscosity p in a vertical tube of length L and radius R. The liquid flows downward under
the influence of a pressure difference and gravity; the coordinate system is that shown in
Fig. 2.3-1. We specify that the tube length be very large with respect to the tube radius,
so that "end effects" will be unimportant throughout most of the tube; that is, we can ig-
nore the fact that at the tube entrance and exit the flow will not necessarily be parallel to
the tube wall.
We postulate that v, = v,(r), vr = 0, v, = 0, and p = p(z). With these postulates it may
be seen from Table B.l that the only nonvanishing components of 7 are rrz= rZr=
-p(dv,/dr).
We select as our system a cylindrical shell of thickness Ar and length L and we begin
by listing the various contributions to the z-momentum balance:
rate of z-momentum in (2~Ar)(#41z=0 (2.3-1)
across annular surface at z = 0
rate of z-momentum out (2~rAr)($,,)J,=~ (2.3-2)
across annular surface at z = L
rate of z-momentum in (2d)($,)(, = (2flL$,)(, (2.3-3)
across cylindrical surface at r
rate of 2-momentum out ( 2 d r + Ar)L)(+J/r+Ar = (2mL$J/r+Ar (2.3-4)
across cylindrical surface at r + Ar
gravity force acting in (2wArL)pg (2.3-5)
z direction on cylindrical shell
 Flow Through a Circular Tube 49

4zz),=o=flux Fig. 2.3-1 Cylindrical shell of fluid

of z-momentum over which the z-momentum bal-
ance is made for axial flow in a cir-
cular tube (see Eqs. 2.3-1 to 5). The
z-momentum fluxes 4, and +,, are
given in full in Eqs. 2.3-9a and 9b.

4rzI r + A r = flux
of z-momentum
out at r + Ar

+ Tube wall

of z-momentum
outatz=L

The quantities +,,and +,, account for the momentum transport by all possible mecha-
nisms, convective and molecular. In Eq. 2.3-4, (Y + Ar) and (r)l,+,, are two ways of writ-
ing the same thing. Note that we take "in" and "out" to be in the positive directions of
the Y- and z-axes.
We now add up the contributions to the momentum balance:

When we divide Eq. (2.3-8)by 2.irLAr and take the limit as Ar + 0, we get

The expression on the left side is the definition of the first derivative of r4,, with respect
to r. Hence Eq. 2.3-7 may be written as

Now we have to evaluate the components 4, and +,, from Eq. 1.7-1and Appendix B.l:

Next we take into account the postulates made at the beginning of the problem-namely,
that vz = v,(r), V, = 0, v g = 0, and p = p(z). Then we make the following simplifications:
50 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

(i) because v, = 0, we can drop the term pqv, in Eq. 2.3-9a; (ii) because v, = v,(r), the term
pvzvz will be the same at both ends of the tube; and (iii) because vZ = vZ(r),the term
-2pdv,/dz will be the same at both ends of the tube. Hence Eq. 2.3-8 simplifies to

in which 9 = p - p g z is a convenient abbreviation for the sum of the pressure and gravi-
tational terms.' Equation 2.3-10 may be integrated to give

The constant C1 is evaluated by using the boundary condition

B.C. 1: at r = 0, =
T , ~ finite (2.3-12)
Consequently C1must be zero, for otherwise the momentum flux would be infinite at the
axis of the tube. Therefore the momentum flux distribution is
1

This distribution is shown in Fig. 2.3-2.

Newton's law of viscosity for this situation is obtained from Appendix B.2 as
follows:

Substitution of this expression into Eq. 2.3-13 then gives the following differential equa-
tion for the velocity:

Parabolic velocity
distribution uz(r)

Linear momentum-
flux distribution
~,,(r) Fig. 2.3-2 The momentum-flux
distribution and velocity distribu-
tion for the downward flow in a
I circular tube.

+
' The quantity designated by 9 is called the modified pressure. In general it is defined by 9 = p pgh,
where h is the distance "upwardv-that is, in the direction opposed to gravity from some preselected
reference plane. Hence in this problem h = -z.
 52.3 Flow Through a Circular Tube 51

This first-order separable differential equation may be integrated to give

The constant C2is evaluated from the boundary condition

B.C. 2: at r = R, v, = 0
From this C, is found to be (Yo - 9 , ) ~ ~ / 4 p L
Hence
. the velocity distribution is

We see that the velocity distribution for laminar, incompressible flow of a Newtonian
fluid in a long tube is parabolic (see Fig. 2.3-2).
Once the velocity profile has been established, various derived quantities can be
obtained:
(i) The maximum velocity v,,,,, occurs at r = 0 and is

(ii) The average velocity (v,) is obtained by dividing the total volumetric flow rate by
the cross-sectional area

(iii) The mass rate of flow w is the product of the cross-sectional area ,rrR2,the density
p, and the average velocity (v,)

This rather famous result is called the Hagen-~oiseuille~equation. It is used, along

with experimental data for the rate of flow and the modified pressure difference,
to determine the viscosity of fluids (see Example 2.3-1) in a "capillary viscometer."
(iv) The z-component of the force, F,, of the fluid on the wetted surface of the pipe is
just the shear stress 7,,integrated over the wetted area

This result states that the viscous force F, is counterbalanced by the net pres-
sure force and the gravitational force. This is exactly what one would obtain
from making a force balance over the fluid in the tube.

G. Hagen, Ann. Phys. Chern., 46,423442 (1839);J. L. Poiseuille, Comptes Rendus, 11,961 and 1041
(1841).Jean Louis Poiseuille (1799-1869) (pronounced "Pwa-zd-yuh," with d is roughly the "00" in
book) was a physician interested in the flow of blood. Although Hagen and Poiseuille established the
dependence of the flow rate on the fourth power of the tube radius, Eq. 2.3-21 was first derived by E.
Hagenbach, Pogg. Annalen der Physik u. Chemie, 108,385-426 (1860).
52 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

The results of this section are only as good as the postulates introduced at the begin-
ning of the section-namely, that v, = v,(r) and p = p(z). Experiments have shown that
these postulates are in fact realized for Reynolds numbers up to about 2100; above that
value, the flow will be turbulent if there are any appreciable disturbances in the sys-
tem-that is, wall roughness or vibration^.^ For circular tubes the Reynolds number is
defined by Re = D ( V , ) ~ /where
~, D = 2R is the tube diameter.
We now summarize all the assumptions that were made in obtaining the Hagen-
Poiseuille equation.
(a) The flow is laminar; that is, Re must be less than about 2100.
(b) The density is constant ("incompressible flow").
(c) The flow is "steady" (i.e., it does not change with time).
(d) The fluid is Newtonian (Eq. 2.3-14 is valid).
( e ) End effects are neglected. Actually an "entrance length," after the tube entrance,
of the order of L, = 0.035D Re, is needed for the buildup to the parabolic profile.
If the section of pipe of interest includes the entrance region, a correction must
be a ~ p l i e dThe
. ~ fractional correction in the pressure difference or mass rate of
flow never exceeds L,/L if L > L,.
(f) The fluid behaves as a continuum-this assumption is valid, except for very di-
lute gases or very narrow capillary tubes, in which the molecular mean free path
is comparable to the tube diameter (the "slip flow region") or much greater than
the tube diameter (the "Knudsen flow" or "free molecule flow" regime).5
(g) There is no slip at the wall, so that B.C. 2 is valid; this is an excellent assumption
for pure fluids under the conditions assumed in (0.See Problem 2B.9 for a dis-
cussion of wall slip.

.
Glycerine (CH20H.CHOH CH20H)at 26.5"C is flowing through a horizontal tube 1 ft long
and with 0.1 in. inside diameter. For a pressure drop of 40 psi, the volume flow rate w / p is
Determination of 0.00398 ft3/min. The density of glycerine at 26.5"C is 1.261 g/cm3. From the flow data, find the
Viscosity from viscosity of glycerine in centipoises and in Pa. s.
Capillary Flow Data
SOLUTION
From the Hagen-Poiseuille equation (Eq. 2.3-211, we find

dyn/cm2)(0.05 in. X
Ibf/in.2 12 in.
ft3 X -
0.00398 - min
1 ---
min 60 s

A. A. Draad [Doctoral Dissertation, Technical University of Delft (199611in a carefully controlled

experiment, attained laminar flow up to Re = 60,000. He also studied the nonparabolic velocity profile
induced by the earth's rotation (through the Coriolis effect). See also A. A. Draad and F. T. M.
Nieuwstadt, J. Fluid. Mech., 361,207-308 (1998).
9.H. Perry, Chemical Engineers Handbook, McGraw-Hill, New York, 3rd edition (1950), pp. 38S389;
W. M. Kays and A. L. London, Compact Heat Exchangers, McGraw-Hill, New York (19581, p. 49.
Martin Hans Christian Knudsen (1871-19491, professor of physics at the University of
Copenhagen, did key experiments on the behavior of very dilute gases. The lectures he gave at the
University of Glasgow were published as M. Knudsen, The Kinetic Theory of Gases, Methuen, London
(1934);G. N. Patterson, Molecular Flow of Gases, Wiley, New York (1956).See also J. H. Ferziger and H. G.
Kaper, Mathematical Theory of Transport Processes in Gases, North-Holland, Amsterdam (19721, Chapter 15.
 s2.4 Flow Through an Annulus 53

To check whether the flow is laminar, we calculate the Reynolds number

4(0.00398 -.)(2.54
min
?
in. X 12 ~ft~ ( ' ~6' 10"s) ( 1 . 2 6 1 cm3

in.
= 2.41 (dimensionless) (2.3-24)
Hence the flow is indeed laminar. Furthermore, the entrance length is
L, = 0.035D Re = (0.035)(0.1/12)(2.41)= 0.0007 ft
Hence, entrance effects are not important, and the viscosity value given above has been calcu-
lated properly.

EXAMPLE 23-2 Obtain an expression for the mass rate of flow w for an ideal gas in laminar flow in a long cir-
cular tube. The flow is presumed to be isothermal. Assume that the pressure change through
Compressible Flow in the tube is not very large, so that the viscosity can be regarded a constant throughout.
a Horizontal Circular
lkbe6 SOLUTION
This problem can be solved approximately by assuming that the Hagen-Poiseuille equation
(Eq. 2.3-21) can be applied over a small length dz of the tube as follows:

To eliminate p in favor of p, we use the ideal gas law in the form plp = po/po,where po and po
are the pressure and density at z = 0. This gives

The mass rate of flow w is the same for all z. Hence Eq. 2.3-27 can be integrated from z = 0 to
z = L to give

where pa,,
1
= + pL) is the average density calculated at the average pressure pa,, =
2@0 + P L ) .

52.4 FLOW THROUGH AN ANNULUS

We now solve another viscous flow problem in cylindrical coordinates, namely the
steady-state axial flow of an incompressible liquid in an annular region between two
coaxial cylinders of radii KR and R as shown in Fig. 2.4-1. The fluid is flowing upward in

L. Landau and E. M. Lifshitz, Fluid Mechanics, Pergamon, 2nd edition (1987), 917, Problem 6. A
perturbation solution of this problem was obtained by R. K. Prud'homme, T. W. Chapman, and J. R.
Bowen, Appl. Sci. Res, 43,67-74 (1986).
54 Chapter 2 Shell Momentum Balances and Velocity Distributions in Laminar Flow

Fig. 2.4-1 The momentum-flux distribution

and velocity distribution for the upward
flow in a cylindrical annulus. Note that the
momentum flux changes sign at the same
value of r for which the velocity has a
Velocity
distribution maximum.

Shear stress
or momentum-
flux distribution

the t u b e t h a t is, in the direction opposed to gravity. We make the same postulates as in
52.3: v, = v,(r), v, = 0, v, = 0, and p = p(z). Then when we make a momentum balance
over a thin cylindrical shell of liquid, we arrive at the following differential equation:

This differs from Eq. 2.3-10 only in that 9 = p + pgz here, since the coordinate z is in the
direction opposed to gravity (i.e., z is the same as the h of footnote 1 in 52.3). Integration
of Eq. 2.4-1 gives

just as in Eq. 2.3-11.

The constant C, cannot be determined immediately, since we have no information
about the momentum flux at the fixed surfaces r = KR and r = R. All we know is that
there will be a maximum in the velocity curve at some (as yet unknown) plane r = AR at
which the momentum flux will be zero. That is,

When we solve this equation for C, and substitute it into Eq. 2.4-2, we get

The only difference between this equation and Eq. 2.4-2 is that the constant of integration
C, has been eliminated in favor of a different constant A. The advantage of this is that we
know the geometrical significance of A.
We now substitute Newton's law of viscosity, T,, = -p(dv,/dr), into Eq. 2.4-4 to ob-
tain a differential equation for v,