Problem 1

Consider a fluid (of density ρ) in incompressible, laminar flow in a plane narrow slit of length L and width W
formed by two flat parallel walls that are a distance 2B apart. End effects may be neglected because B 
W  L. The fluid flows under the influence of both a pressure difference ∆p and gravity.

Figure 1.1: Fluid flow in a plane narrow slit.

a) Using a differential shell momentum balance, determine expressions for the steady-state shear stress
distribution and the velocity profile for a Newtonian fluid (of viscosity µ).
b) Obtain expressions for the maximum velocity, average velocity and the mass flow rate for slit flow.

Solution

1.a Differential Shell Momentum Balance in Rectangular Cartesian Coordinates.

A shell momentum balance is used below to derive a general differential equation that can be then employed
to solve several fluid flow problems in rectangular Cartesian coordinates. For this purpose, consider an incom-
pressible fluid in laminar flow under the effects of both pressure and gravity in a system of length L and width
W , which is at an angle β to the vertical. End effects are neglected assuming the dimension of the system in
the x-direction is relatively very small compared to those in the y-direction (W ) and the z-direction (L).

Figure 1.2: Differential rectangular slab (shell) of fluid of thickness ∆x used in z-momentum balance for flow
in rectangular Cartesian coordinates. The y-axis is pointing outward from the plane of the computer screen.

1
 Since the fluid flow is in the z-direction, ux = 0, uy = 0, and only uz exists. For small flow rates, the viscous
forces prevent continual acceleration of the fluid. So, uz is independent of z and it is meaningful to postulate
that the velocity uz = uz (x) and the pressure p = p(z). The only nonvanishing components of the stress tensor
are τxz = τzx , which depend only on x.
Consider now a thin rectangular slab (shell) perpendicular to the x-direction extending a distance W in the
y-direction and a distance L in the z-direction. A ’rate of z-momentum’ balance over this thin shell of thickness
∆x in the fluid is of the form:

Rate of z-momentum = In Out + Generation = Accumulation

At steady-state, the accumulation term is zero. Momentum can go ’in’ and ’out’ of the shell by both the
convective and molecular mechanisms. Since uz (x) is the same at both ends of the system, the convective
terms cancel out because (ρu2z W ∆x)|z=0 = (ρu2z W ∆x)|z=L . Only the molecular term (LW τxz ) remains to
be considered, whose ’in’ and ’out’ directions are taken in the positive direction of the x-axis. Generation of
z-momentum occurs by the pressure force acting on the surface (pW ∆x) and the gravity force acting on the
volume [(ρg cos β)LW ∆x].
The different contributions may be listed as follows:

• rate of z-momentum in by viscous transfer across surface at x is LW τxz (x)

• rate of z-momentum out by viscous transfer across surface at x + ∆x is LW τxz (x + ∆x)
• rate of z-momentum in by overall bulk fluid motion across surface at z = 0 is ρW ∆xu2z (z = 0)
• rate of z-momentum out by overall bulk fluid motion across surface at z = L is ρW ∆xu2z (z = L)
• pressure force acting on surface at z = 0 is p0 W ∆x
• pressure force acting on surface at z = L is pL W ∆x
• gravity force acting in z-direction on volume of rectangular slab is (ρg cos β)LW ∆x

On substituting these contributions into the z-momentum balance, we get

LW τxz (x) − LW τxz (x + ∆x) + (p0 − pL )W ∆x + (ρg cos β)LW ∆x = 0 (1.a.1)

Dividing equation (1.a.1) by LW ∆x yields

τxz (x + ∆x) − τxz (x) p0 − pL + ρgL cos β

= (1.a.2)
∆x L
On taking the limit as ∆x → 0, the left-hand side of the above equation is exactly the definition of the
derivative. The right-hand side may be written in a compact and convenient way by introducing the modified
pressure P , which is the sum of the pressure and gravitational terms. The general definition of the modified
pressure is P = p + ρgh , where h is the distance upward (in the direction opposed to gravity) from a reference
plane of choice. The advantages of using the modified pressure P are that (i) the components of the gravity
vector g need not be calculated; (ii) the solution holds for any flow orientation; and (iii) the fluid may flow as
a result of a pressure difference, gravity or both. Here, h is negative since the z-axis points downward, giving
h = z cos β and therefore P = pρgz cos β. Thus, P0 = p0 at z = 0 and PL = pL ρgL cos β at z = L giving
p0 pL + ρgL cos β = P0 PL ≡ ∆P . Thus, equation (1.a.2) yields
dτxz ∆P
= (1.a.3)
dx L
This first-order differential equation may be simply integrated to give
∆P
τxz (x) = x + C1 (1.a.4)
L
Here, C1 is an integration constant, which is determined using an appropriate boundary condition based
on the flow problem. Equation (1.a.4) shows that the momentum flux (or shear stress) distribution is linear
in systems in rectangular Cartesian coordinates. Since equations (1.a.3) and (1.a.4) have been derived without
making any assumption about the type of fluid, they are applicable to both Newtonian and non-Newtonian
fluids. Some of the axial flow problems in rectangular Cartesian coordinates where these equations may be used
as starting points are given below.

2
1.b Determination of solution.
Since the problem is a laminar flow in a plane narrow slit, the obvious choice for a coordinate system is Cartesian
coordinates. The fluid flow is in the z−direction, therefore the velocity vector is of the form ~v = (0, 0, vz ).
Further, vz is independent of z and it is meaningful to postulate that vz is a function only for x, vz = vz (x),
and also for the pressure p = p (z). From these, and from the definition of the stress tensor, its nonvanishing
components are τxz = τzx , which depend only on x.
Using the differential shell momentum balance method, consider now a thin rectangular slab (shell) perpen-
dicular to the x-direction extending a distance W in the y-direction and a distance L in the z-direction. A ‘rate
of z-momentum’ balance over this thin shell of thickness ∆x in the fluid is of the form:

Rate of z-momentum = In Out + Generation = Accumulation

At steady-state, the accumulation term is zero. Momentum can go ’in’ and ’out’ of the shell by both the
convective and molecular mechanisms. Since vz is a function only of x, it is the same at both ends of the slit,
and the convective terms cancel out because (ρvz vz W ∆x)|z=0 = (ρvz vz W ∆x)|z=L . Only the molecular term
(LW τxz ) remains to be considered. Generation of z-momentum occurs by the pressure force (ρW ∆x) and
gravity force (ρgW L∆x). On substituting these contributions into the z-momentum balance, we get

(LW τxz )|x − (LW τxz )|x+∆x + (p0 − pL )W ∆x + ρgW L∆x = 0 (1.b.1)
Dividing the above equation by LW ∆x yields

τxz |x+∆x − τxz |x (p0 − pL ) + ρgL

= (1.b.2)
∆x L
On taking the limit as ∆x → 0 the left term of the above equation is the definition of the derivative dτdxxz . The
right-hand side may be compactly and conveniently written by introducing the modified pressure P , which is
the sum of the pressure and gravitational terms. The general definition of the modified pressure is P = p + ρgh,
where h is the distance upward (in the direction opposed to gravity) from a reference plane of choice. Since
the z-axis points downward in this problem, h = z and therefore P = pρgz. Thus, P0 = p0 at z = 0 and
PL = pL ρgL at z = L, giving p0 pL + ρgL = P0 PL ≡ P . Then equation (1.b.2) becomes
dτxz ∆P
= (1.b.3)
dx L
Equation (1.b.3) on integration leads to the following expression for the shear stress distribution tensor:
∆P
τxz = x + C1 (1.b.4)
L
To find the velocity, we substitute τxz from (1.b.4) into Newton’s law of viscosity, and get:
dvz ∆P
−µ = x + C1 (1.b.5)
dx L
The above first-order differential equation is simply integrated to obtain the following velocity profile:
∆P 2 C1
vz = − x − + C2 (1.b.6)
2µL µ
It is worth noting that equations (1.b.3) and (1.b.4) apply to both Newtonian and non-Newtonian fluids,
and provide starting points for many fluid flow problems in rectangular Cartesian coordinates.
Using the no-slip boundary conditions at the two fixed walls, vz (x) |x=B = vz (x) |x=−B = 0, the integration
B2
constants are C1 = 0 and C2 = ∆P 2µL .
Then the final expresions for the shear tensor and the velocity are:
∆P
τxz = x (1.b.7)
L
∆P B 2 x2
 
vz = 1− 2 (1.b.8)
2µL B

3
 It is observed that the velocity distribution for laminar, incompressible flow of a Newtonian fluid in a plane
narrow slit is parabolic.
From the velocity profile, various useful quantities may be derived. The maximum velocity occurs at x = 0
d 2 vz
(where dvdx = 0, dx2 < 0). Therefore,
z

∆P B 2
vz max = vz |x=0 = (1.b.9)
2µL
The average velocity is obtained by dividing the volumetric flow rate by the cross-sectional area,
RB
B
vz W dx ∆P B 2
Z
1 2
vz avg = −B
RB = vz dx = = vz max (1.b.10)
W dx 2B −B 3µL 3
−B

Thus, the ratio of the average velocity to the maximum velocity for Newtonian fluid flow in a narrow slit is
2
3.
The mass rate of flow is obtained by integrating the velocity profile over the cross section of the slit as
follows:
Z B
w= ρvz W dx = 2ρW Bvz avg (1.b.11)
−B

and substituting vz avg from (1.b.10) we have the final expresion for the mass rate of flow.

2∆P B 3 ρW
w= (1.b.12)
3µL
The flow rate vs. pressure drop (w vs. ∆P ) expression above is the slit analog of the Hagen-Poiseuille
equation (originally for circular tubes). It is a result worth noting because it provides the starting point for
creeping flow in many systems (e.g., radial flow between two parallel circular disks; and flow between two
stationary concentric spheres). Finally, it may be noted that the above analysis is valid when B  W . If the
slit thickness B is of the same order of magnitude as the slit width W , then vz = vz (x, y), i.e., vz is not a
function of only x. If W = 2B, then a solution can be obtained for flow in a square duct.

4
Problem 2
Steady, laminar flow occurs in the space between two fixed parallel, circular disks separated by a small gap 2b.
The fluid flows radially outward owing to a pressure difference (P1 − P2 ) between the inner and outer radii r1
and r2 , respectively. Neglect end effects and consider the region r1 ≤ r ≤ r2 only. Such a flow occurs when a
lubricant flows in certain lubrication systems.

Figure 2.1: Radial flow between two parallel disks.

a) Simplify the equation of continuity to show that rur is a function of only z.

b) Simplify the equation of motion for incompressible flow of a Newtonian fluid of viscosity µ and density ρ.
c) Obtain the velocity profile assuming creeping flow.
d) Sketch the velocity profile ur (r, z) and the pressure profile P (r).
e) Determine an expression for the mass flow rate by integrating the velocity profile.
f) Derive the mass flow rate expression in e) using an alternative short-cut method by adapting the plane
narrow slit solution.

Solution

2.a Simplification of continuity equation.

Since the steady laminar flow is directed radially outward, only the radial velocity component ur exists. The
tangential and axial components of velocity are zero, so the velocity vector becomes ~u = (ur , 0, 0).
For an incompressible flow, the continuity equation gives ∇ ~ · ~u = 0. In cylindrical coordinates, this is
expanded as follows:

∂ 1 ∂uθ ∂uz
(rur ) + + = 0⇒
∂r r ∂θ ∂z
∂
(rur ) = 0 ⇒
∂r
rur = f (θ, z)

We are expecting the solution to be symmetric around the z-axis, therefore the solution must be independent
of θ, hence

rur = f (z) (2.a.1)

This proves that rur is a function of z only.

2.b Simplification of the equation of motion for a Newtonian fluid.

For a Newtonian fluid, its equation of motion is the Navier-Stokes equation,

D~u ~ + µ∇2 ~u
ρ = −∇P (2.b.1)
Dt
in which P includes both the pressure and gravitational terms. On noting that ur = ur (r, z), equation (2.b.1)
can be separated into its coordinate components and for steady flow they may be simplified as given below.

5
 ∂ 2 ur
 
∂ur ∂P
r − component: ρ ur = − +µ 2 (2.b.2)
∂r ∂r ∂z
∂P
θ − component: 0 = (2.b.3)
∂θ
∂P
z − component: 0 = (2.b.4)
∂z
From (2.b.3) and (2.b.4) it is easy to prove that P = P (r). Recall that rur = f (z) from the continuity
equation. Substituting ur = f /r and P = P (r) in equation (2.b.2) then gives

f 2 (z) dP (r) µ d2 f (z)

−ρ 3
=− + (2.b.5)
r dr r dz 2
This is the equation of motion for an incompressible, steady laminar flow of a Newtonian fluid of viscosity
µ and density ρ.

2.c Determination of the velocity profile assuming creeping flow.

The equation of motion (2.b.5) is an inhomogeneous and generally non-linear differential equation (it depends
on the form of f (z)). It has no solution unless the nonlinear term (that is, the f 2 term on the left-hand side)
is neglected. This is the ‘creeping flow’ assumption, and under this assumption, equation (2.b.5) becomes an
homogeneous equation. The latter can be written as

dP (r) d2 f (z)
r =µ (2.c.1)
dr dz 2
The left term of (2.c.1) is a function of only r, while the right term is a function of only z. The equality can
be true only if both terms are equal with a constant (a0 ∈ R), so that

dP (r) d2 f (z)
r =µ = a0
dr dz 2
Thus the original partial differential equation has been decoupled into two ordinary differential equations

dP (r) a0
= (2.c.2)
dr r
d2 f (z)
µ = a0 (2.c.3)
dz 2
To solve this system, all we need to do is integrate each equation (equation (2.c.3) needs to be integrated
twice). We thus find

 
r2
P (r2 ) − P (r1 ) = a0 ln (2.c.4)
r1
a0 2
f (z) = z + a1 z + a2 (2.c.5)
2µ
where a1 and a2 are integration constants.
By substituting a0 from equation (2.c.4) into equation (2.c.5), we finally get that

P (r2 ) − P (r1 ) 2
f (z) =   z + a1 z + a2 (2.c.6)
2µ ln rr21

The no-slip boundary conditions of our problem dictate that

6
 ur (r, z = +b) = ur (r, z = −b) = 0
If we insert these into (2.c.6), the result is that

z2
 
P (r2 ) − P (r1 ) 2
f (z) =   b −1 (2.c.7)
2µ ln rr21 b2

therefore as we have assumed creeping flow, the velocity profile is:

P (r2 ) − P (r1 ) 2 z 2
 
ur (r, z) =   b − 1 (2.c.8)
2µr ln rr21 b2

Furthermore the expression for the pressure P (r) is given from the solution of (2.c.2) and is
 
ln rr2
P (r) = P (r1 ) + [P (r2 ) − P (r1 )]   (2.c.9)
ln rr21

where we have substituted a0 from (2.c.4).

2.d Sketch of the velocity and pressure profiles.

The velocity profile is described by equation (2.c.8) and the pressure profile by equation (2.c.9). We produce a
sketch for each using gnuplot, with the following numerical values:

b = 0.1 , µ=1
r2 = 5 , r1 = 1
P (r2 ) = 1 , P (r1 ) = 5

4.5

4
0.014
0.012
ur 0.01
3.5 0.008
0.006
0.004
3 0.002
P

0
-0.002
2.5

2 1
1.5 0.1
2
2.5 0.05
1.5 3
3.5 0
r
4 -0.05 b
4.5
1 5 -0.1
1 1.5 2 2.5 3 3.5 4 4.5 5
r

(a) Pressure profile. (b) Velocity profile.

Figure 2.2: Velocity and pressure profiles.

Regardless of the precise numerical values, the graphs are representative of the fluid’s behavior when the
pressure is greater at the center. The fluid begins to flow outwards rapidly, but both the velocity and pressure
drop as the fluid approaches the outer perimeter. The fluid’s velocity also diminishes as one approaches either
plate, and remains highest at the central area between the plates.

7
2.e Mass flow rate by integration of the velocity profile.
The mass flow rate w is rigorously obtained by integrating the velocity profile using
Z
w = n̂ · ρ~udS

where n̂ is the unit normal to the element of surface area dS and ~u is the fluid velocity vector. For the radial
flow between parallel disks, n̂ = r̂, ~u = ur r̂, and dS = 2πrdz. Then, substituting the velocity profile and
integrating gives

Z b
w = ρ (~n ~u) dS
−b
Z b
= ρur dS
−b
Z b
= 2π ρur (r, z)rdz
−b

and using (2.a.1) and (2.c.7) for rur we finally have for the mass flow rate that

4π P (r1 ) − P (r2 ) 3
w=   b ρ (2.e.1)
3µ ln rr12

Mass flow rate using short–cut method by adapting narrow slit solution
The plane narrow slit solution may be applied locally by recognizing that at all points between the disks,
the flow resembles the flow between parallel plates provided ur is small (that is, the creeping flow is valid).
The mass flow rate for a Newtonian fluid in a plane narrow slit of width W , length L and thickness 2B is
given by

2∆P B 3 W ρ
w= .
3µL
In this expression, ∆P/L is replaced by −dP/dr, B by b, and W by 2πr. Note that mass is conserved, so
w is constant. Then, integrating from r1 to r2 gives the same mass flow rate expression [equation (2.e.1)], as
shown below.
r2 P2
b3 r 4π P1 − P2 3
Z Z
dr
w = 4π (−dP ) ⇒ w=  b ρ (2.e.2)
r1 r 3µ P1 3µ ln r2
r1

This alternative short–cut method for determining the mass flow rate starting from the narrow slit solution
is very powerful because the approach may be used for non–Newtonian fluids where analytical solutions are
difficult to obtain.

8
Problem 3
A wire – coating die essentially consists of a cylindrical wire of radius κR (κ < 1) moving horizontally at a
constant velocity u along the axis of a cylindrical die of radius R. If the pressure in the die is uniform, then
the polymer melt (which may be considered a non-Newtonian fluid described by the power law model and of
constant density ρ) flows through the narrow annular region solely by the drag due to the axial motion of the
wire (which is referred to as ‘axial annular Couette flow’). Neglect end effects and assume an isothermal system.

Figure 3.1: Fluid flow in wire–coating die.

a) Establish the expression for the steady-state velocity profile in the annular region of the die. Simplify the
expression for n = 1 (Newtonian fluid).
b) Obtain the expression for the mass flow rate through the annular die region. Simplify the expression for
n = 1 and n = 1/3.
c) Estimate the coating thickness δ some distance downstream of the die exit.
d) Find the force that must be applied per unit length of the wire.
e) Write the force expression in d) as a ‘plane slit’ formula with a ‘curvature correction’.
Solution

3.a Shear stress and velocity distribution.

The fluid flows in the z-direction shearing constant-r surfaces. Therefore, the velocity component that exists is
uz (r) and the shear stress component is τrz (r). A momentum balance in cylindrical coordinates gives

d(rτrz ) ∆P
= r (3.a.1)
dr L
Since there is no pressure gradient and gravity does not play any role in the motion of the fluid, ∆P = 0.
Equation (3.a.1) with the right-hand side set to zero leads to the following expression for the shear stress
distribution on integration:
C1
τrz = (3.a.2)
r
The constant of integration C1 is determined later using boundary conditions.
Since the velocity uz decreases with increasing radial distance r, the velocity gradient duz /dr is negative in
the annular die region and the power law model may be substituted in the following form for τrz in equation
(3.a.2).
 n
duz C1
m − = (3.a.3)
dr r
Here, m and n are the consistency index and exponent in the power law model, respectively. (The Newtonian
fluid model has m = µ and n = 1, where µ is the fluid’s dynamic viscosity.) The above differential equation is
simply integrated to obtain the following velocity profile (for n 6= 1):
 n1
rq

C1
uz = − + C2 (3.a.4)
m q
where q = 1 − n1 . The integration constants C1 and C2 are evaluated from the following no-slip boundary
conditions:

9
 uz (r = R) = 0, (3.a.5)
uz (r = κR) = V (3.a.6)

From (3.a.5),
 n1
Rq

C1
C2 =
m q
So,
 n1
Rq h
  r q i
C1
uz = 1−
m q R
and (3.a.6) yields
 n1
Rq

C1
V = (1 − κq )
m q
On substituting the integration constants into equations (3.a.2) and (3.a.4), the final expressions for the
shear stress and velocity distribution (for n 6= 1) are

 n
m qV
τrz = (3.a.7)
r Rq (1 − κq )
uz (r/R)q − 1
= (3.a.8)
V κq − 1
When n = 1, the right-hand sides of equations (3.a.7) and (3.a.8) evaluate to 0/0; therefore, the evaluation
may be done using De l’Hopital’s rule by differentiating the numerator and denominator with respect to n (or
q) and then evaluating the limit as n → 1 (or q → 0). On differentiating equation (3.a.7),
   
q (0/0) 1 1
lim q
= lim q
=
q→0 1 − κ q→0 −κ ln κ − ln κ
Similarly, on differentiating equation (3.a.8),

(r/R)q − 1 (0/0) (r/R)q ln (r/R)

   
ln (r/R)
lim q
= lim q
=
q→0 κ −1 q→0 κ ln κ ln κ
Thus, on simplifying equations (3.a.7) and (3.a.8), the shear stress and velocity for a Newtonian fluid (n = 1
and m = µ) are

µV
τrz = (3.a.9)
r ln (1/κ)
uz ln (r/R)
= (3.a.10)
V ln κ

3.b Mass flow rate.

The mass rate of flow is obtained by integrating the velocity profile over the cross section of the annular region
as follows.
R 1
2πR2 V ρ
Z Z
r h r q i r
w = 2πρ ruz dr = −1 d
κR κq − 1 κ R R R
Integration then gives

10
 1
2πR2 V ρ

1  r 2+q 1  r 2
w= − (3.b.1)
κq − 1 2 + q R 2 R κ

On evaluation at the limits of integration, the final expression for the mass rate of flow (for n 6= 1 and n 6= 13 )
is

2πR2 V ρ 1 − κ2+q 1 − κ2
 
w= q − (3.b.2)
κ −1 2+q 2
When n = 1 (or q = 0), the right-hand side of the above equation gives 0/0. Again, using De l’Hopital’s
rule gives

1 − κ2+q 1 − κ2 1 (2 + q)(−κ2+q ) ln κ − (1 − κ2+q )

    
1 (0/0)
lim q
− = lim q
q→0 κ −1 2+q 2 q→0 κ ln κ (2 + q)2
2κ2 ln κ + (1 − κ2 )
= −
4 ln κ
Substituting this in equation (3.b.2) gives the mass flow rate for a Newtonian fluid (n = 1) as

πR2 V ρ 1 − κ2
 
2
w= − 2κ (3.b.3)
2 ln (1/κ)
When n = 1/3 (or q = 2), the term (1κ2+q )/(2 + q) becomes 0/0. Again using l’Hopital’s rule,

1 − κ2+q (0/0)
 
= lim −κ2+q ln κ = − ln κ


lim
q→−2 2+q q→−2

Substituting this last result in equation (3.b.2) gives the mass flow rate for a fluid with n = 1/3 (q = 2) as

2πR2 V ρ 1 − κ2
 
1
w= q ln − (3.b.4)
κ −1 κ 2

3.c Coating thickness.

Some distance downstream of the die exit, the liquid (polymer melt) is transported as a coating of thickness
δ exposed to air. A momentum balance in the ‘air region’ yields the same shear stress expression [equation
(3.a.2)]. The boundary condition at the liquid-gas interface is τrz = 0 at r = κR + δ, which now gives C1 = 0
and τrz = 0 throughout the coating. Furthermore, equations (3.a.4) and (3.a.6) are valid and give the velocity
profile as uz = V through the entire coating. As the velocity becomes uniform and identical to that of the wire,
the liquid is transported as a rigid coating. The mass flow rate in the air region is given by
Z κR+δ " 2 #
δ
wair = 2πρV rdr = πR2 V ρ κ + − κ2 (3.c.1)
κR R
Equating the above expression for the flow rate wair in the air region to the flow rate w in the die region
gives
r
δ w
= κ2 + −κ (3.c.2)
R πR2 V ρ
On substituting the appropriate expression for w from equations (3.b.4), (3.b.3), or (3.b.2), the final expres-
sion for the coating thickness is obtained. It may be noted that the coating thickness δ depends only on R, κ
and n, but not on V and m.

11
3.d Force applied to the wire.
The force applied to the wire is given by the shear stress integrated over the wetted surface area. Therefore, on
using equation (3.a.7) or equation (3.a.9),
(  n
qV
Fz Fz 2πm Rq (1−κ q) , n 6= 1
= 2πκR [τrz ]r=κR ⇒ = 2πµV
(3.d.1)
L L , n=1
ln 1/κ

3.e Plane slit formula for force.

A very narrow annular region with outer radius R and inner radius (1 − )R, where  is very small, may be
approximated by a plane slit with area = 2π(1 2 )RL ≈ 2πRL and gap = R. Then, the velocity profile in the
plane slit will be linear and the force applied to the ‘plane’ wire is given by
Fplane V Fplane 2πµV
=µ or = (3.e.1)
2πRL R L 
On substituting κ = 1 −  (since R − κR = R for the thin annular gap) in equation (3.d.1), the force applied
to the wire is given by
Fz 2πµV
= (3.e.2)
L − ln (1 − )
The Taylor series expansion given below is next used.
1 1 1 1
ln (1 − ) = − − 2 − 3 − 4 − 5 − . . . (3.e.3)
2 3 4 5
Therefore, the denominator of equation (3.e.2) may be written as

 −1  
1 1 1 1 1 1 1 1 1 1 19 4
− = 1 +  + 2 + 3 + 4 + . . . = 1 −  − 2 − 3 −  − ... (3.e.4)
ln (1 − )  2 3 4 5  2 12 24 720

Equation (3.e.4) may be obtained by using (1 + u)−1 = 1 − u + u2 − u3 + u4 − . . . (which is an infinite

geometric progression with |u| < 1). On substituting in equation (3.e.2), the following force expression is finally
obtained as a ‘plane slit’ formula with a ‘curvature correction’.
 
Fz 2πµV 1 1 1 19 4
= 1 −  − 2 − 3 −  − ... (3.e.5)
L  2 12 24 720
The above problem solution considers only drag flow in a wire-coating die. The problem is solved considering
both drag and pressure by Malik, R. and Shenoy, U. V. (1991) ‘Generalized annular Couette flow of a power-law
fluid’, Ind. Eng. Chem. Res. 30 (8): 1950 - 1954.

12
Problem 4
Consider an incompressible isothermal fluid in laminar flow between two coaxial cylinders, whose inner and
outer wetted surfaces have radii of κR and R, respectively. The inner and outer cylinders are rotating at
angular velocities Ωi and Ωo , respectively. End effects may be neglected.

Figure 4.1: Tangential annular flow between two slowly rotating cylinders.

a) Determine the steady-state velocity distribution in the fluid (for small values of Ωi and Ωo ).
b) Find the torques acting on the two cylinders during the tangential annular flow of a Newtonian fluid.

Solution

4.a Steady-state velocity distribution.

Simplification of the continuity equation
In steady laminar flow, the fluid is expected to travel in a circular motion (for low values of Ωi and Ωo ).
Only the tangential component of velocity exists. The radial and axial components of velocity are zero; so,
~ · ~u = 0.
ur = 0 and uz = 0. For an incompressible fluid, the continuity equation gives ∇
In cylindrical coordinates,
1 ∂ 1 ∂uθ ∂uz ∂uθ
(rur ) + + =0 ⇒ =0
r ∂r r ∂θ ∂z ∂θ
So, uθ = uθ (r, z).
If end effects are neglected, then uθ does not depend on z. Thus, uθ = uθ (r).

Simplification of the equation of motion

There is no pressure gradient in the θ-direction. Therefore, the components of the equation of motion
simplify to

u2θ ∂p
r − component: −ρ = −
r ∂r
 
d 1 d
θ − component: 0 = (ruθ )
dr r dr
∂p
z − component: 0 = − − ρg (z points upwards)
∂z
where in the equation for the θ-component the partial derivatives have been replaced with ordinary ones,
since uθ is a function only of r as was shown in the previous step. The r-component provides the radial
pressure distribution due to centrifugal forces and the z-component gives the axial pressure distribution due to
gravitational forces (the hydrostatic head effect).

Solution of the differential equation and the velocity profile

The θ-component of the equation of motion can be integrated to get the velocity profile:

13
 1 d r2 C1 C2
(ruθ ) = C1 ⇒ ru − θ = C1 + C2 ⇒ uθ (r) = r+ (4.a.1)
r dr 2 2 r
The no-slip boundary conditions at the two cylindrical surfaces are

C1 C2
uθ (r = κR) = Ωi κR ⇐⇒ Ωi κR = κR + (4.a.2)
2 κR
C1 C2
uθ (r = R) = Ωo R ⇐⇒ Ωo R = R+ (4.a.3)
2 R
Solving equations (4.a.2) and (4.a.3) for the integration constants we find that they are given by

 
1 κ
(Ωi − Ωo )κR = C2 − ,
κR R
   
R C1 R
Ωi κR − Ωo = κR −
κ 2 κ

Substituting the above values for C1 and C2 in equation (4.a.1), the velocity profile is obtained as

Ωo − Ωi κ2 Ωi − Ωo κ2 R2
 
κR 2 r κR
uθ (r) = r+ = (Ωo − Ωi κ ) + (Ωi − Ωo ) (4.a.4)
1 − κ2 1 − κ2 r 1 − κ2 κR r
The velocity distribution given by the above expression can be written in the following alternative form:

Ωi κ2 R R
   
Ωo κR r κR r
uθ (r) = − + − (4.a.5)
1 − κ2 κR r 1 − κ2 r R
In the above form, the first term on the right-hand-side corresponds to the velocity profile when the inner
cylinder is held stationary and the outer cylinder is rotating with an angular velocity Ωo . This is essentially the
configuration of the Couette - Hatschek viscometer for determining viscosity by measuring the rate of rotation
of the outer cylinder under the application of a given torque (as discussed below). The second term on the
right-hand-side of the above equation corresponds to the velocity profile when the outer cylinder is fixed and
the inner cylinder is rotating with an angular velocity Ωi . This is essentially the configuration of the Stormer
viscometer for determining viscosity by measuring the rate of rotation of the inner cylinder under the application
of a given torque (as discussed next).

4.b Torques acting on the cylinders.

Determination of momentum flux
From the velocity profile, the momentum flux (shear stress) is determined as

d  uθ  Ωi − Ωo κ2 R2
τrθ = −µr = 2µ (4.b.1)
dr r 1 − κ2 r2
Determination of torque
The torque is obtained as the product of the force and the lever arm. The force acting on the inner cylinder
itself is the product of the inward momentum flux and the wetted surface area of the inner cylinder. Thus,

κ2
Tz = (2πκRL)(κR) (−τrθ |r=R ) = 4πµ(Ωi − Ωo ) R2 L (4.b.2)
1 − κ2
The tangential annular flow system above provides the basic model for rotational viscometers (to determine
fluid viscosities from measurements of torque and angular velocities) and for some kinds of friction bearings.

14
Problem 5
An incompressible isothermal liquid in laminar flow is open to the atmosphere at the top. Determine the shape
of the free liquid surface z(r) at steady state (neglecting end effects, if any) for the following cases:

(a) – case. (b) – case.

(c) – case. (d) – case.

Figure 5.1: Free surface shape of liquid in four cases of tangential flow.

a) when the liquid is in the annular space between two vertical coaxial cylinders, whose inner and outer
wetted surfaces have radii of κR and R, respectively. Here, the inner cylinder is rotating at a constant
angular velocity Ωi and the outer cylinder is stationary. Let zR be the liquid height at the outer cylinder.
b) when a single vertical cylindrical rod of radius Ri is rotating at a constant angular velocity Ωi in a large
body of quiescent liquid. Let zRo be the liquid height far away from the rotating rod.
Hint: Use the result from a).
c) when the liquid is in the annular space between two vertical coaxial cylinders, whose inner and outer
wetted surfaces have radii of κR and R, respectively. Here, the outer cylinder is rotating at a constant
angular velocity Ωo and the inner cylinder is stationary. Let zR be the liquid height at the outer cylinder.
d) when the liquid is in a vertical cylindrical vessel of radius R, which is rotating about its own axis at a
constant angular velocity Ωo . Let zR be the liquid height at the vessel wall.
Hint: Use the result from c).

Solution
Simplification of the continuity equation
In steady laminar flow, the liquid is expected to travel in a circular motion with only the tangential component
of velocity. The radial and axial components of velocity are zero; so, ur = 0 and uz = 0.
For an incompressible fluid, the continuity equation gives ∇ ~ · ~u = 0. In cylindrical coordinates,

1 ∂ 1 ∂uθ ∂uz ∂uθ

(rur ) + + =0 ⇒ =0 (5.1)
r ∂r r ∂θ ∂z ∂θ
So, uθ = uθ (r, z). If end effects are neglected, then uθ does not depend on z. Thus, uθ = uθ (r).

15
 Simplification of the equation of motion
There is no pressure gradient in the θ-direction. The components of the equation of motion simplify to

u2θ ∂p
r − component: −ρ = −
r ∂r
 
d 1 d
θ − component: 0 = (ruθ )
dr r dr
∂p
z − component: 0 = − − ρg (z points upwards)
∂z
where in the equation for the θ-component the partial derivatives have been replaced with ordinary ones, since
uθ is a function only of r as was shown in the previous step.
As discussed later, the r-component provides the radial pressure distribution due to centrifugal forces and
the z-component gives the axial pressure distribution due to gravitational forces (the hydrostatic head effect).

Solution of the differential equation and the velocity profile The θ-component of the equation of
motion is integrated to get the velocity profile:

r2
 
d 1 d 1 d
(ruθ ) = 0 ⇒ (ruθ ) = C1 ⇒ ruθ = C1 + C2 (5.2)
dr r dr r dr 2
and therefore, the velocity profile is described by the equation
r C2
uθ (r) = C1
+ (5.3)
2 r
The no-slip boundary conditions at the two cylindrical surfaces are

C1 C2
uθ (r = κR) = Ωi κR ⇐⇒ Ωi κR = κR + (5.4)
2 κR
C1 C2
uθ (r = R) = Ωo R ⇐⇒ Ωo R = R+ (5.5)
2 R
Solving equations (5.4) and (5.5) for the integration constants we find that they are given by

 
1 κ
(Ωi − Ωo )κR = C2 − ,
κR R
   
R C1 R
Ωi κR − Ωo = κR −
κ 2 κ

Substituting the above values for C1 and C2 in equation (5.3), the velocity profile is obtained as

Ωo − Ωi κ2 Ωi − Ωo κ2 R2
 
κR 2 r κR
uθ (r) = r+ = (Ωo − Ωi κ ) + (Ωi − Ωo ) (5.6)
1 − κ2 1 − κ2 r 1 − κ2 κR r
The velocity distribution given by the above expression can be written in the following alternative form:

Ωi κ2 R R
   
Ωo κR r κR r
uθ (r) = − + − (5.7)
1 − κ2 κR r 1 − κ2 r R
In the above form, the first term on the right-hand-side corresponds to the velocity profile when the inner
cylinder is held stationary and the outer cylinder is rotating with an angular velocity Ωo . The second term on
the right-hand-side of the above equation corresponds to the velocity profile when the outer cylinder is fixed
and the inner cylinder is rotating with an angular velocity Ωi .

16
5.a Free surface shape for rotating inner cylinder and fixed outer cylinder.
Now, consider the case where the inner cylinder is rotating at a constant angular velocity Ωi and the outer
cylinder is stationary. With Ωo = 0, the velocity profile in (5.7) is substituted into the r-component of the
equation of motion to get

ρu2θ ρ Ω2i κ4 R2 R2 r2
 
∂p
= = −2+ 2 (5.a.1)
∂r r r (1 − κ2 )2 r2 R
Integrating the above equation we find the pressure profile as

ρΩ2i κ4 R2 R2 r2
 
p(r, z) = − − 2 ln r + + f1 (z) (5.a.2)
(1 − κ2 )2 2r2 2R2
It is clear that the pressure profile is independent of θ due to the axial symmetry present. The above
expression for the pressure p is next substituted into the z-component of the equation of motion to obtain
∂p df1
= −ρg ⇒ = −ρg ⇒ f1 (z) = −ρgz + C3 (5.a.3)
∂z dz
On substituting the expression for f1 in (5.a.3) we find that

ρΩ2i κ4 R2 R2 r2
 
p(r, z) = − 2 − 2 ln r + − ρgz + C3
(1 − κ2 )2 2r 2R2
The constant C3 is determined from the requirement that at the outer cylinder the liquid height be zR . At
that point, the pressure is equal to the atmospheric pressure, patm , therefore

ρΩ2i κ4 R2
p(r = R, z = zR ) = patm ⇒ patm = (−2 ln R) − ρgzR + C3
(1 − κ2 )2
Subtraction of the above two equations allows for elimination of C3 and the determination of the pressure
distribution in the liquid as

ρΩ2i κ4 R2 R2 r2
 
p(r, z) − patm = − 2 − 2 ln r + + ρg(zR − z) (5.a.4)
(1 − κ2 )2 2r 2R2
Since p = patm at every point across the liquid-air interface (boundary), the shape of the free liquid surface
is obtained as
2 " 2 #
1 Ωi κ2 R
  r   r 2
R
zR − z = + 4 ln − . (5.a.5)
2g 1 − κ2 r R R

5.b Free surface shape for rotating cylinder in an infinite expanse of liquid.
The result in a) may be used to derive the free surface shape when a single vertical cylindrical rod of radius Ri
is rotating at a constant angular velocity Ωi in a large body of quiescent liquid.
On substituting R = Ro and κ = Ri /Ro , equations (5.7) (with Ωo = 0) and (5.a.5) yield

2 " 2 #
Ωi Ri2 Ωi Ri2 r2
    
1 r 1 1 4 r
uθ = − 2 and zRo −z = + 2 ln − 4 (5.b.1)
1 − Ri2 /Ro2 r Ro 2g 1 − Ri2 /Ro2 r Ro Ro Ro

The above equations give the velocity profile and the shape of the free liquid surface for the case where the
inner cylinder of radius Ri is rotating at a constant angular velocity Ωi and the outer cylinder of radius Ro is
stationary. Here, zRo is the liquid height far away from the rotating rod.
In the limit as Ro tends to infinity, the velocity profile and free surface shape for a single vertical rod rotating
in a large body of quiescent liquid are obtained as

Ωi Ri2 Ω2i Ri4

uθ = and zRo − z = . (5.b.2)
r 2gr2

17
5.c Free surface shape for rotating outer cylinder and fixed inner cylinder.
Next, consider the case where the outer cylinder is rotating at a constant angular velocity Ωo and the inner
cylinder is stationary. With Ωi = 0, the velocity profile in equation (5.7) is substituted into the r-component
of the equation of motion to get

ρu2θ ρ Ω2o κ2 R2 κ2 R2 r2
 
∂p
= = −2+ 2 2 (5.c.1)
∂r r r (1 − κ2 )2 r2 κ R
Integration gives

ρΩ2o κ2 R2
 2 2
r2

κ R
p(r, z) = − − 2 ln r + 2 2 + f2 (z) (5.c.2)
(1 − κ2 )2 2r2 2κ R
The above expression for the pressure p is now substituted into the z-component of the equation of motion
to obtain
∂p df2
= −ρg ⇒ = −ρg ⇒ f2 (z) = −ρgz + C4 (5.c.3)
∂z dz
On substituting the expression for f2 in equation (5.c.2), we get

ρΩ2o κ2 R2
 2 2
r2

κ R
p(r, z) = − − 2 ln r + − ρgz + C4
(1 − κ2 )2 2r2 2κ2 R2
Again as in a), the constant C4 is determined from the requirement that at the outer cylinder the liquid
height be zR . At that point, the pressure is equal to the atmospheric pressure, patm , therefore

ρΩ2o κ2 R2
 2 
κ 1
p(r = R, z = zR ) = patm ⇒ patm = − − 2 ln R + 2 − ρgzR + C4
(1 − κ2 )2 2 2κ
Subtraction of the above two equations allows for elimination of C4 and the determination of the pressure
distribution in the liquid as

ρΩ2o κ2 R2
 2 2
r2
  
κ R r 1
p(r, z) − patm = − − 1 − 2 ln − 2 1− 2 + ρg(zR − z) (5.c.4)
(1 − κ2 )2 2 r2 R 2κ R
Since p = patm at every point across the liquid-air interface (boundary), the shape of the free liquid surface
is obtained as
2   2
r2
  r  
1 Ωo R 4 R 2
zR − z = κ − 1 + 4κ ln + 1 − . (5.c.5)
2g 1 − κ2 r2 R R2
An alternative approach is to obtain the above equation from the result of a) by substituting r = κR in
equation (5.a.5) to get
2 
1 Ωi κ2 R
 
1 2
zR − zκR = + 4 ln κ − κ (5.c.6)
2g 1 − κ2 κ2
The above expression gives the difference in elevations of the liquid surface at the outer and inner cylinders.
Subtracting equation (5.c.6) from equation (5.a.5) yields
2  2
1 Ωi κ2 R  r   r2
  
R 1 2
zκR − z = − 2 + 4 ln − −κ (5.c.7)
2g 1 − κ2 r2 κ κR R2
The cylinders must now be swapped, i.e., the rotating cylinder should be made the outer cylinder rather
than the inner one. So, let κR = Ro and R = βRo . On substituting κ = 1/β and R = βRo in the above
equation, we get
 2  2 2     2 
1 Ωi βRo β Ro 2 r r 1
zRo − z = − β + 4 ln − − (5.c.8)
2g β 2 − 1 r2 Ro β 2 Ro2 β2
On recognizing that β (which is a dummy parameter) can be simply replaced by κ, Ro may be replaced by
R, and Ωi by Ωo , the above equation is identical to equation (5.c.5).

18
5.d Free surface shape for cylindrical vessel rotating about its own axis.
The result in c) may be used to derive the free surface shape when a liquid is in a vertical cylindrical vessel of
radius R, which is rotating about its own axis at a constant angular velocity Ωo .
In the limit as κ tends to zero, equation (5.7) (with Ωi = 0) and equation (5.c.5) yield the velocity profile
and free surface shape as

Ω2o R2
  r 2 
uθ = Ωo R and zR − z = 1− (5.d.1)
2g R
Note that as κ tends to zero, the first and second terms in square brackets in equation (5.c.5) also tend to
zero. The velocity profile suggests that each element of the liquid in a rotating cylindrical vessel moves like
an element of a rigid body. The free surface of the rotating liquid in a cylindrical vessel is a paraboloid of
revolution [since equation (5.d.1) corresponds to a parabola].

19
Problem 6
A plastic resin is in a vertical cylindrical vessel of radius R, which is rotating about its own axis at a constant
angular velocity Ω.

Figure 6.1: Parabolic mirror from free surface shape of rotating liquid.

a) Determine the shape of the free surface z(r) at steady state (neglecting end effects, if any). Let zR be the
liquid height at the vessel wall.
b) Show that the radius of curvature (of the free surface shape) is given by
"  2 #3 /2  2 −1
dz d z 
rc = 1 +  
 dr2 
dr
c) The cylinder is rotated until the resin hardens in order to fabricate the backing for a parabolic mirror.
Determine the angular velocity (in rpm) necessary for a mirror of focal length 175 cm.
Hint: The focal length f is one-half the radius of curvature rc at the axis.

Solution
Simplification of the continuity equation
In steady laminar flow, the liquid is expected to travel in a circular motion with only the tangential component
of velocity. The radial and axial components of velocity are zero; so, ur = 0 and uz = 0. For an incompressible
fluid, the continuity equation gives ∇~ · ~u = 0. In cylindrical coordinates this is written as follows:

1 ∂ 1 ∂uθ ∂uz ∂uθ

(rur ) + + =0 ⇒ =0
r ∂r r ∂θ ∂z ∂θ
So, uθ = uθ (r, z). If end effects are neglected, then uθ cannot depend on z. Thus, uθ = uθ (r).

Simplification of the equation of motion

There is no pressure gradient in the θ-direction. The components of the equation of motion simplify to

u2θ ∂p
r − component: −ρ = −
r ∂r
 
d 1 d
θ − component: 0 = (ruθ )
dr r dr
∂p
z − component: 0 = − − ρg (z points upwards)
∂z
where in the equation for the θ-component the partial derivatives have been replaced with ordinary ones,
since uθ is a function only of r as was shown in the previous step. The r-component provides the radial
pressure distribution due to centrifugal forces and the z-component gives the axial pressure distribution due to
gravitational forces (the hydrostatic head effect).

20
 Solution of the differential equation and the velocity profile
The θ-component of the equation of motion can be integrated to get the velocity profile:

1 d r2 C1 C2
(ruθ ) = C1 ⇒ ru − θ = C1 + C2 ⇒ uθ (r) = r+ (6.1)
r dr 2 2 r
Since the velocity uθ is finite at r = 0, the integration constant C2 must be zero. Because uθ (r = R) = ΩR,
C1 = 2Ω. Thus,

uθ (r) = Ωr (6.2)
The above velocity profile suggests that each element of the liquid in a rotating cylindrical vessel moves like
an element of a rigid body.

6.a Free surface shape for rotating liquid in cylinder.

Next, the velocity profile is substituted into the r-component of the equation of motion to get

∂p ρu2
= θ = ρΩ2 r (6.a.1)
∂r r
Equation (6.a.1) is easily integrated to obtain

ρΩ2 2
p(r, z) = r + f1 (z) (6.a.2)
2
The above expression for the pressure p is now substituted into the z-component of the equation of motion
to obtain
dp df1
= −ρg ⇒ = −ρg ⇒ f1 (z) = −ρgz + C3 (6.a.3)
dz dz
On substituting the expression for f1 in equation (6.a.2),

ρΩ2 2
r − ρgz + C3
p(r, z) =
2
The constant C3 can be found from the following boundary condition:

ρΩ2 2
p(r = R, z = zR ) = patm ⇒ patm = R − ρgzR + C3
2
where zR is the liquid height at the vessel wall and patm denotes the atmospheric pressure.
Subtraction of the above two equations eliminates C3 and gives the pressure distribution in the liquid as

ρΩ2 2
p(r, z) − patm = (r − R2 ) + ρg(zR − z) (6.a.4)
2
Since p = patm at the liquid-air interface, the shape of the free liquid surface is finally obtained as

Ω2 2
zR − z = (R − r2 ) (6.a.5)
2g
The free surface of the rotating liquid in a cylindrical vessel is a paraboloid of revolution [since equation
(6.a.5) corresponds to a parabola].

21
6.b General expression for radius of curvature.
The circle that is tangent to a given curve at point P , whose center lies on the concave side of the curve and
which has the same curvature as the curve has at P , is referred to as the circle of curvature. Its radius (CP
in Figure 6.1) defines the radius of curvature at P . The circle of curvature has its first and second derivatives
respectively equal to the first and second derivatives of the curve itself at P . If the coordinates of the center C
and point P are (r1 , z1 ) and (r, z) respectively, then the radius of curvature CP is given by
p
rc = (r − r1 )2 + (z − z1 )2 (6.b.1)
Now, the radius CP is perpendicular to the tangent to the curve at P . So, the slope of the radius CP is
given by
dr z1 − z
− =
dz r1 − r
Thus, the first and second derivatives of the curve at point P are

dz r − r1 d2 z 1 r − r1 dz 1 + (dz/dr)2
= and = + =
dr z1 − z dr2 z1 − z (z1 − z)2 dr z1 − z
On substituting the above derivatives in equation (6.b.1), the general expression for the radius of curvature
is obtained as
"  2 #3 /2  2 −1
dz d z 
rc = 1 +  dr2  .
  (6.b.2)
dr

6.c Angular velocity necessary for parabolic mirror fabrication.

Let z0 be the height of the liquid surface at the centerline of the cylindrical vessel. Substituting z = z0 at r = 0
in equation (6.a.5) gives

Ω2 R 2
zR z0 = .
2g
Thus, an alternative expression (in terms of z0 ) for the shape of the free surface is

Ω2 2
z = z0 + r (6.c.1)
2g
Differentiating the above expression twice gives

dz Ω2 d2 z Ω2
= r and 2
=
dr g dr g
We can use the above derivatives to calculate the radius of curvature of the parabola. By substituting the
derivatives into equation (6.b.2), the radius of curvature of the parabola is given by
 3 /2
Ω4 r 2

g
rc = 1 + 2 (6.c.2)
g Ω2
Since the focal length f of the mirror is one–half the radius of curvature rc at the axis,
1 g
f= rc (r = 0) =
2 2Ω2
and therefore, the rotational speed necessary to manufacture a parabolic mirror of focal length f is
r
g
Ω= (6.c.3)
2f
On substituting the focal length f = 1.75 m and the gravitational acceleration g = 9.81 m/s2 in the above
equation, the rotational speed may be calculated as Ω = 1.68 rad/s = 1.68 × 60 ÷ (2π) rpm = 16.00 rpm.

22
Problem 7
Consider an incompressible isothermal fluid in laminar flow between two concentric spheres, whose inner and
outer wetted surfaces have radii of κR and R, respectively. The inner and outer spheres are rotating at constant
angular velocities Ωi and Ωo , respectively. The spheres rotate slowly enough that the creeping flow assumption
is valid.

Figure 7.1: Flow between two slowly rotating spheres.

a) Determine the steady-state velocity distribution in the fluid (for small values of Ωi and Ωo ).
b) Find the torques on the two spheres required to maintain the flow of a Newtonian fluid.
c) Simplify the velocity and torque expressions for the case of a single solid sphere of radius Ri rotating
slowly at a constant angular velocity Ωi in a very large body of quiescent fluid.

Solution

7.a The steady-state velocity distribution.

Simplification of the continuity equation
In steady laminar flow, the liquid is expected to travel in a circular motion (for low values of Ωi and Ωo )
with only the tangential component of velocity. The radial and polar components of velocity are zero; so, ur = 0
and uθ = 0.
~ · ~u = 0. In spherical coordinates this is written
For an incompressible fluid, the continuity equation gives ∇
as follows:
1 ∂ 2 1 ∂ 1 ∂
2
(r ur ) + (uθ ηµ θ) + (uφ ) = 0
r ∂r r ηµ θ ∂θ r ηµ θ ∂φ
So, uφ = uφ (r, θ).

Simplification of the equation of motion When the flow is very slow, the term ρ(~u · ∇)~ ~ u in the
equation of motion can be neglected because it is quadratic in the velocity. This is referred to as the creeping
flow assumption (in other words, we ignore convective acceleration). Thus, for steady creeping flow, the entire
left-hand side of the equation of motion is zero and we get

~ + µ∇2 ~u + ρg = −∇P
0 = −∇p ~ + µ∇2 ~u

where P is the modified pressure given by P = p + ρgh and h is the elevation in the gravitational field (i.e.,
the distance upward in the direction opposite to gravity from some chosen reference plane). In this problem,
h = z = r συν θ. The above equation is referred to as the Stokes flow equation or the creeping flow equation of
motion. The components of the creeping flow equation in spherical coordinates simplify to

23
 ∂P
r − component: 0 = −
∂r
1 dP
θ − component: 0 = −
r dθ
   
1 ∂ 2 ∂uφ 1 ∂ 1 ∂(uφ ηµ θ)
φ − component: 0 = r +
r2 ∂r ∂r r2 ∂θ ηµ θ ∂θ

There is no dependence on the angle φ due to symmetry about the vertical axis. The above partial differential
equation must be solved for the velocity distribution uφ (r, θ).

Solution of the differential equation and the velocity profile

The no-slip boundary conditions at the two spherical surfaces are

uφ (r = κR, θ) = Ωi κR ηµ θ (7.a.1)
uφ (r = R, θ) = Ωo R ηµ θ (7.a.2)

From the boundary conditions, the form uφ (r, θ) = f (r) ηµ θ appears to be an educated guess to solve
the partial differential equation for the velocity profile. On substituting this form in the φ-component of the
equation of motion and simplifying, we get

d2 f df
r2 + 2r − 2f = 0 (7.a.3)
dr2 dr
The above ordinary differential equation may be solved by substituting f (r) = rn (i.e. using the power
series method). The resulting characteristic equation is n2 + n − 2 = 0, whose solution is n = 1, −2. Thus,
 
C2 C2
f (r) = C1 r + 2 and uφ (r, θ) = C1 r + 2 ηµ θ (7.a.4)
r r
On imposing the boundary conditions (7.a.1) and (7.a.2), we find that
C2 C2
Ωi κR = C1 κR + and Ωo R = C1 R +
κ2 R2 R2
The integration constants are thus given by

Ωo − Ωi κ3 κ3 R3
C1 = and C2 = (Ωi − Ωo )
1 − κ3 1 − κ3
On substituting for C1 and C2 in (7.a.4), the velocity profile is obtained as

" 2 #
Ωo − Ωi κ3 κ3 R3 1
  
κR 3
 r  κR
uφ (r, θ) = r + (Ωi − Ωo ) ηµ θ = (Ωo − Ωi κ ) + (Ωi − Ωo ) ηµ θ
1 − κ3 1 − κ3 r2 1 − κ3 κR r
(7.a.5)
The velocity distribution given by the above expression can be written in the following alternative form:

κ2 R2 κ3 R
    2 
Ωo κR r R r
uφ (r, θ) = − + Ω i − ηµ θ (7.a.6)
1 − κ3 κR r2 1 − κ3 r2 R
In the above form, the first term on the right-hand-side corresponds to the velocity profile when the inner
sphere is held stationary and the outer sphere is rotating with an angular velocity Ωo . The second term on the
right-hand-side of the above equation corresponds to the velocity profile when the outer sphere is fixed and the
inner sphere is rotating with an angular velocity Ωi .

24
7.b Determination of momentum flux and torque.
From the velocity profile, the momentum flux (shear stress) is determined as

∂  uφ  Ωi − Ωo κ3 R3
τrφ = −µr = 3µ ηµ θ (7.b.1)
∂r r 1 − κ3 r 3
The torque required to rotate the inner sphere is obtained as the integral over the sphere surface of the
product of the force exerted on the fluid by the solid surface element and the lever arm κR ηµ θ. The force
required on the inner sphere is the product of the outward momentum flux and the wetted surface area of the
inner sphere. Thus, the differential torque on the inner sphere surface element is
Ωi − Ωo
dTz = (τrφ )|r=κR (2πκ2 R2 ηµ θdθ)(κR ηµ θ) = 3µ ηµ θ(2πκ2 R2 ηµ θdθ)(κR ηµ θ) (7.b.2)
1 − κ3
On integrating from 0 to π, the torque needed on the inner sphere is
Z π
κ3 R3 3 κ3 R3
Tz = 6πµ(Ωi − Ωo ) ηµ θdθ = 8πµ(Ω i − Ω o ) (7.b.3)
1 − κ3 0 1 − κ3
Note that the integral above evaluates to 4/3.
Similarly, the differential torque needed on the outer sphere surface element is obtained as the product of
the inward momentum flux, the wetted surface area of the outer sphere, and the lever arm. Thus,
Ωi − Ωo
dTz = (τrφ )|r=R (2πR2 ηµ θdθ)(R ηµ θ) = 3µ ηµ θ(2πR2 ηµ θdθ)(R ηµ θ) (7.b.4)
1 − κ3
On integrating from 0 to π, the torque needed on the inner sphere is
Z π
κ3 R3 3 κ3 R3
Tz = 6πµ(Ωo − Ωi ) ηµ θdθ = 8πµ(Ω o − Ω i ) . (7.b.5)
1 − κ3 0 1 − κ3

7.c Limiting case of single rotating sphere.

For the case of a single solid sphere of radius Ri rotating slowly at a constant angular velocity Ωi in a very large
body of quiescent fluid, let R = Ro and κ = Ri /Ro . Then, on letting Ωo = 0, equations (7.a.6) and (7.b.3) give
 3
Ri3 r

1 Ri 1
uφ (r, θ) = Ωi − ηµ θ and Tz = 8πµΩi (7.c.1)
1 − (Ri /Ro )3 r2 Ro3 (1/Ri )3 − (1/Ro )3
The above equations hold when the outer sphere of radius Ro is stationary and the inner sphere of radius
Ri is rotating at an angular velocity Ωi . In the limit as Ro tends to infinity, the results for a single rotating
sphere in an infinite body of quiescent fluid are obtained as

Ri3
uφ (r, θ) = Ωi ηµ θ and Tz = 8πµΩi Ri3 . (7.c.2)
r2

25
Problem 8
An isothermal, incompressible fluid of density ρ flows radially outward owing to a pressure difference between
two fixed porous, concentric spherical shells of radii κR and R. Note that the velocity is not zero at the solid
surfaces. Assume negligible end effects and steady laminar flow in the region κR ≤ r ≤ R.

Figure 8.1: Radial flow between two porous concentric spheres.

a) Simplify the equation of continuity to show that r2 ur = constant.

b) Simplify the equation of motion for a Newtonian fluid of viscosity µ.
c) Obtain the pressure profile P (r) in terms of PR and uR , the pressure and velocity at the sphere of radius
R.
d) Determine the nonzero components of the viscous stress tensor for the Newtonian case.

Solution

8.a Simplification of continuity equation.

Since the steady laminar flow is directed radially outward, only the radial velocity component vr exists. The
other two components of velocity are zero; so, uθ = 0 and uφ = 0. For an incompressible flow, the continuity
~ · ~u = 0. In spherical coordinates, this is written in fully expanded form as follows:
equation gives ∇
1 ∂ 2 1 ∂ 1 ∂ ∂ 2
(r ur ) + (uθ ηµ θ) + (uφ ) = 0 ⇒ (r ur ) = 0
r2 ∂r r ηµ θ ∂θ r ηµ θ ∂φ ∂r
After integrating the above simplified continuity equation, we find that r2 ur = f (θ, φ). However, due to the
symmetry present in the problem, there is no dependence expected on the angles θ and φ. In other words, r2 ur
must be equal to a constant, C. This is simply explained from the fact that mass (or volume, if density ρ is
constant) is conserved; so,
w
ρ(4πr2 ur ) = w is constant, and C = .
4πρ

8.b Simplification of the equation of motion for a Newtonian fluid.

The equation of motion is

D~u ~ −∇ ~ · τab + ρ~g

ρ = −∇p (8.b.1)
Dt
The above equation is simply Newton’s second law of motion for a fluid element. It states that, on a per
unit volume basis, the mass multiplied by the acceleration is because of three forces, namely the pressure force,
the viscous force, and the gravity force. For an incompressible, Newtonian fluid, the term ∇~ · τab = µ∇2 ~u and
2
the equation of motion yields the Navier - Stokes equation. On noting that r ur = constant from the continuity
equation, the components of the equation of motion for steady flow in spherical coordinates may be simplified
as given below.

26
  
∂ur ∂p
r − component: 0 = ρ ur =− − ρg συν θ
∂r ∂r
1 dp
θ − component: 0 = − + ρg ηµ θ
r dθ
∂p
φ − component: 0 =
∂φ
Note that the modified pressure P may be defined as P = p + ρgh, where h is the elevation or the height
above some arbitrary datum plane, to avoid calculating the components of the gravitational acceleration vector
~g in spherical coordinates. Here,

P = p + ρgr συν θ,
giving
∂P ∂p ∂P ∂p
= + ρg συν θ and = − ρgr ηµ θ.
∂r ∂r ∂θ ∂θ
Then, the components of the equation of motion in terms of the modified pressure are

 
∂ur ∂P
r − component: 0 = ρ ur =− (8.b.2)
∂r ∂r
dP
θ − component: 0 = (8.b.3)
dθ
∂P
φ − component: 0 = (8.b.4)
∂φ

Note that equations (8.b.2) – (8.b.4) could have been directly obtained from the following form of the
equation of motion:

D~u ~ −∇ ~ · τab
ρ = −∇P (8.b.5)
Dt
in which the modified pressure P includes both the pressure and gravitational terms.
From equations (8.b.3) and (8.b.4), P ’s partial derivatives with respect to θ and φ vanish, therefore P is a
function of only r. Substituting P = P (r) and ur = ur (r) in equation (8.b.2) then gives
dP dur
= −ρur . (8.b.6)
dr dr

8.c Pressure profile.

To calculate the pressure profile, we start by substituting ur = C/r2 into equation (8.b.6), which then gives

dP C2
= 2ρ 5
dr r
Integration then produces

ρC 2
P (r) = − + C1 (8.c.1)
2r4
To determine the integration constants, C and C1 , we make use of the boundary conditions. Since the
pressure is P (r = R) = PR and the velocity is ur (r = R) = vR , the integration constants may be evaluated as

C2
C1 = PR + ρ and C = R2 uR
2R4

27
 Then, substitution in equation (8.c.1) yields the pressure profile as
"  4 # "  4 #
ρC 2 R 1 2 R
P (r) = PR + 1− = PR + ρuR 1 − (8.c.2)
2R4 r 2 r

8.d Determination of nonzero components of the viscous stress tensor.

For an incompressible, Newtonian fluid, the viscous stress tensor is given by τab = −µ(∂b ua + ∂a ub ). Since the
only velocity component that exists is ur (r) = C/r2 , the nonzero components of the viscous stress tensor are
dur 4µC ur 2µC
τrr = −2µ = 3 and τθθ = τφφ = −2µ =− 3 (8.d.1)
dr r r r
Note that the shear stresses are all zero. The normal stresses in equation (8.b.6) are partly compressive and
partly tensile. Not only is the φ-component of ∇ ~ · τ directly zero, but the r-component and the θ-component
are zero too as evaluated below.

~ · τ ]r 1 d 2 τθθ + τφφ 4µC 4µC

[∇ = (r τrr ) − =− 4 + 4 =0 (8.d.2)
r2 dr r r r
~ · τ ]θ 1 ∂ τφφ σφ θ 2µC σφ θ 2µC σφ θ
[∇ = (τθθ ηµ θ) − =− + =0 (8.d.3)
r ηµ θ ∂θ r r r

Since [∇~ · τ ] = 0, viscous forces can be neglected in this flow and the Euler equation for inviscid fluids may
be directly used.

28
Problem 9
A parallel – disk viscometer consists of two circular disks of radius R separated by a small gap B (with R  B).
A fluid of constant density ρ, whose viscosity µ is to be measured, is placed in the gap between the disks. The
lower disk at z = 0 is fixed. The torque Tz necessary to rotate the upper disk (at z = B) with a constant
angular velocity Ω is measured. The task here is to deduce a working equation for the viscosity when the
angular velocity Ω is small (creeping flow).

Figure 9.1: Parallel – disk viscometer.

a) Simplify the equations of continuity and motion to describe the flow in the parallel – disk viscometer.
b) Obtain the tangential velocity profile after writing down appropriate boundary conditions.
c) Derive the formula for determining the viscosity µ of a Newtonian fluid from measurements of the torque
Tz and angular velocity Ω in a parallel – disk viscometer. Neglect the pressure term.

Solution

9.a Simplification of the continuity equation.

In steady laminar flow, the fluid is expected to travel in a circular motion (for small values of Ω). Only the
tangential component of velocity exists. The radial and axial components of velocity are zero; so, ur = 0 and
uz = 0.
For an incompressible fluid, the continuity equation gives ∇~ · ~u = 0.
In cylindrical coordinates,
1 ∂ 1 ∂uθ ∂uz ∂uθ
(rur ) + + =0 ⇒ =0
r ∂r r ∂θ ∂z ∂θ
So, uθ = uθ (r, z).
Next, we simplify the equation of motion. This is none other than the Navier–Stokes equation. For a
Newtonian fluid, the components of the Navier – Stokes equation may be simplified as given below.

u2θ ∂p
r − component: 0 = −ρ =− (9.a.1)
r ∂r
 
∂ 1 ∂
θ − component: 0 = (ruθ ) (9.a.2)
∂r r ∂r
∂p
z − component: 0 = − − ρg (9.a.3)
∂z
The symmetry inherent in the problem suggests that there is no pressure gradient in the θ-direction; so,
p = p(r, z).

9.b Velocity profile.

The no-slip boundary conditions at the two disk surfaces are

uθ (r, z = 0) = 0 (9.b.1)
uθ (r, z = B) = Ωr (9.b.2)

29
 It is possible to make an educated guess for the form of the tangential velocity taking a clue from equation
(9.b.2). Thus, it is reasonable to postulate uθ (r, z) = rf (z). Substituting this form in equation (9.a.2) then
gives

d2 f (z) df (z)
2
=0 ⇒ = C1 ⇒ f (z) = C1 z + C2 (9.b.3)
dz dz
Equation (9.b.1) gives f (0) = 0, therefore C2 = 0. On the other hand, equation (9.b.2) gives f (B) = Ω, and
therefore C1 = Ω/B. On substituting f (z) = Ωz/B, the tangential velocity profile is ultimately obtained as
Ω
rz
uθ (r, z) = (9.b.4)
B
The above result is expected and could have been guessed by considering a fluid contained between two
parallel plates (with the lower plate at z = 0 held stationary and the upper plate at z = B moving at a constant
velocity V ). Then, the linear velocity distribution is given by u = V z/B, which is the velocity profile at any r
in the parallel–disk viscometer. On noting that V = Ωr at the upper plate, equation (9.b.4) is obtained.

9.c Determination of torque.

From the velocity profile in equation (9.b.4), the momentum flux (shear stress) is determined as
∂uθ µΩ ∂  uθ 
τzθ = −µ =− r and τrθ = −µr =0 (9.c.1)
∂z B ∂r r
The differential torque dT~ is rigorously obtained as the cross product of the lever arm r and the differential
force dF~ exerted on the fluid by a solid surface element. Thus, dT~ = ~r × dF~ , where dF~ = n̂ · (pδδ + τ )dS, ~r is the
vector drawn from the axis of rotation to the element of surface area dS, n̂ is the unit normal to the surface, p is
the pressure, δ is the identity (unit) tensor, and τ is the viscous stress tensor. For the parallel–disk viscometer,

dT~ = ~r × (n̂ · τ ) dS = rδδ r × (−δδ z ) · τzθδ z δ θ (2πrdr) (9.c.2)

The pressure term is neglected in equation (9.c.2). Since δ z · δ z = 1 and δ r × δ θ = δ z , we obtain

dT~ = r(−τzθ )(2πrdr)δδ z

Thus, the z-component of the differential torque is dTz = r(−τzθ )(2πrdr), which is simply the product of
the lever arm, the momentum flux and the differential surface area. Integrating over the area of the upper disk
after substituting for τzθ from equation (9.c.1) yields
R R
πµΩR4
Z Z
2 2πµΩ
Tz (z = B) = − τzθ (z = B)2πr dr = r3 dr = (9.c.3)
0 B 0 2B
Equation (9.c.3) provides the formula for determining the viscosity µ of a Newtonian fluid from measurements
of the torque Tz and angular velocity Ω in a parallel–disk viscometer as µ = 2BTz /(πΩR4 ). Note that the shear
stress is not uniform [as given by equation (9.c.1)] in a parallel–disk viscometer, whereas it is uniform throughout
the gap in a cone–and–plate viscometer.

30
Problem 10
Consider a fluid (of density ρ) in incompressible, laminar flow in a plane narrow slit of length L and width
W formed by two flat parallel walls that are a distance 2B apart. End effects may be neglected because
B  W  L. The fluid flows under the influence of a pressure difference ∆p, gravity or both.

Figure 10.1: Fluid flow in plane narrow slit.

a) Determine the steady-state velocity distribution for a non-Newtonian fluid that is described by the Bing-
ham model.
b) Obtain the mass flow rate for a Bingham fluid in slit flow.

Solution

10.a Shear stress distribution.

For axial flow in rectangular Cartesian coordinates, the differential equation for the momentum flux is
dτxz ∆P
= (10.a.1)
dx L
where P is a modified pressure, which is the sum of both the pressure and gravity terms, i.e., ∆P = ∆p +
ρgL συν β. Here, β is the angle of inclination of the z-axis with the vertical.
On integration, this gives the expression for the shear stress xz for laminar flow in a plane narrow slit as
∆P
τxz = x (10.a.2)
L
It must be noted that the above momentum flux expression holds for both Newtonian and non-Newtonian
fluids (and does not depend on the type of fluid). Further, τxz (x = 0) = 0 in slit flow based on symmetry
arguments, i.e., the velocity profile is symmetric about the midplane x = 0.

The Bingham model

For viscoplastic materials (e.g., thick suspensions and pastes), there is no flow until a critical stress (called
the yield stress τ0 ) is reached. To describe such a material that exhibits a yield stress, the simplest model is
the Bingham model given below.

duz
η → ∞ or = 0, if |τxz | ≤ τ0 (10.a.3)
dx

τ0 duz
η = µ0 + or τxz = −µ0 ± τ0 , if |τxz | ≥ τ0 (10.a.4)
±duz /dx dx

31
 Here, η is the non-Newtonian viscosity and µ0 is a Bingham model parameter with units of viscosity. In
equation (10.a.4), the positive sign is used with τ0 and the negative sign with duz /dx when τxz is positive. On
the other hand, the negative sign is used with τ0 and the positive sign with duz /dx when τxz is negative.

Inner plug-flow region

Let τxz = +τ0 at x = +x0 . Then, from equation (10.a.2), x0 is given by τ0 = (∆P/L)x0 . Let the inner region
(−x0 ≤ x ≤ x0 ) where the shear stress is less than the yield stress (−τ0 ≤ τxz ≤ τ0 ) be denoted by subscript
i. Since τxz is finite and τxz = −η(duz /dx), the Bingham model [as per equation (10.a.3)] gives (duzi /dx) = 0.
Integration gives uzi = C1 (constant velocity), which implies the fluid is in plug flow in the inner region.

Outer region
Let the outer region be denoted by subscript o. For the region x0 ≤ x ≤ B, the velocity decreases with
increasing x and therefore duz /dx ≤ 0. On the other hand, for the region −B ≤ x ≤ −x0 , the velocity decreases
with decreasing x and therefore duz /dx ≥ 0. Thus, the Bingham model according to equation (10.a.4) is given
by

duzo τ0
τxz = −µ0 + τ0 , or η = µ0 + , for x0 ≤ x ≤ B (10.a.5)
dx −duzo /dx
duzo τ0
τxz = −µ0 − τ0 , or η = µ0 + , for − B ≤ x ≤ −x0 (10.a.6)
dx +duzo /dx
To obtain the velocity profile for x0 ≤ x ≤ B, equations (10.a.2) and (10.a.5) may be combined to eliminate
τxz and get
duzo ∆P τ0
=− x+ , for x0 ≤ x ≤ B (10.a.7)
dx µ0 L µ0
Integration gives
∆P 2 τ0
uzo = − x + x + C2
2µ0 L µ0
Imposing the boundary condition that uzo (x = B) = 0 then yields

∆P B 2 τ0 B
C2 = −
2µ0 L µ0
Thus,

∆P B 2
  x 2  τ B 
0 x
uzo (x) = 1− − 1− . (10.a.8)
2µ0 L B µ0 B

Velocity distribution
For the inner plug-flow region, it was earlier found that uzi = C1 . To determine C1 , it may be noted that
uzi (x = x0 ) = uzo . Then, equation (10.a.8) along with τ0 = (∆P/L)x0 may be simplified to give

∆P B 2  x0 2
C1 = 1−
2µ0 L B
Thus, the final results for the velocity profile are

∆P B 2  x0 2
uzi (x) = 1− , for |x| ≤ x0 (10.a.9)
2µ0 L B
∆P B 2
  x 2  τ B  
0 |x|
uzo (x) = 1− − 1− , for x0 ≤ |x| ≤ B (10.a.10)
2µ0 L B µ0 B
The velocity profile is flat in the inner region as given by equation (10.a.9) and is parabolic in the outer
region as given by equation (10.a.10).

32
10.b General expression for mass flow rate.
Since the velocity profile is symmetric about the midplane x = 0, the mass flow rate may be obtained by
integrating the velocity profile over half the cross section of the slit, as shown below.
Z B Z B
w= ρuz W dx = 2 ρuz W dx (10.b.1)
−B 0

Rather than insert two separate expressions from equations (10.a.9) and (10.a.10) for uz and integrate in
two regions, it is easier to integrate by parts.
" #
B Z B du Z B 
duz

z
w = 2W ρ uz x − x dx = 2W ρ x − dx (10.b.2)

0 0 dx 0 dx
The first term in the square brackets above is zero at both limits on using the no-slip boundary condition
(uz (x = B) = 0) at the upper limit. From equation (10.a.2), x/B = τxz /τB where τB = ∆P B/L is the wall
shear stress. Thus, a general expression for the mass rate of flow in a plane narrow slit is

2W B 2 ρ τB
Z  
duz
w= τxz − dτxz . (10.b.3)
τB2 0 dx

Mass flow rate for Bingham fluid For a material with a yield stress, the lower limit of integration is
reset to τ0 as per equation (10.a.3). Then, substituting equation (10.a.5) and integrating yields

2W B 2 ρ τB 2W B 2 ρ 1 3
Z  
3 1 2 3
w= τxz (τxz − τ0 )dτxz = (τ − τ0 ) − (τ0 τB − τ0 ) (10.b.4)
µ0 τB2 τ0 µ0 τB2 3 B 2
The final expression for the mass flow rate is
"  3 #
2∆P W B 3 ρ 3τ0 1 τ0
w= 1− + (10.b.5)
3µ0 L 2τB 2 τB

Note that τB = ∆P B/L is the wall shear stress and τ0 is the yield stress. Since no flow occurs below the
yield stress (that is, when τB ≤ τ0 ), the above expression is valid only for τB > τ0 . For τ0 = 0 and µ0 = µ, the
Bingham model simplifies to the Newtonian model and equation (10.b.5) reduces to the Newtonian result, i.e.,
w = 2∆P B 3 W /(3µL).

33
Problem 11
Consider a fluid (of density ρ) in incompressible, laminar flow in a long circular tube of radius R and length L.
End effects may be neglected because the tube length L is relatively large compared to the tube radius R. The
fluid flows under the influence of a pressure difference ∆p, gravity or both.

Figure 11.1: Fluid flow in circular tube.

a) Determine the steady-state velocity distribution for a non-Newtonian fluid that is described by the Bing-
ham model.
b) Obtain the mass rate of flow for a Bingham fluid in a circular tube.

Solution

11.a Shear stress distribution.

For axial flow in cylindrical coordinates, the differential equation for the momentum flux is
d ∆P
(rτrz ) = r (11.a.1)
dr L
where P is a modified pressure, which is the sum of both the pressure and gravity terms, i.e., ∆P = ∆p +
ρgL συν β. Here, β is the angle of inclination of the tube axis with the vertical.
On integration (with the condition that τrz must be finite at r = 0), this gives the expression for the shear
stress τrz for steady laminar flow in a circular tube as
∆P
τrz =
r (11.a.2)
2L
It must be noted that the above momentum flux expression holds for both Newtonian and non-Newtonian
fluids (and does not depend on the type of fluid).

The Bingham model

For viscoplastic materials (e.g., thick suspensions and pastes), there is no flow until a critical stress (called
the yield stress τ0 ) is reached. To describe such a material that exhibits a yield stress, the simplest model is
the Bingham model given below.

duz
η → ∞ or = 0, if |τrz | ≤ τ0 (11.a.3)
dr

τ0 duz
η = µ0 + or τrz = −µ0 ± τ0 , if |τrz | ≥ τ0 (11.a.4)
±duz /dr dr

Here, η is the non-Newtonian viscosity and µ0 is a Bingham model parameter with units of viscosity. In
equation (11.a.4), the positive sign is used with τ0 and the negative sign with duz /dx when τxz is positive. On
the other hand, the negative sign is used with τ0 and the positive sign with duz /dx when τxz is negative.

34
 Inner plug-flow region
Let τrz = +τ0 at r = r0 . Then, from equation (11.a.2), r0 is given by τ0 = (∆P/2L)r0 . Let the inner region
(r ≤ r0 ) where the shear stress is less than the yield stress (τrz ≤ τ0 ) be denoted by subscript i. Since τrz is
finite and τrz = −η(duz /dr), the Bingham model [as per equation (11.a.3)] gives (duzi /dr) = 0. Integration
gives uzi = C1 (constant velocity), which implies the fluid is in plug flow in the inner region.

Outer region
Let the outer region be denoted by subscript o. For the region r0 ≤ r ≤ R, the velocity decreases with
increasing r and therefore duz /dr ≤ 0 and τrz ≥ τ0 . Thus, the Bingham model according to equation (11.a.4)
is given by
duzo τ0
τrz = −µ0 + τ0 , or η = µ0 + , for r0 ≤ r ≤ R (11.a.5)
dr −duzo /dr
To obtain the velocity profile for r0 ≤ r ≤ R, equations (11.a.2) and (11.a.5) may be combined to eliminate
τrz and get
duzo ∆P τ0
=− r+ , for r0 ≤ r ≤ R (11.a.6)
dr 2µ0 L µ0
Integration gives
∆P 2 τ0
uzo = − r + r + C2
rµ0 L µ0
Imposing the boundary condition that uzo (r = R) = 0 then yields

∆P R2 τ0 R
C2 = −
4µ0 L µ0
Thus the velocity distribution in the outer region is

∆P R2
  r 2  τ R 
0 r
uzo (r) = 1− − 1− (11.a.7)
4µ0 L R µ0 R
For the inner plug-flow region, it was earlier found that uzi = C1 . To determine C1 , it may be noted that
uzi (r = r0 ) = uzo . Then, equation (11.a.7) along with τ0 = (∆P/2L)r0 may be simplified to give

∆P R2  r0 2
C1 = 1−
4µ0 L R
Thus, the velocity of the plug flow in the inner region is

∆P R2  r0 2
uzi (r) = 1− , for |r| ≤ r0 (11.a.8)
4µ0 L R

Here, r0 = 2Lτ0 /∆P is the radius of the plug-flow region in the central part of the tube. The velocity profile
is parabolic in the outer region as given by equation (11.a.7) and is flat in the inner region as given by equation
(11.a.8).

11.b General expression for mass flow rate.

The mass flow rate may be obtained by integrating the velocity profile over the cross section of the circular
tube as shown below.
Z R
w = 2πρ uz rdr (11.b.1)
0

Rather than insert two separate expressions from equations (11.a.7) and (11.a.8) for uz and integrate in two
regions, it is easier to integrate by parts.

35
 " #
R Z R 
r2 R 1
Z 
2 du z 2 duz
w = 2πρ uz  − r dr = πρ r − dr (11.b.2)
2 0 2 0 dr 0 dr
The first term in the square brackets above is zero at both limits on using the no-slip boundary condition
(uz (r = R) = 0) at the upper limit. From equation (11.a.2), r/R = τrz /τR where τR = ∆P R/2L is the wall
shear stress. Thus, a general expression for the mass rate of flow in a circular tube is

πR3 ρ τR 2
Z  
duz
w= τrz − dτrz . (11.b.3)
τR3 0 dr

Buckingham - Reiner equation for mass flow rate of Bingham fluid For a material with a yield
stress, the lower limit of integration is reset to τ0 as per equation (11.a.3). Then, substituting equation (11.a.4)
and integrating yields

πR3 ρ τR 2 πR3 ρ 1 4
Z  
4 1 3 4
w= τ (τrz − τ 0 )dτrz = (τ − τ ) − (τ0 τ − τ ) (11.b.4)
µ0 τR3 τ0 rz µ0 τr3 4 R 0
3 R 0

The final expression for the mass flow rate is

"  4 #
π∆P R4 ρ 4τ0 1 τ0
w= 1− + (11.b.5)
8µ0 L 3τB 3 τR

The above expression is known as the Buckingham - Reiner equation. Note that τR = ∆P R/(2L) is the
wall shear stress and τ0 is the yield stress. Since no flow occurs below the yield stress (that is, when τR ≤ τ0 ),
the above expression is valid only for τR > τ0 . For τ0 = 0 and µ0 = µ, the Bingham model simplifies to the
Newtonian model and equation (11.b.5) reduces to the Hagen - Poiseuille equation, i.e., w = π∆P R4 ρ/(8µL).

36
Problem 12
Consider a fluid (of density ρ) in incompressible, laminar flow in a plane narrow slit of length L and width
W formed by two flat parallel walls that are a distance 2B apart. End effects may be neglected because
B  W  L. The fluid flows under the influence of a pressure difference ∆p, gravity or both.

Figure 12.1: Fluid flow in plane narrow slit.

a) Determine the steady-state velocity distribution for a non-Newtonian fluid that obeys the power law model
(e.g., a polymer liquid).
b) Obtain the mass flow rate for a power law fluid in slit flow.

Solution

12.a Steady-state velocity distribution.

For axial flow in rectangular Cartesian coordinates, the differential equation for the momentum flux is
dτxz ∆P
= (12.a.1)
dx L
where P is a modified pressure, which is the sum of both the pressure and gravity terms, i.e., ∆P = ∆p +
ρgL συν β. Here, β is the angle of inclination of the z-axis with the vertical.
On integration, this gives the expression for the shear stress τxz for laminar flow in a plane narrow slit as
∆P
τxz = x (12.a.2)
L
It must be noted that the above momentum flux expression holds for both Newtonian and non-Newtonian
fluids (and does not depend on the type of fluid). Further, τxz = 0 at x = 0 in slit flow based on symmetry
arguments, i.e., the velocity profile is symmetric about the midplane x = 0.
For the region 0 ≤ x ≤ B, the velocity decreases with increasing x and therefore du dx ≤ 0. On the other
z

hand, for the region B ≤ x ≤ 0, the velocity decreases with decreasing x and therefore du dx ≥ 0. Thus, the
z

power law model is given as

  n
m − du

dx
z for 0 ≤ x ≤ B,
τxz = n (12.a.3)
−m duz

for − B ≤ x ≤ 0

dx

Velocity distribution
To obtain the velocity distribution for 0 ≤ x ≤ B, equations (12.a.2) and the first of (12.a.3) are combined
to eliminate τxz and get

37
  1/n
duz ∆P x
− = (for 0 ≤ x ≤ B) (12.a.4)
dx mL
On integrating and using the no-slip boundary condition (uz = 0 at x = B), we get
 1/n   x 1+1/n 
∆P B B
uz = 1− (for 0 ≤ x ≤ B) (12.a.5)
mL 1 + 1/n B
Similarly, for B ≤ x ≤ 0, equations (12.a.2) and the second of (12.a.3) may be combined to eliminate τxz
and obtain
 1/n
duz ∆P x
= − (for − B ≤ x ≤ 0) (12.a.6)
dx mL
Again, integrating and using the no-slip boundary condition (uz = 0 at x = B) gives
 1/n   x 1+1/n 
∆P B B
uz = 1− − (for − B ≤ x ≤ 0) (12.a.7)
mL 1 + 1/n B
Equations (12.a.5) and (12.a.7) may be combined and represented as a single equation for the velocity profile
as follows:
 1/n "  1+1/n #
∆P B B |x|
uz = 1− (for |x| ≤ 0) (12.a.8)
mL 1 + 1/n B
Note that the maximum velocity in the slit occurs at the midplane x = 0.

12.b General expression for mass flow rate.

Since the velocity profile is symmetric about the midplane x = 0, the mass flow rate may be obtained by
integrating the velocity profile over half the cross section of the slit as shown below.
B 1/n Z 1
B2W ρ
Z   x 1+1/n   x 
∆P B
w=2 ρuz W dx = 2 1− d (12.b.1)
0 mL 1 + 1/n 0 B B
which results in
1/n
2B 2 W ρ

∆P B
w= (12.b.2)
2 + 1/n mL
When the mass flow rate w in equation (12.b.2) is divided by the density ρ and the cross-sectional area
(2BW ), an expression for the average velocity is obtained.
For n = 1 and m = µ, equation (12.b.2) reduces to the Newtonian result, i.e., w = 2∆P B 3 W ρ/(3µL).

38
Problem 13
Consider a Newtonian liquid (of viscosity µ and density ρ) in laminar flow down an inclined flat plate of length
L and width W . The liquid flows as a falling film with negligible rippling under the influence of gravity. End
effects may be neglected because L and W are large compared to the film thickness δ.

Figure 13.1: Newtonian liquid flow in a falling film.

a) Determine the steady-state velocity distribution.

b) Obtain the mass rate of flow and average velocity in the falling film.
c) What is the force exerted by the liquid on the plate in the flow direction?
d) Derive the velocity distribution and average velocity for the case where x is replaced by a coordinate x0
measured away from the plate (i.e., x0 = 0 at the plate and x0 = δ at the liquid-gas interface).

Solution

13.a Shear stress distribution.

For axial flow in rectangular Cartesian coordinates, the differential equation for the momentum flux is
dτxz ∆P
= (13.a.1)
dx L
where P is a modified pressure, which is the sum of both the pressure and gravity terms, that is, ∆P =
∆p + ρgL συν β. Here, β is the angle of inclination of the z-axis with the vertical. Since the flow is solely under
the influence of gravity, ∆p = 0 and therefore ∆P/L = ρg συν β. On integration,

τxz = ρgx συν β + C1 (13.a.2)

On using the boundary condition at the liquid-gas interface (τxz = 0 at x = 0), the constant of integration
C1 is found to be zero. This gives the expression for the shear stress τxz for laminar flow in a falling film as

τxz (x) = ρgx συν β (13.a.3)

It must be noted that the above momentum flux expression holds for both Newtonian and non-Newtonian
fluids (and does not depend on the type of fluid).
The shear stress for a Newtonian fluid (as per Newton’s law of viscosity) is given by
duz
τxz = −µ (13.a.4)
dx

Velocity distribution
To obtain the velocity distribution, there are two possible approaches.
In the first approach, equations (13.a.3) and (13.a.4) are combined to eliminate τxz and get a first-order
differential equation for the velocity as given below.
duz ρg συν β
=− x (13.a.5)
dx µ

39
 On integrating, uz = ρgx2 συν β/2µ + C2 . On using the no-slip boundary condition at the solid surface
(uz = 0 at x = δ), C2 = ρgδ 2 συν β. Thus, the final expression for the velocity distribution is parabolic as given
below.

ρgδ 2 συν β
  x 2 
uz = 1− (13.a.6)
2µ δ
In the second approach, equations (13.a.1) and (13.a.4) are directly combined to eliminate τxz and get a
second-order differential equation for the velocity as given below.

d2 uz
−µ = ρg συν β (13.a.7)
dx2
Integration gives
duz ρg συν β
=− x + C1 (13.a.8)
dx µ
On integrating again,

ρgx2 συν β
uz = − + C1 x + C2 (13.a.9)
2µ
On using the boundary condition at the liquid-gas interface (at x = 0, τxz = 0 ⇒ duz /dx = 0), the constant
of integration C1 is found to be zero from equation (13.a.8). On using the other boundary condition at the solid
surface (at x = δ, uz = 0), equation (13.a.9) gives C2 = ρgδ 2 συν /2µ. On substituting C1 and C2 in equation
(13.a.9), the parabolic velocity profile in equation (13.a.6) is again obtained.

13.b Mass flow rate.

The mass flow rate may be obtained by integrating the velocity profile given by equation (13.a.6) over the film
thickness as shown below.

δ 1
ρ2 gW δ 3 συν β
Z Z   x 2   x 
w = ρuz W dx = 1− d ⇒ (13.b.1)
0 2µ 0 δ δ
ρ2 gW δ 3 συν β
w = (13.b.2)
3µ

When the mass flow rate w in equation (13.b.2) is divided by the density ρ and the film cross-sectional area
(W · δ), an expression for the average velocity < uz > is obtained. Also, the maximum velocity uz,max occurs
at the interface x = 0 and is readily obtained from equation (13.a.6). Thus, equations (13.a.6) and (13.b.2) give

ρgδ 2 συν β 2
< uz >= = uz,max (13.b.3)
3µ 3

13.c Force on the plate.

The z-component of the force of the fluid on the solid surface is given by the shear stress integrated over the
wetted surface area. Using equation (13.a.3) gives

Fz = LW τxz (x = δ) = ρgδLW συν β (13.c.1)

As expected, the force exerted by the fluid on the plate is simply the z-component of the weight of the liquid
film.

40
13.d Change of coordinates.
Consider the coordinate x0 starting from the plate (i.e., x0 = 0 is the plate surface and x0 = δ is the liquid-gas
interface). The form of the differential equation for the momentum flux given by equation (13.a.1) will remain
unchanged. Thus, equation (13.a.2) for the new choice of coordinates will be

τx0 z = ρgx0 συν β + C1 (13.d.1)

0
On using the boundary condition at the liquid-gas interface (τx0 z = 0 at x = δ), the constant of integration
is now found to be given by C1 = gδ συν β. This gives the shear stress as

τx0 z = ρg(x0 − δ) συν β (13.d.2)

Note that the momentum flux is in the negative x0 -direction. On substituting Newton’s law of viscosity,
duz ρg συν β
= (δ − x0 ) (13.d.3)
dx0 µ
On integrating,
 
ρg 0 1 02
uz = δx x συν β + C2
µ 2
On using the no-slip boundary condition at the solid surface (uz = 0 at x0 = 0), it is found that C2 = 0.
Thus, the final expression for the velocity distribution is
"  2 #
ρgδ 2 συν β x0 1 x0
uz = − (13.d.4)
µ δ 2 δ

The average velocity may be obtained by integrating the velocity profile given in equation (13.d.4) over the
film thickness as shown below.
Z 1 " 0  0 2 #  0 
1 δ 2
Z
ρgδ συν β x x x
w= uz dx0 = − d (13.d.5)
δ 0 µ 0 δ δ δ

Integration of equation (13.d.5) yields the same expression for < uz > obtained earlier in equation (13.b.3).
It must be noted that the two coordinates x and x0 are related as follows: x + x0 = δ. By substituting
 x 2   0 2  0   0 2
x x x
= 1− =1−2 +
δ δ δ δ
equation (13.d.4) is directly obtained from equation (13.a.6).

41
Problem 14
Consider a liquid (of density ρ) in laminar flow down an inclined flat plate of length L and width W . The
fluid flows as a falling film with negligible rippling under the influence of gravity. End effects may be neglected
because L and W are large compared to the film thickness δ.

Figure 14.1: Non-Newtonian liquid flow in a falling film.

a) Determine the steady-state velocity distribution for a non-Newtonian fluid that obeys the power law model
(e.g., a polymer liquid). Reduce the result to the Newtonian case.
b) Obtain the mass flow rate for a power law fluid. Simplify for a Newtonian fluid.
c) What is the force exerted by the fluid on the plate in the flow direction?

Solution

14.a Shear stress distribution.

For axial flow in rectangular Cartesian coordinates, the differential equation for the momentum flux is
dτxz ∆P
= (14.a.1)
dx L
where P is a modified pressure, which is the sum of both the pressure and gravity terms, that is, ∆P =
∆p + ρgL συν β. Here, β is the angle of inclination of the z-axis with the vertical. Since the flow is solely under
the influence of gravity, ∆p = 0 and therefore ∆P/L = ρg συν β. On integration,

τxz = ρgx συν β + C1 (14.a.2)

On using the boundary condition at the gas-liquid interface (τxz = 0 at x = 0), the constant of integration
C1 is found to be zero. This gives the expression for the shear stress tensor τxz for laminar flow in a falling film
as

τxz = ρgx συν β (14.a.3)

It must be noted that the above momentum flux expression holds for both Newtonian and non-Newtonian
fluids (and does not depend on the type of fluid).
In the falling film, the velocity decreases with increasing x and therefore duz /dx0. Thus, the power law
model is given as
 n
duz
τxz = m − (14.a.4)
dx

Velocity distribution
To obtain the velocity distribution, equations (14.a.3) and (14.a.4) are combined to eliminate τxz and get
 1/n
duz ρg συν β
− = x1/n (14.a.5)
dx m

42
 On integrating,
 n1 1
x n +1

ρg συν β
uz = 1 + C2
m n +1

On using the no-slip boundary condition at the solid surface (uz = 0 at x = δ),
1/n 1
δ n +1

ρg συν β
C2 = 1
m n +1
Thus, the final expression for the velocity distribution is
 1/n   x  n1 +1 
ρg συν β δ
uz = 1 1− (14.a.6)
m n +1 δ
Equation (14.a.6) when simplified for a Newtonian fluid (n = 1 and m = µ) gives the following parabolic
velocity profile:

ρgδ 2 συν β
  x 2 
uz = 1− (14.a.7)
2µ δ

14.b Mass flow rate.

The mass flow rate may be obtained by integrating the velocity profile over the film thickness as shown below.
δ 1/n 1
δ2 W ρ  x  n1 +1   x 
Z  Z 
ρgδ συν β
w= ρuz W dx = 1 1− d (14.b.1)
0 m n +1 0 δ δ
or,
1/n
δ2 W ρ

ρgδ συν β
w= 1 (14.b.2)
n +2
m
Equation (14.b.2) for a Newtonian fluid (n = 1 and m = µ) gives

ρ2 gW δ 3 συν β
w= (14.b.3)
3µ
When the mass flow rate w in either equation (14.b.2) or equation (14.b.3) is divided by the density ρ and
the film cross-sectional area W ·δ, an expression for the average velocity < uz > is obtained. Also, the maximum
velocity uz,max occurs at the interface x = 0 and is readily obtained from equation (14.a.6) or equation (14.a.7)
respectively. Thus, equations (14.a.6) and (14.b.2) give
 1/n 1
δ ρgδ συν β n +1
< uz >= 1 = 1 uz,max (14.b.4)
n +2 m n +2
For n = 1 and m = µ, equation (14.b.4) reduces to the Newtonian result given below:

ρgδ 2 συν β 2
< uz >= = uz,max (14.b.5)
3µ 3

14.c Force on the plate.

The z-component of the force of the fluid on the solid surface is given by the shear stress integrated over the
wetted surface area. Using equation (14.a.3) gives

Fz = LW τxz (x = δ) = ρgδLW συν β (14.c.1)

As expected, the force exerted by the fluid on the plate is simply the z-component of the weight of the liquid
film.

43
Problem 15
Consider a fluid (of constant density ρ) in incompressible, laminar flow in a tube of circular cross section,
inclined at an angle β to the vertical. End effects may be neglected because the tube length L is relatively very
large compared to the tube radius R. The fluid flows under the influence of both a pressure difference ∆p and
gravity.

Figure 15.1: Fluid flow in circular tube.

a) Using a differential shell momentum balance, determine expressions for the steady-state shear stress
distribution and the velocity profile for a Newtonian fluid (of constant viscosity µ).
b) Obtain expressions for the maximum velocity, average velocity and the mass flow rate for pipe flow.
c) Find the force exerted by the fluid along the pipe wall.

Solution
Flow in pipes occurs in a large variety of situations in the real world and is studied in various engineering
disciplines as well as in physics, chemistry, and biology.

15.a Shear stress distribution.

For a circular tube, the natural choice is cylindrical coordinates. Since the fluid flow is in the z-direction, ur = 0,
θ = 0, and only uz exists. Further, uz is independent of z and it is meaningful to postulate that the velocity
and pressure are both functions of only z, that is uz = uz (r), p = p(z). The only nonvanishing components of
the stress tensor are τrz = τzr , which depend only on r.
Consider now a thin cylindrical shell perpendicular to the radial direction and of length L. A ‘rate of
z-momentum’ balance over this thin shell of thickness ∆r in the fluid is of the form:

Rate of z-momentum = In Out + Generation = Accumulation

At steady-state, the accumulation term is zero. Momentum can go ‘in’ and ‘out’ of the shell by both the
convective and molecular mechanisms.

Since uz (r)
is the same at both ends of the tube, the convective terms
cancel out because ρu2z 2πr∆r |z=0 = ρu2z 2πr∆r |z=L . Only the molecular term (2πrLτrz ) remains to be
considered, whose ‘in’ and ‘out’ directions are taken in the positive direction of the r-axis. Generation of z-
momentum occurs by the pressure force acting on the surface (p2πr∆r) and gravity force acting on the volume
[(ρg συν β) 2πr∆rL]. On substituting these contributions into the z-momentum balance, we get

(2πrLτrz ) |r − (2πrLτrz ) |r+∆r + (p0 − pL )2πr∆r + (ρg συν β)2πr∆rL = 0 (15.a.1)

Dividing the above equation by 2πL∆r yields

(rτrz )|r+∆r − (rτrz )|r p0 − pL + ρg συν β

= r (15.a.2)
∆r L
On taking the limit as ∆r → 0, the left-hand side of the above equation is the definition of the first derivative.
The right-hand side may be written in a compact and convenient way by introducing the modified pressure P ,
which is the sum of the pressure and gravitational terms. The general definition of the modified pressure is

44
P = p + ρgh, where h is the height (in the direction opposed to gravity) above some arbitrary preselected datum
plane. The advantages of using the modified pressure P are that (i) the components of the gravity vector g need
not be calculated in cylindrical coordinates; (ii) the solution holds for any orientation of the tube axis; and (iii)
the effects of both pressure and gravity are in general considered. Here, h is negative since the z-axis points
downward, giving h = z συν β and therefore P = p − ρgz συν β. Thus, P0 = p0 at z = 0 and PL = pL ρgL συν β
at z = L, giving p0 − pL + ρgL συν β = P0 − PL ∆P . Thus, equation (15.a.2) gives
d ∆P
(rτrz ) = r (15.a.3)
dr L
Equation (15.a.3) on integration leads to the following expression for the shear stress distribution:
∆P C1
τrz = r+ (15.a.4)
2L r
The constant of integration C1 is determined later using boundary conditions.
It is worth noting that equations (15.a.3) and (15.a.4) apply to both Newtonian and non-Newtonian fluids,
and provide starting points for many fluid flow problems in cylindrical coordinates.

Shear stress distribution and velocity profile

Substituting Newton’s law of viscosity for τrz in equation (15.a.4) gives
duz ∆P C1
−µ = r+ (15.a.5)
dr 2L r
The above differential equation is simply integrated to obtain the following velocity profile:
∆P 2 C1
uz = − r − ln r + C2 (15.a.6)
4µL µ
The integration constants C1 and C2 are evaluated from the following boundary conditions:
Condition 1: At r = 0, τrz and uz must be finite.
Condition 2: At r = R, uz = 0.
From Condition 1 (which states that the momentum flux and velocity at the tube axis cannot be infinite),
R2
C1 = 0. From Condition 2 (which is the no-slip condition at the fixed tube wall), C2 = ∆P
4µL . On substituting
C1 = 0 in equation (15.a.4), the final expression for the shear stress (or momentum flux) distribution is found
to be linear as given by
∆P
τrz = r (15.a.7)
2L
Further, substitution of the integration constants into equation (15.a.6) gives the final expression for the
velocity profile as
  r 2 
∆P 2
uz = R 1− (15.a.8)
4µL R
It is observed that the velocity distribution for laminar, incompressible flow of a Newtonian fluid in a long
circular tube is parabolic.

15.b Maximum velocity, average velocity, and mass flow rate

From the velocity profile, various useful quantities may be derived.

• The maximum velocity occurs at r = 0 (where duz /dr = 0). Therefore,

∆P 2
max (uz ) = uz (r = 0) = R (15.b.1)
4µL

45
 • The average velocity is obtained by dividing the volumetric flow rate by the cross-sectional area as shown
below.

RR R R
∆P R2
  r 2 
uz 2πrdr
Z Z
2 ∆P 1
< uz >= 0R R = 2 ruz dr = r 1− dr = = max (uz ) (15.b.2)
2πrdr R 0 2µL 0 R 8µL 2
0

Thus, the ratio of the average velocity to the maximum velocity for Newtonian fluid flow in a circular
tube is 12 .
• The mass rate of flow is obtained by integrating the velocity profile over the cross section of the circular
tube as follows.
Z R
w= ρuz 2πrdr = πR2 ρ < uz > (15.b.3)
0

Thus, the mass flow rate is the product of the density ρ, the cross-sectional area (πR2 ) and the average
velocity < uz >. On substituting < uz > from equation (15.b.2), the final expression for the mass rate of
flow is
π∆P ρ 4
w= R (15.b.4)
8µL
The flow rate vs. pressure drop (w vs. ∆P ) expression above is well-known as the Hagen-Poiseuille
equation. It is a result worth noting because it provides the starting point for flow in many systems (e.g.,
flow in slightly tapered tubes).

15.c Force along the tube wall.

The z-component of the force, Fz , exerted by the fluid on the tube wall is given by the shear stress integrated
over the wetted surface area. Therefore, on using equation (15.a.7),

Fz = 2πRLτrz (r = R) = πR2 ∆P = πR2 ∆p + πR2 ρgL συν β (15.c.1)

where the pressure difference ∆p = p0 − pL . The above equation simply states that the viscous force is balanced
by the net pressure force and the gravity force.

46
Problem 16
A fluid (of constant density ρ) is in incompressible, laminar flow through a tube of length L. The radius of the
tube of circular cross section changes linearly from R0 at the tube entrance (z = 0) to a slightly smaller value
RL at the tube exit (z = L).

Figure 16.1: Fluid flow in a slightly tapered tube.

Using the lubrication approximation, determine the mass flow rate vs. pressure drop (w vs. ∆P ) relationship
for a Newtonian fluid (of constant viscosity µ).

Solution
Hagen-Poiseuille equation and lubrication approximation
The mass flow rate vs. pressure drop (w vs. ∆P ) relationship for a Newtonian fluid in a circular tube of
constant radius R is, as was derived in the previous problem
π∆P ρ 4
w= R (16.1)
8µL
The above equation, which is the famous Hagen-Poiseuille equation, may be re-arranged as
∆P 8µw 1
= (16.2)
L πρ R4
For the tapered tube, note that the mass flow rate w does not change with axial distance z. If the above
equation is assumed to be approximately valid for a differential length dz of the tube whose radius R is slowly
changing with axial distance z, then it may be re-written as
dP 8µw 1
− = (16.3)
dz πρ R4 (z)
The approximation used above where a flow between non-parallel surfaces is treated locally as a flow between
parallel surfaces is commonly called the lubrication approximation because it is often employed in the theory of
lubrication. The lubrication approximation, simply speaking, is a local application of a one-dimensional solution
and therefore may be referred to as a quasi-one-dimensional approach.
Equation (16.3) may be integrated to obtain the pressure drop across the tube on substituting the taper
function R(z), which is determined next.

Taper function
As the tube radius R varies linearly from R0 at the tube entrance (z = 0) to RL at the tube exit (z = L),
the taper function may be expressed as
z
R(z) = R0 + (RL − R0 )
L
On differentiating with respect to z, we get
dR RL − R0
= (16.4)
dz L

Pressure P as a function of radius R

47
 Equation (16.3) is readily integrated with respect to radius R rather than axial distance z. Using equation
(16.4) to eliminate dz from equation (16.3) yields
8µw L dR
−dP = (16.5)
πρ RL − R0 R4
Integrating the above equation between z = 0 and z = L, we get
Z PL Z RL
8µw L dR
−dP = (16.6)
P0 πρ RL − R0 R0 R4
and therefore,
 
P0 − PL 8µw 1 1 1
= − 3 (16.7)
L πρ 3 (RL − R0 ) R03 RL
Equation (16.7) may be re-arranged into the following standard form in terms of mass flow rate:

3λ3
   
π∆P ρ 4 3(λ − 1) π∆P ρ 4
w= R0 = R (16.8)
8µL 1 − λ−3 8µL 0 1 + λ + λ2
where the taper ratio λ = RR0 . The term in square brackets on the right-hand side of the above equation may
L

be viewed as a taper correction to equation (16.1).

48
Problem 17
A fluid (of constant density ρ) is in incompressible, laminar flow through a tube of length L. The radius of the
tube of circular cross section changes linearly from R0 at the tube entrance (z = 0) to a slightly smaller value
RL at the tube exit (z = L).

Figure 17.1: Fluid flow in a slightly tapered tube.

Using the lubrication approximation, determine the mass flow rate vs. pressure drop (w vs. ∆P ) relationship
for a fluid that can be described by the power law model (e.g., a polymeric liquid).

Solution
Hagen-Poiseuille equation and lubrication approximation
The mass flow rate vs. pressure drop (w vs. ∆P ) relationship for a power law fluid in a circular tube of
constant radius R is
1/n
πR3 ρ

∆P R
w= 1 (17.1)
n +3
2mL
The above equation, which is the power law fluid analog of the Hagen-Poiseuille equation (originally for
Newtonian fluids), may be re-arranged as
 n
∆P 3n + 1 w 1
= 2m 3n+1
(17.2)
L n ρπ R
For the tapered tube, note that the mass flow rate w does not change with axial distance z. If the above
equation is assumed to be approximately valid for a differential length dz of the tube whose radius R is slowly
changing with axial distance z, then it may be re-written as
 n
dP 3n + 1 w 1
− = 2m 3n+1
(17.3)
dz n ρπ R (z)
The approximation used above where a flow between non-parallel surfaces is treated locally as a flow between
parallel surfaces is commonly called the lubrication approximation because it is often employed in the theory of
lubrication. The lubrication approximation, simply speaking, is a local application of a one-dimensional solution
and therefore may be referred to as a quasi-one-dimensional approach.
Equation (17.3) may be integrated to obtain the pressure drop across the tube on substituting the taper
function R(z), which is determined next.

Taper function
As the tube radius R varies linearly from R0 at the tube entrance (z = 0) to RL at the tube exit (z = L),
the taper function may be expressed as
z
R(z) = R0 + (RL − R0 )
L
On differentiating with respect to z, we get
dR RL − R0
= (17.4)
dz L

49
 Note that the taper function is not affected by the type of fluid (Newtonian or power-law); it is a purely
geometrical characteristic of the system.

Pressure P as a function of radius R

Equation (17.3) is readily integrated with respect to radius R rather than axial distance z. Using equation
(17.4) to eliminate dz from equation (17.3) yields
 n
3n + 1 w L dR
−dP = 2m (17.5)
n ρπ RL − R0 R3n+1
Integrating the above equation between z = 0 and z = L, we get
Z PL  n Z RL
3n + 1 w L dR
−dP = 2m (17.6)
P0 n ρπ RL − R0 R0 R3n+1
and therefore,
 n  
P0 − PL 3n + 1 w 1 1 1
= 2m − (17.7)
L n ρπ 3n(RL − R0 ) R03n 3n
RL
Equation (17.7) may be re-arranged into the following standard form in terms of mass flow rate:
1/n  1/n
πρR3

∆P R0 3n(λ − 1)
w= 1 0 (17.8)
n +3
2mL 1 − λ−3
where the taper ratio λ = RR0 . The term in square brackets on the right-hand side of the above equation may
L

be viewed as a taper correction to equation (17.1).

50