The slow motion of a sphere through a viscous fluid towards

a plane surface

HOWARD BKENNER
Department of Chemical Engineering, New York University, New York 53, New York

(&ceivecl 31 J~nuavy 1961)

Abstract-Bipolar co-ordinates are employed to obtain “ exact ” solutions of the guat,ions of

slow, viscous flow for the steady motion of a solid sphere towards or away from a plane surface
of infinite extent. Two cases are considered : (i) the plane surface is rigid and fluid adheres to
its surface ; (ii) the plane is a free surface on which the tangential stresses vanish. Deformation
of the surface in the latter case is neglected. Numerical results are provided for the corrections
to Stokes’ law necessitated by the presence of the plane boundary at a finite distance from the
particle. Application of the results to end-effect correlations in the falling-ball viscometer are
discussed.
RBsumB-Des coordonnCes bipolaires sont utilisCes pour obtenir des solutions “ exactes ” des
Cquations relatives Q 1’Ccoulement lent et visqueux d’une sph&e solide en mouvement uniforme
au voisinage ou & distance d’unc surface plane infinie. Deux cas sont consid&% : (1) La surface
plane est rigide et le fluidr y adh&re ; (II) elle est libre et les forces tangentielles disparaissent.
On neglige dans ce dernicr cas la d&formation de la surface. Des rCsultats numeriques sont
don&s pour les corrections de la loi de Stokes nCcessitCes par la pr&ence d’un plan de sCparation it
une distance finie de la particule. Suit une discussion de l’application des r&ultats aux relations
donnant l’effect des ext&mit& dans les viscosimktres a chute de bille.
Zusammenfassung-Mit Polarkoordinaten werden “ exakte ” Liisengun der Gleichungen fiir
langsames viskoses Fliessen fiir die stationgre Bewegung einer festen Kugel gegen eine ebene,
unendlich grosse Oberflilche, oder von ihr hinweg, erhalten. Es werden zwei Fiille betrachtet :
(I) Die ebene Ober&che ist fest und mit Fliissigkeit behaftet ; (II) Es liegt eine freie Oberfllche
vor, in der die Tangentialspannungen verschwinden. In diesem Fall wird die Verformung der
Oberfl%che vernachliissigy. Numerische Ergebnisse werden angegeben zur Korrektur des
Stokes’schen Gesetzes, die infolge der Gegenwart einer ebenen GrenzflBche in endlicher Entfernung
von dem Teilchen notwendig ist. Die Anwendung der Ergebnisse zur Korrektur von Endeffekten
im Kugelviskosimeter wird diskutiert.

tions of this nature emphasize corrections arising

STOKES’ law, for the resistance of a sphere moving from the fact that the fluid is laterally bounded.
slowly through a viscous liquid, finds wide They do not, however, provide insight into the
application in the interpretation of low Reynolds equally interesting case where the fluid is bounded
number sedimentation phenomena. This well- longitudinally, that is, perpendicular to the
known relation applies only to fluid media which direction of motion of the particle.
extend to infinity in all directions. In most real In this context we propose to investigate the
situations, however, the fluid is externally bounded motion of a spherical particle towards or away
by rigid walls and/or a free surface. The presence from a single plane surface in an otherwise un-
of these boundaries at finite distances from the limited fluid. Two distinct cases are of interest :
particle necessitate corrections to Stokes’ resis- (i) the plane surface is rigid as, for example, when
tance formula. it constitutes the bottom of the container in
The case in which a sphere falls along the axis which the particle falls ; (ii) the plane is a free
of an infinitely long circular cylinder has been surface as, for example, when it corresponds to
elaborated in great detail by many authors, most the interface between a liquid and the atmosphere.
notably by HABERMAN and SAYRE [l]. Investiga- The rigid wall case has been treated by LORENTZ

242
 The slow motion of a sphere through a viscous fluid towards a plane surface

[2] for the situation in which the sphere radius, motions it is permissible to neglect the quadratic
6, is small compared to the instantaneous distance, term v * V v in comparison to the viscous term,
h, of its midpoint from the plane. He finds that Y vav. This can be rigorously justified for
the resistance of the particle is greater than would sufficiently small values of the particle Reynolds
he predicted by Stokes’ law by the amount number, bU/v.
The fluid motion is inherently unsteady, since
the distance separating the particle from the_
bounding plane continually changes. However,
It is our intention to provide “ exact ” solutions as is easily established, the time-dependent term
,of these problems, unrestricted with regard to the in the equations of motion, 3 V/B t, becomes
ratio b/h. negligibly small by comparison to the viscous
term for sufficiently small values of the dimension-
1. EQUATIONSOF MOTION less group lJb2/v h.
We consider here the problem of a solid sphere Thus, whenever these two dimensionless criteria
of radius b moving with constant velocity U are met, the equations of motion are of the form
towards a plane surface, z = 0. The instantaneous
VP= KV’V
distance of the sphere centre from the plane is
denoted by h, as depicted in Fig. 1. The origin where K is the fluid viscosity. These are the so-
of co-ordinates, 0, is located at the point of called creeping motion equations. They are to
intersection of the sphere axis with the plane. be considered simultaneously with the continuity
Attention is confined to the semi-infinite domain, equation.
z > 0. Related problems dealing with the slow rotation
of two spheres perpendicular to their line of
2
centres [S, 41 and the slow translation of two spheres
parallel to their line of centres [5] have been solved
exactly in bipolar co-ordinates.
Because the motion is axisymmetrical, the
present problem is most readily treated by means
of Stokes’ stream-function, #. In terms of this
Radius b function the velocity components in cylindrical
u Infinite co-ordinates [p, z] are
h ,/plane, Z =0
I
1 a+ 124
vp=___,vz=-__’ (1.1)
P 3i-i P 3P

Neglecting the effects of fluid inertia and the time

derivatives of velocity, the differential equation
FIG. 1. Schematic sketch. satisfied hy the stream-function is [6]

Qb4($V = 0, (1.2)
The fluid motion is, in general, governed by the
where the differential operator @s has the follow-
Stokes-Navier equations,
ing form in cylindrical co-ordinates:

@p++p3
(- 1
3
(1.3)
JP P3P -
and continuity equation,
For the class of problems at hand we describe
v-v=0
the fluid motion in bipolar co-ordinates. The
for incompressible fluids. For sufficiently “ slow ” bipolar co-ordinates (5, 7) of a point in a meridian
Chem. En@?k!. Sri. Vol. 16, NOR. .7 and 4. tkremt~cr, 1901.
 HOWARD BEENNER

plane are defined by the conformal transformation where for brevity we have put
PI /L = cos 7. (1.8)
2 + i p = ic cot 3 (7 + i gy, Here,
or, equivalently, [7]
IY,([)=a,cosh(n.-_)E+b,sinh(?t--&)f+
p + i (z + c) + c, cash (n + 3) [ + d,, sinh (n + $) 5, (1.9)
[+iv=ln
p$-’
and C;iy (p) is the Gegenbauer polynomial of
where c is a positive constant. In the present order n. + 1 and degree - 4. These latter func-
application it is only necessary to consider the tions are related to Legendre polynomials via the
situation for which z > 0 and p >, 0, correspond- relation
ing to the range of values co > 5 2 0, w > q > 0. c--1/2
n
(p) = PC-2Cd - p?a
(CL)
On the basis of the foregoing we find 2n - 1

sin 7 sinh 5 The constants h, bn, . . . . are to be determined

p=c ,2=c 9 (1.4)
cash [ - cos 7 cash f - cos T from the boundary conditions.
As shown by STIMSONand JEFFERY, the fric-
and thus
tional force, F, in the positive z direction, opposing
(z - c coth 6)” + p2 = (c cosech 02, the motion of the sphere is

x2 + (p - c cot q)2 = (c cosec 7)“.

F= xi?
r K‘12)n=r
2bsinha (a, + bn + cn + d,), (1.10)
The surfaces evolved by rotating the curves
5 = constant in the p, z plane about the z-axis where K is the viscosity.
are therefore a family of spheres of radii c cosech 5 In the subsequent development we shall require
whose centres lie along the z-axis at the points an expansion of the term (cash 5 - p)s’2 p2
(p = 0, z = c coth 5). When 5 = 0 the sphere having the same general form as the right-hand
degenerates into the plane z = 0. side of (1.7). Upon combining (1.4), (1.6)and
If we denote by .$ = a > 0 the solid sphere of (1.8), there is obtained
radius b on which we ultimately wish to satisfy
(cash t - #a p2
the boundary conditions, then = (1 - $) (cash 6 - P)-‘/P.
b2 sinh2 a
b = c cosech a. But
cash 5 = B [exp (0 + exp (- 81,
Furthermore, since the centre of this sphere lies
at a distance h from the plane z = 0, then so that

h = c coth a. (cash f - I_L)-~‘~

= d(2) exp (4 5) x
x [ 1 - 2 exp (5) ~1+ exp (2 fl]-112
Solving these simultaneously we obtain
If, in the potential expansion
a=cosh-l($=ln(:+,/[(i)a-l]}, (1.5)
(r12- 2 rl‘r2p + 722)-l/2 = 2 Pfi (p) & ,
and k=O r2
c = b sinh a. (1.6) valid for re > ~1, we put rr = 1 and re = exp (5)
A solution of (1.2)in bipolar co-ordinates, (5‘ > 0), then
suitable for satisfying boundary conditions on (cash 5 - P)-“’ = d(2) k!. pk (/J)
the sphere and plane, has been given by STIMSON
and JEFFERY [5]. With slight modifications their exp [ - (k + 4) 81.
However,
solution is
(2k + I) (I - P2) Pk (cl) =
(cash I - /1P2 # =nfO LJ, (5) G-;/12 (P), (1.7)
(k + 1) (k + 2) G:/i (CL)- k (k - 1) GY2 (CL),

244
 The slow motion of a sphere through a viscous fluid towardsa planesurface

from which we find, by appropriately altering the and

summation indices,

(cash .$ - p)s/a p2 = d(2) b2 sinh* a 2 ,n (n + 1)

[y (cash5 - ~~~~~c=cl $1
?Z=l r. -I
= >+ 4 U (cash 5 - CL)312
p21 * (2.9)
exp[- (n - $1 5]_ exp[- (n + 8 El f=o
1
i 2n - 1 2n + 3 1
In analogous fashion the evanescence of velocity
c?LYla (p) (1.11)
at the solid plane wall leads us to the boundary
which is in the desired form.
conditions
2. SOLID PLANE SURFACE [(cash 5 - 11.F (c1]+, = 0 (2.10)
and
Here and in the sequel the fluid motion is
referred to a co-ordinate system at rest with res- ;-; (cash 5 - cL)s/2+] = 0. (2.11)
pect to the plane. On the hypothesis of no relative f-0

motion at fluid solid interfaces the boundary The conditions (2.8)-(2.11), in conjunction with
conditions on the surface of the sphere, [ = a, (1.7) and (1.11) lead to the following four simul-
are taneous equations to determine the constants
VP = 0, v, = - u. (2.1) a,,, b,,, . . . . . :

Again, the boundary conditions at the solid plane a, cash (n - fr) a + bn sinh (n - 4) a +
surface, 8 = 0, are
+ c, cash (n + 3) a + d,, sinh (n + +) CC
VP = 0, v, = 0. (2.2) = b2 sinha a U n (n + 1)
These can be expressed in terms of the stream d(2)
function in the following way : By the chain-rule
1 - -1 4) al _
exp [ -2n(n exp [ -2n (m
+ +3 9 4 1, (2.12)
(n - 4) [unsinh (n - 1) a +
d2 3P + 4 cash (n - 4) a] +
= p VP r7 - P vz 3-I, 3 (2.3)
+(fi++)[cnsinh(n++)a+d,cosh(n++)a]
and
b2 sinh2 a U n (n + 1)
=-
(2.4)
2 d(2)

Thus, on the sphere surface, {exp[-(~-~)a]-eexp[-(n+9)a]}9 (2.13)

a, + Ga = 0, (2.14)
[gg,;. [$hup2]i_.t2e5)
= and
(n - Q) bn + (n + 8 & = 0. (2.15)
and

=[A4up21 (2*6)
(g$),=. &.’
The solution

c, = -a, =
of this set of equations is

Since v and .$ are orthogonal the former of these

two conditions is satisfied by

and
As can be verified by direct differentiation the d __(2n-l)b =_b2sinh2aUn(n+1)
following is entirely equivalent to (2.6) and (2.7) : n
(2n+3)” ~42) (2n + 3)

I
.
#]~=a
[(c-h 5 - CL)~‘~ 2sinh(2n+l)a+(2n+l)sinh2,a_l (2.17)
= [ 3 U (cash 5 - /L)%‘*P"]~_~, (2.3) 4 sinh2 (n + &) a - (2n + 1)2 sinh2 a
Chem.Engn&Sri.Vol. 19,Nos.3 and 4. December,1901.
245
 HOW-D BRENNER

If we denote by X the correction which must be for small b/h, which agrees exactly with the value
applied to Stokes’ law as a result of the presence given by LORENTZ [2] obtained by the method
of the solid wall then of “ reflexions.”

F=Brr~hUh, (2.18)
3. FREE SURFACE
and from (1.19) When the plane, z = 0, towards which the
4 co sphere falls is a free surface the boundary condi-
fi(n + 1)
A=-sinhu 2 tions on the plane are that the normal component
3 n=i (2n. - 1) (2n +3)

1
of velocity and the tangential stresses vanish :
Zsinh(%+l)cr+(2n+l)sinh2cr
4 sinhs(n. + +) GC- (2% + 1)2sinl12 a - 1 (2.19)
and
where the parameter u is given in terms of the
ratio of sphere radius, b, to the distance of its
centre from the plane, h, in equation (1.5). This
formula is virtually identical to a. similar expres-
sion given by STIMSON and JEFFERY [5] for the
Stokes’ law correction in the case of two equal
The latter equation is automatically satisfied by
size spheres falling parallel to their line of centres,
virtue of the prevailing symmetry. Furthermore,
except that the lead term in brackets is inverted.
in view of (3.1) the vanishing of the first stress
Incidentally, as pointed out by FAX~N [8], there
component is assured by setting
is a typographical error in their manuscript and
the multiplier of their expression for h should be
4 rather than $.
Using tabulated values of the exponential and Now,
(3 13 2 z=o =
0. (3.2)

hyperbolic functions [9, 10, 111, we have made 1 J# 2*3t 3*37

accurate calculations
range of interest.
of X for values of b/h in the
These are presented in Table 1.
2),=----~--

PJP ( 1 ---+-__.
P JebP h3P )

Since 5 = 0 along the plane z = 0 then the

derivative 3 t/b p vanishes on this plane and the
Table 1. Stokes’ Law corrections .for a solid plane,
condition (3.1) is met by (3 #/3 ~)~=e = 0 or,
equation (2.19)
since 5 and 7 are orthogonal,

--q-L- b/h x (&=ll = 9. (3.3)

O 1 1 co
In regard to (3.2) we have from (1.1) that an equiv-
0.5 l-1276260 0*88681885 9.2517663 alent condition imposed on the stream function is
1.0 1+430806 064805428 3~0360641
l-5 2.3524096 0.42509603 1.8374749
2-o 3*7621957 0.26580222 1.4128629
2.5 6.1322895 0.16307123 1.2219882
3.0 lo.067662 0329327927 1.1252465 But, with the aid of (3.3) and the relation
co co 0 1

When the sphere is far removed from the wall where J is the Jacobian of the transformation,
it is sufficient to retain only the first term in
cod1 6 - cos 7j
(2.19) and to neglect exp (- a) compared to I J I=) 5
c
exp (a). This results in
96 the vanishing of the tangential stress is easily
shown to correspond to

246
 The slow motion of a sphere through a viscous fluid towards a plane surface

3b
(3.4) B=l+;Ehv (3.14).

valid for small b/h.

The boundary conditions expressed by (3.3) and
(3.4) are equivalent to
Table 2. Stokes’ law corrections for n free surface,
[(cash 5 - /L)~‘~((I]~=~ = 0 (3.5) equation (3.13)
-
and
a h/b
[s2(c-h $1
5 - 14~~’ E=. = 0. (3.6)
B

0 1 03

From (1.7), these result in the simultaneous 0.5 1*1276260 3.98670

1-o 1.5430806 1.97369
equations
1.5 2.3524096 1.4636
a, + cn = 0 (3.7) 2.0 3.7621957 1.247131
and 2.5 6.1322895 1.1388563
(n - 4)” a, + (n + $)2 c, = 0. 3.0 lo.067662 1~0803758
(3.8)

The surface conditions

from their previous
on the sphere are unaltered
values. Thus, to the above
co W

I 1

we append equations (2.12) and (2.13). The Independent confirmation of the present results
solution of this set of four simultaneous equations by a different method of computation is provided
is by the work of FAX~N and Darr~ [12]. These
a, = cn = 0, (3.9) authors studied the slow motion of two spheres of
unequal size each moving with arbitrary constant
B = r_sinh2 cc Un (n + 1)
n velocity parallel to the line of centres, using a
d/(2) (2m-1) successive approximation technique known as the

19
4cosh2(n++)a+2(2n+1)sinh2a_1 method of “ reflexions.” If in their treatment
_- (3.10)
2sinh(2n+1)a-(2n+1)sinh2a the radius of each sphere is b, the centre-to-centre
distance is Zh, and if the spheres move towards
and
each other with the same velocity, U, their
d = b2 sinh2 u Un (ti + 1) expression for the frictional drag, F, experienced
n by either sphere assumes the form
d(2) (2n + 3)

1.
1_4cosh2(n+$)a-2(2n+l)sinh2a
(3.11)
2sinh(2n+l)a-(2n+l)sinh2a

The frictional force experienced by the sphere is

therefore
F=srrtcbUj? (3.12) 5,331 b ’ 19,821 b 8
+ --
16,384 0h + -- 0- +
65,536 h
where the Stokes’ law correction is
76,115 b Q
+ . .... (3.15)
B = $ sinh $ (2 n “‘T)& ‘n)+ 3) + 262,144 0h

But, when two equal spheres approach one another

4cosh2 (n + 4) a +(2n +1)2sinh2 a
___- 1 (3.13) with the same velocity, the plane midway between
2sinh(2?% + 1)a -(2n+ 1)sinh2a I them is a plane of symmetry on which the normal
Values of /? calculated from the above are pre- velocity and tangential stresses vanish. Thus,
sented in Table 2. (3.13) and (3.15) should be comparable. Values
When the sphere is far from the free surface, of the Stokes’ law correction calculated from
the St,okes’ law correction obtained from (3.13) is (3.15) are presented in Table 3. At the smaller

247
 HOWARD BRENNER

ratios of h/b this series converges too slowly to greatest. This contention can be demonstrated
yield accurate results. The numbers tabulated by invoking results from ideal fluid theory [6]
are estimated to be correct to about 1 digit in which state that a sphere moving perpendicular
the last significant figure. These results should to a wall is repelled by the wall whether the
be compared with those given in Table 2. The particle motion is directed towards or away from
agreement is excellent. it, the magnitude of the force being the same in
either case. Thus, the forces of inertia hinder the
particle in the former case and assist it during
Table 3. Stokes’ law corrections for a free surface,
the latter. From this we may infer that when
equation (3.15)
inertial effects are sensible the particle resistance
is least in the case where the sphere recedes from
h/b B the surface and vice-versa.
- It is a matter of experience that the proximity
1
1.1276260 3 of a boundary to a moving particle enhances the
1*5430306 1.96 range of particle Reynolds numbers, U~/Y, over
2.3524QQ6 1.4634 which the creeping-motion equations provide a
3.762195’7 1.247126
valid description of the flow - i.e. the range in
6.1322895 1.13385607
which a direct proportionality exists between
lo.067662 1.080375850
oc, 1 particle resistance and velocity. For example, in
CARTY’S [KS] experimental study of a ball rolling
within a viscous fluid down an inclined plane, this
proportionality was observed to persist up to
4. DISCUSSION particle Reynolds numbers of about 26. On this
The preceding calculations show that the effect basis it is reasonable to expect the present analysis
of a stationary obstacle in the path of a particle to yield correct values at Reynolds numbers
is to increase the resistance of the latter beyond substantially above O-5, the value normally cited
that which it would experience in an unbounded as the upper limit of Stokes’ law for an un-
medium when moving at the same velocity. bounded fluid.
Alternatively, if a given force (e.g. gravity) be
acting the effect is to decrease the sedimentation
velocity below that given by Stokes’ law. Further-
more, the increased resistance is less in the case
of a free surface than in the case of a solid surface,
which is unable to yield to the stresses. The
inclusion of inertial effects would not, in all
probability, modify these qualitative conclusions.
On the other hand, the present results show
that for both types of surfaces the resistance is
increased by the same amount regardless of
whether the particle is moving towards or away
from the plane. It is here that the omission of
inertial terms in the equations of motion leads,
Radius
in the general case, to an unrealistic inference. b
It seems more natural to expect that the resistance
be different, according as the sphere is approaching
or receding from the surface. Of the two possi-
bilities, the most plausible conjecture is that the
resistance suffered by an approaching sphere is FIG. 2. Falling-ball viaeometer.

248
 The slow motion of a sphere through a viscous fluid towards a plane surface

5. FALLING-BALI,VISCOMETER falling between two infinite parallel rigid planes,

It is of interest to attempt an application of the motion of the particle being parallel to the
our results to the falling-ball viscometer. In such walls, as in Fig. 3. Let Ii, and Zs, respectively,
a device a spherical particle falls along the axis of denote the distance of each plane from the
a circular cylinder, as in Fig. 2. Of importance mid-point of the sphere, and let 2 1 be the distance
are the corrections to Stokes’ law occasioned by between walls.
the proximity of the cylindrical boundary,
container bottom and free surface to the particle.
Dimensional analysis indicates that the Stokes’
law correction resulting from these three effects

t
is of the form

-0
6rrKbU
F =f($,f,f)
Of primary interest is the situation wherein each Radius

of the above ratios are small. The separate correc- b

tions required for each of the boundaries alone are : 1U

(i) cylindrical boundary correction [l] :

f ($9 030) = 1 + 2.105 ($) + 0 [ (&)“I

(ii) container bottom correction :

FIG.3. Sphere falling between parallel planes.

(iii) free surface correction : LORENTZ [2] provides a solution in the case
where a sphere moves parallel to a single plane
+,o,~)=l+~(~)+O[(;)2] wall, valid for small values of b/la.

His correction to Stokes’ law is

It is tempting to assume that these corrections
can be applied separately in the limit, in which F
=l+;(;)+o[(y] (i=1,2)
event a Taylor series expansion would give 6nKbU

F Thus, were it correct to simply superimpose the

= 1 + 2.105 b +
6rrtcbU (& ) individual corrections, the resistance of a sphere
falling between two plane walls would be given by
+;(;)+g;)+..... (5.1)
F
6aKbU
=l+;+(;+;)+o[(y] (5.2)
This is essentially the point of view adopted by
LADENBURG [la] in an oft-cited paper dealing Now, FAX&N [15] has obtained a detailed solu-
with corrections to the falling-ball viscometer. tion of the problem at hand, valid for small b/l.
The conception is, however, fundamentally un- When 11= 1s FAX~N’S solution is
sound, as it gives cognizance only to interactions
F b 2
between particle and boundaries while failing to = 1 + 1.004; + 0
6r/cbU [( 7 )I
take account of the interactions among the
boundaries themselves. while for 11= 3 12, the coefficient in the above
To demonstrate this contention consider the expression becomes l-3052. The corresponding
related, but more tractable, problem of a sphere coefficients derived from (5.2) are l-125 and l-500,
C’hm.En@%.Sci.Vol.16,Nos.s and 4. tkxxmber, 1961.
249
 HOWARD BRENNER

respectively. As these differ from the correct b = radius of sphere

b, = coefficient in equation (1.9)
values we may infer that it is not generally
c = positive parameter defined by equation (1.6)
permissible to superpose individual corrections. c, = coefficient in equation (1.9)
Although the technique of superposing solutions c,-‘/Z (CL)= Gegenbauer polynomial of order m and degree
is fundamentally in error, it appears from the - 4 with argument p
two numerical examples cited above that the errors d,, = coefficient in equation (1.9)
F = force on sphere
incurred may not be too serious. It therefore
h = distance from sphere centre to plant
appears worthwhile to bring equation (5.1) to boundary
fruition, despite. its shortcoming. This calculation h, = initial distance
differs from LADENBURG’S [la] principally in that hf = final distance
the free surface correction was not known to him. tl = distance from sphere centre to free surface
i = d-1 = imaginary number
In its stead he utilized the same correction as
J = Jacobian of a co-ordinate system transform-
for a rigid surface. ation
The instantaneous velocity of the ball is k = summation index
U = - dh/dt. Let hi and hf, respectively, denote 1 = one-half distance between walls
the initial and final distance of the ball above the I,, 1, = distance from sphere centre to walls
n = summation index
bottom of the container and let a = H + h be
p = dynamic pressure of fluid
the depth of liquid in the cylinder. The duration
P,(p) = Legendre polynomial of order m wit,h
of the experiment is t. Upon substituting into
argument ~1
equation (5.1) and performing the necessary rl, r2 = distance from origin to points 1 and 2,
integrations, bearing in mind that F is constant, respectively
we eventually obtain R, = cylinder radius
t=time
F U = instantaneous velocity of sphere
PN
6rrKbU u = average velocity of sphere

($)’
’ +2*105 (i) (hJa) y (hf/d
lJ, (0 = function
(1.9)
of argument

v = fluid velocity vector

8 defined in equation

9 (hi/a)
s In (hfla)
I i ‘n 1 - (h//u)
1 - (ha/a) 1 (5.3)
vP = velocity component
vz = velocity component
z = co-ordinate
in radial direction
in longitudinal
measured along symmetry
direction
axis
where 0 = (hi - hf)/t denotes the average velocity
z’;; = p -- component of stress-vector acting
of fall of the ball during the experiment. By way across plane z = constant
of example, when the speed of fall is timed over .z + = 4 - component of stress-vector acting
the middle third of the length of the viscometer across plane z = constant
we have hi/a = 2/3 and hf/a = l/3, whence Greek Letters
C(= parameter dellned by equation (1.5)
F /I = Stokes’ law correction factor for a free
P N 1 + 2.105 $9 + 3.90 $
6rrtcbU surface, defined by equation (3.12)
7 = bipolar co-ordinate
For a typical viscometer whose depth-to-radius K = dynamic viscosity of fluid
ratio, a/&, is 10: 1 this implies that the correc- X = Stokes’ law correction factor for a solid
tion for end-effects will be roughly one fifth of surface, defined by equation (2.18)
-
/.L= CO8q
the correction for the cylindrical boundary alone.
C = biDolar co-ordinate
Achmozuledgements-The author would like to thank p = cylindrical co-ordinate
JOHN RAPPEL of New York University for his useful p = density of fltid
suggestions. Y = kinematic viscosity of fluid
+ = azimuthal cylindrical co-ordinate
NOTATION o2 = differential operator defined by equation
n = depth of liquid in viscometer (1.3)
an = coefficient in equation (1.9) I+$= Stokes’ stream function

250
 of a sphere through a viscous fluid towards a plane surface
The slowInotion

REFERENCES

PI HABERMAN W. and SAYRE R. M. Motion of Rigid and Fluid Spheres in Stationary and Moving liquids inside Cylin-
drical Tubes. Report 1143. David W. Taylor Model Basin, U.S. Navy Department, Washington D.C., October 1958.
PI LORENTZ H. A. Abh. theoret. Phys. 1907 1 23.
t31 JEFFERY G. B. PTOC. Land. Math. Sot. (Series 2) 1915 14 327.
[41 WADEIWA Y. D. J. Sci. Engng. Res. India 19.58 2 245.
[51 S'I'IMSON M. and JEFF-JZRY G. B. Proc. Roy. Sot. 1926 A 111, 110.
PI MILNE-THOMSON L. M. Theoretical IIydrodynumics f3rd ed.) pp. 168, 513, 553. Macmillan, New York 1950.
PI JEFFERY G. B. PTOC. Roy. Sot. 1912 A 87, 109.
PI FAX~N H. 2. angezv. Math. Mech. 1927 7 79.
PI Table of IZyperbotic Sines and Cosines, Appl. Math. Ser. Nos. 36 (Nov. 30 1953) and 45 (Nov. 15 1955). National
Bureau of Standards (U.S. Commerce Department) Washington D.C.
WI Tables of the Exponential Function ezp (z) Appl. Math. Ser. No. 14 (June 29 1951). National Bureau of Standards
(U.S. Commerce Department) Washington, D.C.
Pll Tables of the Descending Ezponential, Appl. Math. Ser. No. 46 (Nov. 15 1955). National Bureau of Standards
(U.S. Commerce Department) Washington D.C.
P21 FAXBN H. and DAHL H. Ark. Mat., Astron. Fys. 1925 19A No. 13.
P31 CARED J. J. Jr. Resistance Coe$lcients for Spheres on a Plane Boundary. B.S. Thesis, Massachusetts Institute of
Technology 1957.
P41 LADENBURG R. Ann. Phys. 1907 23 447.
P51 FAX~N H. Ann. Phys. 1922 68 89; Ark. Mat., As&on. Fys. 1924 18 No. 29 ; 1925 19 A No. 13.

Ch. Engng. Sci. Vol. 16, Nos. 3 and 4. December, 1961.

251