Force at a Point in the Interior of a SemiInfinite Solid

Raymond D. Mindlin

Citation: Physics 7, 195 (1936); doi: 10.1063/1.1745385

View online: http://dx.doi.org/10.1063/1.1745385
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v7/i5
Published by the American Institute of Physics.

Additional information on Physics

Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
l\[ A Y. 1936 PHYSICS VOLUME 7

Force at a Point in the Interior of a Semi-Infinite Solid

RAYMOND D. MINDLIN, Department of Civil Engineering, Columbia University
(Received March 13, 1936)

A solution of the three-dimensional elasticity equations for a homogen:?us i.sotro~ic soli.d

is given for the case of a concentrated fo~ce acti~g in the i~ter.ior of a .seml-Infimte solid. This
represents the fundamental solution having a smgular pomt In a solid bounded by .a. pla~e.
From it may be derived, by a known method of synthes~s, the s~lut.ions for t~e sen:lI-mfin~te
solid which correspond to the solutions known as nuclei of stram In the solid of Indefimte
extent.

1. INTRODUCTION result with the original Kelvin single-force solu-

tion, the Boussinesq formulas are obtained.
O NE of the fundamental results in the theory
of elasticity is the Kelvin solution! for a
The Kelvin solution may be used in studying
stresses due to a force applied at a great distance
force applied at a point in a solid of indefinite
from a boundary while the Boussinesq solution is
extent. It is well known that stress distributions
applicable in the case where the force acts at the
for a number of problems of practical importance
surface. The solutions described in this paper fill
may be obtained from the Kelvin solution by
in the gap between the two by giving the stresses
methods of synthesis and superposition. 2 For
for the case where the force is applied near the
example, the problem, first solved by Lame, of
surface. Such a condition is approached in a
the stresses in a spherical container under uni-
number of practical problems such, for example,
form internal or external pressure may be
as in the case of a guy wire anchor. Again, by
treated by superposing stresses derived from
integrating these new solutions along a line, we
Kelvin's results. First we form, from the Kelvin
may approximate the conditions produced by a
single force solution, the stresses for a double
friction pile or an anchor rod.
force, that is, a pair of equal and opposite forces
Solutions of the elasticity equations such as the
acting at neighboring points. By superposing
single force, double force, center of compression,
three mutually perpendicular double forces, we
and line of centers of compression are referred to
arrive at the solution known as the center of
as nuclei of strain. 2 The number of nuclei of
compression. Finally, the addition of a unifo.rm
strain which may be obtained by synthesis from
tension or pressure to the center of compreSSlOn
the Kelvin solution is unlimited and further
leads to the solution of Lame's problem.
combinations lead to the solutions for such
The classical problem of Boussinesq dealing
important cases as Southwell's problem of the
with a normal force applied at the plane boundary
spherical cavity in an unlimited solid under
of a semi-infinite solid 3 has found practical· ap-
simple tension 4 and Cerruti's problem of a force
plication in the study of the distribution ?f
applied tangentially at the plane boundary of a
foundation pressures, contact stresses, and III
semi-infinite solid.·
certain problems of soil mechanics. It is known
Each of the nuclei of strain obtained from
that Boussinesq's problem may also be solved
Kelvin's results is a solution applicable to a solid
by superposing solutions derived from Kelvin's
of indefinite extent. With the solution for a
results. To do this we form a line of centers of
force in the interior of a semi-infinite solid, as
compression by integrating the stresses for a
described in this paper, we may apply the method
single center of compression. By combining this of synthesis to develop a new series of nuclei of
strain for which the stresses vanish on the plane
1 Sir W. Thomson, Cambridge and Du~lin Math. ].
(1848). See also A..~. H. Love,. Mathematzcal Theo;y· of boundary of the semi-infinite solid. Such solutions
Elasticity fourth editIOn (Cambndge, 1927), p. 183, and may be termed half-space nuclei of strain.
S. Timoshenko, Theory of Elasticity (New York, 1934), p.
4 R. V. Southwell and H. ]. Gough, Phil. Mag. 1, 71
321.
2 Love, p. 186, and Timoshenko, p.
323" . . (1926).
3 ] . Boussinesq, Applications ~es Potentwls . . . (Pans, 5 V. Cerruti, Ace. Lincei. Mem. fis. mat., Roma 13, 81
1885). Also Love, p. 191, and Tlmoshenko, p. 331. (1882). Also Love, p. 241.
195

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
196 RAYMOND
The notion of such half-space nuclei of strain
suggests some interesting possibilities. By em-
ploying these half-space nuclei as unit solutions,
it may be possible to find solutions analogous to
those obtained from the Kelvin nuclei but with
the additional feature that a new plane boundary
is introduced. For example, the solution anal-
ogous to Southwell's problem would pertain to a
spherical cavity near the surface of a semi-
infinite solid; the solution analogous to the
problem of Boussinesq would be that for a force
applied at the surface of a body bounded by two
perpendicular planes (i.e., the transmission of
force through "quarter-space"); the solution
analogous to the hollow sphere under uniform
pressure would apply to an eccentrically hollow
sphere.
The present paper deals with the fundamental
solution for the single force in the interior of a
semi-infinite solid. The problem is divided into
two parts: (1) force normal to the boundary;
(2) force parallel to the boundary.
The discovery of the solutions followed a
study of H. M. Westergaard's interpretation6
of the problems of Kelvin, Boussinesq and
Cerruti in terms of Galerkin vectors and the
results obtained are shown to include these
problems for certain limiting conditions.
2. THE GALERKIN VECTOR
The components of stress in an elastic iso-
tropic solid were expressed by B. Galerkin 7 in
terms of partial differential coefficients of three
functions satisfying a fourth-order equation.
P. F. Papkovitch 8 and H. M. Westergaard 9 have
expressed the displacements and stresses in
terms of a vector stress function whose scalar
coefficients are the Galerkin functions. The dis-
placements are given, in Westergaard's form of
the Galerkin vector, b y 9
U= (1j2G)[2(1-M)~X - (ajax) div F],
v=(1j2G)[2(1-M)~Y-(ajay)
W= (1j2G)[2(1- M)~Z - (ajaz) div F],
6 H.. M. Westergaard, Lectures on Elasticity while
Visiting Professor at the University of Michigan (Summer
Session, 1934).
7 B. Galerkin, Comptes rendus 190, 1047 (1930).
8 P. F. Papkovitch, Comptes rendus 195, 513, 754
(1932).
9 H. M. Westergaard, Bull. Am. Math. Soc. 41, 695
(1935).
Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
D. MINDLIN
where: X, Y, Z are the Galerkin functions,
i, j, k constitute an orthogonal system of unit
vectors
F=iX +j Y +kZ is the Galerkin vector,
LI. is Laplace's operator,
G is the modulus of rigidity,
p. is Poisson's ratio.
The stresses are given b y 9
O"x = 2(1-,u) a~x jax+ (,u~ - a 2j ax 2) div F,
O"y= 2(1- ,u)a~ y j aY+(M~ - a 2j ay2) div F,
O"z=2(1-,u)a~Zjaz+(,u~- a 2jaz 2 ) div F,
Tyz = (1-M)(a~ y j az+a~Zj ay)
-(a 2 jayaz)divF, (2)
Tzx= (1- M)(a~Zj ax+a~Xjaz)
-(a 2 jazax) div F,
Txy= (1- ,u)(a~X j ay+a~ Y j ax)
-(a 2 jaxay) divF.
For the case of zero body force, the Galerkin
vector must satisfy the biharmonic equation,
(3)
It was observed by Professor Westergaard 6
that the coefficient of a single axially symmetrical
component of the Galerkin vector is identical
with Love's function lO for strain symmetrical
about the corresponding axis.
3. NUCLEI OF STRAIN IN TERMS OF GALERKIN
VECTORS
The solutions given in sections 4 and 5 were
obtained by superposition of nuclei of strain for
an unlimited solid. It has been found con-
venient, in applying the method of superposition,
to express the nuclei of strain in terms of Galerkin
divF], (1) vectors.
In the following list of functions, A, B,
C, ... H are constants and R= (X2+y2+Z2)!.
It may be observed that each of the functions
satisfies the biharmonic equation and that the
stresses derived from them vanish at infinity.
10 Love, p. 274; Timoshenko, p. 309.
 FORCE IN THE INTERIOR OF A SOLID 197

GALERKIN VECTORS FOR NUCLEI OF STRAIN IN AN (3) kH[y log (R+y)-RJ (type G(3) along y axis from
UNLIMITED SOLID y = 0 to y = - (0)
A. Single force (Kelvin's problem). (4) iH[y log (R+y) - RJ (type G(1) along y axis from
(1) iAR (single force in x direction) y=O to y= - 00)
(2) JAR (single force in y direction) (5) jH[z log (R+z)-RJ (type G(2) along z axis from
(3) kAR (single force in z direction) z = 0 to z = - 00)
(6) kH[x log (R+x) - RJ (type G(3) along x axis from
B. Double force
x = 0 to x = - 00)
(1) iBx/ R (double force in x direction)
(2) jBy/R (double force in y direction)
(3) kBz/ R (double force in z direction) 4. FORCE NORMAL TO THE BOUNDARY OF A
C. Double force with moment SEMI-INFINITE SOLID
(1) iCz/ R (double force in x direction with moment
about yaxis) The semi-infinite solid is considered to be
(2) jCx/ R (double force in y direction with moment bounded by the plane z=O, the positive z axis
about z axis) penetrating into the body. A force P is applied
(3) kCy/R (double force in z direction with moment
at point (0, 0, +c) and acts in the pOSitive z
about x axis)
(4) iCy/R (double force in x direction with moment
direction (Fig. 1).
about z axis) Since the stress will be symmetrical about the
(5) jCz/ R (double force in y direction with moment z axis, we need consider only the k component of
about x axis) F. Transforming to cylindrical coordinates
(6) kCx/ R (double force in z direction with moment r, 0, z and noting that Z is independent of 0, we
about y axis)
find, from Eqs. (1) and (2), the following ex-
D. Line of double forces with moment
pressions for the displacements and stresses III
(1) iDz log (R+x) (type C(1) along x axis from x=O
to x= - 00) symmetrical cylindrical coordinates :10
(2) jDx log (R+y) (type C(2) along y axis from y=O
to y= - 00)
u= -(1/2G)(a 2Z/araz),
(3) kDy log (R+z) (type C(3) along z axis from z=O
to z= - 00)
(4) iDy log (R+x) (type C(4) along x axis from x=O ITr = (a/ aZ)[M~Z - a2z/ ar 2J,
to x= - 00) (4)
(5) jDz log (R+y) (type C(5) along y axis from y=O ITo = (a/ az) [M~Z - (l/r) (az/ ar) J,
to y= - 00)
(6) kDx log (R+z) (type C(6) along z axis from z=O
to z= - 00)
E. Center of compression or dilatation
(1) iE log (R+x)
(2) jE log (R+y) (All three yield the same stresses)
(3) kE log (R+z)
F. Line of compression or dilatation of constant strength
(1) iF[x log (R+x)-RJ (along x axis from x=O to 4-------~~-\------~~
x=-oo)
(2) jF[y log (R+y)-RJ (along y axis from y=O to
y= - 00)
(3) kF[z log (R+z)-RJ (along z axis from z=O to
z=-oo)
G. Doublet (double center of compression-dilatation)
(1) iG(I/R) (axis of doublet parallel to x axis)
(2) jG(1/R) (axis of doublet parallel to yaxis)
(3) kG(I/ R) (axis of doublet parallel to z axis)
H. Linearly varying line of doublets with strength propor-
tional to distance from the origin
(1) iH[z log (R+z)-RJ (type G(l) along z axis from
z = 0 to z = - 00)
(2) jH[x log (R+x)-RJ (type G(2) along x axis from FIG. 1. Force normal to the boundary in the interior of a
x = 0 to x = - 00) semi-infinite solid.

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
198 RAY:\'10ND D. MINDLIN

We first apply the Kelvin solution at point
(0,0, +e) by replacing R with R l = (rz+(z-e)Z)!
in solution A (3) of section 3. We note that
stresses (}"z and Trz are produced on the plane
z=o. The boundary conditions for z=o are
[O"zJz=o= [Tr.]z=O=O. (5)
It is therefore necessary to add further solutions
to remove the boundary stresses and these solu-
tions must introduce no new singularities in the
region z::';O. Nuclei of strain of types A(3), B(3),
E(3), F(3) and G(3), applied at (0, 0, -e), are
found to satisfy these conditions if the proper
values are assigned to the constants A, B, E, F
and G.
To transfer nuclei of strain from. the origin to
(0, 0, -e), we replace z with (z+e) and R with
R 2 = (r2 + (z+e)2)! in the formulas of section 3.
The values of the constants are obtained from
the boundary 'Eqs. (5) and the equilibrium
condition
P= - j""21rrO"zdr, (z>e). (6)
o

This procedure leads us to the required Galerkin vector:

F= [Pk/87r(1- J.t)J{R l +(3 -4J.t)R 2-2e(z+e)/R z -4(1- 2J.t)e log (R 2+z+e)
+4(1-J.t)(1-2J.t)[(z+e) log (R 2 +z+e)-R zJ+2e z/R 2 1. (7,
The solution for a force in the interior of a semi-infinite solid and normal to the boundary is seen
to be compounded of six nuclei of strain which, in an unlimited solid, represent: (1) a single force at
(0,0, +e); (2) a single force at (0, 0, -e); (3) a double force at (0, 0, -e); (4) a center of compression
at (0, 0, -e); (5) a line of centers of compression extending from z= -e to z= - 00 ; (6) a doublet at
(0,0, -e).
Collecting terms in Eq. (7) we obtain: 11
F = [Pk/87r(1- J.t)J (R l +[8J.t(1- J.t) -lJR2 +4(1-2J.t)[(1- J.t)z- J.teJ log (R 2 +z+e) - 2ez/R2 1. (8)

Eqs. (8) and (4) yield the displacements and stresses:

Pr [z-e (3-4J.t)(z-e) 4(1-J.t)(1-2J.t) 6eZ(z+e)]
U=----+ +,
167rG(1- J.t) R l 3 R 23 R 2(R 2+z+e) R 25

W=
P [3--4J.t
-+
8(1-J.t)2- (3-4,u) (z-e)2 (3 -4,u)(z+e)Z-2ez 6eZ(z+e)2]
+--+ +,
167rG(1-J.t) Rl R2 R l3 R 23 R 25
P [(1-2J.t)(Z-e) (1-2J.t)(z+7e) 4(1-,u)(1-2J.t) 3r 2(z-e)
O"r = +------
87r(1-J.t) R l3 R 23 R 2 (R 2+z+e) R l5
2
+6e(1-2J.t)(z+e)2-6e (z+e) -3(3 -4J.t)r 2(z-e) 30er 2Z(z+e)], (9)
R 25 R27
2
P(1-2J.t)[(Z-e) (3-4J.t)(z+e)-6e 4(1-J.t) 6e(z+e)2 6e (z+e) ],
0"9=
87r(1- J.t)
--+
R 3 R 23 R 2(R 2 +z+e)
+---
R (1- 2J.t)R 25
l 25
O"z= P [ (1-2,u)(z-e) + (1-2,u)(z-e) 3(z-e)3
87r(1-J.t) R l3 R 23 R l5
3(3 -4J.t)z(z+e)2- 3e(z+e)(5z -e) 30eZ(z+e)3],
R~ R? .
Trz= Pr [_1-2J.t +1-2J.t 3(z-e)2 3(3-4J.t)z(z+e)-3e(3z+e) 30eZ(z+e)2].
87r(1- J.t) R l3 R 23 R l5 R 25 R27
----
Jl R. D. Mindlin, Comptes rendus 201,536 (1935).

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
 FORCE IN THE INTERIOR OF A SOLID 199

f}fJ iz·-
tJ.6

/11 S' ~ 1.9 Iz 0 I, 12 8 4£

I a1 r - r--- r-....
---
1---- i---
l6u -F /
""
z (1 \ 12

R r£~
--t;?p
-Ill
I"

\
at

/
~C
'-LUI \L
\ -4L ~
p
'"
\'VU
-liD

FIG. 2. Normal stress on planes parallel to boundary for the FIG. 3. Vertical deflection of the surface for the case of a
case of a force normal to the boundary (I' = 0.3). force normal to the boundary (1'=0.3).

Fig. 2 shows the distribution of normal stress on horizontal planes, and Fig. 3 shows the vertical
deflection of the surface of the semi-infinite solid.
When C-7oo all terms in Eqs. (9) containing R2 vanish and the solution becomes that for Kelvin's
problem where the force is applied at (0, 0, +c) in the positive z direction. The corresponding Galerkin
vector is
F=k[PRI/87r(1-,u)]' (10)

When C-70 Eqs. (9) give the stresses and displacements for Boussinesq's problem.I 2 These stresses
are represented by the Galerkin vector

F=k(P/27r)[2,uR+(1-2,u)z log (R+z)]. (11)

The solution for the problem of Boussinesq was given by Professor Westergaard 6 in the form of the
Galerkin vector
(12a)

combined with the potential function

cp= -[(1-2,u)/27rJP log (R+z). (12b)

The relation between cp and F is given by

grad cp = 2 (1- ,u)AF - grad div F. (13)

As a further verification of the solution, we may compare the stresses with those obtained for the
corresponding two-dimensional case by E. Melan. 13 While the approach to the solution in three
dimensions was simplified through the use of cylindrical coordinates, the comparison with Melan's
formulas is facilitated by referring the components of stress to rectangular coordinates. Eqs. (9) then
become:

12 The stresses and displacements for the problems of 13 E. Melan, Zeits. f. angew. Math. und Mech. 12, 343
Boussinesq and Cerruti are given by F. Vogt, Del Norske (1932). This comparison was suggested by Professor S.
Videnskaps Akademi i Oslo, Math.-Naturv. klasse, No.2, Timoshenko.
pp. 5, 20 (1925).

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
200 RAYMOND D. MINDLIN

(Jx= P r(1-2M)(Z-C) 3x 2(z-c) + (1-2M)[3(z-c)-4M(Z+C)]

87r(1- M) _ RIo R I5 R 23
3(3 -4M)X2(Z -c) - 6c(z+c)[(1- 2M)Z- 2MC] 30cx 2z(z+c)

4(1-M)(1-2 M)( x2
1------
R 2(R 2+z+c) R 2(R 2+z+c)
(Jy=~_[(1-2M)(Z-C) _ 3y2(Z-~+ (1-2M)[3(z-c) -4M(Z+C)]
87r(1- M) R I3 R I5 R 23
3(3 -4M)y2(Z-C) - 6c(z+c)[(1- 2M)Z- 2MC] 30 cy2Z(Z+C)

4(1-M)(1-2M)( y2 y2)]
- R 2(R 2+z+c) 1- R 2(R 2+z+c) - R22 ' (14)

(1-2M)(Z-C) (1-2M)(Z-C) 3(Z-C)3

(Jz P [ -----+-----
87r(1- M)

3(3 -4M)Z(Z+C)2- 3c(z+c) (Sz - c) 30CZ(Z+C)3],

R 25 R27
(1- 2M) (1- 2M) 3(Z-C)2 3(3 -4M)Z(Z+C) - 3c(3z+c) 30CZ(Z+C)}
Py [
Tyz = 87r(1- M) R I3
+ R 23 R I5 R 25 R27
(1- 2M) (1- 2M) 3(Z-C)2 3(3 -4M)Z(Z+C) - 3c(3z+c) 30CZ(Z+C)2}
Px [
Tzx= 87r(1-M) R I3
+ R 23 R I5 R 25 R27

Pxy [ 3(z-c) 3(3-4M)(Z-C) 4(l-M)(1-2M\ 1 1) 30CZ(Z+C»).

Txy + ---+-
87r(1- M) R I5 R 25 R 2 2(R 2+z+c) R 2+z+c R2 R27

We consider a uniform distribution of forces S. FORCE PARALLEL TO THE BOUNDARY OF A

of magnitude p per unit of length along a line
through (0, 0, +c) parallel to the y axis. If db is
a small element of this line at a distance b from
the Z axis, the stresRes due to the uniform pressure
p acting on db are found by substituting pdb for
P and (y-b) for y in Eqs. (14). If, now, we
integrate these formulas with respect to b
between the limits - 00 and + 00, we obtain the
corresponding solution for plane strain. By taking
account of the relations between the elastic
constants for plane strain and plane stress and
also noting an interchange of axes, this pro-
cedure leads directly to the stresses given by
Melan.
SEMI-INFINITE SOLID
In this case there is no axial symmetry and we
employ rectangular coordinates (x, y, z), and
Eqs. (1) and (2). The force P is applied at (0, 0,
+c) and acts in the positive x direction (Fig. 4).
We again begin by applying the Kelvin solu-
tion at (0, 0, +c), this time using type A(1). To
remove the stresses on plane z=O, introduced
by the single force, we employ nuclei of strain of
types A(1), C(6), D(6), G(l) and H(1), again
substituting (z+c) for zandR 2 = (X 2+y2+ (Z+C)2)!
for R. The constants associated with these
solutions are determined from the boundary
conditions [(Jz]z~o= [Tyz]z~O= [TzxJz~O= ° (15)

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
 FORCE IK THE INTERIOR OF A SOLID 201

and the equilibrium condition
p= -4f."'f'" Tzxdydx,
o 0
The result of these operations is the Galerkin
vectorll
F = [P /87r(1-,u) J(ilR I + R 2- 2e 2/ R2
+4(1-,u)(1- 2,u)[(z+e) log (R 2 +z+e) -R2JI
+k! 2ex/ R 2 + 2 (1- 2,u)x log (R 2 +z+e) I).
Thus, for the case of a force in the interior of a
semi-infinite solid, parallel to the boundary, the
solution is obtained by superposition of six nuclei
(z>e). (16) of strain for the unlimited solid: (1) a single force
at (0, 0, +e); (2) a single force at (0, 0, -e);
(3) a doublet at (0, 0, -c); (4) a semi-infinite line
of doublets extending from Z= -e to z= - if.;
with strength proportional to the distance from
Z= -e; (5) a double force with moment at
(0, 0, -c); (6) a semi-infinite line of double
forces with moment extending from Z= -e to
(17) z=-oo.
The displacements and stresses, obtained from
Eqs. (1), (2) and (17), are found to be:

Px [(1-2,u) (1-2,u) 3(z-e)2 3(3-4,u)(z+e)2 6e( SZ(z+e)2)]

O"z= ---- - +- e+(1-2,u)(z+e)+--- ,
87r(1-,u) R I3 R 23 R I5 R 25 R 25 R22

1"z= Pxy [_ 3(z~e) 3(3-4,u),(z+e) +~(1-2,u+ 5Z(z+e»)],

87r(1-,u) RIO R2 0 R 25 R22
TZ£=--~-[- (1-2,u)(z-e) + (1-2,u)(z-e) _ 3x 2(z-e) _ 3(3-4,u)x 2(z+e)
87r(1-,u) R I3 R 23 R I5 R 25

Py [
Txy= 87r(1-,u)

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
202 RAYMOND D. MINDLIN

lu

~'" ..
'"
+-----~;7,I~~------~~ ff1 dI
\
'/
C Iii \ I. 1\
-t .,J/ (J
~
f"tIU i'-
ref
~ la:'
'" \ -tl., /,
'/

\j
'"
-<2J

FIG. 4. Force parallel to the boundary in the interior of a FIG. 5. Normal stress along lines parallel to the x axis on
semi-infinite solid. planes parallel to the boundary for the case of a force
parallel to the boundary (,u = 0.3).
Fig. 5 shows the distribution of normal stress on
horizontal planes, and Fig. 6 shows the vertical ......
deflection of the surface of the solid along the
~ ~'.r-..
y ••

x axis.
When c~ 00 all terms in Eq. (18) containing
R2 vanish and the solution becomes that for '4111 l
I
\
Kelvin's problem with the force apolied at , "- >-.....
(0, 0, +c) in the positive x direction. The cor-
responding Galerkin vector is
- .'~
f'..
-f r"
11.4,,£
-4_£
(19) \ / ."N
1'--'
When c~O Eqs. (18) give the stresses and ,*0.0"

displacements for the problem of Cerruti,12

These results are represented by the Galerkin FIG. 6. Vertical deflection of the surface along the x axis for
vector the case of a force parallel to the boundary (,u=0.3).

F= [P/471'(1-M)]Ci/R+2(1-M)(1-2M) combined with the potential function

X[zlog (R+z)-R]\
cf> = [(1- 2M)P /271' ][x/ (R +z) J. (21b)
+k {(1- 2M)X log (R +z) l). (20)
Following the same procedure as in the case of
The solution for the problem of Cerruti was the normal force, the stresses given in Eqs. (18)
given by Professor Westergaard 6 in the form of may be reduced to Melan's formulas for the two·
the Galerkin vector dimensional analog,13
The author is indebted to Professor H. M.
F= [P/471'(1-M)][iR Westergaard for valuable criticisms of the
+k(1-2M)X log (R+z)] (21a) manuscript.

Downloaded 14 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions