Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Philosophical
Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences.
http://www.jstor.org
CONTENTS
PAGE PAGE
INTRODUCTION 357 5. THE EXPANSIONS FOR FORCES IN A STRIP 372
1. THE GENERAL METHOD OF SOLUTION 358 6. A PAIR OF CIRLES IN A STRIP 384
2. Two EQUAL CIRCLES IN AN INFINITE
PLANE 361 7. TWO PAIRS OF CIRCLES IN A STRIP 390
3. Two PAIRS OF CIRCLES IN AN INFINIrE CONCLUSION 391
PLANE 363
SUMMARY 391
4. AN INFINITE DOUBLE ROW OF CIRCLES IN
AN INFINITE PLANE 367 REFERENCES 392
INTRODUCTION
X(z) -- Zf(Tz),
where T is one of the transformations of the group.
Consider a uniform plate of infinite extent or bounded by parallel straight lines and
containing a number of equal circular holes having an invariancy of the type men-
tioned. It is supposed to be in a state of generalized plane stress defined by a stress
function X. This function must be biharmonic, i.e. it must be a solution of the equation
V4X 0. (1 1)
rr I_.1
= I2X - _ r - Aldx\0
r2dO2 rr dr - r \ r 0d
(c) if there are straight boundaries parallel to the x-axis, the stressesyy and xy must
vanish on them.
We first construct a set of biharmonic functions having singularities at the centres of
the circles and which are invariant under the transformation which leaves the circles
unchanged. If there are straight boundaries the condition (c) must be satisfied as well.
Two methods may be conveniently employed to obtain these functions.
When the plate is infinite we start with a complex harmonic function
Wo(z)= uo-iv, (1-2)
having the requisite singularities and invariance. Differentiation of wo(z) s times leads
to the function
Ws(Z)= U,-iV,. (1-3)
The real and imaginary parts of the complex functions defined by equations (1-2)
and (1.3) will be functions satisfying the conditions, and have the form
00c
us = r-scosss + rt(SYn
cosnO+ s,nsinnO),
n=l
co (1'5)
vs r-s sins + rn(SyncosnO+ s4,sinnO).
n=l
U. = r-S+2 cos sO+ r"{(sAn+ sA'r2) cos nO+ (sB, + sB'r2) sin nO},
n=0
oo
(1O6) n
V = r-S+2 sin sO+ rn{(sCn+ SCr2)cos nO+ (sD + sDr2) sin nO}.
n=O
44.2
These functions will usually be sufficient for our purpose (see Howland 1935) of
finding the stress function for the problem. In certain cases further functions may be
required. They will be given later when the various configurations of circles are con-
sidered. They are found by considering the stress functions for forces acting, in suitable
directions, at the centres of the circles. Leaving, for the moment, the method of con-
structing the functions when the plate is of finite breadth, we next have to combine the
stress functions given above in such a way that the boundary conditions (b) on the
circles are satisfied. The functions are such that if the conditions are satisfied on one
circle they will be automatically satisfied on each of the others. In each case considered
below we shall construct all the functions sufficient to solve the most general problem,
but to indicate the method of solution we assume here some simplification. Suppose
symmetry conditions are such that our functions are even in both x and y. Our co-
efficients are then all zero except those of the even cosines and terms independent of 0.
Suppose further that Xois a stress function which would give the stresses in the plate if the
holes were not present, i.e. the stress function giving the infinity conditions. We then
assume as our final stress function
The constants L, M have to be chosen so that the conditions (b) are satisfied. Hence
we substitute this Xinto the equations for rr and rOand equate to zero after putting r - a
(the radius of the circle). This leads to a double infinity of linear equations between the
constants. They are of the form
-
L2n P2n+ 2nhoLo + (2nh2sL2s + 2i2s M2s),]
s=1 I
(1l8)
where Pe, Q2 are known constants depending on o0,and the new coefficients 2h2s, etc.
depend upon the coefficients 2sA2 etc., and upon a. A formal solution of the equations is
given by O 0co
L2n = L(n (2
M2= 2 ,29)
r=O r=0
Mr) - 2n
2jL(r-1) - (2 jLs 2(l)i 2nsM-) ).
s -]-n2sMs j)
l-2n7O- L n=l
The validity of this solution is established if we prove that the series in (1.9) and (1-10)
are convergent. This can be done in particular cases if suitable inequalities are found for
2n) etc. In every case so far investigated the series have been proved
the coefficients 2A2,s
convergent provided the ratio of the radius of a circle to some other typical length, such
as the distance between two centres, is not too large (see e.g. Knight I934).
For the problems in which the plate is of finite breadth the sets of stress functions are
most conveniently found by a different method. We start with a fundamental stress
function representing forces at the centres of the circles which leave the straight edges
free from stress. This is expanded in a series about one centre. The coefficients in these
series contain a, the modified value of Poisson's ratio used in generalized plane stress
(Filon 1903, p. 67; Love 1927, p. 138). These functions are biharmonic functions for all
values of o-, and this may be treated as an arbitrary constant provided the functions
obtained from the fundamental one are of the type giving single-valued displacements
(Knight 1934, p. 256). Having given a a particular value we obtain by successive
differentiation functions of type (1.6).
In the most general problem four sets of functions will be required, two sets u, and
two vs. These may be found from two fundamental functions by giving a two special,
suitable values. The method of solution is the same as before, for the infinite plate; but it
is, in general, complicated by the fact that four sets of unknown coefficients will have to
be evaluated by the successive approximations.
Let the co-ordinate systems associated with the circles as shown in fig. 1 be defined by
z x+iy, z' x'+iy'. (2-1)
y
FIG. 1
We then have invariance for reflexion in a line parallel to the y axis midway between
the centres. If the centres are at a distance b apart, the relation between z and z' is
Z'- - ( + b).
Define subsidiary variables by
= z/b, g'
z'l/b--( 1) (22)
=x/b, a y/b, peio=C. J
Harmonic functions of the required type are
-w0 =log C+ log ',
(
Ws )]_- (log ) + (log ')
- g-s+(_)'s(1+)-s, (s? 1).
i(
Ws -S + I
n=O
(_)n--+s n---. )l (2.4)
If we write w =- u -ivs, we obtain, by equating real and imaginary parts, the func-
tions
u0 -- -log p + --- ) p."cos nO)
n=1
n=l n
,~~r ,(2-5)
vu= So +..
COS sin
(-) n n P ,
n
~~~~~~~~~~n=O
),,,(2.6)
v5 p-s sin sO (-)"n+s pn sin n9.
Suitable biharmonic functions are obtained from these by writing, when s> 2,
V2- 2qul
0o
n
-sin 20(+ (-) (1 -p2) pn sin nO.
n=0
Thus the case s =2 is not special and the coefficients in the series may be obtained by
putting s - 2 in the equations (2)8).
The biharmonic functions U1 and V1 will not be required since they correspond to
zero stresses, but functions U0 and V0 may be required in any particular problem. Some
1-- 2or
x1(C)= logp- 2s, (2.9)
1-2or
..- psin0logp+20pcoso+0
.- i-cro pnsin nO.
(OD+ ODnp2) (2-12)
n=0
An- OD'
OA' n
=- -_ (2.13)
I o-2(n+ Y)'
Let the centres of the four circles be at the points given by z = p ? iq, where
z = x+iy defines a rectangular co-ordinate system. At the four centres we take four
subsidiary co-ordinate systems as shown in fig. 2. We then have invariance for re-
flexions in the axis of x and y. The four systems associated with the circles are
Z4 -- X4+ iZ4 = - -
Z2 = x2 + iY2 -- --p- q, -p q.
The complex harmonic functions given previously (Howland and McMullen 1936)
are defined by
- W logZ1Z2Z3Z4, (3-1)
(- nn+s
-n-+
(- )n +s - (n+s)
s- 1_) (
Z
Z lS
ln (2;h)" (2iq)
i-J (21)n
21 XI
x 4
.1
x3
y. 3
FIG. 2
00
(3-4)
uo= 0 +I 0 sinn),
cosnO-+j-
rn(?on
n=l
where the coefficients have the following alternative forms according as n - s is even or
odd:
ccos
s _ (n+s-_-l{ L (_-)(s-n) (n+s)
&[ n 1-)(2p)+$s (2q)n+s (21).s
3 /n+s-1\ )2-(""- 0}
sin (n+s)
n5" \ n {(F(2q)"+s
~(21)n+s
(3.7)
sin -1\ (_)2(s-n-) sin (n+-s) 0}
(2q)n+s (21)n+s
The coefficients in (3-4), i.e. when s = 0, are not special and are given by the above if
(n-) is interpreted as i/n.
Wenow define biharmonic functions in the following way.
We now define biharmonic functions in the following way.
where the four terms represent the contributions from the four singularities taken
in the same order as above. Then
-
y (Us(1) + U2) -3) U4))
will be a biharmonic function with the right type of invariance. It is evident that if we
now change the signs of all the terms in the equations (3 6), (3 7) which contain powers
of 2q or 21, and if we denote by Sa', etc. the modified values of s5,, etc., the new function
is
r- +2 sin sO+ r{ (sC + sC r2) cos nO+ (D + SD; r2) sin n0}, (3.8)
n=O
Vol. 238. A. 45
Cn
SC +s-2
n sin(n+s 2)0
n ) (21)n+ -2'
(3.9)
COS (n+s-2) 0
? /e+,y-2U+s(-)2(5-n)
n (2p)n+~- (2q)^+s-2 (2l)n+s-2
__ n-
-- (2)+S-n) (21)n-+-s)
(3.10)
( n+s
_ -2) 1 cos (n+s-s2)0
."? '[ n (2p)ns-2
p (2l )"+s-2
0
(n+s--1) 1 , cos(n+s)
SD'-"n +
\ n+l 'n
h(2p)II (21)"
cos (n+s- 2) -)
n+s even: sA n (ns 2) (2- 2 (-)-2) '
+
n+s-2
nv (2p)n+s-2 (2q)n+s-2 (21
1 (s- )(sn+)cos s
)
sAln _n+s-1 (7
n+
+ /l (2)n+s (2)" (21)n s |
(3.12)
sB _~ sin (n+s--2)o
(n+s--2
n- n ) (21)n+s-2
_
n
sBJn- +sns
/n
I sin (n +s) .
n +1 (21)n4?s
odd: A - - -s- 2\
(n
n n-Vs
s odd: SA l (){+ 1 cos (n+s-2)\
2/)n - }3
=cos 20 + r{(2A - 2A, r2) cos nO+ (2B +2Bs r2) sin nO}.
n=0
sin 20 + rn{(2C,A- 2C' r2) cos nO- (2D, + 2Dn r2) sin nO}.
/1=0
n-O
The coefficients 2An,etc. are not special and are given by the general expressions
above.
As in the previous section U1, V1 are not required and the functions U0, V0 are
obtainable, if wanted, from the expansions of
4 4
Uo- 2 n=rl(Zn). Vo X(Z?,).
n=l
Let the circles be arranged as in fig. 3 with their centres at the points (p na, 0),
(Vnat-p, q) referred to axes OXY. Let O'X'Y' be axes with origin at O' whose co-
ordinates referred to OXY are (p, q).
If z =x---iy, z'x'+ iy',
then z' = p+ iq-z. (4.1)
Harmonic functions having the right sort of invariance are defined by
- w - log sin Trz/a-1 log sin 7rz'/a
where - 0+
-a 0,
and a2 - 0,
?a2ltn+I
2n/n (4-5)
FIG. 3
00c
c= ot 7TO.
The expansion for ws follows at once and is
00
an = n + an
22 - - + 2s --1 I s
?-2?i 12n , 2 ii 2s 1
12n
-- --
2sa,n 2n 0 (4-8)
l
2s+1'4 1- -2 2+2 )+ I1)
a^n+l 2n + I'2n+ 2s+21
0 -- logp + E pn(OnCcos
nO+ o sin
nO),
n=O
09
- (4.10)
U p-s cos so + E pn(Sn cos n0 + s, sin nO),
n=O
Explicit forms for the coefficients are easily written down in terms of the real and
imaginary parts off,(c), but are rather complicated in character. We shall content
ourselves by considering three special cases in which there is some simplification.
Theparallelposition
The first special case is that for p = 0 when the lattice of centres is rectangular. This
will be called the parallel position.
i/ ji
f?
FIG. 4
sYn
- sf (n + s1) (ns) b+, n + s odd.
2s-A = ((222n+2s)-2s) b2
2 -
A2n+ -- 2n -1 2+2s (2n + 2s) b2n+2sb}
2n3--I12n
2s
-t-4 (4.15)
sA=
n sAn = 0 n-s odd
sB= SB 0, n+s even,
-B b (4.)
( s-) (n+s-2) , n+s odd,
BI
n
y?{n+s
-)
I (n s)s,)
b,, ? n+s odd.
n+sodd
- -( +nl ) l
SC-B SB sC sB',,
n nA (4.17)
sD -- A,
A- UD A. J
The functions U2, V2 are not special, while U0, V0 may be obtained in a manner
similar to that used previously. They are not of particular interest here, for in most
problems with these boundaries u0 may be used instead of JU0,etc.
The alternateposition
The next special case in which there is some simplification is that in which p = 1a.
The circles are arranged as shown in fig. 5.
FIG. 5
Let 7T2nf2 ()/ (2n) ! b , 7T2n+ f2n+l (C)/(2n+ 1)!= ib2,. (4-19)
Consequently the harmonic and biharmonic functions for this set of boundaries are
obtained from the equations (4-13)-(4.17) when bnhas been replaced by b'.
A rowofpairs of circles
A third special case is that for q 0. The boundaries are now pairs of circles with
their centres collinear (fig. 6).
i-i
FIG. 6
The biharmonic functions have the same form as before with the special coefficients
gi ven by
2SA2= (2n- 2s
{202 + (2n+2s-2) dn+22 },
2s+lA -
2n- 2sl -22n+2s 2 ) d22
2n+l
2n+) K d2+ 2j,
(4.22)
-
2sA = (2n2 r 1 {2 -2 (2n)
2s+ 2.)d2 +2},
2n
- 2s+l
2s+
2s+A IA I 21d\I
-O2p+ 2n2s2 ) 2ss
2) (2 +2s+ d2n2+2}
2n+l^ = _22n+
2n2 J.
){-+2s-2+(21+2.+2)-,
(4-23)
For the problem of circular holes in a strip we take as our fundamental singularity a
force acting at a point which is to be the centre of the hole. Moreover, this force must be
given by a stress function which produces zero stresses on the parallel straight boundaries.
We have, therefore, to consider the expansion of such a function about its centre and
about other points in the strip. When this has been done suitable singularities will be
taken and combined to give the required functions in the various cases.
Y /
1e'
h
x -I - - x
/
Y
FIG. 7
The force P, acting at (o, h), is taken in direction OX; then writing
Pb
X 4r(-) - {(l-2S)og p+2(1-o-) (~-a) 0+0}, (5'3)
where - X1+X2 +X3 +X4,
C
o~
SC- /cS
X2 = fcsC s, (B + B2) sin u du,
(5-4)
(5-4)
f0 u1cC- (c +us) ) S
X3= --2 (c- (B1-B2) sin u6 du,
?c)
CS- ~sC (BI-B) sin u du.
X4 y- cs/usC
s -sih c cosusinhu, C
S=sinhl, coshu,
(5.5)
Z = sinh22u+ ,2' -sinh 2u - 2u;
To obtain the expansion about the singularity, i.e. the point (o, h), write
= psin9, -- a = ' = p cos ,'
C' =cosh u', S' - sinh u', (57)
c2 - cosh 2au, s2 - sinh 2cau. J
Then substituting the values of B1, etc. into equations (5.4), we get, after some
reduction,
??sin-
u{ S'+ U3 U4 q'S'} du,
X +X2 =
U1C2SU ,'C'+
(5.8)
X3- X4= si
U2 {VC' +V2S' V3'C' +V4'S'}du,
4
Vol.
238.A.
Vol. 238. A. 46
where
U1 = 2au2s2-- (1 -+2) Uc2- (1
-a
2) u2-us2 "Ue-2u
U3 - ---e-2u) -2 c ( --2)
2(2s US2 (C2- 1))-(- s2,
UT4=- (2u- -e-2") (C2+ ) U-aU2,--(1--2ff) (2 - 1) u,
V2 - (2u +e-2u)) Z- ( -a2) u2s2-au- (1-2- Sa) [Ca (c2- l)- (2+ 1 4- e-2) S2],
V= 1(2u-l+e- 2) a2-U2(C2 )- (
- 2) US2
V34V
2
=
(2u-1 +e-2I ) us2(2-1 2)12--( --2S) )e2, -
-
,??
_
e)2?>
IaU(C2)
ujcz -- uS2 --(1 -2o)u(C2 - 1).
.
si2 +n- i (2n +)! (2-2)!
Let
Let
f
f
oO gl
du= I,
ze-O f ~ 2u
du - J1,
-I---
and denote with primes the corresponding integrals in which the Z is replaced by Z'.
Then provided the order of integration and summation may be changed we obtain
the series
-
00 A
+
XI (A' + B'p2) p sin nO,
~nl~~ I I1~ ~ (5.11)~~
x.- + X4 =1(An
' + Bp2) p sin nO,
where
C
{2
2(2n)! (Inn -1-C2n)
(I--12n) C2--D2
-IzgC2n 2 a+Jan- +J2-)}
2n--(
1)2, (5.12)
tan1 --2 rr)
+- I2n + C2n) - n(I22n-1- 2n-1) -aSS2n -2 D2n- 1+J2n- 1)
-l1)!
(2n --)
--
B2n = 2(2 1) {(I2n+l 2nl) S2 l 2n T2n(1 2n
2
2(2n+ 1)!
- 2aS+
Az2n+l-(l= -a2) I 2n+
1 -( a2) 1 a2n
(2n + 1)!{(
) C2n- )--l)-(D2 13
:=
2n1+l 2(2n1-2n{(I 2n+2-I2n+2)--l1(C2n+-I2+)-aS2n+
+ (1-2<r)
2(D2;z)}'-J2n1)- (CG2n+-I-l2
- ) 0pcos 0
X={--4P= -) (1 - 2P) p sin 0logp +2(
+ I [(A' +A")+ (B' +B n)p2]pn i n (5.14)
46-2
It will be noted that some of the integrals in the above coefficients are divergent at the
lower limit. The convergence of the integrals (5-8) has been discussed (see Howland
and Stevenson 1933), but the coefficients we shall use require some investigation. The
divergent integrals occur when n is zero or unity in the coefficients A', A' AA',B". The
first and second of these, since they multiply terms producing zero stresses, may be put
equal to zero. Consider A', this is
and I--
CCl-1+ f -(osh-
,---- --1)u du
.------ a
which converge.
In the other coefficients one other combination occurs which has to be considered.
This is
D' _ -J, - f (cosh2(a~u- ) ue-d2u,
which presents no difficulty.
GO (-_n-1 p \n
- =1 n sinn(-)
K
where lo, 11 no, nl, p0, may be taken as zero while the other coefficients are
etc., the sign of a being changed throughout. x X3 are obtained in a similar manner.
This leads to the expressions:
n=O (2n)!
00
and so on, the integrals (5-10) with their integrands multiplied by cos flu being denoted
by italics,those multiplied by sinflu, by italicswith bar. Primes as before will be used to
denote the corresponding integrals in which 2 is replaced by Z".
Then
L2n
L2n S2n
!2,2n__(
{
.L + 02)C2nC2_ -, (
(. 7I__2n_-- 1T-na n_S
2-n= 2
2)
221
+n(C2n-l1+,2n)-1n(C2n-2 +22n-2 D2 + -2 1+J2n-2)
+ 1 12nl12
+- ({8(2n)
22-+-)-2-2n (82n-1 -
T2n-1)-(X(C2n-12n)}
(
(--i) i -221 (- - 2n (5n-23)1}
S2 1 III
(2n + 1). l 2n+I2n- 2'2n-
l1+ n2n--
M2n (n
2 (2n2 ){C2n+ 1 +2n+ 1- (C2n 1+ 2n -+ D2n- + 2n) - S2n+ 1
1)!
-(+1-20-) (C2n+i2n)},
M~(2n+ 2-{ 'I-2n+2- (S2n+ T2n+ 1)_ (C2n2 - 2n+2)(1 -2o)S2n+1};
--(1- 20')
[aC2n-1 I- (n -2?)S2n-2 -S2n-
I-
1 T2n - 2+ 2n-1] },
N'
2n+
(2n- + 1)
2n-+I - 2-
+2({2CS2n+ a2 ?D
2n-n(C1-{2n1+ I-- TI2122n- n-1 + 2nl
2
n-2} 1) (5.24)
2(2n-- -)!
+ + J2+
Pnl = (2 2)! {2n2 2n +2 -2(C 2nl- D2n +2n- 1)
-L2n = { + 2a T2 -1 -- 2) C2n
-+- I
( l-a2)n +
-(- l- l)
(2n)
-
l(C2n-2- 2n-2 --IS2n+ J2n-2)
2n-2 1
-L" =-^g
-- {2aC& -aD2-(1 a2) S2n+1 +4an}
L(2z+ ,_2n +l S(2n2-1)-S 2n+1
l-0'2nl}, _
+2(2n)}2*) ( 5, 25)
- 0') -- 2- (c2n -
-M2n == 22- (I-- n--,
! { 2n+l (1XS2
+ l1 I02n+1+ Ir)0.n2 -22n-I2n + 2n)- 2n+ n1' ,
-( (-f)?{X(z(C2n-l-2n I) - 2n
{2'- -2n-1 --22n-}_
- }((2)'+(2n -- +)- n+ S n+
M-,.~ 1
2(2n+~l)!{G~+-I 2
S2) (
nI-T),
a-2(nn- -
= (2n1-
--N2'n+l 2an+l +l aT2n--(1
(l +2) C2+--(1 - --2) I2n+1}
+
(2n+){i
+ Q2(2){2n-I2'n-2(2n-1-D2n- 2n-1 l
l+-2n-l)-?S2n} t (5.26)
C
- t
1) 2n 2-2 C2 n
I2n - n_) -l
1n- (2 i )
(2
(2n+t-1)!
= 2(2/
--Pn (
+1) {n+1--S2n-- T2n) -~ +l+n+l)
Sn}' (-2-)
1
I
/(2n _2 2ni- _ {_
-(1-20f) (6C2n+l-I2n+1)}
The second expansion for a longitudinal force has therefore the form
Pb o
= -)(( { -
PX
47r(l ) pn[(ln +mnp2) cos nO+ (nn+-pnp2) sin nO]
In the first few coefficients of the series we find divergent integrals of the same
character as those found in (5-14). These have been investigated and no unusual
combination is found. We omit the details here as also the discussion of those of the two
following expansions.
at (o, h) expandedabout(o, h)
(c) Transverseforce
The transverse force is one acting perpendicular to the parallel edges of the strip, i.e.
in direction OY. If it acts at the point (o, h), it will have as its stress function
Pb
X" 4ff 2c-) (q- a) log p + 2(1--cr)O0 I5"}, (5'28)
( -cr)(I
4(1-P O
cr
where w here- x-' %%3+
and ,,
"usS -(s ucC (P-P) CO du,
X2 =-y
and=u ---sC (P- ' -- P2) d
cos uudu,
t (5-29)
x u 2cC -(c- us) S (PI + P2) cos
.......S
usdu,
X3=j --'a't
with P1 = 2(1-)+u(l-)]
P2 = [2(1- o)+u(l+a)] e-u(1l+),
(5.30)
P = [(1-2r) )+u(1-a)] e-u(-a )
The reduction of 0" to its final series form is carried out in the same way as in case (a).
We omit the details of this and give the stress function in its required expansion. This is
x"=Pb
4((1- -2or) pcosOlogp+2(1--) OpsinO
- (1-2 -)
[(I2l
- C2n-l)
+-
S2n-l (--) S2-2
-2 S
T2n-2]},
= 0) I2n+1- (1 -
2) C2n+l + (n-1) D2n-lI2n-lJn
C2n+l (2n 1)! {(l- (C2n-1
D2n
= I2n+l) +S2n+l+ (S2n-T'2n) + (1-2) S2n},
2(2n+ li{-a(C2n+l+
Vol. 238.
v
A. 47
C2= ( {(n - 1)
a-C2"n-1+ nal'- 1+22^XC'n-2+5aD' - (1 +--a2) SI
+(n-)
(n- )(2n +D_ln +I2n +J- ) +2 a2- (n -i) aS2n +aT (5.33)
A force P acting at (-a, -Ih) in direction O'Y' has the stress function
+
(2n+
22
( 2)=! 1 'a)log )
(+ 2( -) t , (534)
)(-[n{(C
where the accented co-ordinates are those used in case (b). The function i"'is similar to
D"but with 6', i.e. - (5+f), for 6, etc. In fact, the process of obtaining the required
series follows closely that used in (b). We shall content ourselves with giving the final
form. We find that
-
X=
2n2 b 1 (x2
((12o p- y) cos -- (z + w p2) sin nO]
n=O
yn = -Yn,
-
(5.37)
z = ( )- {(n--2) flsin n-- 2na cos n)}/Kn
2n(n- ),
n =- Wn;
= 1
X +l(2n+l(1-2)2n+l (1 C2n+l+ (n-- ) (C2n-I2n)
-
+2(n-21) (C2n_ --
1 -D 2-2n
)1 2n 2n-1) + S2n+l- (n --) (aS2n -( T2n (5.38)
-
= 2(2n + + - 2n+1-I2n+1I+ Jl2n+1)
2) {2n+ 2 2n+2 2 (C2n+1
-Y
-Y2n+l+1-D
+ (1 -- 2) (C2n+l-I2n+l)};
1 = -- -) Cn 2D2n - (nr- ) (1
2Z2n+ (2a 11 {2C2+1- 72n- a2) S2n+1
-
(2n -)! Sn + (i) n- - T2n (540)
-
-(1-20f) I(X-2n C2n) nS2n- 1+52n--T2n- }
WI I + 1 2nD2 2
_ +1l-(c2n-D2,.-7I2+:-,2)--
(c2-7.-15 -2f)
_
2n2
Z2n+l (2n+1)!( 22n12n+I
--321)S2n2n
+
+ (n--
n -)f{+-(2
_,
l-?2 1
2n-1)
S2
( -+2)
'n-I-
-.,
n?2
n-1)'
- +41 (5.41t)
(5
= +(1-2o)+2)! {a(2n-2-2n+2)
[a(C+2nI2n+2+(2
nl+ +2n+ _2n-22n2 2n}n l
W2)+l 2 + 2n) ---_-
n2n+1 - 2}- - (
2n+)
W2n
= 2-(2-1)1 Z+i+'i+i(C+ -+ J2)-a.+ (1-2o-) (-' l'-, J,
The general case when the circles are arbitrarily placed in a strip can be considered
with the aid of the expansions obtained in the previous sections, provided there is
symmetry about a point on the axis of the strip. The algebra, although straightforward,
is very heavy, and we shall content ourselves with the investigation of two special
problems. The first of these is when the centres of the circles lie on the axis.
case
(a) First symmetrical
Let the centres be at the points (0, 0), (--a, 0), the strip as before being defined by
y b-+b. Using the notation of the previous sections, the centres are given by 6 = 0,
function is Xo X-X ( 1)
Xo- X-X 6-1
where X is given by equation (5-14) and X' by (5-27) after putting x zero.
Consider X first. Since a is zero, the coefficients (5-12) and (5-13) have simplified
values while the integrals (5.10) are related by the equations
Cn= I, Dn Jn, Sn = Tn= . (6 2)
Consequently all the coefficients vanish except
n+l = (-+
+ (2 (I-I2-l 2 2 J2n-l)
-){-2
+(1-2ar) (212n--2n2_l-IJ2n-_)}, (6.3)
2n+l = 2(I2n+l -J2n+l)--(1--20) I2n+l}'
(2n+?2)!{I2n+2
yb
p P
X4 (.,x a
-,?
FIG. 8
Taking the simplified X' in the same way, we find that K- f and = Te
T and the
appropriate values of the coefficients are
(_12 ,m ()n
2 2n )n+l(n-)
-1)2-) 2-1'
(2n m2-- m2n 2(2n + 1) f2n+1
(-)n(2n -1) (-)"+1
>
n2n+l -4n(n)2 P2n+l P2n+1 4(n_l)2+2, (6-4)
12n (_ )n+1
"2n - 2n+ 1
2(22n- 1) 4n4-2n;
the others vanishing.
The integrals (5-22) have their particular values and the remaining coefficients
derived from (5-23)-(5-26) become zero with the exception of
2n (2n)! {22n+2ni2n-n(2n-2+J2n-2)
+(1--2o') (22n_l-J2n-2-[2n- ]]'2n 2)},
N
(2ntl)!{22n122(2n2-1 2-)
-
P2n-f-l = (2n 2)2n 2-2 (2n+ 1 J2n+ )-( 1-2C) 12n )1};
47-3
0O
OB2 -_ _2n+2
-1
_)n+V
2n
+1 (2
(2n +2) f2n+3
2n+3
+
U2s+- p-2s-1 sin (2s+ 1) 0 f+ (2s+lA2s 1A^p2) p2ncos 2nO
n=O
o0
2sA 2n = 1 - (2n-
2('2n+2s + 2nS2s) 1)) (I2n+2s-1+2n+2s-1)
+I2n+2s-l)
(2s- 1) (2n) !{2(I2+2s12n+2s
2-1\2n-1
_fT +J2n n2-
-(J2+2s2-1HI 2 +2?-)}F ()n?sl(2fl?2, i ___+2
1
2A' __ + +, ,,_n-s(2n, + 2sl\ ) 2n+2
2n 2n+2s+l1
(2s-1)! (2n + 1)! 2n2n+ 2 +2 2n+2s+2
1 (6.13)
-
(( n 2n + 2s\ 2n
2n+ I9 2n+2s+1'
*2s
(1A !2n)
!2 {2'2n+2s+l -(2n- 1) 2n+2s -J2n +2s}
2+(2n)--(I2n-1
+2n-[2(( 2n-1 +-I2n-1)--(J2n-l +J2n-1)]}+ (2n
- 2n+2)} -
D2n+ (2n + 2) 2n+3- (I2n+2
-2 2n+3
n=0O
Jr2n()n2 + 2s-1 1
2
2n 1-l s2n+2s.
=
( 2) (4-'lS 2+2+2 ({ 2n+2s+-2) (I2n+2s+l+I2n+2s+l)
22n ((2s) +2 + 1
~~~~2sh)~~~~ /
-
2'n+1 2) ( {2 n1 +2 +3-(2n+2s+2+J2n+2-s+2)}
(2s) ! (2n + 2)!i
+( )n+(2n +2s+2 _-_
2fn+2 !12n+2s+3
-(J2\ (J2n+2s+3~T^2n+2s+3})\\
++ j2,,+s)}+ ( _),n-ts ) 2n+2s+3 (2n+3)
2n - 3 J 2n+2s+4
It is clear that the functions we have obtained are even in y but odd in x (if x, y are
co-ordinates with origin at the centre of symmetry). They are therefore suitable, with
the addition of the appropriate X giving the infinity conditions, for a problem such as
that of a strip under pure tension at infinity. It is obvious that other sets of functions,
e.g. even in x, odd in y, can be obtained should they be required by choosing the appro-
priate expansions from 5 (c) and 5 (d).
(b) Secondsymmetricalcase
We may next conveniently consider the other symmetrical case in which the circles
are situated as shown in fig. 9.
yI b
FIG. 9
There remains one more configuration of circles in a strip which may be considered
with the expansions we have developed. The circles, four in number, are situated in the
strip as shown in fig. 10. The algebra is necessarily heavy, for we have no simplification
x,
from the
such aX( is the contribution ingularity at , , , 3, 4.
traightforward
Suppose and
act at th no serious dificulty. We content
cpresents shall ourselves by indicating ow the fnda-
mental functions
wilacd
byieing
can be set up. Four sets of functions
X3 ----r?3 4X-4----L X4
)
der from one27)
is obtained two
by stress of the si or
functions
changing
FIGC 10O
FIG. 10
First let
First X= %X(1-X-
- X(3) +X(4)
A(1)-x(2)-x(3)- {-V (7'1)
(7.1)
wherex(r)is the contributionfromthe singularityat 0^, r = 1y2, 3, 4. Supposeforcesto
act at the centresin directionsOrX,.
X^1)
X(l) is given
is given by (5?14).
(5.14).
X(2)is obtained from (5 27) when aGis replaced by -o in the various coefficients, and 6
is replaced by 71-6.
X(3)
X3 isis given by (5-27).
given by (5-27).
%(4) is obtained from (5 27) by changing the sign of 0 throughout, with 4- = 0.
CONCLUSION
Although the functions set up have been obtaine(l with the object of providing
machinery for the solution of problems in generalized plane stress, they are of course
suitable for the solution of the biharmonic equation in general. The only restriction is
the symmetry of the boundaries and the conditions on them. When the boundaries are
partly straight lines, the functions obtained will have to be adjusted to the appropriate
conditions of the new problem, e.g. the slow motion of a viscous fluid (cf. Howland and
Knight I932).
I wish to express my thanks to ProfessorG. Temple who has read the paper in manu-
script and made a number of suggestions. Also to the Ordnance Committee for per-
mission to submit the paper for publication.
SUMMARY
No numerical work is included but the expansions of the necessary functions have
been determined. They may be used for any problem where the biharmonic equation
has to be solved with the appropriate boundaries. The method of solution when the
required functions have been established is indicated.
REFERENCES