The London, Edinburgh, and Dublin Philosophical

Magazine and Journal of Science

Series 7

ISSN: 1941-5982 (Print) 1941-5990 (Online) Journal homepage: http://www.tandfonline.com/loi/tphm18

XXVI. Heat conduction in composite circular

cylinders

J.C. Jaeger

To cite this article: J.C. Jaeger (1941) XXVI. Heat conduction in composite circular cylinders,
The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 32:213,
324-335, DOI: 10.1080/14786444108521306

To link to this article: http://dx.doi.org/10.1080/14786444108521306

Published online: 29 Apr 2009.

Submit your article to this journal

Article views: 24

View related articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tphm18

Download by: [University of California, San Diego] Date: 29 December 2015, At: 08:59
 [ 324 j

XXVI. Heat Conduction in Composite Circular Cylinders.

By J. C. JAEGER *

[Received September 28, 1939.]

Downloaded by [University of California, San Diego] at 08:59 29 December 2015

THe, conduction of heat in composite linear and spherical solids has been
extensively studied ~. The object of this paper is to present some results
for circular cylinders of two materials ; it is to be regarded as a sequel
to two papers, " A Problem in Conduction of H e a t " :~, and "Some Two-
dimensional Problems in Conduction of Heat with Circular S y m m e t r y " §,
which will be cited as I. and II., respectively. The method used here is
that of the Laplace Transformation as developed in the paper I., which
may be consulted for a fuller exposition; the solutions given here are
formal, but in all cases it can be verified that the solutions obtained in
the form of line integrals satisfy the conditions of Theorem II. of paper I.,
and the verification that they satisfy the given differential equations and
boundary conditions follows as in § 9 of that paper.
The algebra is complicated, but is greatly simplified by the systematic
use of cylinder functions ; the principal results used are collected in § 2.
In § 3 the hollow composite cylinder with zero initial temperature and
boundaries kept at constant temperatures for t ~ 0 is discussed; in § 4
the cylindrical surface source in the hollow cylinder with boundaries
kept at zero; in § 5 the corresponding results for the solid composite
cylinder are given without proof. A brief discussion of the region outside
a circular cylinder is given in § 6.
2. I t is convenient to introduce functions defined as follows :--
D(x, y)=-Io(x)Ko(y)--Ko(x)Io(y), . . . . . (1)
~r+s
I)r,,(x, y)---- axray~D(x, y) . . . . . . . . . (2

where for brevity Dl(X, y) will be written for D1, 0(x, y).

• Communicated by the Author.

t Cf. Carslaw, ' Conduction of Heat ' (2nd ed., Macmillan, 1921), §§99, 105,
108 ; Churchill, Duke Math. Journal, ii. p. 405 (1936) ; Math. Annalen, cxv.
p. 720 (1938).
:~ Carslaw and Jaeger, Proc. Cambridge Phil. Soc. (in the Press).
§ Carslaw and Jaeger, Proc. London Math. Soc. (in the Press).
 Heat Conduction in Composite Circular Cylinders. 325
These functions are connected with the cylinder functions *
C(x, y)=Jo(x)Yo(y)--Yo(X)Jo(y) . . . . . . (3)
~r+s
C~,,(x, y ) = Ox, Oy---------~C(x,y) . . . . . . . . . (4)
by the relations
D(ix, iy)=--½7rC(x, y) . . . . . . . (5)
i~+~D~,~(ix, iy)-=--~rrC~,s(x, y) . . . . . . . (6)
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

The following properties are trivial:

D'+2'~(x' Y)~-x1 D ~+l'~(x' y)--D~,~(x, y)=O . . . . . (7)

1
D,,~+~(x, y)~- ~D~,s+l(x, y)--D~,~(x, y)----0 . . . . . (8)

D(x, y)------D(y, x) . . . . . . . . . . . . . (9)

Do, i ( x , y ) = - - D l , o(Y,X) . . . . . . . . . . . (10)
1
~)1, 0(X, X ) = - - . . . . . . . . . . . . . . . (11)
X
D(x, y)Di(x , z)--D(x, z)Dl(x, y ) = 1D(z, y) . . . . . (12)

D(x, y)D1, l(x, z)--Do, l(x, z)Dl(x , y ) = 1Ol(Z , y). (13)

1
D(x, y)D~, l(x, y)--Do, l(x, y)Dl(x, y ) - ~ - - . . . . . (14)
xy
1
lo(x)D~(x, Y)--Io'(x)D(x, y ) = xIo(y) . . . . . . . (15)

Io(x)Dl(y, z)--Io(z)Dl(y, x)-~Io'(y)D(x, z) . . . . . (16)

Io(x)D(y, z)+Io(y)D(z, x)÷Io(z)D(x, y)=O . . . . . (17)
Equations (15), (16), (17) are valid also if the I are replaced b y K.
3. Hollow cylinder of one material from r = a to r = b and of another from
r=-b to r=-c. The initial temperature of the whole zero. The surface r=-a
kept at Vo, and r = c at zero, for t:>O.
Let the temperature, conductivity, specific heat, and density in
a ~ r ~ b be Vl, K1, Cl, and Pl, and let Kl=K1/clp 1 ; let the corresponding
quantities in b ~ r ~ c be v~, K2, ca, P2, and K2.
* It should perhaps be remarked that C2a(a, fl) has been used [Titehmarsh,
Proc. Load. Math. Soc. (2) xxii. p. 15 (1924)] for the function J21(a)Y23(fl)
--Y23(a)J23('fl), but notations for cylinder functions are not well standardized,
and the notation of (3) and (4) is particularly convenient for the type of problem
under consideration.
 326 Dr. J. C. Jaeger on Heat
Then we have to solve

,qk~;~-~t; ar/=-~ -, a<r<b, t>O, . . . . (18)

IB2v~ 1 av2\ Ov~

K2(~r2-]-rW)~--~., b<:r<:c,t>O . . . . (19)
with vl=vz, r=b, t>0 . . . . . . . . (20)
av~ ~v2
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

Ki~r=K2~r, r=b, t:>O . . . . . . (21)

vl=Vo, r-~a, t>0 . . . . . . . . (22)
v~'-O, r=c, t>0 . . . . . . . . (23)
vl=0 when t ~ O (a<r<:b) ; v2~O when t----O (b~r<:c). (24)
Multiplying by e-~t, p > 0 , integrating with respect to t from 0 to ~ , and
writing
Vl=f: e-~Vldt, ~2=f: e-'tv2dt,
we obtain the subsidiary equations
d2~1~ 1 d-v1
dr 2" r ~ - - q l a ~ l = 0 ' a<r•b, . . . . . (25)
d2~ 1 a~ 2 3- "

dr'--'~ r'-~r --q2 v2= u , b<r<c, . . . . . (26)

to be solved with
"vl--~ V o/P , r--~a, . . . . . . . (27)

~0, r--~.c, . . . . . . . (28)

v1=v2, r~b, . . . . . . . (29)

K d'Vl - c~2
1~ =~h'2~' r-~b, . . . . . . . (30)

where q12=p/K1, q~-~p/K,,. . . . . . . . (31)

The general solution of (25) which satisfies (27) is

71---- V° D(rql' bql) t-A D(rql, aql),
P D(aql, bql)
and the general solution of (26) which satisfies (28) is
v2--~BD(rq2, cq~).
Equations (29) and (30) then require
AD(bqi, aql)--BD(bq2, eq2)~0,
KlVo
1~
KIqlADi(bql, aql)--K2q2BDl(bq2, eq2)= bp D(acl 1, bql)
 Conduction in Composite Circular Cylinders. 327

Solving for A a n d B and substituting, we obtain finally

- Vo
Vl:p- ~ ~NiDl(bql, rqi)O~bq2, cq2)--K2q2D(bql, rql)D~(bq2, cq2)}, (32)
K1Vo ,
-v~= bpA(p)Dtrq2, cq2). . . . . . . . . . . . . . . . (33)
where
A(p)=KiqlDl(bql, aql)D(bq ~, cq2)--K~q~D(bqi, aql)Dl(bq2, cqe). (34)
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

To evaluate v2 we have from the inversion theorem for the Laplace

Transformation :
K1V° ()'+~® eUD(r~2' ~ ) d2 . . . . . . (35)
v~: 2~rib J),-i® 2A(h)
where ~tl=~V/(X/K1) , and ~=~/(2/K2). . . . . . (36)

Fig. 1.
8' B

' A

The integrand of (35) has simple poles at 2----0 and ~--KI~s2~ where
± a e , s = 1, 2 . . . . . are the roots of *

KiCI(b~ , a~)C(Kb~z, Kc~)--KK~C(b~, a~)Cl(Kb~, Kc~)=0, . (37)

where K = %/(K~/K~). . . . . . . . . (38)
Consider the integral of the integrand of (35), taken round the contour
of fig. 1, consisting of the line AB distant ~ from the imaginary axis and
portion of a circle/" which does not pass through a n y pole of the integrand.
The integral over F tends to zero t, as its radius tends to infinity through
* For details of the method of proof see II. § 2.
t C1 is written throughout for C1, 0.
 328 Dr. J. C, Jaeger on Heat
a sequence of values which avoid poles of the integrand. Thus the

fpoles
Y+~ in (35) may be replaced by 2~i times the sum of the residues at the
of its integrand.
The residue at )~= 0 is
b log(r/c)
K 1 log (b/c)--K 2 log (b/a) . . . . . . . (39)
To evaluate the residue at ~ = - - K I ~ z we have
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

2~ d~--(~
~) =b(K~/~t~--K~/~)D(b/z~, a/~a)D(b/~, c/~)
+b/z~(Kl~K~)Dl(b/~l, a/~l)D~(b/ze, c/z~)
+c/~{Kg~Do, ~(b/~, c/~)D~(b/~, a/zl)--K~lz~D(bbh, a/~l)
x D~, ~(b~, c~)}
-[-a/~{K~/~D(b/~, e/@D~, ~(b/~, a/~)--K~/~Do, ~(b/z~, a/~l)
× DI(b~2, c~2)}. . . . . . . . (40)
Now, when 2 = - - ~ % ~ ,
K~/~D~(b/~, c/~e)D(b/~e, e/~) ___C(~b~,, Kc~s) _ K,~C~(~b~,, ~e~,)
: a , say,
KI~D~(b/~, a/~) D(b/z~,a~l ) C(b~,, a~,) K~C~(b~,,a~,)
(41)
Introducing this in (40) and using (14), we obtain

--3Vj~=_~. - •

Kia K21 K , (42)

-~-b \_~2/~£12(Si--U2)O12(b~i' a~i)°'@ b no-.]h=-:¢s. °
1

2bK2F(~,)C(Kb~s, Kc~s)C(b:¢,, a~s) '

where
1
- - =C2(Kb~,, Kc%){ ¼K~b2rr2(K~K2-- K1)% 2C2(b%, a% )
F(~,)
~¼1r~b2Kl(K1--K2)%2C~2(b~,, a%)--i-K1K2}--K22CZ(b%, a~,). (43)
Thus, the residue of the integrand of (35) at )~= --K% 2 is
--~bK~e-'~"'tC(~:r%, Kc~,)C(Kb%, ~:c~,)C(b~,, a%)F(%).. (44)
From (39) and (44) we have finally
K 1 log (r/e)
v~=V°K 1 log (b/c)--K~ log (b/a)
o0
--TrK1K2V 0 2: e-K'~gtC(Kr%, Kc%)C(Kb%,Kc%)C(b%, a%)F(%). (45)
$51
 Conduction in Composite Circular Cylinders. 329

In the same way we find from (32), using (12),

T K 1 log (b/c)--K 2 log (b/r)
vl=v°K-~l I-@ (b/e)--K 2 log (b/a)
o0
--TrK1KsV 0 1 e-~'tO(r~, a~,)O~(Kb~,, Kc~,)F(%). (46)
s=l
For the same problem but with r = a maintained at zero and r-=c at V1
for t~O we find, in place of (32) and (33),
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

Vl= K~V1 D(rql, aql),

bpzJ(p)
- V 1
Vs--p-ff~p) {KiqlD(bqs, rq~)D~(bq~, aq~)--K~q2D(bqv aql)D~(bqs, rq~)},

where A(p) is given by (34). Hence

K s log (r/a)
Vl~
- - V 1 K 1 log (b/c)--K~ log (b/a)
+TrKs2Vi Z e-K'as'tC(ras, a~s)C(Kb~s, Kco:,)C(b~s, a~s)F(~,). (47)
*=1
vs=V1 K1 log (b/r)--Ks log (b/a)
K 1log (b/c)--K 2 log (b/a)
oo
+~K2W1 Z e-~uC(Kr%, Kc~,)C~(b~v a%)F(~,), (48)
s=l
where tne ~s are the roots of (37).
The solution for the cylinder with r = a maintained at V o and r=c at
V1 for t > 0 is obtained by adding (46) to (47) and (45) to (48).
4. Hollow cylinder of one material from r = a to r=b and of another from
r=b to r==c. The surfaces r-~a and r=c kept at zero for t > 0 . A n in-
stantaneous cylindrical surface source * of strength Q at t=O over r=r',
a<r'<b.
Here in the notation of § 3 we have to solve (18), (19), (20), (21) with
vl~-0, when r=a, t > 0 , . . . . . . . . (49)
vs----0, when r : c , t > 0 . . . . . . . . . (50)
and vl-~ul+wl, . . . . . . . . . . . . . (51)

where u~= Q - ~ exp [ - (r~+r'2)/4Klt]Io (rr'

47rKlt \2~1t] , . . . .
(52)
and w~ satisfies
/~2W 1 1 ~'t6'i "~ OWl.
~l~r~ O r ] = Ot ' a<r<~b, t>O. (53)
and Jim w~=-O, a < r ~ b . . . . . . . . (54)
t---)-0

* of. Ii., § lo.

SER. 7, VOL. 32, NO. 213--OCT. 1941. 2A
 330 Dr. J. C. Jaeger on Heat

I t is known that *
El= Io(r'ql)Ko(rql), r>r',
(55)
Q
2rrKiKo(r'q,)Io(rq,), r <r"
J
- -

The subsidiary equations, formed as in § 3, are

d~ 1 d~ 2 - 0
-~r2"+ r ~ - - q ~ = ' b<r<c, . . . . . (56)
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

1dr ql~wl= O, a<r<b, . . . . .

d~-Wl2dr ~-rl d~ (57)

v~-----O, r=c, . . . . . . . . (58)

Vl=0, ~'--a, . . . . . . . . (59)
~1='~, r=b, . . . . . . . . (60)
Ik cl~l --K d~ r=b, . . . . . . . (61)
1--d7- 2dr,
where ql and q2 are defined in (31).
The solution of (56) satisfying (58) is
v2=CD(rq2, cq2). (62)
As solution of (57) we take
wl=AD(rql, bql)+ BD(rql, aql). (63)
Then, using the vMue (55) of ul, (59) requires
Q
AD(aql, bql)+ ~ Io(aql)Ko(r'qI)=O'
~nd (60) and (61) give g - %

CD(bq~, cq~)--BD(bql, aql)--- -~K Io(q,r')Ko(bql),

KlqlQ
Keq~CDl(bqe, cq~)--KlqlBD, (bql, a~l ) = 2qTt¢---~I°(r'ql)K°'(bql) + ~ - ~

Solving for A, B, C and substituting in (62) and (63) we obtain on reduction

QK1
v2= 2¢r•lb.4 (p) D(r'ql, aq~)D(rq~, cq2). . . . . . (64)
where A(p) is given by (26),
-vl-----2Tr~.Q(p) aql)[KlqlD(bqv cq2)Dl(bql, r' ql)
_Keq~Dl(bq~ , cq2)D(bql , r'ql)] , when a<r<r' <b, (65)
and when a < r ' < r < b we interchange r and r' in (65).
* The result follows from Watson, ' Theory of Bessel Functions, § 13.7.
 Conduction in Composite Circular Cylinders. 331
From (64) and the inversion theorem for the Laplace Transformation

QK 1 [y+i® e~tD(r't% a/~,)D(r/% c/x2)Adz~

(A) (66)
v2= 47r2iK1b Jy-i~
where, as before,

The integrand of (66) is a single-valued function of ;~ with simple

poles at 2~:--K1%~, where the % are the roots of (37).
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

Then, as in § 3, it follows that the line integral in (66) may be replaced

by 27ritimes the sum of the residues at these poles. These are immediately
evaluated on using (42) and we obtain
1
v 2 : - - ~IrQK 1K 2 Z %2e-K~s'tF(%)C(b%, a%)C(r %, age)
XC(Kb%, Kc%)C(~r~s,Kc%). (67)
Similarly, from (65), we obtain, when a < r < r ' ,

vi=~Tr bK2Q 27 %ae-~sUF(%)C(b%, a%)C(r%, a%)C(Kb%, Kc%)

s=l
x {K~C(Kb%,~c%)Cl(ba~, r'%)-K~KCd~b~ ~, ~c%)C(b~, r'%)}.
This becomes, on using (41),
v : ~ - - 17rK~K2Q ~ %2e-K~dtF(%)C(r%, a%)C(r'%, a%)C2(Kb%, Ke~), (68)

and this, being symmetrical in r and r', holds for both a < r < r ' and
r'<r<b.
The solution of the problem of an instantaneous cylindrical surface
source of strength Q at r', b<r' <c, is obtained in the same way.
We find
QK2' D(r'q2 , cq~)D(rql , aql), . . . . . . . . . . (69)
(p)
v2= QD(rq2' cq2) ~K D'b aql)Dl(bq~, r'q2)
21rK~A(p) '/ ~q~ (q~,
--K~q~D~(bq~, aql)D(bq~, r'q~)], when c > r > r ' > b , (70)
where A(p) is given by (34).
Then, from the inversion theorem, we obtain
~QK~ Z e-~2~%~F(%)C(b~, a%)C(r%, a~)
Vl= 4~2 ~=1
×C(~r'%, ~c%)C(~b%,~c%), (71)

V2~---
wQK~2~I
492 s ~= l %ee-~,~.,uF(%)C2(b%,a~s)C(~r~s, ~c%)
×C(~r'gs, ~c%). (72)
2A2
 332 Dr. J. C. Jaeger on Heat
If we put Q-~27rr'f(r')dr' in (67), (68), (71), (72) and integrate with respect
to r' from a to c, we obtain the solution of the problem in which the
initial temperature of the cylinder is f(r) and the surfaces r = a and r = c
are kept at zero. Assuming that f(r) is such that the orders of integration
and summation may be interchanged, we have

Vl---~-- 7r2KIK2Z %~e-~'tF(%)C(r%, a~)C(Kb%, ~c%)g(%), . (73)

s~l
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

1
v2=-- ~ KxK~ ~ o:s~e-~,~uF(~s)C(b~, a~s)C(~r~s, ~c:~)g(~s), , (74)
where
g(%)----C(Kb%, Kcc~)r br'f(r')C(r'%, a~s)dr'~ C(b%, a%)
Jg

× r'f(r')C(Kr'~,,Kc~)dr'. (75)
b

5. The composite solid cylinder of one material, vl, Ki, cl, 01, K1, for
0 <~r<a, and of another, v2, K2, c2, P2, Ki, for a < r < b .
The results corresponding to those of §§ 3 and 4 are as follows :--
(i.) Initial temperature zero. The surface r = b kept at Vo for t>0.

vl----Vo+TrK2~Vo ~ e-~suXo(r%)Jo(a%)C(s:a~, Kbgs)Fl(~s), (76)

v2:Vo~-TrKe2Vo Z e-~*'tJo2(a:cs)C(trr~s, Kb%)Fl(~s), (77)

where :~%, s--~l, 2 , . . . , are the roots of

K1Jo'(a~)C(Kag, Kb~)--KK2Jo(ao:)Cl(s:ao:, Kb~)=0. (78)
where K=V(K1/Ke), and

~-~1(:¢s)= 17rua2~,2C2(Ka:¢,,Kb:cs)[K2(K2K2--Ki)Jo~(aas)
+ Ki(K1--K2)Jo'2(a~,)]- K~2Jo2(a~s). (79)
(ii.) The surface r = b kept at zero for t>0. Instantaneous cylindrical
surface source of strength Q .at t-~O over r ~ r ' , O<r' <a.

v ~ = - 4 , Q K ~ K 2 ~ %~e-~s'tJo(r'~,)Jo(re,)C2(~a:¢ ,, s:bgs)Fl(~,), (80)

ve------ 17rQK~Ke ~ o~,~e-~'~s~tjo(r'~,)Jo(a%)e(~zae ,, ~b~,)

× C(~r~,, ~b~,)F~(~,), (81)
where F~(%) and the e, are given in (79) and (78).
 Conduction in Composite Circular Cylinders. 333

(iii.) The surface r=b kept at zero for t ~ O. Instantaneous cylindrical

surface source of strength Q at t:-O over r-~r', a<r' <b.

~QK22K1 ~ ~82e-K'~,'tJo(a~s)Jo(r~a)C(Kr'~s, Kb~s)

Vl=--' 4K2 s=1
×C(Ka%, Kb%)FI(%), (82)

V2___--- - ~QK2~'K1
_ _ Z ~,2e-~dtJo2(a~s)C(Kr%, ~b%)C(Kr'~ s, Kbgs)Fl(~,).
4K2 s= 1
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

(83)
(iv.) The surface r=b kept at zero for t~O. The initial temperature of
the solid f(r).
I 2
Vi=---~Tr K1K 2 ~ g~e-~,~SUJo(r%)C(Ka%, Kbo~s)Fl(O~s)gi(o~s), (34)
S=1

v2~ -- ~1 Ir ~KiK2s =X 1 ~2e-~'~dtJo(a%)C(Kr%,

" Kb~s)Fl(~)gl(%), " (85)

where

g l ( ~ ) : C ( K a ~ , Kb:~s)~:r'f(r')Jo(r'%) dr'
K2~¢ 1 rb
q- ~ J o ( a ~ 8 ) Jar'f(r')C(Kr'~, Kb%)dr'. (86)

6. The region bounded internally by the cylinder r-=a. From r : a to

r= b the solid is of one material, vl, Pl, cl, K1, K1, and for r ~ b of another,
V2, P2, C2, K 2 , / ~ 2 -

(i.) The surface r : a maintained at temperature V o for t~O ; the initial

temperature of the whole zero.
Proceeding as in § 3, the Laplace Transforms of the temperatures in
a < r < b and r ~ b respectively are found to be
- V o
V ~ = p 2 ~ ) (Klq~Ko(bq~)Dl(bq~ , rq~)--Keq2Ko'(bqe)D(bqx, rql)}, (87)

- K~V o , ,
V~--PG-d-~)
K°(rq~)'--,l- " . . . . . . . . . . . . . (88)
where
A(p)=K~q~Ko(bqi)D~(bq~ , aq,)--Keq=Ko'(bqz)D(bq~, aql), (89)
and q~= V(p/~), q~= V'(p/~).
v~ and va are determined from these b y the use of the inversion
theorem. This gives
K~Vo (~'+ ~® e~
v~= 2~rib J~,_i~ Xd(2t)"K°(rl~)d2' " . . . . (90)
where p.~= V ( ~ / K 1 ) , ~£2~ - V(*~/K2).
 334 Dr. J. C. J a e g e r on H e a t

T h e integrand of (90) has a b r a n c h point at the origin, so we use the

c o n t o u r of fig. 2 consisting of the line A B distant 7 from the i m a g i n a r y
axis, arcs AA'C and B B ' F of a circle F whose radius 1~ will t e n d to ~ ,
t h e lines CD and E F on which arg 2 equals - - ~ and ~ respectively, a n d
a small circle a b o u t t h e origin whose radius e will t e n d to zero. T h e inte-
gral o v e r / " tends to zero as the radius tends to infinity *. Also t h e r e are
no poles of the i n t e g r a n d within the contour. Thus, the line integral
in (90) m a y be replaced b y the limits of the integrals over CD, E F , a n d t h e
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

small circle as R-->oo a n d e-->0.

T h e small circle gives V 0.

Fig. 2.
B'
L
_ 8

¢
)

A' A

On CD we p u t 2-~Klu2e -i~ and on E F , 2-~Klu2e~: and obtain finally

v2=V0 + 4K~Vo (~ e-K,u'tdu _ _
~r~b Jo u 2 { ¢ ~ ( u ~ ( u ) ) [ K 1 C l ( b U ' au)( Kru)
--K2KC(bu, au) u, Kru)], (91)
where
¢ ( u ) = K f l o ( K b u ) C l ( b u , au)--~K~Jo'(Kbu)C(bu , au),
(92)
¢(u)=K1Yo(Kbu)Cl(bu,.au)--KK2¥o'(Kbu)C(bu, au), ~ "
and ~ = V(~/K2).
Similarly, we obtain
8KIK2V 0 e_~=, t C(ru, au) du
vl----V°-~ ~3b2 o f" ¢2(u)÷~b~(u) u3 . . . . (93)

* For the method of proof and a more detailed discussion of the procedure
of this section, see I.
 Conduction in Composite Circular Cylinders. 335
(ii.) The surface r----a maintained at zero for t~O. A n instantaneous
cylindrical surface source of strength Q at t--O over r ~ r ' , a~,r' ~b.
2K,K,Q (® e_~,~,t C(r'u, au)C(ru, au) du (94)
v,- ~ Jo ¢~(u)+~(u) u'

2----~-T Jo - ~-~~)- . . . . . . . . (95)

where
x(r, u)-~K,C(Kbu, Kru)C,(bu, au)--K2KC,(Kbu , Kru)C(bu, au), . (96)
Downloaded by [University of California, San Diego] at 08:59 29 December 2015

and ¢(u) and ~b(u) are defined in (92).

(iii.) The surface r = a maintained at zero for t~O. A n instantaneous
cylindrical surface source of strength Q at t--O over r ~ r ' , r' ~ b .
K2K1Q ~ e_~U,t C(ru, au)x(r' , u) du . . . . . (97)
v,~-- lr2bK2 o ¢2(u)+~b2(u)
K,Q (~ e_~uux(r,, u)x(r' u )
~ ) udu . . . . . . . (98)

(iv.) The surface r-~a maintained at zero for t>O. The initial tempera-
ture of the whole f(r). It is assumed that f(r) is such that the orders of
integration may be interchanged.
4K,K2 ;= -K,uu C(ru, au) . . du
vl= .2b~ Joe ¢~(--~)gt~; T, . . . . (99)

eKl (~ e_~.~ x(r, u) g(u)du, (1oo)

v2~-- ~rb Jo Ce(u)-~b*(u) . . . .

where
b • - t 7rbK1 f~
g(u)=
f a
r'f(r')C(r'u, au)dr ~- 2--K---~K u J r'f(r')x(r' , u)dr'.
" 1K2 - b
(101)

The University of Tasmania.