Journal of Engineering Physics and Thermophysics, Vol. 72, No.

3, 1999
A METHOD OF PAI RED I NTEGRAL EQUATI ONS
I N THE REGI ON OF LAPLACE TRANSFORMS FOR
SOLVI NG NONSTATI ONARY HEAT CONDUCTI ON
PROBLEMS WI TH MI XED DI SCONTI NUOUS
BOUNDARY CONDI TI ONS
N. I. Yurchuk, V. P. Kozl ov, and
P. A. Mandr i k
UDC 517. 968, 536. 24
On the basis of the method developed, the solutions of f our problems of mathematical physics are obtained
f or an infinite plate (a plane layer of thickness z = h) with assignment of mi xed discontinuous boundary
conditions ( BC) on one of the surfaces z = 0 of the plate and unmi xed BC on the other surface z = 17.
In t he pr esent wor k a devel opment and ext ensi on of t he met hod of pai red i nt egral equat i ons ar e suggest ed
to solve parabol i c part i al di fferent i al equat i ons with mi xed boundar y condi t i ons (BC). A circle r -- R on t he sur f ace
z -- 0 is a line of di scont i nui t y of mi xed BC. Mi xed di scont i nuous BC in pr obl ems of mat hemat i cal physi cs ar e t aken
to mean pr obl ems di f f er ent from t hose of t he Di ri chl et and Ne uma nn t ype. In ot her words, t he t er m "mi xed BC"
means t hat wi t hi n t he limits of vari at i on of one vari abl e (in this case, wi t hi n t he limits of vari at i on of t he cyl i ndri cal
coor di nat e r on t he sur f ace z -- 0 of t he pl at e) het er ogeneous (mi xed) BC of t he first and second ki nd can oper at e
in t he cor r espondi ng r anges of vari at i on of this vari abl e (see Fig. 1). On t he ot her pl at e surface z = h t he two t ypes
of unmi xed BC, namel y, ei t her of t he first or of t he second ki nd, ar e consi dered.
Below we show t hat a solution of t he heat - conduct i on equat i on for an infinite pl at e under t he act i on of
mi xed BC on one of its surfaces z -- 0 is r educed in a classical form to a solution of t he cor r espondi ng i nt egral
equat i ons in t he regi on of L- t r ansf or ms.
Pr obl em No. 1. It is requi red to solve t he di fferent i al equat i on
Orr (r, z, v) + r-lOr (r, z, ~) + Ozz (r, z, ~) = a- l ot (r, z, r) , 0 < r < oo , 0 < z < h, r > 0, (1)
where r and z ar e cyl i ndri cal coordi nat es; v is t he time; a > 0 is t he t her mal - di f f usi vi t y coefficient; O(r, z, T) = T(r,
z, r) - T O (To = const is t he initial t emper at ur e of t he infinite pl at e at T = 0).
Mi xed BC for Eq. (1) are given at z = 0:
O( r , O, v ) =f ( r , v ) , O<r <R, z =O, " t ' >O; (1.1)
- Oz ( r , O, z ) =O , R<r < oo, z =0, T>0. (1.2)
For f (r, T) > 0 we as s ume t he exi st ence of t he Lapl ace integral.
A BC of t he first ki nd is pr escr i bed on t he surface z = h:
O( r , h, T) =O, 0_<r < oo, z =h, ~>- 0. (1.3)
The origin of coor di nat es r = z = 0 is sel ect ed at t he cent er of t he circle r = R on t he surface z = 0.
Bel arusi an St at e Uni versi t y, Mi nsk, Belarus. Tr ansl at ed from I nzhener no- Fi zi cheski i Zhur nal , Vol. 72, No.
3, pp. 555-571, May- June, 1999. Original article submi t t ed March 20, 1998.
534 1062-0125/99/7203-0534522. 00 9 Kl uwer Academi c/ Pl enum Publ i shers
Region ~ <tr ' l < o~, z = 0
/ / / - - - - T @ ' _
( /
k l 1
"z
Re g i o n l r l a 0 , z =h j
_ , 2 r
Region 0<l r i <A ), z = 0
Fig. 1. Di a g r a m not a t i on f or r egi ons of b o u n d a r y condi t i ons in pr obl e ms Nos .
1-4 on t he s ur f aces of a n i nf i ni t e pl at e of t hi cknes s h. I n pr obl e ms Nos . 1 a n d
3 mi x e d BC (1. 1) a n d (1. 2) ar e pr e s c r i be d on t he s ur f a c e z -- 0, a n d i n
pr obl e ms Nos . 2 a nd 4, mi xe d BC (2. 1) a nd (2. 2) on it. Un mi x e d BC (1. 3)
a nd (3. 1) can ope r a t e on t he ot her s ur f ace z = h.
Ta ki ng i nt o account b o u n d e d n e s s condi t i ons f or t he t e mp e r a t u r e on t he axi s (r = 0, 0 < z < h, T > 0) a n d
at i nf i ni t y ( r - ~ ~ , 0 < z < h, ~- > 0) a n d a ppl yi ng t he L- t r a n f o r m t o Eqs. (1) a n d ( 1. 1) - ( 1. 3) , we p r e s e n t t he
sol ut i on f or t he t r a ns f or m 0( r , z, s) = L[O(r, z, 3) ] = L[-T(r, z, 3) - T O ] in t he f or m
0 ( r , z , s ) = T o ( r , z , s ) - - - r ~ = ] ~ ( p , s )
s 0
' i n ' z > < / " + : ) ]
' ~
JO (pr) d p , Re s > 0 , (1. 4)
wher e Jo(pr) is a Bessel f unct i on of t he f i r st ki nd of zer o or der ; A( p, s) is t he unknown f u n c t i o n - t r a n s f o r m, whi ch
mus t be de t e r mi ne d.
When h -~ oo, f r om Eq. (1. 4) we obt a i n a sol ut i on f or 0( r , z, s) = LI T( r , z, ~) - TO] of a s e mi i nf i ni t e b o d y
[11:
0 ( r , z , s ) lh_,= = A( p , s ) exp - z + Jo(Pr) dp"
0
( 1. 5)
a n d t he der i vat i ve wi t h r es pect to t he nor ma l to t he s ur f ace
0
Jo (pr) ap.
( ~ . 6 )
at z = 0 is as fol l ows:
_ O z ( r , O , s ) = ~ - ~( p, s) V [ ( p 2 + S ) jO(Pr) dp. (1. 7)
0
Accor di ng to Eq. ( 1. 4) , t he exces s val ue of t he t e mp e r a t u r e - t r a n s f o r m at z = 0 has t he f or m
535
O( r , 0, s) = ~0 - A ( p ' s ) t a n h ( h V ( p 2 + S ) ) J O ( p r ) d p = ~ O [1 - g ( p ' s ) ] ' A ( p ' s ) J O ( P r ) d p '
( 1 . 8 )
wher e
(p, s) = (1. 9)
is a known f unct i on.
Whe n h --, oo a nd ~(p, s) --, 0, we f i nd t he wel l - known sol ut i on f or a hal f - s pace at z -- 0 wi t h mi xe d BC
(1. 1) a nd (1. 2) [1 ]. Ta ki ng i nt o account mi xed BC (1.1) a nd (1. 2) on t he s ur f ace z -- 0, upon appl yi ng t he
L- t r a ns f or m t o t hem, f r om Eqs. (1. 7) and (1.8) we obt ai n pai r ed i nt egr al equat i ons in t he r egi on of L- t r a ns f or ms ,
f r om whi ch t he unknown f unct i on A( p, s) must be det er mi ned:
.~ [ 1 - g(p, s)]- A(p, s)J O(P r )dp=T f (r , s), 0 < r < R, R e s > 0 ;
o
( 1 . 1 0 )
~ ~ A ( p , s ) V ( p 2 + S ) J o ( P r ) d p = O , R < r < o o , Re s > 0 , (1. 11)
0
wher e t he known f unct i on is
oo
7 ( r , s ) = L I f ( r , ~ ) l = f e x p ( - s T ) f ( r , r ) d r ,
o
r E ( O , R ) , z = 0 , r > 0 .
To sol ve Eqs. (1. 10) a nd (1. 11), we i nt r oduce a not he r unknown f unct i on ~1 (t, s) by me a ns of t he r el at i on
It t ur ns out t hat t hi s s ubs t i t ut i on ens ur es sat i sf act i on of t he homoge ne i t y of one of pai r ed equat i ons ( 1. 11) , si nce
subst i t ut i ng Eq. (1. 12) i nt o ( 1. 11) , we have (R < r < co, Re s > 0)
oo
f No (pr) dp
o
R
536
since [ 1- 4]
~ ~ l ' t ( l , s ) d l ~ P J o ( P r ) sin ( t ~ ( p 2 + S ) )
o o v ( ~+s )
d p = O ,
( 1 . 1 3 )
O, O < t < r ,
Vp- - s Tz
, O < r < t .
( 1 . 1 4 )
In or de r to obt ai n now a n equat i on f or det er mi ni ng t he funct i on ~ l (t, s ) , we s ubs t i t ut e t he val ue of A( p,
s) given by f or mul a (1.12) into t he fi rst pai r ed i nt egr al equat i on (1. 10):
R
7 ~ ( p , s ) [I - g ( p , s ) ] J 0 ( P r ) d p = f ~1 ( t , s ) dt 7
0 0 0
PJo (pr)
_
o o , / ( ~ + s )
rf ~ l ( t , s ) d t 7 P J o ( P r ) ( V ( p 2 ~ ) )
+ ~f ~ l ( t , s ) d t ~ P J ~ cos ( t V ( p 2 + S ) ) d p _
R (pr) ( V ( p 2 s ) )
f ~z (t, s ) d t 7 g, (P, s) PJo cos t +
=
Y ~1 ( t , s ) ar - f ~ ( t , s )
o ~ z - z 5 z ~ VW2 - - r r )
dt-
o 0 , / ( ~ + : )
= ] ( r , s ) , 0 < r < R , R e s > 0 . (1. 15)
In t he derivation of Eq. (1.15) t he value of t he following di scont i nuous i nt egral [1-4 ] is used:
537
PJo(Pr) cos ( t 4 ( p 2 + S ) ) d p =
sin
47Z-:TZ
, 0<l <r ,
, 0 < r < t .
(1.16)
Equat i on (1.15) is al r eady a basis for det ermi nat i on of the unknown anal yt i c function ~1 (t, s). The solution
of this equat i on can be obt ai ned by represent i ng this function in t he form of a functional seri es I1 ] expanded in
hal f-i nt eger powers of t he paramet er s. However, in t he present work we will fi nd anot her way to solve t hi s equat i on.
Let us assume t hat t here is a solution for Eq. (1.15). We reduce this equation to a si mpl er form, namel y,
a Fr edhol m- t ype equat i on of t he second kind for real variables t hat contain t he complex par amet er s. For this, we
r e p l a c e r b y /.t i n f o r mu l a ( 1 . 1 5 ) , mu l t i p l y b o t h i t s s i d e s b y t h e i n t e g r a t i n g f a c t o r cos
(X/s/a(r2 - / t 2) ) . 21udt ~/ ~r 2- kt 2 and i nt egrat e t he expressi on obt ai ned over/z within t he limits from zero to r:
f f exp ~ d t -
o ~ o v ~
r C O S /
f
0
2~d~
f sin dt -
- f f T1 ( t , s) dt 7 g (P, s)
0 ~ 0 0
x PJ o( Pt Q c o s ( t V~ + s / a ) d p = f O< r < R Re s > 0
~ - - , , 9
v~z+ s/a o ~ 0.17)
In t he l eft -hand si de of Eq. (1.17) we change the or der of i nt egrat i on by using t he wel l -known Dirichlet
formula (for i t erat ed integrals). Thi s l eads us to the following integral equation (0 < r < R, Re s > 0):

rf "T1 (t, s) dt - ~ f T1 (t, s) dt Si (t
0 0
+ r ) ] - S i [ g [ ( ~ ) ( t - r ) ] + ~ ~ ( p , s )
2 cos
( t g f ( p 2 + S ) l sin I r g [ ( p 2 + - ~ )
2 s =J ~
p + - - 0 ~ (1.18)
a
538
where Si(~) = f (sin x / x ) d x is t he integral sine [2-7 1.
o
In deriving Eq. (1.18) we use t he values of t he following integrals:
r ( , / ( : ) )
f = ar - 2 Si ~ (1.19)
o e-/z =Sz ~ - - c - ~
exp 2~d~
( r ~ ) co~ ( r ~ '
, ( r
f = S i ( r + t ) +
o r e - T: ~
sin
+ S i ( r ( r - t ) ) - 2 S i ( r ~ ' ,
( 1. 2o)
.
f = s i ( t + r ) - S i ( t - - r ) ,
o r ~
(i.21)
sin 2~d~
( ,/ ( ~ ) Jo (p/x) 2/xd/x
r
f = (1.22)
COS
Differentiation of equality (1.18) with respect to r gives an integral equation in t he region of L- t r ansf or ms
(with t he par amet er s) for det ermi ni ng t he unknown funct i on ~ol (r, s):
1 R
~1 (r, S) -- -~ f ~I (P, S) K1 (r, p, h, s) dp = F1 (r, s ) , 0 < r < R, Re s > 0,
0
(1.23)
where t he kernel of integral equation (1.23) has the form
" ~ 1 ( r , p , h , s ) = s i n (p - r) x / s/ a + s i n (p + r) ff s / a
+
p - r p + r
+ hh cosh (x) cos x + cos x dx ; (1.24)
and
539
2 d ~ (1.25)
F1 (r, s ) - Jr d r 0
2~d~
is a known funct i on in (0, R) at z = 0.
Having subst i t ut ed t he value of ~1 (P, s) from Eq. (1.23) into (1.12), we find t hat
{ }
F1 ( t ' S) + ~ f ~1 (/9' S) h ' l ( l ' P, h, s) dp . ( 1 . 2 6 )
0
Substitution of Eq. (1.26) into solution (1.4) and application of t he inversion formula of t he Laplace i nt egral
give t he value of t he t emper at ur e field in an infinite plate (i.e., t he solution of Eq. (1)') with prescri bed mi xed BC
(1.1), (1.2) and unmi xed BC (1.3) in t he form
1 p d p
0 (r, z, s) = T (r, z, r) - T O = ~-~ f exp (st) d s ~ Jo ( pr ) x
f c o s t
o
+
a t F1 ( t , s ) + ~ f ~1 ( , o , s ) 2 1 ( t , p, h, s) dp ,
0
r _ > 0 , O<_r <_ ~ , O < _ z < _ h , a > Re s > 0 , a = c o n s t , (1.27)
where ~o 1 (/9, S) is t he solution of integral equation (1.23) with t he L- par amet er s.
When h --, ~, t he value of T1 (r, s) shoul d be sought from integral equation (1.23), t he kernel of whi ch will
have t he form
D
lim K 1 ( r , p , h , s ) = K 1 ( r , p , s ) =
hooo
sin ( p - r) ~/ s / a sin (t9 + r) ~/ s / a
+ (1.28)
p - r p + r '
and t he solution of Eq. (I) for a half-space (h --, oo) with assi gnment of mi xed BC (1.1) and (1.2) is wri t t en from
Eq. (1.27) as follows:
1 a+i~ 7 ( 4 ( P 2 s ) )
0 (r, z, s) [h-,~ - 2~i f exp (st) d s exp - z + x
a-i e~ 0
{~ 1 R _ }
X 1 ( t , s ) + ~ f ~1 ( p , X ) K l ( t , p , s ) d p , O < _ r < oo, Re x > 0 , r >_0.
0
(1.29)
540
When s --> 0 ( r --, oo), f r om Eq. (1. 29) we obt ai n t he wel l - known s t a t i ona r y sol ut i ons of t he Lapl ace equat i on [ 8- 10]
f or a hal f - s pace wi t h t he cor r es pondi ng mi xed BCs at z = 0.
Pr obl e m No. 2. It is r equi r ed to sol ve Eq. (1) unde r unmi xe d BC (1. 3). Th e i ni t i al t e mpe r a t ur e fi el d of t he
i nf i ni t e pl at e is uni f or m a nd cor r es ponds t o pr obl em No. 1, i.e., T ( r , z, 0) = TO = const .
Th e mi xed BCs at z -- 0 have t he f or m
- O z ( r , O , r ) = - ~ f ~ , O < r < R , z = 0 , T , A> 0 ; ( 2. 1)
0 ( r , 0 , ~ ) = 0 , R < r < o o , z = 0 , T > - 0 , ( 2. 2)
whe r e q ( r , z) > 0 is a pr es cr i bed f unct i on of t he he a t - f l ux dens i t y i n t he ci rcl e r = R, z = 0 t hat has t he Lapl ace
t r a ns f or m ~( r , s) = L [ q ( r , r) ]; 2 is t he t her mal - conduct i vi t y coef f i ci ent of t he pl at e.
Th e sol ut i on of Eq. (1) in t he r egi on of L- t r a ns f or ms wi t h BC (1.3) on t he s ur f ace z = h is r e pr e s e nt e d by
f or mul a ( 1. 4) . Ta ki ng i nt o account mi xed BC (2. 1) a nd (2. 2), f or de t e r mi na t i on of t he unknown f unct i on A( p, s)
we come t o t he fol l owi ng pai r ed i nt egr al equat i ons:
(p, s) p2 + JO (p r) dp = , 0 < r < R , Re s > 0 ,
o
( 2. 3)
~ - A ( p ' s ) t a n h ( h V ( p 2 + S ) l J ~ R< r < oo, Re s > 0 . ( 2. 4)
Th e sol ut i on of t hes e equat i ons will be car r i ed out by means of t he s ubs t i t ut i on
A ( p, s) = p 1 ~ ~2(t,s) sin ( tV [ ( p2+S ) )
~
a t . ( 2 . 5 )
Just as f or pr obl em No. 1, it is not di ffi cul t to show t hat s ubs t i t ut i on (2. 5) ens ur es sat i sf act i on of t he
homoge ne i t y of Eq. (2.4) in accor dance wi t h t he val ue of di s cont i nuous i nt egr al ( 1. 14) .
Subs t i t ut i ng Eq. (2.5) i nt o ( 2. 3) , we obt ai n an i ni t i al i nt egr al equat i on wi t h t he L- p a r a me t e r to f i nd t he
unknown f unct i on ~2( t , s) u n d e r pr es cr i bed BC (2.1) and (2.2) on t he pl at e s ur f ace z = 0:
~ ~ 2 ( t ' s ) d t { o si n ( t V( p 2 + S ) )
Jo (p r) pdp +
+7
o
exp
( - h V ( p 2 + S ) ) sin ( t V [ ( p 2 + ~ ) )
Jo (p r ) pdp } =
(r, s)
= 2 , 0 < r < R, Rc s > 0 . (2. 6)
541
J us t as in pr obl e m No. 1, we r e duc e Eq. (2. 6) to an i nt egr al equat i on of t he t ype ( 1. 23) . Us i ng t he wel l -
known r e l a t i on f or Bessel f unct i ons pJo(pr) = ( 1 / r ) ( d / d r ) [ r J l ( p r ) ] , whi ch we s ubs t i t ut e i nt o Eq. ( 2. 6) , a n d
i nt e gr a t i ng its l ef t - a nd r i ght - ha nd si des over r wi t hi n t he l i mi t s f r om zer o to r, we obt ai n:
R
r f "~2(t , s) dt
0 0
Jl (pr) dp +
o
( t , s )
J
o xrT-=3-z
exp
exp
( - h V ( p 2 + S ) ) sin ( ' V ( p 2 + S ) )
o g g - - z
(pr) dp ] =
+ f ~ 2 ( t , s ) si n t d t + r f ~ z ( t , s ) dt x
0 o
O0

0
exp
=~( q ( p , s ) p d p , 0 < r < R, Re s > 0 .
0
(2. 7)
I n t he der i vat i on of Eq. (2.7) we us ed t he val ue of t he f ol l owi ng di s c ont i nuous i nt egr al [ 1-4 ]:
O r t si n s (t 2 _ r 2)
, r >t;
t > r ;
( 2 . 8 )
I nt e gr a l equat i on (2. 7) can a l r e a d y ser ve as a basi s f or de t e r mi na t i on of t he unknown f unct i on ~2(r, s) in
t he r egi on of L- t r a n s f o r ms . For t hi s, it is ne c e s s a r y to r e pr e s e nt t hi s f unct i on as t he c or r e s pondi ng f unct i onal s er i es
e x p a n d e d in ha l f - i nt e ge r power s of t he p a r a me t e r s, as was done in [1 ].
Bel ow we s ugges t a n o t h e r way of de t e r mi ni ng t he unknown f unct i on ~2( r , s). By a na l ogy wi t h pr obl e m No.
1, we r epl ace r by p in Eq. ( 2. 7) , mul t i pl y t he l eft - a nd r i ght - ha nd si des of Eq. (2. 7) by a s i mi l ar i nt e gr a t i ng f a c t or
to pr obl e m No. 1, a nd t hen i nt e gr a t e over dp wi t hi n t he l i mi t s f r om zer o to r. As a r esul t , we come to t he f ol l owi ng
i nt egr al equat i on wi t h t he L- p a r a me t e r :
542
1 R
~ z ( r , s ) - ~ f ~2( I ~ K2 (r, p, h, s) dp =F z ( r , s ) , 0 < r < R , Re s > 0 ,
0
(2.9)
wher e
K 2 ( r , p , h , s ) =
p - r p +r
h si nh (x) cos x - cos x
hC-/7-d
dx ;
(2. 10)
t he known f unct i on F2(r, s) E (0, R) , z - 0 has t he form
~ 0
(2. 11)
In deri vi ng Eq. (2. 9), we used t he val ues of i nt egr al s (1. 19)-(1. 21) a nd t he cal cul at ed i t er at ed i nt egr al s
1 r /~
- f f ~ ( p , s ) p d p =
o ~ o
f ~ (/9, s) sin p d p ,
0
(2. 12)
f 2/x2 J l (p~t) d/~ =
o
= 2p -- ~)-~ - .
(2.13)
The val ue of Eq. (2.11) follows from Eq. (2.12) upon di f f er ent i at i on of t he l at t er wi t h respect to r wi t h
al l owance for t he const ant fact or n - 1 and cancel l at i on of t he left- and r i ght - ha nd si des of Eq. (2.9) by r.
Subst i t ut i on of t he value of ~2(t , s) from Eq. (2.9) i nt o Eq. (2.5) yi el ds | he unknown f unct i on A( p, s) for
t he pr esent probl em:
( p, s ) = P 1 f sin t p2 + dt x
o
543
{ , R }
x F2 (t , s) + ~ f ~2 ( p, s) K2 ( t , p , h, s) @ .
0
( 2. 14)
Us i ng Eq. (2. 14) in sol ut i on (1. 4) a nd a ppl yi ng t he i nver si on f or mul a of t he Lapl ace i nt egr al , we obt a i n
t he val ue of t he t e mp e r a t u r e fi el d O(r, z, T) at a ny poi nt of an i nf i ni t e pl at e wi t h a s s i g n me n t of mi xe d BC (2. 1) a n d
(2. 2) at z -- 0 a n d u n mi x e d BC (1. 3) at z = h:
1 o+i oo
0 (r, z, ~) = T (r, z, T) - T O = 2ari f exp ( s t ) ds x
a- i oo
x ~ Jo (pr) pdp
0 s i n h [ h V ( p 2 + S ) ] V ( p 2 + s )

x f si n t + dt (t, s) + -~ f ~2 (P, s) KZ (t, p, h, s) dp ,
0 0
O<r <_ o o , T>_O, O<z <_ . h , a > Re s > 0 , a = c o n s t . ( 2. 15)
wher e ~ 2 ( r , s ) is t he sol ut i on of i nt egr al equat i on (2. 9) wi t h t he L- p a r a me t e r s .
Whe n h --- oo, t he val ue of ~2( r , s) s houl d be de t e r mi ne d f r om i nt egr al equat i on ( 2. 9) , t he ke r ne l of whi ch
will have t he f or m
l i m K'2 ( r , p , h , s ) = K 2 ( r , p , s ) =
h ~
si n (t9 - r) ~I s / a si n (p + r) ~ s / a
m
p - r p +r '
( 2. 16)
and t he sol ut i on of Eq. (1) f or a ha l f - s pa c e (h --, oo) wi t h a s s i gnme nt of mi xe d BC (2. 1) a n d (2. 2) is wr i t t e n f r om
Eq. (2. 15) as
1 a+i~176 "~ ( V ( S ) )
O( r , z , T ) lh_. ~ = ~ f e x p ( s r ) ds exp - z p 2 +
a - i ~ 0
X
x J~ pdp ~ s in ( t V ( p 2 + S ) ) d t x
[ F 2 1 R
X ( t , s ) +- ~ f ~2 ( P, s ) Kz ( t , p , s ) d p , O<_r<_ oo, z___O, Re s > O, r>__O.
0
( 2. 17)
Wh e n s - - , oo (r--* oo), f r om Eq. (2. 17) we obt a i n t he we l l - known s t a t i o n a r y s ol ut i ons of t he La pl a c e
equat i on [ 8- 10 ] f or a ha l f - s pa c e wi t h t he c or r e s pondi ng mi xe d BCs at z = 0.
Pr o b l e m No. 3. I t is r equi r ed t o sol ve Eq. (1) wi t h mi xed BC (1. 1) a nd (1. 2) on t he s ur f a c e z = 0 of a pl at e.
Th e i ni t i al condi t i on is r e t a i ne d as in pr obl e ms Nos . 1 a n d 2. Th e di f f er ence f r om t h e m will be in t he a s s i g n me n t
of unmi xe d BC (1. 3) at z = h, whi ch is r e pr e s e nt e d as
Oz ( r , h , T ) =O , 0 _ < r < oo, z =h , T>__O, (3. 1)
i. e. , when z = h, 0 _< r < oo, and r > O, t her e is i deal heat i nsul at i on on t hi s s ur f ace of t he pl at e over t he e nt i r e
dur at i on of heat exchange.
5 4 4
Th e sol ut i on of Eq. (11 in t he r egi on of L- t r a ns f or ms wi t h mi xed BC (1.1) a nd (1.21 on t he pl at e s ur f ace
z = 0 a nd unmi xe d BC (3.11 on t he pl at e sur f ace z = h f or t he excess t e mp e r a t u r e - t r a n s f o r m 0( r , z, s) --- L[O(r, z,
T) ] can easi l y be obt ai ned in t he fol l owi ng form:
-O (r, z, s) = "T (r, z, s) - TO = ~ -B (p, s) Jo (Pr) d p , Re s > 0 , ( 3. 2)
wher e B( p, s) is t he unknown f unc t i on- t r a ns f or m t hat must be de t e r mi ne d.
When h --, ~, f r om Eq. (3. 2) we obt ai n sol ut i on (1.5) f or a hal f - s pace [1 ]. Account i ng f or mi xe d BC ( 1. I )
a nd (1.2) on t he pl at e sur f ace, upon appl yi ng t he L- t r a ns f or ma t i on to t hem, we come to t he f ol l owi ng pai r ed i nt egr al
equat i ons wi t h t he L- pa r a me t e r s:
B (p, s) Jo ( pr ) dp = f (r, s ) , 0 < r < R, Re s > 0 ; ( 3. 3/
0
~ - B ( p ' s ) V ( p 2 + S ) t anh ( h V f ( p 2 + S ) ) JO(pr) R < r < oo, Re s > 0 , (3. 4)
wher e t he known f unc t i on- t r a ns f or m is
] ( r , s ) = L [ f ( r , v ) ] , r ~ ( O , R ) , z = 0 , r > 0
To sol ve pai r ed equat i ons (3.3) a nd (3.4) we use t he subst i t ut i on
at, (3.5)
si nce it ens ur es sat i sf act i on of t he homoge ne i t y of Eq. (3.4) f or R < r < oo.
Af t er t he subst i t ut i on of Eq. (3.5) i nt o Eq. ( 3. 3) , we have a n i nt egr al equat i on i n t he r egi on of L- t r a n s f o r ms
f or f i ndi ng t he new unknown f unct i on ~3( r , s):
0 ~ e x p - ~ d t - fo ~ s i n
dt +
R
+ f ~3 (t, s) dt ~ Jo (pr) pdp =
0 0 si nh ( h ~ f ( p 2 + S ) ) ~ t r ( p 2 + S )
=/ ( r , s ) , 0 < r < R, Re s > 0 .
(3. 6)
Equat i on (3. 6) is obt ai ned by means of di scont i nuous i nt egr al s (1. 14) a nd (1. 16). Usi ng t he ma t he ma t i c a l
me t hod t hat was appl i ed by us to pr obl em No. 1, we r educe Eq. (3.6) to an i nt egr al equat i on wi t h t he L- pa r a me t e r :
1 R
~ 3 ( r , s ) - - ~ f ~3(/ 9, s) K3 (r, p, h, s) dp = F 1 ( r , s ) , 0 < r < R, Re s > 0 , (3. 7)
0
545
wher e t he ker nel of (3. 7) has t he f or m
sin (p - r) 4 s / a si n (t 9 + r) ~/ S/ a
K'3 (r, p, h, s) = +
p - r p +r
- h si nh (x) cos x + cos d x ,
hf s / a
( 3 . 8 )
Th e known f unct i on Fl ( r , s) ~ (0, R) , z = 0 is de t e r mi ne d by f or mul a ( 1. 25) .
Th e s ubs t i t ut i on of t he val ue of ~3( t , s) f r om Eq. (3.7) i nt o Eq. (3. 5) gives
( p , s ) = f cos t _ _ + dt
p2 + t anh h 2 +
F1 (t, s) + ~ f ~3 (P, s) K'3 (t, p, h, s) dp . (3. 9)
0
Usi ng Eq. (3. 9) in sol ut i on (3.2) and appl yi ng t he i nver si on f or mul a of t he Lapl ace i nt egr al , we obt ai n t he val ue
of t he t e mpe r a t ur e f i el d O(r, z, r ) at a ny poi nt of an i nf i ni t e pl at e wi t h i mpl ement at i on of mi xed BC (1. 1) a n d (1. 2)
on t he s ur f ace z -- 0 of t he pl at e a nd unmi xe d BC of t he t ype (3.1) on t he ot he r s ur f ace z = h:
1 o+i oo
0 ( r , z , ~ ) = T ( r , z , r ) - T 0 = ~ f e x p ( ~ ) d s x
o- / oo

f Jo (pr) pdp
x f c o s t p2 + dt F l ( t , s ) +' ~ f g 3 ( P, s ) Ka ( t , p , h , ' s ) d p ,
0 0
0 < _ r < ~ , r > 0 , O<_z <_h, o > Re s > 0 , o = c o n s t , ( 3. 10)
wher e ~3( r , s) is d e t e r mi n e d f r om t he sol ut i on of i nt egr al equat i on (3. 7) wi t h t he L- pa r a me t e r . Whe n h --- oo, t he
val ue of ~3( r , s) = ~ l ( r , s) (see pr obl em No. 1), a nd cor r es pondi ngl y t he sol ut i on of Eq. (1) f or a hal f - s pace
coi nci des wi t h f or mul a ( 1. 29) .
I n t he s t a t i o n a r y s t at e ( r - - , oo), s- - , 0, we obt ai n t he we l l - known s t a t i o n a r y s ol ut i ons [ 8- 10] f or a
semi i nf i ni t e body wi t h t he cor r es pondi ng mi xed BC at z = 0.
Pr obl e m No. 4. It is r equi r ed to sol ve Eq. (1) wi t h mi xed BC (2. 1) a nd (2. 2) on t he s ur f ace z = 0 of a pl at e
and unmi xe d BC of t he t ype (3.1) on t he ot he r sur f ace z = h of t he pl at e.
Th e i ni t i al c ondi t i on coi nci des wi t h pr obl ems Nos. 1-3. Th e s ol ut i on of Eq. (1) u n d e r BC (3. 1) is
de t e r mi ne d by f or mul a (3. 2).
Usi ng f or Eq. (3. 2) mi xed BC (2.1) and (2.2) at z = 0, upon appl yi ng t he L- t r a ns f or ma t i on to t hem, we
obt ai n t he fol l owi ng pai r ed i nt egr al equat i ons wi t h t he L- pa r a me t e r f or f i ndi ng t he unknown f unc t i on- t r a ns f or m
B( p, s):
546
-B (p, s) p2 + t anh h p2 + Jo (Pr) dp = , O < r < R , Re s > 0 ;
0
(4.1)
7
- B ( p , s ) J O ( P r ) d p = O , R < r < oo, Re s > 0 .
0
(4.2)
To solve t hese equat i ons, we car r y out t he subst i t ut i on
P ~ ~ 4 ( t , s ) s i n ( t d ( p 2 + s ) )
at, (4. 3)
whi ch ensur es sat i sf act i on of t he homogenei t y of Eq. (4.2). When h --, 0% t anh( h~/ p2 + s / a ) --, 1 a nd we come to
t he cor r espondi ng pai r ed i nt egr al equat i ons wi t h t he L- par amet er for i nvest i gat i ng t he t emper at ur e fi el ds i n a
hal f - space [1 ] wi t h pr escr i bed BC (2.1) and (2.2) at z = 0. Subst i t ut i on of B(p, s) from Eq. (4.3) i nt o Eq. (4. 1),
per f or med as in pr obl em No. 2, gives an i nt egr al equat i on wi t h t he L- pa r a me t e r for det er mi nat i on of t he unknown
f unct i on- t r ans f or m ~4(r, s):
t-~ 4 ( t , S) ( d ( S ) ) R ( t , S) ( d ( S ) )
0 ~ e x p - ~ a t - f o v ~TZ- 7 2 - @4 si n ~ d t +
+ f ~4 ( / ' S) si n t dt -- r f ~4 ( / , S) dt x
0 o
1 .~ ( 4 . 4 )
=~- ~( u , s ) / ~d / . t , 0 < r < R, Re s > 0 .
0
Expr essi on (4.4) is t he basi c i nt egr al equat i on wi t h t he L- par amet er for f i ndi ng t he unknown f unct i on- t r ans f or m
~4(r, s). Usi ng t he above- descr i bed mat hemat i cal met hods, t hi s expr essi on can be r educed to a si mpl er f or m t hat
is si mi l ar to t he Fr edhol m form, but wi t h t he L- par amet er :
1 R
~4 (r, S) -- ~ f ~4 (/9, S) ~"4 (r, p , h, s) dp = FX (r, s ) , 0 < r < R, Re s > 0 ,
0
(4.5)
wher e t he known f unct i on- t r ans f or m Fz( r , s) is det er mi ned in (0, R) , z = 0, T > 0, Re s > 0 by f or mul a (2.11).
The ker nel K4( r , p , h, s) is given by t he expr essi on
K 4 ( r , p , h , s ) = s i n (/9 - r ) ~/ s / a _ sin (t9 + r ) X / s/ a +
p - r p + r
+ h cosh (x) cos x - cos d x ,
h~f s / a
(4.6)
547
Havi ng de t e r mi ne d t he val ue of ~4( r , s) f r om Eq. (4.5) a nd s ubs t i t ut i ng ~4( t , s) i nt o f or mul a ( 4. 3) , we
f i nd t he unknown f unct i on- t r ans f or m:
{ 1 R }
X F2 (/ ' s) at- ~ f ~4 (/9, s ) ~' 4 ( l , p , h, s) d/9 .
0
(4. 7)
Us e of Eq. (4. 7) i n sol ut i on (3.2) a nd appl i cat i on of t he i nver si on f or mul a of t he Lapl ace i nt egr al al l ow us
to de t e r mi ne t he s ought t e mpe r a t ur e fi el d at a ny poi nt 0 < r < o o , 0 _.< z < h, r ___0 of an i nf i ni t e pl at e wi t h mi xe d
BC (2. 1) a nd (2. 2) on t he pl at e s ur f ace z = 0 a nd unmi xe d BC (3. 1) on t he ot he r pl at e s ur f ace z = h:
1 a + i o o
O( r , z , r ) = T ( r , z , r ) - - T 0 = ~ f e x p ( ~ ) d s x
o- t oo
cosh
o
cosh
x f si n t p 2 + d t (t, s ) + ~ f ~ 4 (t 9, s ) K 4 ( t , p , h , , s ) d p ,
0 0
O_ < r _ < o o , T > O, O < _ z < _ h , a > Re s > O, c r = c o n s t ,
( 4 . 8 )
s) shoul d be de t e r mi ne d f r om t he sol ut i on of i nt egr al equat i on (4. 5) and t he val ue of t he f unc t i on- t r a ns f or m ~4( r ,
with t he L- pa r a me t e r .
Whe n h -~ 0% f r om Eq. (4. 8) we have sol ut i on (2. 17) f or a hal f - s pace u n d e r mi xed BC (2. 1) a n d (2. 2) at
z -- 0, f or ~4( r , s) - ~z( r , s) , a nd K 4 ( r , p , h , s) f or h ~ oo coi nci des wi t h t he val ue of ker nel (2. 16) K2( r , p, s).
Th e c o n s i d e r e d pr obl e ms of ma t he ma t i c a l phys i cs wi t h mi xe d BC a r e s ui t a bl e f or var i ous pr act i cal
appl i cat i ons:
a) in des i gni ng opt i mum cool i ng s ys t ems in r adi oel ect r oni c equi pment ;
b) in s pace t echnol ogy ( unde r condi t i ons of local heat i ng or cool i ng) ;
c) i n l as er medi ci ne f or i mpr ovi ng t he oper at i ng saf et y of l as er i nst al l at i ons;
d) f or cont r ol of t he l as er t echnol ogy i n t he pr oduct i on of LIC;
e) in t he devel opment of new met hods a nd devi ces of nondes t r uct i ve moni t or i ng f or i dent i f i cat i on al l t he
t her mophys i cal char act er i s t i cs of mat er i al s by t e mpe r a t ur e me a s ur e me nt s i n t i me a nd space di r ect l y on a b o u n d a r y
sur f ace of t he body u n d e r i nvest i gat i on;
f) f or sol vi ng i nver se pr obl ems of mat hemat i cal physi cs on t he basi s of i nvest i gat i on of t he di r ect pr obl em
wi t h t he ai m of opt i mum r egul ar i zat i on.
I n t he pr e s e nt wor k f or t he fi rst t i me appl i cat i on of t he me t hod of pai r ed nons t a t i ona r y i nt egr al equat i ons
is i mpl ement ed f or t he i ndi cat ed cl ass of pr obl ems of mat hemat i cal physi cs by r educi ng t hem to i nt egr al equat i ons
with an L- pa r a me t e r . A s t at i onar y var i ant of sol vi ng t he Lapl ace equat i on fol l ows f r om t he pr e s e nt e d sol ut i ons as
a par t i cul ar case wi t h s --- 0 (r ~ oo) [4, 7-9 ].
Th e me t hod s ugges t ed can al so be used to solve t he Hel mhol t z equat i on [9, 1 ] u n d e r t he c ons i de r e d mi xed
BC in t he case of r eal k = Re v'-~-~, or to solve t he Kl e i n - Go r d o n equat i on f or pur e i ma gi na r y k = iw = l m v ~ / a ,
si nce t he anal yt i cal r esul t of t hei r sol ut i on l eads to i dent i cal i nt egr al equat i ons (1. 29) and (2. 17).
548
In conclusion we note that practical application of the presented solutions for the mixed problems of
mathematical physics with the mixed BC considered requires, as the final result, determination of the unknown
functions-transforms -~i(r, s) in the complex plane with the prescribed accuracy from integral equations (1.23),
(2.9), (3.7), and (4.5). But up to now the theory of solving the latter has not been developed. However, some ideas
to solve these integral equations with an L-parameter have appeared.
First, they can be solved by representing the analytical function ~i ( r , s) in the form of the corresponding
functional power series expanded in half-integer powers of the parameter s [ 1, 5-7 ]. Here the prescribed functions-
transforms [(r, s) -- L[ f ( r , ~) ] and ~(r, s) = L[ q( r , r ] in the circle r = R on the surface z = 0 should be expanded
in the corresponding power series in s, the expansion coefficients of which will be known. Equating the left and
right terms of these expansions in Eqs. (1.23), (2.9), (3.7), and (4.5), for identical powers of the parameter s we
obtain a system (infinite or finite) of corresponding integral equations with degenerate kernels, which can be solved
by well-known methods [12 ].
Second, our preliminary investigations of the nonstationary kernels Ki (r, [9, h, s) present in Eqs. (1.23),
(2.9), (3.7), and (4.5) have shown that they are bounded and square integrable within the intervals 0 < r < R,
0 < p < R for Re s > 0, and consequently, in order to solve integral equations (1.23), (2.9), (3.7), and (4.5), one
can apply a method of successive approximations [12 ] developed for a Fredholm equation of the second kind for
real variables.
And finally, due attention should be given to approximate computer methods for solving the obtained
integral equations with an L-parameter by constructing effective computational algorithms.
REFERENCES
1. N.A. Abdel' razak, Methods of Solving Two-Dimensional Nonstationary Heat Conduction Problems with Mixed
and Unmixed Discontinuous Boundary Conditions, Author' s Abstract of Candi dat e' s Dissertation in the
Physical-Mathematical Sciences, Minsk (1996).
2. A.P. Prudnikov, Yu. A. Brychkov, and O. I. Marichev, Integrals and Series: Elementary Functions [in Russian ],
Moscow (1981).
3. A. P. Prudnikov, Yu. A. Brychkov, and O. I. Marichev, Integrals and Series: Special Functions [in Russian ],
Moscow (1983).
4. G. N. Watson, Theory of Bessel Functions, Pt. 1 [Russian translation ], Moscow i1949).
t .
5. G. Bateman and A. Erdelyz, Tables of Integral Transformations, Vol. I. Four i er - Lapl ace-Mel l i n Transforms
[Russian translation ], Moscow (1969).
6. A.V. Luikov, Theory of Heat Conduction [in Russian ], Moscow (1967).
7. V. P. Kozlov, Two-Dimensional Axisymmetric Nonstationary Heat-Conduction Problems [in Russian l, Minsk
(1986).
8. Ya. S. Uflyand, Method of Paired Equations in Problems of Mathematical Physics [in Russian ], Leningrad
(1977).
9. N. A. Virchenko, Paired (Triplet) Integral Equations [in Russian l, Kiev (1989).
10. I. Sneddon, Mixed Boundary-Value Problems in Potential Theory, Amsterdam (1966).
11. Yu. V. Gandel ' , Reduction of a Class of Paired Integral Equations to a Fredholm Equation of the Second Kind,
in: Problems of Mathematical Physics and Functional Analysis, Kiev (1976).
12. P. P. Zabreiko, Integral Equations [in Russian], Moscow (1968).
549