Also, the problem of contact of impressing a system strip of stamps in an elastic half-space of a logarithmic series in the case of two symmetric intervals is solved.
Applied Mathematics and Computation 118 (2001) 95±111 www.elsevier.com/locate/amc
Spectral relationships for integral operators in contact problem of impressing stamps M.A. Abdou Department of Mathematics, Alexandria University, Alexandria, Egypt
Abstract A method is given for solving the problem of mechanics of continuous media between a ®nite system of stamps varying width and an elastic half-space in a three dimensional formulation. The friction within the region of contact is neglected. The problem is solved in the Mathieu function form. Also, the problem of contact of impressing a system strip of stamps in an elastic half-space of a logarithmic series in the case of two symmetric intervals is solved. Some important relationships are investigated. Also the spectral relationships of the integral equations with Karlman kernel are obtained. Ó 2001 Elsevier Science Inc. All rights reserved. Keywords: Contact problems; Mathieu function; Potential theory method; Integral equation of the ®rst kind; Logarithmic kernel
1. Introduction The mechanics mixed problems of continuous media have been studied by many authors (see [3,6,10,15]). Protsenko [14] used the potential theory method for solving the problem about the contact of a thin plate in the form of an in®nite strip lying on an elastic frictionless half-space in a three dimensional formulation. Popov [3] used the orthogonal polynomials method and its generalization, for the investigation of complex mixed problems of the mechanics of continuous media. In Ref. [2] Kovalenko studied the mixed problems of the mechanics of continuous media with boundary conditions speci®ed on a circle and he obtained a Fredholm integral equation of the ®rst kind with singular kernel. Aleksandrov and Pozharskii [12] studied the problems concerning the pressing of a thin linear inclusion in the form of a stiness rib into a plate having a wedge shape in planform. In Ref. [7] Abdou solved the three dimensional 0096-3003/01/$ - see front matter Ó 2001 Elsevier Science Inc. All rights reserved. PII: S 0 0 9 6 - 3 0 0 3 ( 9 9 ) 0 0 1 6 7 - 8
96
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
semi-symmetric Hertz contact problem of two rigid surfaces having two p dierent elastic materials occupying the domain w fz 0; x2 y 2 < a2 g and the kernel takes the potential function form. Here, the method of orthogonal function applied for the investigation of complex mixed problems of the mechanics of continuous media, are based on the utilization of spectral relationships that invert the main part of the kernel of the integral equation of the problem under consideration. A suciently general approach to the derivation of spectral relationships that is based on potential theory is proposed. The kernel of the problem is represented in the Weber± Sonin integral formula. We prove that, the kernel of the consideration problem ful®lls a non-homogeneous wave equation. Also, here eigenfunctions are obtained in the problem of impressing a system strip stamps in an elastic halfspace of a logarithmic series in the case of two symmetric intervals. In the end, the spectral relationships of the integral equations with Karlman kernel are obtained. 2. Contact problem with potential kernel Consider a system of a three dimensional contact problem of frictionless impression of a system rigid stamps in the surface of an elastic half-space occupying the domain ÿ1 < x; y < 1; z > 0. The integral equation of such a problem [11] takes the form k Z Z X i1
Xi
h
Pi
n; gdn dg 2
x ÿ n
y ÿ g
2
x; y 2 Xi ; h G
1 ÿ m
i1=2 2phfi
x; y
ÿ1
;
2:1
where fi
x; y di ÿ ai x bi y ÿ fi
x; y. Here, G is the displacement magnitude, m the Poisson's coecient, Pi
x; y are the unknown normal stress under the stamps, Xi are the contact domain between the stamps and the half-plane surface. Also, fi
x; y are the known functions describing the shape of the stamps base. While, di ÿ ai x bi y are the rigid displacements of the stamps under the action of forces Pi and moments Mix ; Miy . We valid Eq. (2.1) under the evident statics conditions Z Z Pi
x; ydx dy Pi
Pi constant;
Xi
yPi
x; ydx dy Mix ; Xi
2:2
Z Z
Z Z
Xi
xPi
x; ydx dy Miy :
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
97
The relationship between Pi ; Mix ; Miy and di ; ai ; bi can be determined from expressing the equilibrium of the stamps on the half-space, when ai bi 0 we set Mix Miy 0. Using the potential theory method [11], we introduce the simple layers potential of densities Pi
x; y distributed over the domain Xi . k Z Z X Pi
n; gdn dg q :
2:3 Vi
x; y; z 2 2 i1
x ÿ n
y ÿ g z2 Xi The functions Vi
x; y; z are harmonic everywhere except on a plane slit in the domain Xi and its vanish at in®nity as Pi Rÿ1
Pi constant; p 2 2 2 R x y z . Moreover the functions Vi
x; y; z are continuous in all space including the domain Xi and its normal derivatives take the form oVi 2pPi
x; y;
x; y 2 Xi ;
2:4 0;
x; y 62 Xi : oz z!0 On the slits, we set the following boundary condition: Vi
x; y; 0 2phfi
x; y
x; y 2 Xi :
2:5
The solution of the integral equation (2.1) is equivalent to the problem (2.3)± (2.5) of determining the harmonic functions Vi
x; y; z. The contact problem concerning the frictionless impression of a system stamps of strip planform in an elastic half-space representing by (2.1), with the aid of Ref. [10], reduces to the integral equation k Z ai X /i
ki ; tK0
ki jt ÿ xjdt phfi
ki ; x;
2:6 i1
ÿai
where K0
t are the Macdonald functions and ki are arbitrary positive number. The solution of the integral equation (2.6), reduces to the following boundary value problem (see (7.1)±(7.5) of [10, pp. 161])
and
o2 Vi o2 Vi 2 ÿ ki Vi 0
x 62 Xi ; z 6 0; ox2 oz 1 oVi
ki ; x; 0 ph /i
ki ; x; jxj < ai ; 0; jxj > ai oz Vi
ki ; x; 0 fi
ki ; x; Vi
ki ; x; z ! 0
jxj 6 ai ;
x2 z2 ! 1;
where we assumed Vi
x; y fi
ki ; x coski y;
Pi
x; y /i
ki ; x cos ki y
2:7
2:8
2:9
98
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
and we introduced the potential functions q k Z ai 1 X K0 ki
x ÿ t2 z2 /i
ki ; tdt: Vi
ki ; x; y 2ph i1 ÿai
2:10
Let us introduce the local Cartesian coordinates xi hi zi , placing their origins in the middle of the segments ÿai ; ai
i 1; 2; . . . ; k and let us perform the variable substitution xi ai xi and zi ai zi . Passing now in each local coordinate system xi ; hi ; zi to the elliptic cylinder coordinates according to the formulae xi ai cosh ni cosgi ;
zi ai sinh ni sin gi
0 6 gi 6 2p; 0 6 ni < 1:
2:11
Using Eq. (2.11) in (2.6)±(2.9), we have o2 Vi o2 Vi ai k2i
cosh2ni ÿ cos 2gi ; 2 ÿ 2 ogi on2i
Vi 0
2:12
Vi
ki ; 1; gi 0;
2:13
and Vi
ki ; 0; gi fi
ki ; ai cosgi ; while Eq. (2.8) becomes /i
ki ; ai cos gi
h oVi ai j sin gi j on n0
for all values of gi :
2:14
Assuming the functions fi
ki ; x to be such that it can be expanded into a uniformly convergent series of periodic Mathieu functions (see Ref. [9]) in the intervals ai 6 xi 6 ai 1 X k2i a2i fi
ki ; ai cosgi cin Cen
gi ; ÿqi qi
2:15 4 n0 we will seek the solution of (2.12) in the form Vi
ki ; ni ; gi Ui
ni Wi
gi :
2:16
Using (2.16) in (2.12), we arrive to study the Mathieu equations Wi 00
ai 2qi cos2gi Ui00 ÿ
ai 2qi cosh 2ni Ui
0Wi
gi fi
ki ; ai cosgi ;
Wi 0; Ui 0
ai are constants; Ui
1 0:
2:17
Furthermore, following the general theory of Mathieu function [9], we write the general solution (2.16) of the boundary value problem (2.12) formulated to satisfy the condition of (2.17) in the form Vi
ki ; n; g
1 X rin Fe Kn
n; ÿqi Cen
g ÿ qi : n0
2:18
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
99
The coecients rin are found from the set of in®nite system ri`
n X 1 X rsn T`n
s; i s6i n0
ci` ; Fe Kn
n; ÿqi
Qn`
s; iCe`
0; ÿqi ; T`n
s; i Fe K`
0; ÿqi
2:19
where cin are the coecients of expansion of the function of (2.15). Dierentiating (2.18) and using the result in (2.14), we obtain /i
ki ai cos g
1 X h Fe Kn0
0; qi cin Cen
g; ÿqi ai j sin gj n0 Fe Kn
0; ÿqi
2:20
Using (2.20) in (2.6), we arrive at the following spectral relationship for the Mathieu functions: k Z ai Ce cos ÿ1 n ÿ q X n i ai q K0
ki jn ÿ xjdn 2 2 i0 ÿai ai ÿ n pFe Kn
0; ÿqi ÿ1 x Ce cos ; ÿ q
2:21 n i : Fe Kn0
0; qi ai Equation (2.21) represents the spectral relationships for a system of Fredholm integral equation of the ®rst kind with Macdonald kernel. 3. Weber±Sonin integral formula Using the polar coordinates Eq. (2.1) takes the form k Z ai Z p X Pi
q; /q dq d/ p 2phfi
r; h: 2 q2 ÿ 2rq cos
h ÿ / r ÿp ÿa i i1 To separate the variables, one assumes cosmh; cosmh; fi
r; h fim
r Pi
q; h Pim
r sin mh; sin mh: Using (3.2) in (3.1), we have X Z ai Lm
r; qPim
qq dq 2phgim
r; i0
ÿai
3:1
3:2
3:3
where Z Lm
r; q
p
cos m/ d/ p : 2 r q2 ÿ 2rq cos / ÿp
3:4
100
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
To write Eq. (3.4) in the Weber±Sonin form, ®rstly we use the following relation (3(a)): Z 2p cosm/ d/ 2p
am zm F
a; m a; m 1; z2
3:5 2 a m!
1 ÿ 2z cos / z 0 and 4z ÿ2a 1 1 2 F a; a 2 ÿ b; b 2 ; z
1 z F a; b; 2b; 1 z2
3:6 C
m a : jzj < 1 Re a > 0;
am C
a Hence, the kernel (3.4) takes the form ! p 2 pC
m 1=2
rqm 4rq 1 1 F m 2 ; m 2 ; 2m 1; ; Lm
r; q 2m1 2 m
r q
r q
3:7 where F
a; b; c; z is the Gauss hypergeometric function and C
x is the Gamma function. Secondly, using the famous relation (3(b)) Z 1 aa ba 2ÿb F
a
1 ÿ b=2 Ja
axJa
bxxÿb dx 2aÿb1 C
a 1C
1 b=2
a b 0
3:8 F a
1 ÿ b=2; a 1=2; 2a 1;
4ab=
a b2 : Eq. (3.7) takes the form Z 1 Jm
tqJm
trdt Lm
r; q 2p 0
Jm
x is the Bessel function:
Using the following notations p r q u ; v ; wi
r rPim
r; ai ai
p hi
r h rgim
r;
we can write Eq. (3.3), with the aid of (4.9), in the form k Z 1 X K
u; vwi
udu hi
v; i1
ÿ1
where k
u; v
p uv
Z 0
1
Jm
tuJm
tvdt:
3:9
3:10
3:11
3:12
Eq. (3.11) represents a system of Fredholm integral equation of the ®rst kind with kernel (3.12) in the form of Weber±Sonin integral formula.
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
101
4. Partial dierential equation To prove that the kernel (3.12) ful®ls a non-homogeneous wave equation, dierentiate it with respect to u and v, respectively, we obtain o o 1 1 1 ÿ K
u; v ÿ K
u; v ou ov 2 u v Z 1 p t Jm0
tuJm
tv ÿ Jm
tuJm0
tv dt:
4:1 uv 0
The second derivatives take the form 2 o o2 1 o 1o 3 1 1 ÿ K
u; v ÿ ÿ K
u; v ÿ K
u; v ou2 ov2 u ou v ov 4 u2 v 2 Z p 1 2 00 t Jm
tuJm
tv ÿ Jm
tuJm00
tv dt: uv 0
4:2 Using the relations (3(b)) n n Jn0
z Jnÿ1
z ÿ Jn
z Jn0
z Jn
z ÿ Jn1
z; z z in (4.1) and (4.2) we have 2 o o2 ÿ K
u; v
h
u ÿ h
vK
u; v; ou2 ov2 where
ÿ h
x m2 ÿ 14 xÿ2
m 6 12 :
4:3
4:4
Eq. (4.4) represents a non-homogeneous wave equation.
5. Contact problem with logarithmic kernel Consider the system of the integral equation k Z X i1
ai ÿai
k
t
1 2
k
x ÿ t /i
tdt pfi
x k
Z
1 ÿ1
tanh u jut e du u
under the static conditions Z ai /i
tdt P < 1: ÿai
j
k 2
0; 1; p ÿ1;
5:1
5:2
5:3
102
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
The boundary value problem (5.1)±(5.3) represents the contact problem of a strip occupying the region 0 6 y 6 h, made of material satis®es Hook's law (see Ref. [10]). The strip, in the absence of mass force, lies without fraction on a rigid support, a system of rectangular stamps is impressed into the boundary of a strip y h. Assume the frictional forces in the contact area between the stamps and the strip are small, so it can be neglected. Also, assume the width of the area of contact is independent of the magnitude of the force applied. As in Ref. [3, pp. 32], we can write the kernel in the form Z 1 1 tanh u jut pt xÿy e du ÿ ln tanh ; k 2
0; 1 : t k
t 2 ÿ1 u 4 k
5:4 If k ! 1 and
x ÿ y is very small, so that it satis®es the condition tanh a ' a, then we have pt 4k d ln :
5:5 ln th ln jtj ÿ d 4 p Here the kernel (5.2) takes the form k
t ÿ ln jx ÿ yj d: So (5.1) becomes k Z ai X 1 ln d /i
tdt pfi
x; jx ÿ tj i1 ÿai under the condition (5.3). We introduce the logarithmic potential function 2 3 Z k a i X 1 6 7 Ui
x; y 4 log q d 5/i
tdt: 2 ÿa 2 i i1
x ÿ t y
5:6
5:7
Eqs. (5.7) and (5.3) reduce to the Dirichlet boundary value problem o2 o2
x; y 62
ÿai ; ai ; ox2 oy 2 Ui
x; yjy0 pfi
x
x 2
ÿai ; ai ; p 1 r x2 y 2 ; Ui
x; y ' P ln d r 1 P ln d ! finite term
as r ! 1: r DUi
x; y 0;
D
5:8
The solution of the integral equation (5.6) is equivalent to the solution of the Dirichlet problem (5.8). After the function Ui
x; y in (5.8) has been constructed, the density of the potential /i
x will be determined from the formula
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
/i
x ÿ
1 oUi
x; y lim sgn y p y!0 oy
x 2
ÿai ; ai :
Assume the density source function 1 Wi
x; y Ui
x; y ÿ P ln d : r
103
5:9
5:10
So, Eq. (5.8) can be written as DWi
x; y 0
x; y 62
ÿ ai ; ai ; Wi
x; yjy0 pf
x ÿ P
ln jxj ÿ d
x 2
ÿ ai ; ai ; Wi
x; y ! 0
as r ! 1:
Consequently Eq. (5.9) is transferred to 1 oWi
x; y ÿ pP d
x ; /i
x ÿ lim sgn y p y!0 oy
5:11
5:12
where d
x is the Dirac-delta function. We construct the solution of the boundary value problem (5.11) by the method of conformal mapping (see Ref. [1]), that transforms a given complicated region into a simpler one. To this end, we note that the mapping function p ai ai
5:13 z W
n
n nÿ1
n qe jh ; z x jy; j ÿ1 2 2 maps the region in (x, y) plane into the region outside the unit circle c, such that w0
n does not vanish or becomes in®nite outside the unit circle c. The mapping function (5.13) maps the upper and the lower half-plane
x; y 2
ÿai ; ai into the lower and the upper of the semi-circle q 1, respectively. Moreover the point z 1 will be mapped onto the point n 0. Using the parametric equation of (5.13) and under the condition (5.2), we can rewrite the density source of the logarithmic potential of Eq. (5.10) in the form 2q d Wi
q; h Ui0
q; h ÿ P ai
5:14 ai 1 ai 1 q qÿ coshi ; sin hi Ui0
q; h Ui 2 2 q q In view of Eq. (5.14) the boundary value problem of (5.11) is transformed to o2 Wi 1 oWi 1 o2 Wi 0
q 6 1; ÿ p < h < p; oq2 q oq q2 oh2
104
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
Wi
0; hi 0;
2 Wi
1; hi f0i
hi ÿ P ln d ; a
f0i
hi fi
ai cos hi :
5:15
Consequently, after using the chain rule, Eq. (5.12) is transformed to oWi :
5:16 /i
ai cos hi
pai j sin hi jÿ1 P oq q1 6. Fourier series method To solve the Dirichlet problem of (5.15), we use the Fourier series method (see Ref. [13]) Wi
q; hi
1 X n0
ain
1 p
Z
ain qn cos nhi ; p
ÿp
ÿp 6 hi 6 p;
fi0
h cosnh dh;
a0
1 2p
Z
p
ÿp
fi0
hdh:
6:1
Substituting (6.1) in (5.15), then using the dierentiating result in (5.16) (see Ref. [8]), we obtain ÿ1 p coshi
pai sin hi ; n 1; 2; . . . ;
6:2 /i
ai cos hi ÿ1 P
pai sin hi ; n0 and P
ÿ1 Z p 2 fi0
hdh: 2p ln d ai ÿp
Finally, substitute (6.2) in (5.1), we have the following relationship a l Zi X 1 Tn
t=ai p
log 2=ai d; n 0; ln d p dt p 2 T
x=ai ; n P 1; 2 jx ÿ tj ai ÿ t n n i1 ÿai
where Tn
x is the Chebyshev polynomial of the ®rst kind. We can derive some important relationships from Eq. (6.4): (i) If, n 2m, xi sin ni =2 ; ai sin ai =2 and if n 2m + 1,
ti sin gi =2 sin ai =2 ai
6:3
6:4
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
xi tan ni =2 ; ai tan ai =2
105
ti tan gi =2 ai tan ai =2
ÿ ai < ni ; gi < ai ; ai p; m 0; 1; 2; . . . ; i 1; 2; . . . ; k; the integral operator (6.4) becomes ! k Z ai X 1 ln d Pi
gdg: KP j 2j sin nÿg i1 ÿai 2
6:5
6:6
Hence, with the aid (6.5), the spectral relations (6.4) take the following form " # sin g=2 k Z ai T cos
g=2dg X 2m sin ai =2 1 sin n=2 p l2m T2m ln d sin ai =2 j 2j sin nÿg 2
cos g ÿ cosai i1 ÿai 2
6:7 and n Z X
!
ai
1 d ln nÿg 2 sin i1 ÿai 2 tan n=2 ; l2m1 T2m1 tan ai =2
where
l2m
tan g=2 T2m1 tan cos
g=2dg ai =2 p 2
cosg ÿ cosai
ÿ1
p
2m ; ÿp ln
sin ai =2 d ;
6:8
m P 1; m0
and ÿ1
l2m1 p
2m 1 ;
m P 1:
6:9
(ii) Dierentiating (6.4) with respect to x, we have k 1X p i1 k Z X i1
Z
X Tn
t=ai dt p aÿ1 i Unÿ1
x=ai 2 2 t ÿ x ÿ t a ÿai i0 i ai
aj
ÿaj
n P 1;
dt p 0
jxj < ai ;
t ÿ x a2i ÿ t2
where Un
x=a is the Chebyshev polynomial of the second kind. Using (6.7), (6.8) in (6.10), we have k 1X 2 i1
tan g=2 g ÿ n Tn tan ai =2 cos
g=2 p dg cot 2 2
cosg ÿ cosai ÿai
Z
ai
6:10
106
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
8 0; n 0; jnj < ai ; > > > < tan n=2 cosec
ai =2U2mÿ1 tan ai =2 ; n 2m; m 1; 2; . . . ; > > > : cosec
ai =2U2mÿ1 tan n=2
ÿ1m sin ai tan ai 2mÿ2 ; n 2m ÿ 1 tan ai =2 1 cos ai =2 4
6:11 and tan g=2 cot gÿn Tn tan sec g=2 2 ai =2 p dg 2
cos g ÿ cos ai ÿai 8 ÿ tan n=2 < cosec
ai =2 sec2 n2 Unÿ1 tan
n P 1; ai =2 ÿ : sec a2i tan n2 ; n 0; jnj < ai :
k 1X 2 i1
Z
ai
6:12
The reader must know that, the previous relationships can be derived with the aid of the following relations: dt cos
n=2 cos
n=2 sin
a=2 sec
g=2 p p dn a sin
gÿn 2
cos g ÿ cosa
t ÿ x a2 ÿ t2 2 and
6:13
g n nÿg g g cos cos cos ; 2 2 2 2 2 g n n gÿn n cos cos cos : cos 2 2 2 2 2 cos
7. Contact problem with Karlman kernel Consider a system, in a half-space, of contact problem of frictionless impression of a system rigid stamps in the surface of elastic whose modulus changes according to a power low. Such problem [10] can reduce to the following integral equations n Z ai X /i
tdt
0 6 l < 1;
7:1 l fi
x jx ÿ tj i1 ÿai under the conditions Z ai /i
ydy P < 1; ÿai
7:2
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
we introduce, the general Karlman potential function k Z ai X /i
tdt h il=2 Ui
x; y: i1 ÿai
x ÿ t2 y 2
107
7:3
The solution of Eq. (7.1) with (7.2) reduces to the boundary value problems l oUi o2 o2 0
x; y 62
ÿ ai ; ai ; D 2 2 ; DUi ox oy y oy Ui
x; yy0 fi
x; Ui
x; y ' Prÿl
r ! 1; r
p x2 y 2
7:4
The complete solution of (7.1) is given by l
/i
x k lim sin yjyj y!0
oUi ; oy
C
l=2 x 2
ÿai ; ai ; k p 2 pC
1 l=2
7:5
where C(t) is the Gamma function To obtain the solution of the boundary value problems (7.4), ®rstly, we assume Ui
x; y jyj
ÿl=2
Vi
x; y;
7:6
to eliminate the term oUi =oy. Secondly we use the transformations mapping (5.13), to obtain the boundary value problem in polar coordinates in the form " # 1 1 2 Vi0
q < 1 DVi0
q; h l
2 ÿ l 2 4q sin 2 hi
q2 ÿ 1 ÿl=2 ai 1 qÿ sin hi Vi0
q; hi jq1 fi
ai cos hi 2 q Vi0
q; hjq0 0 where
Vi0
q; hi Vi
ÿp < hi 6 p;
7:7
ai 1 ai 1 q qÿ cos hi ; sin hi Vi
x; y q q 2 2
and D
o2 1 o 1 o2 : oq2 q oq q2 oh2
7:8
The solution of (7.7) can be obtained by separating the variables, for this aim, we assume Vi0
q; h R
qwi
h:
7:9
108
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
Hence, we have
" # d2 Ri dRi q2 2 l
2 ÿ l q ÿ a i Ri 0 q dq do2
q2 ÿ 12 2
0 6 q 6 1
7:10
and d2 wi l
2 ÿ l 2 ai wi 0 dh2 4 sin 2 h
ÿp < hi 6 p;
7:11
where a2i are the constants of separation. As in Ref. [9] the general solution of (7.10) and (7.11), respectively take the form Ri
q qnl=2
1 ÿ q2
l=2
F
l=2; n l; n 1 l=2; q2 ;
R
0 0; 0 6 q < 1; n 0; 1; 2; . . . ; n ai ÿ l=2
7:12
and wi
h jai sin hi j
l=2
Cnl=2
cos hi
ÿp < hi 6 p; n 0; 1; 2 . . .:
7:13
Cnl=2
t
is a Gegenbauer where F
a; b; c; z is the hypergeometric function, while polynomial. Using (7.12) and (7.13) in (7.6), with the aid (7.9), we have Ui0
q; hi qnl F
l=2; n l; n 1 l=2; q2 Cnl=2
coshi ; 1 1 1 1 q qÿ coshi ; sin hi Ui
x; y; Ui0
q; h U 2 q 2 q
0 6 q < 1; ÿp < hi < p; n 0; 1; 2; . . .
7:14
The complete solution of the problem, can be obtained as ®rstly writing Eq. (7.5) in polar coordinates /
ai cos hi
C
l=2
ai sin hi lÿ1 l oUi0 ; p l1 1l lim
1 ÿ q2 q!1 oq p2 C 2
0 < hi < p:
7:15
Secondly using (7.14) after dierentiating with respect q in (7.15), ®nally, we have C
lC
n 1 l=2 lÿ1
ai sin hi Cnl=2
cos hi ; /i
ai cos hi p l p2 C
1 l=2C
n l 0 < hi < p:
7:16
Hence, inserting (7.16) in (7.1), after using the second equations in (7.4), we obtain the following spectral relationships:
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111 k Z X i1
ai ÿai
Cnl=2
t=ai dt l
1ÿl=2
jx ÿ tj
a2i ÿ t2
kn Cnl=2
x=ai ;
kn pC
n ln!C
l cos
pl=2ÿ1
109
jxj < ai ;
n P 0:
7:17
8. Conclusions From the above results and discussions, the following may be concluded: (1) For the contact problems, when a system of impressing stamps lay on a strip of an elastic material and under certain conditions, most of its reduce to a system of integral equations of the ®rst kind with dierent kernel k Z ai X K
u; v/i
vdv fi
u; K/ i1
ÿai
K : L2
ÿai ; ai ! L2
ÿai ; ai ;
8:1
where K
u; v 2 C
ÿ ai ; ai ÿ ai ; ai ;
i 1; . . . ; k
and the discontinuous kernel of (8.1) must satisfy the relation Z ai Z ai k 2
u; vdu dv m2 < 1: ÿai
ÿai
Also the unknown potential functions /i are continuous in the domain of integration and satisfy the normal condition. Moreover the unknown potential functions satisfy the Lipschitz condition with respect to the second argument. (2) The integral operator (8.1) is a positive compact and self-adjoint operator, so we may write K/ kn /n where kn and /n are the eigenvalues and the eigen functions of the integral operator, respectively. (3) A system of a three dimensional contact problem of frictionless impression of rigid stamps in the surface of an elastic (G, m) half-space occupying the domain ÿ1 < x; y < 1; z 0 reduces to an integral equation of the ®rst kind with symmetric kernel. (4) The potential function kernel reduces to the Weber±Sonin integral formula Z p 1 Jm
tuJm
tvdt; K
u; v uv 0
which represents a non-homogeneous wave equation (see Eq. (4.4)). (5) The value of the kernel (4.12) can be represented in the Legender polynomials as follows:
110
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
K
u; v
1
uvm1=2 X C
n m 1=2Pnm
uPnm
v ÿ1 2 2 n0 C
1 m n
m 2n 1=2
Pnm
x is Legender polynomial;
n 0; 1; 2; . . .
8:2
(6) The contact problem of the zero harmonic symmetric kernel of the potential function is included as a special case when m 0. Also, the contact problem of the ®rst and higher order
m 1; 2; . . . harmonic is included as a special case. (7) A system of plane contact problem of the impression of two symmetrically arranged stamps in an elastic half-space with logarithmic kernel, or with Karlman kernel, represents a Fredholm integral equation of the ®rst kind and its kernel takes the form Z p 1 m t J1=2
tuJ1=2
tvdt; 0 6 m < 1;
8:3 K
u; v uv 0
where m 0 for logarithmic kernel and 0 6 m < 1 for Karlman kernel, also 1=2 for symmetric and skew-symmetric, respectively. (8) This paper can be considered as a generalization of the work of contact problem in continuous media which is discussed in Refs. [8,10,11]. 9. Unlinked references [4,5] References [1] A.G. Sveshnikov, A.N. Tkhonov, The Theory of Functions of Complex Variable, Mir, Moscow, 1982. [2] E.V. Kovalenko, Some approximate method of solving integral equations of mixed problems, Prkil. Math. Mech. 53 (1) (1989) 85±92. [3] G.Ya. Popov, Contact Problems for a Linearly Deformable Base, Kiev-Odessa, 1982. [4] H. Bateman, A. Erdely, Higher Transcendental Functions, vol. 1, Nauka, Moscow, 1963. [5] H. Bateman, A. Erdely, Higher Transcendental Functions, vol. 2, Nauka, Moscow, 1973. [6] I.Ya. Shtearman, Contact Problem of Elasticity Theory, Moscow, 1949. [7] M.A. Abdou, Fredholm integral equation of the second kind with potential kernel, J. Comp. Appl. Math. 72 (1996) 161±167. [8] M.A. Abdou, S.A. Hassan, Boundary value of a contact problem, PV. M.A. 5 (3) (1994) 311± 316. [9] N.V. Machalan, Theory and Application of Mathieu Functions, Mir, Moscow, 1954. [10] V.M. Aleksandrov, E.V. Kovalenko, Problems in The Mechanics of Continuous Media With Mixed Boundary Conditions, Nauka, Moscow, 1986. [11] V.M. Aleksandrov, Development of The Theory of Contact Problems in The USSR, Nauka, Moscow, 1976.
M.A. Abdou / Appl. Math. Comput. 118 (2001) 95±111
111
[12] V.M. Aleksandrov, D.A. Pozharskii, On contact problems for wedge-shaped plates, J. Appl. Math. Mech. 55 (1) (1991) 114±119. [13] V.S. Vladimirov, Equations of Mathematical Physics, Mir, Moscow, 1984. [14] V.S. Prostenko, V.G. Protsenko, Contact problem with several stamps, Dokl. Akad. Nauk. Ukrssr Ser. A, 10, 1972. [15] V.S. Prostenko, V.L. Rvachev, Plate in the form of an in®nite strip on an elastic half-space, Prkil. Math. Mech. 40 (2) (1976) pp. 298±305.