Zheng Liang1992
Zheng Liang1992
Zheng Liang1992
Published by SUT,
Shanghai, China
Ra.vh'igh-Green
ith'ntitv. The content o f the paper provides the powetJid theoriesfor the application of BEM
in thL~ prohh, m.
Ke][ words
I.
Introduction
The research work for analysis of the plates on elastic foundation can date back to last century.
During the period of one hundred years, many famous scholars, such as Timosheko, WoinshyKriger, Reissner, Vlazov et al., have made great efforts to research the analytical method, and a lot
of outstanding achievements have been obtained. But because of the limitations of the analytical
method, these achievements can only solve a small number of engineering problems. In the recent
twenty years, the BEM (boundary element method) has been widely used in many fields. For the
bending problem of plates on Winkle foundation, KatsikadelisI-~.~1used the fundamental solution
derived by Hertz or Wymanr~l and gave an application of BEM. But, to another more complicated
problem
the bending problem of plates on two-parameter foundation, because the fundamental
solution derived by Vlazov was not expressed as elementary function, the application of BEM is
very difficult. In Ref. [5],the fund~mlental solution which is expressed as Fot.rier-Bessel integral form
is handled by numerical integration, but, its accuracy can not be guaranteed effectively, In addition,
the theory of the establishing process for boundary integral equations are still not sufficient.
By means of the Fourier integral transformation of generalized function, the fundamental
solution for the problem which is expressed as Hankle function of complex variable is obtained in
this paper, and the fundamental solution is expanded into power series of real variable. The
uniformly convergence of the series is proved. On the basis of the above work. two boundary
integral equations for bending problem of plates on two-parameter foundation which are suitable to
arbitrary shapesandarbitrary boundary conditions are established, by using of the Raylcign-Green
identity. The content of the paper provides powerful theories for the application of BEM in this
problem.
657
658
II.
Differential Equation
or
Lw=q
(2. ] )
w(s)--w.(s)=O~ O(s)--O.(s)=O
w(s)-w.(s)=O, M(s)-M.(s)=O
V(s)-V.(s)=O,
M(s)-M.(s)=O
U'
Kw
On
M(w)=-D[
Kz
VZ-(1-v)O--g~-]w
T(w)=--D(1-v)
KZw
dnK---7
(2.2)
K
V(w)=Q+_~T=_D[~___nV2+(]._v)
K g-rTK~'
K2 ] w
Ks
Notice that
in which f l = < n , r), O=(r, x ' ) , n means the normal direction of boundary, t means the
tangential direction of the boundary, as shown in Fig. 1.
Fundamental
1. D e r i v a t i o n
659
Solution
of the fundamental
solution
Supposing that there is a unit concentrated force at point P ( ~ , r/) in a infinite plate, then the
fundamental solution must satisfy the following equation:
L w * = a ( x - ~ , y--r~)
(3.1)
w*=
r+==r+~ I
- ~ - J _ = J_: T e
(3.2)
thus,
(3.3)
x -- ~-----r c o s r
col=o9c o s a ]r
y-- r/=rsin~b }
c02----cosina"
r = = ( x - - ~ ) 2 q - ( y - r / ) 2, coZ=co~+co~
dog~dwa=ogdcoda, i[cot(x-~)+co2(y-q) ] = i o . r
Then, Eq. (3.2) can be written as:
.
w =)'-~I
4~
+OO
Jn
~t
+1:~
---x~exp['to'r]dc-oda=~f
J-n
4~"
1
---r--codco.~..exp[it~.r]da
Jo
A~
"
"
1 r+~ Jo(cor).codco
-.
2,r Jo
l__L_[+:o
-- 2~rDJ0
Supposing that
~"
Jo(o~r)
co++Gpco2/D+k codo~
ap
}=
G, - I I i k \
-Tff+T4
(3.4)
z G, \~
) i
- 4D---<0, under
this
660
2" , l ]
2Jra a,--a,L3o
~..2--u~
.2 c~176
,,,
Jo
--a'2
that is
1
w * = 4D d 4 ( k / D ) - ( G , / D f
EH~~176
(3.5)
Eq. (3.5) represents the fundamental solution which is expressed by Hankle function of
complex variable. In the deriving process of the fundamental solution expression (3.5), an integral
formula is applied:
+~oI0(bx)__,..
=i r-r~,)
3r
~01
where H~ '~ indicates the first kind of zeroth order Hankle function.
2. Series e x p r e e i o n o f H a n k l e f u n c t i o n
It can be known from Ref. [8] that
H~ 1>(z) =Y0(z) + i Y o ( z )
(3.6)
in which Yo(z)is the zeroth order Neumann function and ]o(Z)~ Yo(z) can be expressed as the
infinite series:
~--,(-- I) k
Jo(z)----kz.~o -(~-~ ( + ) "
yo(z)=2
(3.7/
j o ( z ) , m y z ---~2 .L;,o
~.~~(--1) k r
(+)"
(3.8)
1
where ~b(k-kl)=! - b - ~ q - . . - + T - C ,
where C B = 0 . 5772] ... is a Euler constant.
Therefore
Hi"
,-~ % 2 k
(3.9)
. /
z=pexpEifl3 ~ p
(3.1o)
( z ) - - - - ( 1 - 2 f l ) k ~ - (--1)k/P~
i,-~-]~2k eos2kfl
----2
i n ~-,@o
p ~-~ (--1)kip
~zk sin2kfl+ 2-~-~' (--1)k
-~!~r
(--~7.) k-2"]
r
I m H 0('(z)
----(I
k-O
661
(3.11)
"
. o0 (__1)k~p.z~
,.0
+ )
eos2kp
(3.i2)
Re and Im indicate separately the real part and imaginary part of the Hankle function.
3. Series e x p r e s s i o n o f t h e f u n d a m e n t a l s o l u t i o n
Assumng that
G, t .T ~1 / 4[ L. l~k-\ ) - L'-~--)
I G , \2 i----(p,exp [ifl,])=
a~ --------~-~-
(3.13)
a]=- 2DG"
(3.14)
2"~/-/4{k-D-]-k
1
.~
(~)'i=(p,
expEifl,])z
then, the following result can be obtained by comparing the above two equations:
Pt=Pz---- 4~/ k / D , flt-{-fls-----=
1~
./~_(k,~
" ~/
T k - ~ ' ] [.- kG,
~ ] .~' ' cosZpt= - ~ / G,
4Dk
sin2fl~=y
(3.15)
where p, and Pz are separately the moduli of at and as, fit and flz separately the modular angles of
a~ and as. Substituting the series expression (3. i 0) of H 0(1~(z) into expression (3.5) of the fundamental solution w* , the imginary part can disappear. So the following expression of w* is
obtained:
7./3 t * ~
[2
Ic-O
..
-~:--b~a' ~--~i~-f~--~---]
sin2kfl~
4 ~o ( _ !)k
. p,r
,k
(3.16)
Eq. (3. i 6) is just the series expression of the fundamental solution for bending problem of
plates on two-parameter foundation. It can be provexi that if G p = 0 in Eq. (3.16), the fundamental
solution becomes
w*=--Kei ( p,r) / (2xDp~ )
(3.17)
because of fl,----.~-,
fh=7
Eq. (3.17) represdnts the fundamental solution for bending problem of plates on Winkle
foundation, it is just the same as the fundamental solution derived by Wymanm, in which Kei (p,r)
is Kelvin function of zeroth order. This conclusion also shows that expression (3.16) of the
fundamental solution is dependable.
By means of th~ Cauchy-Hadamard criteriontg] for convergence of power series, it is known
662
that every term in the right-hand side of Eq. (3.16) is'uniformly convergent for any0~<r< +oc.
It needs to be explained that it is necessary to retain the term In ( p , r / 2 ) o f the series expression,
but not to expand it into power series again, because it Can be known that r---- 0is a singular point of
w* according to this term. It is widely used in the application of BEM
IV.
Integral R e p r e s e n t a t i o n o f S o l u t i o n
Eq. (2.1) can be transferred into an equivalent form of integral equation. Considering the
integral equality:
IIo(w*Lw-wLw*)d~
=II(Dw*v'vZw-G,w*V2W+ kw*w
-- Dwv2V2W* +Gpwv~w * - k w. w* )dg-J
that is,
for any two functions, w and v which are four times continously differentiable inside region ~ ,
and three times on its boundary F , the Rayleigh-Green identity is
II(vV'V'w--wv'v2v)d~
Jr
0n
- -0-~-- V
wt~.
-~-J
(4.2)
where pqfi.Q.
Applying Eq. (2.2), the following equality can be proved:
DIr(W.~n.,V w-w~-~
cl
VZW.
Ow*
z 9 Ow -z .\,r.
On V m+--fffi-V w ] a J
= I r w * V ( w ) d F - I r wV(w*)dF-
(4.3)
663
Thus, the integral representation of solution for bending problem of plates on two-parameter
foundation can be expressed as:
Ow*
in which
[IQqw*d~
OW
w*
(4.41
is an integration term in domain ~ , and the kernel functions in Eq. (4.4) are
Before deriving boundary integral equation, it is necessary to study the properties of the
fundamental solution (when r ~ 0 ) further.
Let
oo
- ~ - - m-~---2_, , , - 7 ~ , ~--f-- I
z #-0 ~,KI)
z /
~--z.~ (---k~yrl)r
k-o
7- , - -
sin2kfll
(5.1)
[J
A----1/4D~/4(k / D )
thus: w*----A.uo(r)
Apparently, while r-~O , the function
lim
sin2kfl,
u0(r)
(5.2)
- (G,/D) =
(5.3)
behaves the following approximation:
Uo(r)f2(fl, - ill)
r--*O
u[ (r)~.-~(fl,-- flD(~)2.rcos2flt + ~ ( ~ ) Z r (
2ln%f-1
)sin2fll
--~-~(2)(-~)Zrsin 2fl,
u"~(r)'~ 4--'-~(flz--flt)('~") z cOs2[3' q-4 \ 2 , \
2 1 1 1 % - t - 3 )sin2fl,
-8~(2)(-~)'sin2fl,
8 .
l p x2 1
a** ( r ) ' - - ~ - s * n Z f l , ~ - )
-7-
u~,~(r).~_8sin2fl, [p,~=7 1r
For integral equation (4.4), let p-.P6F, then, the boundary integral equation usually can be
obtained. But, while P->PEf, every term in Eq. (4.4) except for the integration in domain $'2 will be
664
a singular integral. Therefore, their existence and continuity must be proved separately. That is,
studying the following integrals:
I,=I
I2=j7--tgw
~ * M (w)dr=AI U'o(r)M(w)cos3dF
l~= I r % ~ M (w*) +G,w* 3dF= AIr O~--~n - D~ (cos'fl +vsin'fl)a",
1,=AIrw[V(w. ) + G,---~-]dI"
Ow*
=AIrw{--Dcosfl(u: ++u"O-rl--ru'~)-D(l~ v)
[sinflsin2fl
-7".t]
-7~176
+G,u'o (u)cosB}dF
At point PE/", making a tiny circular arc and taking P as its center, the radius of the circular arc is e.
Thus, on boundaryF0, cos fl = 1, sin fl -- 0. (because of the same direction of r and n ) and r = e
For each integral above, because the first term in the right-hand side of the following identity
Ir (...)dF=lim~r '
(-..)dF+lim~
(...)dF
has not involved singularity, it is only necessary to study the second term in the right-hand side of
the identity.
With the help of the approximation of the function uo(r), the following conclusions can be
proved (because the values of V(w) ,M(w) ,dw/dn and w are all limited):
lim[
r
Jr~
lim{
e~O dro
uo(r)V(w)dI"=O
u~(r)M(w)cosfldF=O
l i m ( Ow ~-D[(cosZfl+vsin'fl)u,~(r)+(ginafl+vcos,fl)_~]
,.-o J On t
+GoU~(r) }dl-'----O
Therefore, the integrals I,, I: and 13 exist, and are continuous as p ~ P E F . They can be integrated
under proper sense.
For integral term I :
I:=limA.J
e-.o
w{-Dcosfl(u~,+)!-u" o-Tuo)
1 ,\
ro
r
D(1-v)
9[ sinflsingfl
X(Ug--~-uO"+r-~U~)+ 2cos2fl( c~
'
665
by means of con fl= 1, sin fl=O, R=r-----e and substituting the approximate expressions of
u~, uxo, u*0 into the above integral term, it can be written as:
G,u~dF
+limA.~
9 ~ "'* 0
~'
Moreover, because of l i r a
A i
e'-'~O
G, . u'odF=O
, therefore
Pe
I : = - AD16sin2fl,(P']21im
Jr "
X 2 ] ~-o It. !wdr
r
16
pl z
l " -- ] " # ~
t ~
ne~
0
o
pto P
Substituting the values of A, p~ and sin2flj into the above equality, the following result is
obtained:
11=
2sr--a w(P)
2:r
(5.4)
in which a is the internal angle of tangent line at point P on boundary F , as shown in Fig. 2. For
smooth boundary,a-----:r. Therefore, integral I 4 exists, but is discontinuous as p~P6.I" because
there is a jump value at point
P, I]=
2~-a w(P)
2x
while p ~ P , the following
obtained:
tln
.I P ~ t l
666
C(P) Ow(P)
Onp =IIo qw~dO+ ~r w~V(w)- Ir to[ V (w~) + Gp ~cgw~
j ]dF
d--~--EM(w'~)+Gpto'~]dF
_ [ Ow~ M ( w ) d F + [
Jr an
jr on
(5.6)
All kernel functions in Eq. (5.5) and Eq. (5.6) are referred to the Appendix.
It needs to be explained that in the above two boundary integral equations, the term induced by
the discontinuity of the twisting moment on non-smooth boundary has not been taken into account.
But it can be easily solved for the case by adding the term into the right-hand side of Eq. (5.5) as well
as Eq. (5.6) directly.
VI.
Conclusions
Acknowledgement
The writers wish to express their thanks to Professor Chen Shan-lin for
his sincere help, encourgrnent and valuable advice.
Appendix
Kernel functions in Eq. (5.5) or Eq. (4.4) are as follows:
w*=Auo(r)
~-n =Aud(r)cosB
M (to')= -
DA[(cosZfl+ vsinZfl)uo', ( r ) +
V(w~176
(s irtZfl+vcosZfl)~]
"uo'(r)]
667
9 l+v\
D(~--V)A{ u~'(r)coscp(cos2~+~_v]
M(w,')=
~'
"
+ sing, s i n2/5']
(r)-r2-T-uu"(r)+2u"'(r))
+
. r.,
cos/3
1' t u " (r)--luo '(r) )[cos(2/3--q~)-l-cos2/3cosq)]
r t"~-o
+2( cos/3
r
u~( r)--r-'7.
in which /3= <n, r>, ~v--<n., r>, /~ is the radius of curvature of the boundary curve, as shown in
Fig. 2 and Fig. 3.
References