Sie sind auf Seite 1von 11

Electronic Journal of Differential Equations, Vol. 2000(2000), No. 15, pp. 111.

ISSN: 1072-6691. URL: http://ejde.math.swt.edu or http://ejde.math.unt.edu


ftp ejde.math.swt.edu ftp ejde.math.unt.edu (login: ftp)
A diusion equation for composite materials

Mohamed El Hajji
Abstract
In this article, we study the asymptotic behavior of solutions to the
diusion equation with non-homogeneous Neumann boundary conditions.
This equation models a composite material that occupies a perforated
domain, in R
N
, with small holes whose sizes are measured by a number r.
We examine the case when r <
N/(N2)
with zero-average data around
the holes, and the case when lim0 r/ = 0 with nonzero-average data.
1 Introduction
As a model for a composite material occupying a perforated domain in R
N
,
the diusion equation with non-homogeneous boundary conditions has been the
object of many studies. In particular, we are interested in the properties of the
solution to the equation
div (A

) = f in

,
(A

(x)u

) n = h

on S

, (1)
u

= 0 on ,
where

is the perforated domain obtained by extracting a set of holes S

from
, f and h

are given functions, and A

is an operator in the space


M(, ; ) =
_
A [L

(R
N
)]
N
2
: (A(x), ) [[
2
,
[A(x)[ [[ R
N
, p.p x
_
which is dened for all real numbers and .
When h

L
2
(S

) and the domain has holes of size r

, solutions to (1) have


been studied by D. Cioranescu and P. Donato [3] for r

= and A

(x) = A(
x

)
with A M(, ; ), and by C. Conca and P. Donato [5] for A = I and r

.
Using the concept of H
0
-convergence introduced by M. Briane et al. [2], P.
Donato and M. El Hajji [6] showed convergence of solutions in not-necessarily

1991 Mathematics Subject Classications: 31C40, 31C45, 60J50, 31C35, 31B35.


Key words and phrases: Diusion equation, composite material,
asymptotic behavior, H
0
-convergence.
c 2000 Southwest Texas State University and University of North Texas.
Submitted October 14, 1999. Published February 22, 2000.
1
2 Diusion equation for composite materials EJDE2000/15
periodic domains. The H
0
-convergence is proven by showing strong convergence
in H
1
() of the distribution, concentrated on the boundary of S

, given by

h
,
H
1
(), H
1
0
()
= h

,
H
1/2
(S), H
1/2
(S)
, H
1
0
(). (2)
This method allows the study the asymptotic behavior of solutions to (1) for
h

L
2
(S

) with r

>
N/N2
(see [6]), and for perforated domains with
double periodicity with h

H
1/2
(S

) and r

/ 0 as 0 (see T. Levy
[8]).
In this article, we study the perforated domains with r

<
N/N2
, and
perforated domains such that r

/ 0 as 0. In these situation the distri-


bution given by (2) does not converge strongly in H
1
(), and so the method
described above can not be applied. In spite of this, we describe the asymptotic
behavior of solutions to (1) using oscillating test functions. This method was
introduced by L. Tartar [10] and has been used by many authors.
2 Statement of the main result
Let be a bounded open set of R
N
, Y = [0, l
1
[.. [0, l
N
[ be a representative
cell, and S be an open set of Y with smooth boundary S such that S Y . Let
and r

be terms of positive sequences such that r

. Let c denote positive


constants independent of . We denote by (r

S) the set of translations of r

S
of the form k
1
+r

S with k Z
N
. Let k
l
= (k
1
l
1
, .., k
N
l
N
) represent the holes
in R
N
.
We assume that the holes (r

S) do not intersect the boundary . If S

is
the set of the holes enclosed in , it follows that there exists a nite set / Z
N
such that
S

=
_
kK
r

(k
l
+S).
We set

= S

, (3)
and denote by

the characteristic function of

. Let V

denote the Hilbert


space
V

=
_
v H
1
(

), v
|

= 0
_
equipped with the H
1
-norm. Let A(y) = (a
ij
(y))
ij
be a matrix such that
A
_
L

_
R
N
__
N
2
,
A is Y-periodic, and there there exist > 0 such that
N

i,j=1
a
ij
(y)
i

j
[[
2
, a.e. y in R
N
, R
N
. (4)
We note that for every > 0,
A

(x) = A(
x

) a.e. x in R
N
. (5)
EJDE2000/15 Mohamed El Hajji 3
In this paper, we study the system
div(A

) = 0 in

,
(A

).n = h

on S

, (6)
u

= 0, on ,
where

is given by (3), A

is given by (5), and h

is given by
h

(x) = h(
x
r

), (7)
where h L
2
(S) is Y -periodic function. Set
I
h
=
1
[Y [
_
S
hd . (8)
We examine the case where r

<
N/(N2)
with I
h
,= 0, and the case where
lim
o
r

/ = 0 with I
h
= 0. The following result describes the asymptotic
behavior of the solution to (6) in the two cases.
Theorem 1 Let u

be the solution of (6). Suppose that one of the following


hypotheses is satised
I
h
,= 0 and
_
_
_
lim
0
r

N/(N2)
= 0 if N > 2,
lim
0

2
(ln(/r

))
1
= 0 if N = 2,
(9)
or
I
h
= 0 and lim
0
r

= 0 . (10)
Then, for every > 0, there exists an extension operator P

dened from V

to
H
1
0
() satisfying
P

L
_
V

, H
1
0
()
_
, (11)
(P

v)
|
= v v V

, (12)
|(P

v)|
(L
2
())
N C |v|
(L
2
())
N , v V

. (13)
such that
(
r

)
N/2
P

u weakly in H
1
0
(), for N > 2
and
P

[(
r

)
1/2
(log

r

)
1/2
u

] u weakly in H
1
0
(), for N = 2,
where u is the solution of the problem
div(A
0
.u) = 0 in ,
u = 0 on ,
4 Diusion equation for composite materials EJDE2000/15
and the matrix A
0
= (a
0
ij
)
ij
has entries
a
0
ij
=
1
[Y [
_
Y
(a
ji

N

k=1
a
ki

j
y
k
dy), (14)
and
j
is a Y -periodic function that satises
div(A
t
(y
j

j
)) = 0 in Y (15)
Remark. One can replace the rst equation of system (6) by
div(A

) = f

in

,
with
(
r

)
N/2
f

f weakly in L
2
() if N > 2,
(
r

)
1
(ln(/r

))
1/2
f

f weakly in L
2
() if N = 2.
Then u will be the solution of
div(A
0
.u) = f in ,
u = 0 on .
This approach has been used in [5] for the case A = I, in [3] when A = I and
r

= , and in [6] for the case where r

>
N/(N2)
using the H
0
-convergence
and some arguments given by S. Kaizu in [7].
3 Proof of the main result
Observe rst that S

is admissible in , in the sense of the H


0
-convergence
([6, 4, 5, 10]). Then there exists an extension operator P

satisfying (14).
On the other hand, the matrix A
0
can be dened by
A
0
= /
Y
(
t
Aw

) =
1
[Y [
_
Y
t
Aw dy, R
N
,
where for every R, w

is the solution of the problem


div(
t
Aw

) = 0 in Y, (16)
with w

y Y -periodic.
For x R
N
, let
w

(x) = w

(
x

). (17)
To simplify notation, let

=
_
(r

/)
N/2
if N > 2,
(r

/)
1
(ln(/r

))
1/2
if N = 2.
(18)
EJDE2000/15 Mohamed El Hajji 5
Taking u

as a test function in the variational formulation of (6), and using the


classical techniques of a priori estimates, one can easily show the existence of
two constants c and c independent of such that
c

|(

)|
L
2
()
c.
Hence, from (13), up to a subsequence,
P

) u weakly in H
1
0
(). (19)
Set now

= A

[P

)]. Then using (19), (11)-(13) and (4)-(5), one shows


that

is bounded in L
2
(), and so up to a subsequence

weakly in L
2
(). (20)
Case where (9) is satised. Let D(). Then from the variational
formulation of (6), one has
_

.dx =

_
S
h

d . (21)
If N() denotes the number of the holes included in , one has then
[

_
S
h

d[ ||
L

()

kK
_
(r(S+k))
[h(
x
r

)[ d(x)
c

N()r
N1

_
S
[h[ d (22)
c

r
N1

N
[S[
1/2
|h|
L
2
(S)
.
From (18), one can write

r
N1

N
=
_
(
r
N2

N
)
1/2
if N > 2,
[
2
(ln(/r

))
1
]
1/2
if N = 2,
and so in virtue of (9),
lim
0

N1

N
= 0, (23)
hence
lim
0

_
S
h

d = 0 .
On the other hand, it is easy to show that

1 strongly in L
p
() p [1, [,
hence _

.dx
_

.dx. (24)
6 Diusion equation for composite materials EJDE2000/15
One can deduce that, as 0 in (21),
_

dx = 0, D().
Consequently
div = 0 in . (25)
It remains to identify the function . Let w

be the function dened by (16)-


(17). Then
w

x weakly in H
1
(), and so L
p
() strong p < 2

where 1/2

= 1/2 1/N, with N 2. Let D(), by choosing w

as a test
function in the variational formulation of (6), one has
_

(w

) dx =

_
S
h

dx. (26)
To pass to the limit as 0 in (26), we set
_

(w

) dx = J

1
+J

2
, (27)
where
J

1
=
_

.w

dx and J

2
=
_

.dx.
Using the results given by G. Stampacchia in [9] (see also [1]), one can deduce
that w

(Y ), so

x strongly in L
2
(). (28)
This with convergence (20), gives
J

1
=
_

dx
_

xdx as 0 . (29)
Now, we may write
J

2
=
_

dx
_
S

dx. (30)
One the one hand,
_

dx =
_

t
A

[P

)] dx
_

t
A

[(

] dx
=
_

t
A

[(

] dx
EJDE2000/15 Mohamed El Hajji 7
because, from the denition of w

,
_

t
A

[P

)] dx = 0 .
From the denition of A
0
,
t
A

A
0
weakly in L
2
(). From (19), up to
a subsequence,
P

) u strongly inL
2
(),
which implies
_

t
A

) dx
_

A
0
udx.
Hence
_

dx
_

A
0
u dx. (31)
On the other hand,
[
_
S

dx[ c|

|
L
2
()
|w

|
L
2
(S)
. (32)
Since |

|
L
2
()
is bounded,
[
_
S

dx[ c|w

|
L
2
(S)
. (33)
Note that
|w

|
2
L
2
(S)
=
_
S
[(w

)(x)[
2
dx
=

kK
_
r(S+k)
[(w

)(x)[
2
dx
=

kK
_
r(S+k)
[(
y
w

)(
x

)[
2
dy
= N()
N
_
r

S
[(
y
w

)(y)[
2
dy
c
_
r

S
[w

[
2
dy .
Since r

/ 0 and w

H
1
(Y ), it follows that
_
rS/
[w

[
2
dy 0 .
Using (33), one has
_
S

dx 0 as 0.
8 Diusion equation for composite materials EJDE2000/15
This, with (30) and convergence (31) imply that
J

2

_

A
0
u dx. (34)
Next we pass to the limit in the right hand of (26). With the same argument
as in (22),
[

_
S
h

d[ c

N()r
N1

_
S
[hw

[ d
c

r
N1

N
|w

|
L
2
(S)
[S[
1/2
|h|
L
2
(S)
.
Since we have shown that
lim
0

r
N1

N
= 0,
from (9) one deduces that

_
S
h

d 0.
Finally, by passing to the limit as 0 in (26), and using (32) and (34) one
obtains
_

xdx
_

A
0
u dx = 0,
hence, from (25) it follows that
_

dx =
_

A
0
udx D(), R
N
,
i.e., = A
0
u.
Case where (10) is satised. Let D(). Then from the variational
formulation of (6),
_

.dx =

_
S
h

d. (35)
The arguments used the proof of (24) can be applied here to obtain
_

.dx
_

.dx.
To pass to the limit in the right-hand side of (35), we introduce N as the
solution to
div N = 0 in S,
N.n = h on S .
EJDE2000/15 Mohamed El Hajji 9
The existence of N is assured by the hypothesis I
h
= 0. Set N

(x) = N(
x k
r

),
for x in (Y r

S)
k
. Then
_
S
h

d =
_
S
.N

dx,
hence

[
_
S
h

d[

||
L
2
(S)
|N

|
L
2
(S)
.
Note that |N

|
L
2
(S)
c(
r

)
N/2
|N|
L
2
(S)
, so

[
_
S
h

d[ c

(
r

)
N/2
||
L
1
(S)
. (36)
Since

(
r

)
N/2
=
_
1 if N > 2
(ln(/r

))
1/2
if N = 2,
it follows from (36), when N = 2, that
lim
0

[
_
S
h

d[ = 0 .
For N > 2, one has

[
_
S
h

d[ c||
L
2
(S)
.
Since

1 strongly in L
p
(), for all p [1, [ and D(), one deduces
that
_

(1

)[[
2
dx 0 .
Hence, by passing to the limit as 0 in (35), one obtains
_

.dx = 0 ,
then div = 0 in .
Let w

be the function dened by (16)-(17) and D(). As in the previous


case, by using w

as a test function in the variational formulation of (6), one


has
_

.(w

) dx =

_
S
h

d .
From (27), (29) and (34), one has
_

.(w

) dx
_

x.dx
_

A
0
.u dx. (37)
10 Diusion equation for composite materials EJDE2000/15
Now we show that

[
_
S
h

d[ 0. (38)
One has

[
_
S
h

d[
=

[
_
S
(w

).N

dx[

[
_
S
.w

.N

d[ +

[
_
S
.w

.N

d[

|N

|
L
2
(S)
|w

|
L
2
(S)
+

|N

|
L
2
(S)
|w

|
L
2
(S)
c

(
r

)
N/2
_
|w

|
L
2
(S)
+|w

|
L
2
(S)
_
c

(
r

)
N/2
_
||
L

()
|w

|
L
2
(S)
+||
L

()
|w

|
L
2
(S)
_
.
c

(
r

)
N/2
_
|w

|
L
2
(S)
+|w

|
L
2
(S)
_
.
Note that
|w

|
2
L
2
(S)
=
_
S
[w

[
2
dx =

kK
_
r(S+k)
[w

[
2
dx
N()
N
_
r

S
[w

[
2
dx c
_
r

S
[w

[
2
dx.
Since w

H
1
(S) and r

/ 0,
lim
0
_
r

S
[w

[
2
dx = 0 .
Hence |w

|
L
2
(S)
0. On the other hand, one has |w

|
L
2
(S)
c. Finally,
as
lim
0

(
r

)
N/2
= 0 ,
one deduces (38). This and (37) completes the proof, using the same arguments
as in the previous case.
Acknowledgments The author would like to thank Professor Patrizia Do-
nato for her help on this work.
References
[1] Brezis H. Analyse fonctionnelle, theorie et applications. Masson, 1992.
EJDE2000/15 Mohamed El Hajji 11
[2] Briane M., A. Damlamian & Donato P. H-convergence in perforated
domains, volume 13. Nonlinear partial Dierential Equations & Their Ap-
plications, Coll`ege de France seminar, H. Brezis & J.-L. Lions eds., Long-
man, New York, ` a paraitre.
[3] Cioranescu D.& Donato P. Homogeneisation du probl`eme de Neumann
non homog`ene dans des ouverts perfores. Asymptotic Analysis, 1:115138,
1988.
[4] Cioranescu D. & Saint Jean Paulin J. Homogenization in open sets
with holes. Journal of Mathematical Analysis & Applications, 71:590607,
1979.
[5] Conca C. & Donato P. Non-homogeneous Neumann problems in
domains with small holes. Modelisation Mathematiques et Analyse
Numerique, 4(22):561608, 1988.
[6] Donato P. & El Hajji M. An application of the H
0
-convergence to
some nonhomogeneous Neumann problems. GAKUTO International Series,
Math. Sc. and Appl., Gakkotosho, Tokyo, Japan, 9:137156, 1997.
[7] Kaizu S. The Poisson equation with nonautonomous semilinear doundary
conditions in domains with many tiny holes. Japan. Soc. for Industr. &
Appl. Math., 22:12221245, 1991.
[8] Levy T. Filtration in a porous ssured rock: inuence of the ssures
connexity. Eur. J. Mech, B/Fluids, 4(9):309327, 1990.
[9] Stampacchia G. Equations elliptiques du second ordre ` a coecients dis-
continues. Presses Univ.Montreal, 1966.
[10] Tartar L. Cours Peccot. Coll`ege de France, 1977.
Mohamed El Hajji
Universite de Rouen, UFR des Sciences
UPRES-A 60 85 (Labo de Math.)
76821 Mont Saint Aignan, France
e-mail: Mohamed.Elhajji@univ-rouen.fr