Professional Documents
Culture Documents
Contaminant Transport in Fractured Porous Media by Fourier Series
Contaminant Transport in Fractured Porous Media by Fourier Series
of Mathematics,
University of Melbourne,
The steady-state transport of a decaying contaminant in a fractured porous rock matrix by twodimensional diffusion and vertical advection is treated by a Fourier sine transform technique for the
cases of a single vertical fracture and a periodic array offractures. The general case of unequal Peclet
numbers in the fracture and matrix regions is reduced to a first-kind Fredholm integral equation in
the transformed interfacial flux. The asymptotic behavior of the solution for large and small vertical
positions relative to the contaminant source is analyzed. For the special case of equal Peclet numbers
an exact analytical expression for the steady-state concentration has been derived for an arbitrary
source distribution. A method of calculating an accurate analytical approximate solution to the general
mismatched Peclet number case is proposed, and its accuracy verified by comparison of its predictions
with the exact numerical solution obtained by a boundary integral method.
Keywords: contaminant,
dispersion,
advection,
Introduction
The occurrence of fractures in porous media is an important factor in the movement of a contaminant by
diffusion and advection. They constitute very effective
pathways, since the internal permeability and porosity
of fractures are greater than those of the surrounding
rock or soil.
The analysis of contaminant transport in fractured
porous media is environmentally important because it
can be used to predict the consequences of leakage of
radioactive or toxic waste from an underground repository. A typical scenario concerns a contaminant
source in the vicinity of a fracture network leading to
a freshwater aquifer. Since such networks may feasibly be of extremely intricate connected or disconnected form, attention here is restricted to mathematically tractable worst-case situations of uniform vertical
fractures of constant width in uniform saturated porous
media. In what we term a full two-dimensional treatment of transport in such systems, the groundwater
velocity is assumed to be vertical and constant in the
fracture and the porous rock matrix. Also it is assumed
that in both regions contaminant diffusion occurs in
both the vertical and horizontal directions. Further,
adsorption onto the matrix face and within the matrix
Address reprint requests to Dr. Landman at the Department of Mathematics, University of Melbourne, Parkville, Victoria 3052, Australia
Received December
160
Appt.
Math.
Modelling,
Contaminant
Transport equation
The differential equation governing the steady-state
transport of a decaying contaminant in a single medium
by two-dimensional diffusion and vertical groundwater
flow is
Kf
D:1
-
(1)
LVf
(24
(2b)
(3d
y = YIL
(3b)
L2A*Rm
L2A*Rf
~
D;
(5c)
22
a=:
(6)
(5b)
Dsn2
D:2
(54
LV
p = -
f = hf
D?l
DE
052
052
D,g+D**~-v~y-A*RC=O
POROUS MATRIX
Y
Fig.la
in the fracture
in the matrix
Fig.lb
(4a)
(4b)
Appl.
Math.
Modelling,
1989, Vol.
13, March
161
Contaminant
case considerations
y = 0,
0,
dC
-+O,
ax
c+O,
x LO
- 0,
ax-
x=0,
y>o
0b)
x-+m,
y>o
(7c)
Y+m,
xro
(7d)
a2X
Kf~
01x51
(8a)
x=0,
y>o
(8b)
(114
eUmyi2
x(x, y)
in the matrix
(1lb)
a*x
+
idx
in the fracture
aY
krn~+$-jmXEO
in the matrix
x=1,
y>o
c+o,
y-m,
OSXl
C(Sf,Y)
= c(s-,Y)
$(s+,Y)
= +-,Y)
(12b)
where
(134
Wb)
The boundary conditions on X(x, y) are precisely the
same as the boundary conditions on c(x, y), namely
equations (7) and (8) for the single fracture and periodic
array of fractures problems, respectively. The interface conditions (9) become
evmy/*x(S+,y)
XkY)
(144
= eufy*x(S-,y)
=
g&*~(S-,
y)
of x(x, y) is
(94
(9b)
u is defined as
(10)
X(x, w)sin oy do
(8d)
(124
162
in the fracture
y=o,
- 0,
ax-
(T=-
eOy*
x(x, y)
ac
of solutions
C(&Y) =
eumY/*$(S+,y)
ac
formulation
(7a)
(15)
of x in the y
(16)
We now take the Fourier sine transform of the transport equations in the fracture and matrix regions for
the two fracture-matrix geometries considered. For both
geometries the fracture region is 0 5 x < S, y > 0.
Taking the Fourier sine transform of equation (12a)
and using the boundary conditions, we find that the
Fourier sine transform X satisfies
$ - (bf(w))2X= ax
-2
ax
0,
x = 0,
gfqx)
w>o
(174
(17b)
Contaminant
where
with
f
by&))
l/2
(18)
q
(
g(x, w) =
>
20
mcb(w)sinh b(w)( 1 - 6)
A(w)cosh bf(o)x
X(x, w) =
+ g(x, w)
bf(o)sinh bf(w)S
(19)
Here gf(x, w) is a particular solution to the inhomogeneous equation (17a) and is found by using variation
of parameters:
(25)
2w
$-rKfbf(w)Sinhbf(w)6
x
g(x, w) =
cash bf(w)(6 - x)
S(x)cosh bf(w)x dx
cash bf(w)x
I
x
S(x)cosh b(w)(6 - x) dx
As stated above,
(20)
A(w) is an unknown function which will be determined
b(w)
w2
From equations (5b) we observe that /3 = 0 corresponds to the fracture and matrix regions having the
same Peclet number.
(26)
(21)
)
X(x,4 = -
p=p_$
112
,-*
bm(w)tx- 6) +
gm(X,
w)
(22)
Single fracture problem
p = 0. In terms of the transform
where
interface conditions
2w
g(x, @) =
x(s+,w)
?TKb(W)
.{
e-bmW(x-*i[
S(x)cosh b(w)(x
- 6) dx
function X the
(14) become
= X(6-,w)
(27a)
:(a+,w) = cs(S-,o)
(27b)
be deterto satisfy
dSldx +
problem
B(w)cosh b(w)( 1 - x)
+ g(x, w)
b(w)sinh b(w)( 1 - 6)
(24)
coth bf(w)S
B(w)
- + g(o) =
A(4
b(w)
bf(4
+ gf(4
B(w) = aA
(284
(28b)
we have defined
gf(w) = gf(S-, 0)
(29a)
g(w) = g(6+) w)
(29b)
(30)
163
Contaminant
6 > 0. Equations
epPYx(a+,
Y) = x(6-,
epPy$(iS+,
y) =
Y)
(314
u$(c-,y)
(3lb)
To proceed, we require
the left-hand sides of
Fourier sine transforms
It can be easily verified
transform of h(y), that
for B(o):
m
/
(35)
6 < 0. Equations
x(6+, Y) = @Yx(~-, Y)
$(S+,y)
(3W
= uepyg(S-,y)
H(w) = ,I
(32)
h(y)sin oy dy
0
h(y) (v > 0) is
+ gf
(w)
B(w) = (TF_~(A)(w)
wH(w) dw
ffi
= -7xfJ
/o [(w - u)2 + v2][(w + w)2 + $1
lr
sine transform
(33)
of equations
Wb)
F&.9(4 = aA
A(w) = ;B(o)
bf(4bm(4[
- g(w) +
(37b)
function X the
interface conditions are again given by equations (27).
Using equations (19) and (24), we obtain the equations
- coth b(o)(l - 6) B(o) + gm(w)
b(w)
= coth bf(w)S
b(w)
A(4
+ g(o)
B(o) = aA
g(41
coth bm(l - S)
p < 0. Proceeding
tions (36)
b
=
B + gm
coth bf(o)S
b(o)
Fp(B)W = PA
B(o) = vF_~(A)(~)
(w)
A(4
+ gf(4
(404
(40b)
(29).
(39)
= F_,
p > 0. Fourier-sine-transforming
(384
(38b)
164
coth bfS
-A
(374
FvW)(w)
4
coth bf8
7A
+ gf (0)
>
(41a)
(41b)
Contaminant
x(x, y) =
5 Res(X(x,
wj))e-ffKf~~22y
j=l
m
[ - X(.x, rein/z)
/
(AVZ
+ X(x, re -i3T2)] e -y dr
+;
a(l)
(T(Km)1/2
(42)
(43)
(A - Af) - KV
Km
+ arl
tan
>
(4)
Single fracture
problem
in the fracture
in the matrix
there is
e - [(A+
c(S-,y)
c(s+,Y)
(45)
25
j=l 1 +
(Kf/Km)(~j/bm(Oj))2
!&,*p-
uf/2ly
+ rSb(wj)
[(vj/bm(wj))2
1(+2l
oc
-y(rvufi2)
&.
(46)
where
problem
k
X(X,
Y)
7~
Res(X(x, wj))e-+~~**Y
( j=l
.cot ((yJS)
(47a)
+ 2m Res(X(x, wX))e~(A+K~n*2PZy
(48)
n=l
r2 - Af
Q=Kf
(47b)
(Am - Af) - dn I*
Km
)
tanh[((Am-~~~Kfn2)(l
-s)]
+(+qtan+=O
(49)
All the ~2 > qk, and they satisfy another transcendental equation ((A51) in Appendix 3). Again, the residues and,
hence, the coefficients of the exponential terms in the summation in equation (48) are independent of of and urn.
Consider the example where S(x) is specified as in equation (45). Then
165
Contaminant
transport
in fractured
porous
media:
et al.
A. Fogden
e - [(A+&/Z)* ~ uf/Z]y
2
Kf
77js
+
sin
qja
Cos
qja
1
a
jCot
(T77i
js
tan
7@
(1
6)
b"(wj)
> 0, j = 1, 2, . . . , k, while
< 0, n = 1, 2, . . . . .
c(x, y) - k(x)e - w
for all x
(51)
((F)*+Am)*
-f+
(53)
--+
+ Af +
Kf7):
(54)
(A - A) - Kfq2 * +
(+6772
Km
(A
Af)
Kf77*
(554
l*
(52a)
. tanh
+
.
tanh
respectively.
166
(52b)
c+* = 0
(55b)
(Note that this approximation has made the single fracture equation quadratic in q2, while the periodic array
of fractures equation remains transcendental.) We in-
Contaminant
+ vfol - Af)
(56)
(57)
for both geometries and at all points in the fracturematrix system (since the maximum value of the dimensionless function c(x, 0) = S(x) is assumed to be
unity). Further, this upper bound can be multiplied by
appropriate horizontal concentration profiles to yield
the separated approximations
01x<6
X>6
for the single fracture problem
e-"Y
C(X,Y)
(58a)
05x<s
6<x11
(58b)
where
A - (a2 + Pa!)
U=
(
Km
I*
(59)
A(Af
+ Kf$
&)
In this equation (Yrepresents the analytical approximation to the reciprocal decay length of the original
problem and is obtained from equations (52a) and (52b)
for the single fracture and periodic array of fractures
geometries respectively, as usual.
In summary, the solution of any general (p f 0)
problem is analytically approximated by the exact solution of the p = 0 matching problem in which the
value of uf = urn is given by equation (60) (while the
choice of boundary condition at y = 0 and the values
167
Contaminant
Of the dimensionless
are all unchanged).
parameterS
u,
6, hf, A",
Kf,
Km
Table l(a).
Common
Darameter values
Parameter
Dimension
Value
;:
Rm
7
P
ef
m
m
m.day-
m
-
L = (~~lh*Rm)2
a:
V
b
Table l(b).
1.780 x 10mg
1.6 x IO-$
1.0
1.0
1.0
1.0
0.5
0.5
0.0
5.0 x IO-5
1.0
0.3
Case
1
2
3
4
em
0.1
0.1
0.01
0.1
0.01
0.01
0.01
0.1
e-my
()
~ P
re
= 2 1 + (Kf//cm)(q/bm(o))2 + cTSbm(w)((~/bm(w))* + l/a2) + ;1(rcArnjli2P2
+ Q*
/
where
two solutions.
b(w) =
while in the second term P and Q are defined by equations (47) and
hf=hf+
Am=,,+
i
0
168
21
pr(r
- 012) &
Contaminant
Table 2.
Parameter
K
Km
6
A
ifrn
LIm
(T
a
I
Dimensionless
parameter
Case 1
1.0
1.0
1.67 x lO-4
2.76 x 1O-4
0.598
1.0
0.0
3.63 x IO5
0.2683 x IO-
0.1279
0.5931
Case 2
Case 3
Case 4
1.0
1.0
1.67 x IO-=
2.76 x 1O-4
0.598
1.0
1.0
1.0
1.67 x 10m4
2.69 x 10m3
0.584
1.0
1.0
1.0
1.67 x 10m4
2.69 x 10m3
0.584
1.0
0.0
3.63 x IO4
0.2022
0.3897
0.5502
0.0
3.72 x IO4
0.2035
0.3851
0.5384
0.0
3.72 x lo3
0.7668
0.8770
0.2396
mation
will, as y ---f ~0, approach the constant value
given by the corresponding percentage error in k(x).
In general, the analytically derived value of (Ydiffers
from the value obtained numerically (typically the magnitude of the discrepancy is less than O.l%), so at large
y the magnitude of the percentage error is an extremely
slowly growing exponential function. Thus, in general,
the percentage error in both approximations will become sizable at large y; however, this does not destroy
the validity of either approximation, since at such distances from the source the exact and approximate concentration levels are equally insignificant. In this paper
we cease plotting the percentage error at the distance
vertically where the concentration falls below 10e4.
In Fogden er al. lo it was found that the dimensionless group that most strongly influences the nature of
the steady-state concentration distribution is the product CTS,representing the effectiveness of the fracture,
relative to the porous matrix, as a pathway for contaminant transport. In particular, we showed there that
as (TS increases, the exponential analytical approximation given by equation (58) becomes increasingly
accurate. Also, in the analytical approximation given
by the exact solution of the p = 0 altered problem,
we clearly expect that, as US increases, the uniform
value u of the Peclet number obtained from equation
(60) approaches the value of the fracture Peclet number
uf (while moving away from that of the corresponding
matrix quantity urn) in the original problem. (This fact
is illustrated by the values of u given in Table 2 for the
four specific cases considered here.) Further, it is obvious from the construction of this second approximation detailed in the previous section that its accuracy will decrease as 1~~- urnI is increased (with all
other dimensionless parameters held constant).
For case 1 the percentage error in the vertical concentration profiles at x = S and x = i predicted by the
two analytical approximations is displayed in Figures
2(a) and 2(b), respectively. With the exception of small
y values in the latter of these two representative
profiles, the approximation given by equation (58a) is very
good in this case, while the agreement between the
numerical solution of the case 1 problem and the solution of its fi = 0 matching problem is excellent for
all y values considered. The corresponding plots for
cases 2 and 3, which are not shown here, retain the
-----------------------------
o-5
-----------
100
200
.----___
300
-0.5
AC%
1
I
I
I
L___
----___
---___
--------
0.5-
---__
--------
--
-__
Appl.
Math. Modelling,
169
Contaminant
AC%
loo-
__r--
__---
,xX
,
50-
//
,/
/
.:
(.
'Y
10
-15.L
-5.
-10.
-20
AL%
400.
Appendix 1: asymptotic
results as w +
I
I
The expression
300. :
F,(H)(w) - H(w)
zoo- \
\\
____________________-___I
10
'
(v>O)
(Al)
= ;I
-15
1
L(w, w)H(o) do
W)
170
aso+=a
Appl.
Math. Modelling,
4ww
[(o - w)Z + zq[(o + w)2 + 91
(A3)
ZJ) + 2(w4
+ VW
+ v4)*
(A9)
if gf or gm are not identically zero. Therefore in this
case gf and gm decrease slower than 6(wd3) as w + 03
unless S(S) = dS(S)ldx
= 0 or S(6+) = dS(S+)ldx
= 0, respectively. (This method assumes that all the
derivatives of S(X) are finite at the boundaries of the
fracture and matrix regions.) The asymptotic forms for
A and B as w -+ 00 are obtained from (A6) and (A7),
using (AS) and (A9) for the two geometries. In both
cases we find that
A(w) -;B(o)
2[S(6+) -
7T[(Kf)*
S(S-)I + o i
CJ.(Km)*]
o-+w
(AlO)
bB(o) -
WV
(-44)
v2, du
= H(w)
4~) -
as 0 + 00,
I*
geometries,
o)sin oy dw
(All)
of the interfacial
flux as
%(x,
y)
lim lim
yOxL+6f ax
(-46)
for the single fracture problem and
Single fracture
problem
ax
A(w) - ; B(o)
(x, w) = B(w)e-bm()(x-S) +
(x, w)
(A121
(Al3)
where
B
ws(6+)
- S(S-)I
co
?T[(Kf)*
C(Km)*]
(AI4)
171
Contaminant
e ~ bm(4(x_ 6)
(x, y) = cum*
as
sin
dw
Periodic
array of fractures
problem
0
m
+
I
(x,
w)sin wy do
(A15)
x) + agm
ax (x, 6.))
6)
(A221
my
my
dw
(B(w) - B,)e-b@)C-S)sinwy dw
+
0
m
_I5
(x, w)sin wy dw
sinh b(w)(l - x)
sinh b(o)(l - 6)
do
(A231
cc
=
I
e-~*+~*~sinoydw
(A16)
where
X-S
(A17)
ff=(Km)112
Using a generalized
tries, we get
0
=--=1 z
cuz*+ 1
Y
y* + (Y2
(A19)
Hence as x --, S+ ,
J0
bm(d(x ~ 6) sin
=--2 w
7~F(o)
!_ Sf(x, w)
uf
SYX, 0)
(A24)
wy
do
Y
y2 + (X - 6)2/Km
WO)
172
X(x, w)
I(a,y)-_Je-sinpzdp
m
I-
notation
Here Sf and S are functions in terms of integrals involving the source distribution S. Only their properties
will be required for this discussion; explicit expressions for them are in Appendix 4 for both geometries.
Contaminant
Single fracture
(A25)
(A26)
X(x, w)e+do
1, --m
(~27)
$(Kf(hyw))*
(h
Af))
(A28)
Let
V(w) = 5 + in
(A29a)
(b(o))2 = /.&I+ iy
(A29b)
(A29c)
Then
/-h =
$(Kf@
s*)
(A - A)),
u1
/_&I +
CL2 =
(/A: +
2
vy* l*
$26~
problem
=kIm
(A30a)
VI
v2 = 5-L
(A30b)
6431)
(A33)
(A34a)
(A34b)
Kf(cf2
T*))
iKf2577
(A35)
(A36)
From (A34), p2 = v2 = 0 is not a solution, so we may
consider & + 4 > 0. Then multiplying (A34a) by p2
and (A34b) by u2 and adding them gives
(& + &sinh 256 + a(5p2 + nv2)(cosh 256
- cos2$)
= 0
(A37)
173
Contaminant
urn- Af) -
KfV22 + +q
Km
tan
q6
(A39)
which is equation (44) in the main text. Also, if k is
the positive integer defined by the inequality
(A40a)
then (A39) has exactly k solutions,
. . . ) k, such that
say vj, j = 1, 2,
Fig. Ita
(A40b)
Furthermore,
j=
1,2 7 . . . 3k-l
(A40c)
X(x, w)eio dw
-R
R-rCYJ
>
Fig. l+b
Figure 4. Closed contour used in the Fourier inversion integration: (a) single fracture problem; (b) periodic array of fractures
problem
-m
bf(wj) =
iqj
and
+i
[-X(x,
_f
(A)*
The Wjdependence
wj
dF(tij)ldw
174
Appl.
(~42)
Periodic
array of fractures
problem
(A44)
S(X, Oj)
Modelling,
Proceeding
of the terms
and
Math.
wi
S(X, wj)
dF(mj)/dw
(A43)
Contaminant
to the equation
transport
in fractured
media:
+ (T((&
x
+ vvhinh2~2(1
- 6) + (71~2 - 4%)
cos
2776) = 0
(A47)
From their definitions it is easy to verify that qp2 5y 1) 0 for all 5, rt 2 0. Hence as sinh u 2 u and
(sin U( I u for all u 2 0,
(4%~+ qV2)sinh2p2(l - S)
+ (77~~- Sv2)sin 2v2(l - 6) 2 2(1 - S)&.L: + v$)
(A4g)
Since p2 = y = 0 is excluded by the second condition
above, each term on the left side of (A47) is strictly
positive for 5 > 0. Thus there are no solutions of the
system (A46) for 5 > 0, so only the case c = 0 need
be considered.
Set 5 = 0; then equations (A46) reduce to
v2 sin 2$(cosh
which is equation (49) in the main text. On the other hand, if n > ((A 0, and v2 = [Z_Q)~,
so (A49b) is satisfied and (A49a) becomes
_
(Am
Af)
tan
Km
>
WO)
+7
n?rl2
(-)l-6
Af)l~f)"2,
then v1 = 0, pi < 0, p2 =
[(
(A51)
= II2
(A49b)
+g77tan$=o
Kf$
(A49a)
(Am-~~-KfT2)12tanh[((hm-~~-Kf~2)2(~
__@I
et al.
A. Fogden
porous
nrrl2
and b, = 6
l/2
<k;
= 1,2,.
. . .
Appl.
Math.
Modelling,
1989,
175
Contaminant
Kf7#)*
(j
=
1, 2, . . . , k) which all lie on the Im(o)
axis between ki(hf)* and I*,
with jq+rI >
[ql, and an infinite number of simple poles at k w,* =
ki(Af + K~T$*)*,
n
=
1, 2, . . . , lying on the Im(w)
axis beyond + i(A
with lo,*+r[ > [o$ Hence, X(x, W)
is an analytic function in the complex w plane except
for these simple poles. Therefore we evaluate the integral in (A27) by closing the contour in the upper halfplane, as shown in Figure 4(b), since the integrand is
exponentially decaying there. Hence we obtain
m
X(x, w)eioY dw
(5
2mj
~ (*+
oj))e
~f~2)zY
j=l
(A53)
-m
=
Or
Sk w,)
dF(w,.)ldw
7TK
~,*))e~(~+~b~*2)2Y
Res(X(x,
(A521
n=l
Sf(x, w) = f
cash bf(w)x
I
x
sinh bf(w)(S - x)
S(x)cosh 64~)~ dx
6
-
fpyx,
w)
cash bf(w)6
I
x
S(x)sinh b(o)(x - x) dx
us
e-bm(-)(x-s)
(A54)
cc
+ u bf(W)~~~f(w)F
(sinh P(w)(x
- 6)
x
x
S(x)e-bm()(-S)dx -
I
s
I
6
S(x)sinh b(o)(x - x) dx
(A55)
>
For the periodic array of fractures problem with p = 0, Sf and S in (A24) are
1
Sf(x, w) = f
S(x)cosh bf(o)(S - x) dx
cash bf(w)(S -- x)
6
. (sinh
bf(w>(S
x)
(A56)
Contaminant
S(x, W) = (+5 cash b(w)( 1 - x) 1 S( x )cash bf(w)x dx + cash bf(w)6 cash b(w)(l - x)
(
- 1
1
+
(+ s
(A57 1
>
2
3
4
5
6
I
8
9
10
11
12
in Porous
Media.
Elsevier, New
York, 1972
Appt.
Math. Modelling,
177