Professional Documents
Culture Documents
Classification of The Group Invariant Solutions For Contaminant Transport in Saturated Soils Under Radial Uniform Water Ows
Classification of The Group Invariant Solutions For Contaminant Transport in Saturated Soils Under Radial Uniform Water Ows
Abstract
1 Introduction
Prediction of water and contaminant movements in soils is important to the
theory of soil salinity and underground water pollution. Among others, the
sources of contamination are movement of agricultural and industrial con-
taminants. Some exact solutions are constructed for constraint assumptions,
as such, investigation and study of these problems is extremely dicult and
challenging [1], even when the problem is given in terms of linear partial
dierential equations [2].
Broadbridge et al. [2, 3, 4] applied the classical Lie point symmetries
to analyze the convection dispersion equations arising in solute transport
theory. A compendium of exact (group-invariant) solutions is given in [5].
Other researchers used dierent techniques to obtain exact solutions (see e.g.
[6, 7, 8]). Chen et al. [9] constructed analytical solutions for two-dimensional
advection-dispersion equation with transverse dispersivities depending lin-
early on the spatial variable. Yadav et al. [10] constructed analytical solu-
tions for solute transport in semi-innite porous domain. Symmetry reduc-
tions and group invariant solutions were constructed for models describing
motion of a polytropic gas [11].
1
Exact analytical solutions for contaminant transport in porous media
have been constructed for example in [12, 13, 14, 15]. Chrysikopoulos and
Sim [12] developed the one-dimensional stochastic model describing virus
transport in homogeneous, saturated semi-nite porous media. The ideas in
[12] are extended to three dimensional problems in [13, 14]. Experimental
investigations of acoustically enhanced colloid transport in water-saturated
packed columns [15]. In this paper we focus on theoretical macroscopic
deterministic models given in terms of partial dierential equations.
Numerical models are able to simulate complex reactive transport phe-
nomena, but can be time consuming to construct and subject to numerical
discretization errors [7]. Also, available packages have signicant disagree-
ment in their prediction of solute transport [16]. Therefore, exact solutions
are very important because they are needed both as validation tests for nu-
merical schemes and also provide insight into the water and solute transport
processes.
In this paper, we construct group-invariant solutions and classify them
according to the one-dimensional optimal systems. This is a signicant ad-
vancement of the work by [4] and [6]. We consider contaminant transport
under radial uniform water ows. The resulting convection-dispersion equa-
tion given in terms of stream function is analyzed when dispersion coecient
is proportional to the water pore velocity. It turns out the Lie point symme-
tries are admitted when the dispersion coecient is a constant in one case
and depending on water pore velocity in the other. In Section 2, we pro-
vide the mathematical models. We briey discuss the symmetry methods
in Section 3. The problem at hand is analyzed in Section 4 and the group
invariant solutions are classied according to the one-dimensional optimal
systems in Section 5 and some physically realistic solutions are discussed.
lastly, some concluding remarks are given in Section 6.
2 Mathematical model
The theoretical background in this section is obtained from Hillel [17]. The
macroscopic deterministic model, which is based on local conservation laws
is given by
(c)
= (D(v)c) (cV), (2.1)
t
where D(v) denotes the coecient of hydrodynamic dispersion and is
given in cartesian coordinates. Experimental and theoretical observations
show that the dispersion coecient can take the power law form D(v) = v p
with being a proportionality constant and 1 p 2 (see e.g. [18]). Here,
v is the pore water velocity. Given uniform water ow in saturated soils,
then the continuity equation is given by V=0. Uniform saturated water
ow implies that the soil water content = s , where s is the water content
2
at saturation. By Darcys law, which states that water ux is proportional
to the gradient of the hydraulic pressure head, that is V = ks , where
is the total hydraulic pressure head and ks is the hydraulic conductivity at
saturation (see e.g. [17]), as such we obtain the Laplaces equation 2 = 0.
Under these assumptions, Equation (2.1) then becomes
c
= (D(v)c) + k c, (2.2)
t
where k = ks /s and v = |k|. Note that this problem becomes extremely
dicult to solve exactly when v must be the modulus of the potential ow
velocity eld for an incompressible uid (see e.g. [2]). However, one may
transform Equation (2.1) from cartesian coordinates (x, y) to the streamline
coordinates (, ) using Laplace preserving or conformal transformations
[19], see also [2]. The resulting solute transport Equation is given by
c c
= v 2 [D(v)c] + v 2 , (2.3)
t
( )
where =
, . The velocity potential is and is a conjugate
harmonic stream function such that x = y and y = x . Also, the
functions and satisfy the Laplace equation. In radial water ows, the
velocity potential is given by = (Q/s ) logr and the stream function
= (Q/s ) arctan(y/x), where Q is the source strength (pumping rate).
Introducing the normalized concentration, velocity potential and time given
by C = c/cs , = logR and = t/ts respectively with cs , ts and R being
the concentration, time at soil saturation and the distance or radius from
the point source, we may write Equation (2.3) as
C C
= v 2 [D(v)C] + v 2 . (2.4)
3
such as Equation (2.4) describing contaminant transport under radial water
ows, we seek transformations of the form
= + 1 (, , C) + O(2 )
= + 2 (, , C) + O(2 ) , (3.1)
C = C + (, , C) + O(2 )
X [2] (Equation(2.4))|Equation(2.4) = 0,
4
4.1 Constant dispersion coecient
Given p = 0, then the dispersion coecient becomes a proportionality con-
stant, . Given dispersion coecient as an arbitrary constant = 0, Equa-
tion (2.4) admits nite two extra symmetries
X3 = 2 ,
( ) (4.1)
2
1 e
X4 = 2 + C .
2 4 C
Further, symmetries may be admitted when = 1 and = 1. Note that
= 1 implies negative dispersion coecient which is not physically real-
istic, and as such we focus on the case = 1. Given p = 0 and = 1, then
Equation (2.4) admits nite four extra symmetries given by
X3 = e ,
X4 = 2 ,
( 2 ) (4.2)
e
X5 = + C ,
2 C
( )
2 e 2
X6 = + C .
2 4 C
5
Note that the admitted symmetry structure and number is not aected by
the constant , that is we obtain the same symmetries up to the specied
value.
6
admitted by a k th-order partial dierential equation
( )
u(k) = F x, u, u(1) , ..., u(k1) ,
[Xi , Xj ] X1 X2 X3 X4 X5 X6
X1 0 0 0 2X1 2X3 4X4 2X2
X2 0 0 0 0 0 0
X3 0 0 0 X3 X2 2X5
X4 2X1 0 X3 0 X5 2X6
X5 2X3 0 X2 X5 0 0
X6 2X2 4X4 0 2X5 2X6 0 0
7
Three cases arise.
Case 1. If (X) > 0, then we choose to be a real root of a1 2a 2
4 +4 a6 =
0 and = a6 /(8a6 2a4 ). This implies ae1 = ae6 = 0 and ae4 = (X) = 0.
Thus, X is equivalent to the vector X e = X4 + ae2 X2 + ae3 X3 + ae5 X5 . Acting on
e
X by adjoint maps generated by X5 and X3 , namely Ad (exp (ae5 X5 )) and
Ad (exp (ae3 X3 )) , as such X5 and X3 in X e vanish. No further simplications
are possible, therefore X is equivalent to a multiple of X4 + aX2 for some
a R provided (X) > 0.
Case 2. If (X) < 0, we set = 0 and = a4 /4a1 so that ae4 = 0. One
may then assume both the coecients of X1 and X6 to be unity. Thus, X
is equivalent to the vector X e = X1 + X6 + ae2 X2 + ae3 X3 + ae5 X5 . Acting on
e
X by adjoint maps generated by X5 and X3 , namely Ad (exp ((ae3 /2) X5 ))
and Ad (exp ((ae5 /2) X3 )) , as such X5 and X3 in X e vanish. No further
simplications are possible, as such X is equivalent to X1 + X6 + aX2 , a R
given (X) < 0.
Case 3. (X) = 0. Two subcases arise. If not all the coecients a1 , a4 , a6
are zero, then we are free to choose and such that ae1 = 0 and ae4 =
ae6 = 0. In this case X is equivalent to a multiple of X e = X1 + ae2 X2 +
ae3 X3 + ae5 X5 . Acting on Xe by adjoint maps generated by X1 and X3 , namely
Ad (exp ((ae3 /2ae5 ) X5 )) and Ad (exp ((ae2 /ae5 ) X3 )) , ae5 = 0, as such X2
and X3 in X e vanish. Therefore X is equivalent to a multiple of X1 X5 . If
ae5 = 0, then acting on X e by Ad (exp ( (ae3 /2) X5 )) and so X is equivalent
to a multiple of X1 + aX2 , a R. If all the coecients a1 , a4 , a6 are zero,
and assuming a3 = 0 (say a3 = 1), then acting on X by Ad (exp (ae2 X5 ))
and Ad (exp ((ae5 /2) X5 )) yield X3 , a multiple of X. If a3 = 0 but a5 = 0,
acting by any group generated by X1 , namely, Ad (exp (X1 )) gives a nonzero
coecient in front X3 implying that this case is similar to the case when
a3 = 0. Thus, the only remaining vectors are multiples of X2 . The set of
one-dimensional optimal system is given by
{ }
X4 + aX2 , X1 + X6 + aX2 , X1 X5 , X1 + aX2 , X2 , X3 , a R. (5.5)
8
corresponding to the appropriate admitted symmetry generator if and only
if u = H(x) satises
d 1 d 2 d
1
= 2
= ... = .
dx dx u
Classication of the group-invariant by the elements of the optimal system
in (5.5) and (5.6) are listed in Tables 3 and 4, respectively. Wherever they
appear k1 and k2 are arbitrary constants, 1 F1 (b, c; z) and U (b, c; z) are the
conuent hypergeometric functions, Ai(z) and Bi(z) are the Airy functions,
while Jn (z) and (z) represent the Bessel function of rst kind and Euler
gamma function respectively and erf(z) represents an error function. A
well documented review of such functions is presented by Abramowitz and
Stegun [27].
C(, ) 0, as . (5.8)
2 3 G + (4 2 1)G = 0,
and hence ( )
1
G = c1 + c2 erf .
2
9
5
=1
4.5 =5
=10
4 =20
3.5
C
2.5
1.5
0.5
0
0 5 10 15
R
z
2
e d is the error function [27]. In terms of the original
2
erf(z) =
0
variable and subject to the boundary conditions, we obtain
( )
e
C = C0 erfc . (5.10)
2
Example (b)
The X6 -invariant solution is given in functional form as
( 2 )
1 e
C = exp G(), (5.12)
4
where
C 0, and C = ( ), .
10
8
J
3
0
10 20 30 40 50 60 70 80 90 100
C = C0 , R=0 (5.14)
and
1 dC
C+ = 0, R = Ra . (5.15)
R dR
11
0.4
0.35
0.3
0.25
C
0.2
0.15
0.1
0.05
0
0 5 10 15 20 25 30 35 40
1
=1
0.9 =2
=5
0.8
0.7
0.6
C
0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4
R
12
1
0.9
0.8
0.7
0.6
C
0.5
0.4
0.3
0.2 p=1
p=1.5
0.1 p=2
0
0 1 2 3 4 5
R
The boundary condition (5.14) implies that pollutants are supplied at the
origin, and boundary condition (5.15) correspond to the assumption that
pollutants are not carried though at some distance Ra , rather it accumulates
here. We obtain the exact solution
{( ) ( p )}
1 1 R
C = C0 1 + exp , (5.16)
p
where is given by
( ) ( )
Rap2 Rap
=1 1+ exp .
p
13
1
0.9
0.8
0.7
0.6
C 0.5
0.4
0.3
0.2 =1
=2
0.1
=100
0
0 1 2 3 4 5
R
0.8
0.6
C
0.4
0.2
0
6
5
4 4
3
2 2
1
0 0
p
14
1
=0.1
0.9 =0.5
=1
0.8
0.7
0.6
C 0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4
R
0.14
=1
=1.5
0.12 =2
0.1
0.08
C
0.06
0.04
0.02
0
0 1 2 3 4 5
R
15
Solution (5.17) is depicted in Figures 8 and 9.
In his work, Philip (1994) considered the instantaneous point source for
contaminant dispersion during radial water ow in porous media. Exact
solutions were constructed for the two- and three- dimensional models with
dispersion coecient depending on Peclet number. Here, we consider mod-
els in stream functions coordinates. Exact close-form (similarity) solutions
are constructed using the elements of the one dimensional optimal systems.
These new solutions may be viewed as representing the continual supply of
contaminant at a point (source) which are dispersed radially.
Acknowledgements
RJM wishes to thank the National Research Foundation of South Africa
under REDIBA program, for the continued generous nancial support.
Conict of interests
The authors declare that there is no conict of interest regarding the pub-
lication of this article.
References
[1] F. Catania, M. Massab`o and O. Paladino, Estimation of transport
and kinetic parameters using analytical solutions of the 2D advection-
dispersion-reaction model, Environmetrics, 17: 199-216, 2006.
16
[3] P. Broadbridge, R.J. Moitsheki and M.P. Edwards, Analytical solutions
for two dimensional solute transport with velocity-dependent disper-
sion, Environmental Mechanics: Water, Mass and Energy transfer in
the biosphere, 129: 145-153, 2002.
[4] R.J. Moitsheki, P. Broadbridge and M.P. Edwards, Group invariant so-
lutions for two dimensional solute transport under realistic water ows,
Quaestiones Mathematicae, 29(1), 73-83, 2006.
[5] R.J. Moitsheki, Invariant solutions for transient solute transport in sat-
urated and unsaturated soil, PhD thesis, University of Wollongong,
2004.
[8] J.E. Houseworth and J. Leem, A quasilinear model for solute transport
under unsaturated ow, Vadose Zone Journal, 8(4): 1031-1037, 2009.
[9] J-S Chen, C-F Ni and C-P Liang, Analytical power series solution-
s to the two-dimensional advection-dispersion equation with distance-
dependent dispersivities, Hydrological Processes, 22: 4670-4678, 2008.
[10] R.R. Yadav, D.K. jaiswal, H.K. Yadav and Gulrana, Analytical solu-
tions for temporally dependent dispersion through homogeneous porous
media, International Journal of Hydrology Science and Technology,
2(1): 101-115, 2012.
[11] E.M.E Zayed and H.A. Zedan, Autonomous forms and exact solutions
of equations of motion of polytropic gas, International Journal of The-
oratical Physics, 40(6): 1183-1196, 2001
[14] Y. Sim and C.V Chrysikopoulos, Analytical solutions for solute trans-
port in saturated porous media media with semi-innite or nite thick-
ness, Advances in Water Resources, 22(5): 507-519, 1999.
17
[15] J.M. Thomas and C.V Chrysikopoulos, Experimental investigation
of acoustically enhanced colloid transport in water-saturated packed
columns, Journal of Colloid and Interface Science, 308: 200-207, 2007.
[16] J. Woods, C.T. Simmons and K.A. Narayan, Verication of black box
groundwater models, in EMAC98: Proceedings of the Third Biennial
Engineering Mathematics and Applications Conference, Institution of
Engineers Australia, Adelaide, 523-526, 1985.
[17] D. Hillel, Environmental soil physics, Academic Press, New York, 1998.
[21] A.C. Hearn, Reduce users manual version 3.4, Santa Monica, Califor-
nia: Rand Publication CP78, The Rand Corporation, 1985
[24] N.H. Ibragimov, Elementary Lie group analysis and ordinary dieren-
tial equations, Wiley, New York, 1999.
[25] G.W. Bluman and S.C. Anco, Symmetry and integration methods for
dierential equations, New York: Springer-verlag, (2002).
[26] G.W. Bluman, A.F. Cheviakov and S.C. Anco, Applications of sym-
metry methods to partial dierential equations, New York: Springer-
verlag, (2010)
18
Table 2: Adjoint representation for the base vectors.
Ad X1 X2 X3 X4 X5 X6
X1 X1 X2 X3 2X1 + X4 2X3 + X5 42 X1 + 2X2 4X4 + X6
19
X2 X1 X2 X3 X4 X5 X6
X3 X1 X2 X3 X3 + X4 X2 + X5 2 X2 + X6 2X5
X4 2
e X1 X2 e X 3 X4 e X5 e2 X6
X5 X1 + 2X3 2 X2 X2 X2 + X3 X4 + X5 X5 X6
X6 X1 2X2 + 4X4 + 42 X6 X2 X3 + 2X5 X4 + 2X6 X5 X6
Table 3: Group-invariant solutions for Equation (2.4) with constant D.
a
[ ( )
3 1
( a 1 1 )]
X4 + aX2 C = 2 G(), where = e2 a<0 C = a/2 21 1e k21 F1 a+1 , ;
2 4 e 2 + k 1 1 F 1 2 , 2 ; 4 e2 .
[ ( 2 )]
and G satises a=0 C = k2 2 k1 erf 1 2
[ 2 e
a/2 1 1
( 1a 3 1 ) ( a 1 1 )]
3 G + 14 (6 1)G 81 aG = 0. a>0 C= 2 e k 2 1 F1 2 , ;
2 4 e 2 + k 1 1 F 1 2 , 2 ; 4 e2 .
[ ]
X1 + aX2 C = ea G(), where = a<0 C = ea k1 cos ae k2 sin ae
and G satises a=0 C = e k1 +[ k2 . ]
G + G ae2 G = 0 a>0 C(, ) = ea k1 cosh ( ae ) ik2 sinh ( ae ) .
2 3
X1 + X5 C = e( e + 3 ) G(), where ( )( )
= 2 + e and G satises C = exp e + 32 3 Ai( 2 + e )k1 + Bi( 2 + e )k2
G G = 0.
20
[ [ ]
1 2 a
X1 + X6 C = exp 2(1+4 2 )
a<0 C = exp e2 (1+4 2 ) 4 ln (1 + 4 ) + 2 tan
1 (2 )
e [ ( )
] 2 2
a 1 3+ia 3 i
+aX2 14 ln (1 + 4 2 ) + 2 tan1 (2 ) G(), ei/2e (1+4 ) 2e2 (1+4 2) k F
2 1 1 4 , ;
2 e2 (1+4 2 )
( )]
i/2 e 2 (1+4 2 ) 1 3+ia 3 i
where = e 1 + 4 2 +e 2e2 (1+4 2 ) 1
k U 4 , 2 ; e2 (1+4 2 ) .
[ ]
1 2
and G satises a=0 C = exp e2 (1+4 2 ) 4 ln (1 + 4 )
[ ( ) ( )
1
6 G + 2 5 G + (1 a 2 )G = 0, 1 J 1 2 2 k1 34
2e 1+4 2 4 2e (1+4 )
( ) ]
1
(5)
+ 2e 11+ 2 J 1 2e2 (1+4 2) k 2 4 .
4
[ ]
1
a>0 C = exp e2 (1+4 2) 4 ln (1 + 4 2 ) + a arctan (2 )
2
[
2 2 1
( 3ia 3 i
)
ei/2e (1+4 ) 2e2 (1+4 2 ) k2 1 F1 4 , 2 ; e2 (1+4 2 )
]
2 2 1
( 3ia 3 i
)
+ei/2e (1+4 ) 2e2 (1+4 2) k 1 U 4 , ;
2 e2 (1+4 2 ) .
Table 4: Group-invariant solutions for Equation (2.4) given D(v) = e2 .
[ ] ( )
8 3 3
X1 + X4 C = exp 3 + e2 2 G(), C = exp 8
3 + e 2 2
{ 2 2 [ ( )
4 +e 1+4(4 2 +e2 )
21
where = 4 2 + e2 e 4 k1 Ai (4)4/3
( )]}
1+4(4 2 +e2 )
and G satises +k2 Bi (4)4/3
.
1
G 2 G 4 G = 0.
[ ( )
1
2
X1 + aX2 C= ea G(), a<0 C(, ) = ea
exp 1 4a)e
4 (1 k1
( ) ]
1
where = + exp 4 (1 + 1 4a)e2 k2
[ ( ) ]
e 2
and G satises a=0 C = k2 + exp 2 k1 .
( ) [( )
2 1
e
G + 2+ G + a e4 G = 0, a>0 a
C = e exp 2 k1
4 (1 1 + 4a)e
( ) ]
1
2
+ exp 4 (1 + 1 + 4a)e k2 .