Environmental Technology

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/tent20

Contaminant dispersion with axial input sources in soil

media under non-linear sorption

Rakesh Kumar Singh, Tapan Paul, Nav Kumar Mahato & Mritunjay Kumar
Singh

To cite this article: Rakesh Kumar Singh, Tapan Paul, Nav Kumar Mahato & Mritunjay Kumar
Singh (2023) Contaminant dispersion with axial input sources in soil media under non-linear
sorption, Environmental Technology, 44:13, 1903-1915, DOI: 10.1080/09593330.2021.2016992

To link to this article: https://doi.org/10.1080/09593330.2021.2016992

Published online: 28 Dec 2021.

Submit your article to this journal

Article views: 191

View related articles

View Crossmark data

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tent20
ENVIRONMENTAL TECHNOLOGY
2023, VOL. 44, NO. 13, 1903–1915
https://doi.org/10.1080/09593330.2021.2016992

Contaminant dispersion with axial input sources in soil media under non-linear
sorption
Rakesh Kumar Singha, Tapan Paulb, Nav Kumar Mahatob and Mritunjay Kumar Singha
a
Department of Mathematics and Computing, Indian Institute of Technology (Indian School of Mines), Dhanbad, India; bDepartment of
Mathematics, School of Basic and Applied Sciences, Adamas University, Kolkata, India

ABSTRACT ARTICLE HISTORY

Contaminant transport in a soil formation is described by advection dispersion equation. In this Received 23 August 2021
study, a horizontal and transversal contaminant transport along transient groundwater flow Accepted 1 December 2021
under non-linear sorption is solved numerically to examine the contaminant distribution profiles
KEYWORDS
in finite soil media. The horizontal and transversal pore-water seepage velocities are defined as Advection; dispersion;
varying exponentially with time. Dispersion is considered directly proportional to the first power numerical solution; porous
of the pore-water velocity. Axial input sources varying exponentially with time are assigned medium; sorption isotherm
along the two-dimensional coordinate axes. For approximating the governing solute transport
equation into algebraic equations, Crank–Nicolson (CN) and alternating direction implicit (ADI)
methods are used. Both the numerical solutions are illustrated graphically with the help of
computer software for various hydrological input data. In a special case, the numerical solutions
are also compared with an approximate solution obtained by PDEtool. The comparison is
performed with the help of contour plots. The CN method gives more accurate result than ADI
method for the present model problem.

Introduction
industrial, or irrigational utilities. Such modelling of the
For the study of solute or contaminant transport in a solute transport in an aquifer, to analyse its fate and
geological formation, it is required to understand the transport, may help to palliate the aquifer contamination
basic ideas and theory behind the movement of con- problem. The solute transport along groundwater flow
taminant through porous media. There are some pro- has been interpreted under various geometric and
cesses operating contaminant transport in the hydrologic conditions in the past. To construct a
groundwater system. The mathematical model of the model, there must be simple assumptions, as the situ-
contaminant transport in porous formation is known ation of transport medium is generally complicated to
as advection dispersion equation (ADE) derived on the be simulated exactly. A quantitative understanding of
law of conservation of mass and Fick’s law [1]. Through the solute transport in an aquifer is of great importance
porous media, groundwater may be contaminated due from the environmental perspective. The solute trans-
to various sources, such as leakage of septic tanks, port problems in a groundwater system are mostly
dumping of medical or industrial waste, pesticides and solved either analytically or numerically. Several analyti-
fertilizers used in agriculture, and so on and therefore, cal and numerical techniques have been proposed in the
the groundwater may become unusable for domestic, past to solve the ADE subject to various geometric

CONTACT Rakesh Kumar Singh [email protected] Mathematics and Computing, Indian Institute of Technology Indian School of Mines
Dhanbad, Dhanbad 826004, India
© 2021 Informa UK Limited, trading as Taylor & Francis Group
1904 R. K. SINGH ET AL.

conditions. Generally, it is difficult to solve the ADE by in cylindrical coordinates with various types of inlet
analytical techniques and even in some complex cases boundary conditions by using Hankel transform and
their solutions are not possible to determine. In such generalized integral transform techniques. Ma et al.
cases, numerical techniques may be compatible. The [11] discussed the alternating direction implicit (ADI)
numerical solution of the ADE has been an interesting method of high-order to solve an unsteady convec-
field of research for many years. Among the various tion-dominated problem and computed numerical
numerical approaches, finite-element method (FEM) errors and CPU run times for different ADI schemes.
and finite-difference method (FDM) are found most Savovic and Djordjevic [12] solved a 1D ADE with vari-
popular as these methods transform the differential able coefficients using explicit FDM in semi-infinite
equations that govern flow (or transport) into a system media. The numerical solutions are obtained under
of algebraic equations. For the multi-dimensional non- different flow conditions in uniform and non-uniform
linear solute transport equation subject to complex con- transport media. Again, Savovic and Djordjevic [13] pre-
ditions, the applicability of analytical approach is often sented numerical solution of a 1D ADE using explicit
found unsuitable and hence numerical method may be FDM in semi-infinite porous medium under the con-
applied as an alternative option [2]. When computers siderations of temporally dependent dispersion along
were not available widely in the past, only manual calcu- a uniform flow and also spatially dependent dispersion
lations (for analytical approach of simplified modelled along a non-uniform flow subject to pulse-type constant
problems) or analogue studies were possible, and evi- input source condition. Moreover, Djordjevic and
dently the numerical techniques having limited value Savovic [14] obtained numerical solution of a 2D linear
were not much popular. Ever since the 1960s, numerical ADE by explicit FDM in non-uniform semi-infinite
modelling has been an active area of research for porous formation. A pulse-type constant input source
dealing with groundwater movement and solute trans- was considered to examine its transport profile. van Gen-
port in porous formations [3]. uchten et al. [15] solved a 1D ADE analytically by consid-
There are various research papers presented in the ering the mass exchange of first-order between the river
past to find the analytical and/or numerical solutions and zones of dead water simultaneously. Wadi et al. [16]
of one-, two-, or three-dimensional ADE subject to suit- applied Laplace transform method to solve a 1D ADE
able geometric conditions. For a higher-order solute subject to transient flow in a river. Singh and Das [17]
transport governing equation with complex type solved the 1D scale-dependent solute dispersion pro-
boundary conditions, analytical solution is generally blems in a non-homogeneous medium by LTT and com-
not suitable to deal with. Wang et al. [4] applied a pared the result with approximate solution derived by
semi-analytical method to solve a two-dimensional explicit FDM. Djordjevic et al. [18] employed explicit
(2D) contaminant transport equation subject to FDM to solve a 2D ADE with variable coefficients
complex geometric conditions in uniform soil for- under the consideration of first-order decay and zero-
mations. Dehghan [5] elaborated some numerical order production in finite porous media subject to peri-
approximations of finite difference to obtain the odic seepage flow. Moreover, the FEM is also widely
approximate solutions of an ADE in three-dimension used for solving the ADE as this method is capable of
subject to the Dirichlet boundary condition. In addition, handling complex geometrical solute transport pro-
the stability conditions of all the approximated linear blems [19–22]. In addition, meshfree method is becom-
equations were discussed by applying the Von ing popular in the field of computational methods
Neumann stability method. Moreover, Ataie-Ashtiani after two decades since when it was initiated. This
and Hosseini [6] derived the numerical errors for the method may be used to solve a partial differential
approximate solution of a 2D advection–dispersion reac- equation that does not require mesh creation in the
tion equation and found that Crank–Nicolson (CN) domain [23,24]. Moranda et al. [25] studied the 1D ADE
method is the most accurate method on the basis of under linear equilibrium sorption and decay subject to
truncation error analysis. Savovic et al. [7] presented temporally dependent mixed boundary condition. To
numerical solution of a diffusion equation using FDM solve the proposed model problem, LTT was used.
for radon transport through a finite soil formation. Guer- Solute transport in a groundwater system is also investi-
rero and Skaggs [8] obtained the exact solution of a one- gated along and against the pore-water flow for input
dimensional (1D) ADE by applying generalized integral source imposed around at the middle location of the
transform method. Kumar et al. [9] also applied the transport medium [26]. Whereas, Banaei et al. [27] inves-
Laplace transform technique (LTT) to solve a 1D ADE tigated numerically and experimentally the behaviour of
subject to the variable coefficients. Chen et al. [10] groundwater flow and contaminant transport in soil for-
derived the exact solutions of a 2D transport equation mations under the consideration of sorption.
 ENVIRONMENTAL TECHNOLOGY 1905

The governing equations were solved separately by At the extreme boundaries, solute fluxes are specified
using FDM. zero as follows:
The main aim of the present work is to solve a 2D con-
∂c
taminant transport equation under non-linear sorption = 0; x = a, y . 0, t . 0 (5)
∂x
by CN and ADI methods. The fate and transport of the
solute concentration is examined for transient ground- ∂c
water flow subject to axial input sources in finite = 0; x . 0, y = b, t . 0 (6)
∂y
uniform soil media. The axial input sources are con-
sidered temporally dependent. According to the In this study, non-linear sorption of solute is considered
authors’ best knowledge, there has not found any such in equilibrium state. The relationship between the liquid
work in the past. and solid phase concentrations is defined as F = kd cq
abc
called Freundlich sorption isotherm and F = 1+ ac called

Mathematical model of the 2D contaminant Langmuir sorption isotherm, where kd [M−q L3q ] is sorp-
transport governing equation tion coefficient, q is non-linearity parameter determined
empirically that ranges from 0 to 1, a [M−1 L3 ] is Lang-
Contaminant transport in a porous medium is math- muir constant, and b is the highest sorption capacity.
ematically described by ADE associated with the liquid The above relationships between F and c are known as
and solid phase concentrations (i.e. c [ML−3 ] and non-linear sorption isotherms [29,30].
F [MM−1 ], respectively) under the consideration of sorp- On using the above non-linear sorption isotherms
tion [28]. The 2D solute transport under sorption in a separately in Equation (1) yields
uniform porous formation, having porosity f and
density r [ML−3 ], is expressed as follows: ∂c ∂2 c ∂2 c ∂c ∂c
R = Dx 2 + Dy 2 − u − v (7)
∂t ∂x ∂y ∂x ∂y
∂c r ∂F ∂ c 2
∂ c ∂c 2
∂c
+ = Dx 2 + Dy 2 − u − v (1) ⎧
∂t f ∂t ∂x ∂y ∂x ∂y ⎪ rqkd q−1
⎪
⎨ 1+ c for Freundlich sorption
f
where, u [LT−1 ] and v [LT−1 ] are the average linear vel- where, R =
⎪
⎪ rab
ocities along horizontal and transversal directions ⎩ 1+ for Langmuir sorption
f(1 + ac)2
respectively, Dx [L2 T−1 ] and Dy [L2 T−1 ] are the horizontal is the retardation factor that describes the retardation
and transversal dispersion coefficients, respectively, x [L] effects caused by non-linear sorption isotherms.
and y [L] are space variables, and t [T] is time variable. For unsteady groundwater flow, the horizontal and
Initially, a constant background solute concentration transversal pore-water seepage velocities are defined,
c0 [ML−3 ] is assumed in the uniform soil medium as respectively, as follows [31]:
follows:
u = u0 f (mt)and v = v0 f (mt) (8)
c = c0 ; 0 ≤ x ≤ a, 0 ≤ y ≤ b, t = 0 (2)
where, a [L] and b [L] are horizontal and transversal finite where, u0 [LT−1 ] and v0 [LT−1 ] are constant initial
spatial lengths, respectively. seepage velocities, and m [T−1 ] is the flow resistance
For assigning the axial input sources, Dirichlet coefficient.
boundary conditions are specified along the x- and Using the dispersion theory i.e. dispersion is straight
y-axes as shown in Figure 1. Such types of sources proportional to t th power of the pore-water velocity,
may be seen in the angular crop field area where fer- where t ranges from 1 to 2 [32], the longitudinal and
tilizers or pesticides are used extensively. The input transversal dispersion coefficients are expressed,
sources are varying exponentially with time. Math- respectively, as follows:
ematically, the boundary conditions are expressed as Dx = Dx0 f t (mt)and Dy = Dy0 f t (mt) (9)
follows:
where Dx0 [L2 T−1 ] and Dy0 [L2 T−1 ] are constant initial
c = c1 exp (−g1 t); x ≥ 0, y = 0, t . 0 (3)
dispersions.
c = c2 exp (−g2 t); x = 0, y . 0, t . 0 (4) Using Equations (8) and (9) for t = 1, Equation (7)
reduces to the following form:
where, c1 [ML−3 ] and c2 [ML−3 ] are constant input con-
centrations along x- and y-axes, respectively, R ∂c ∂2 c ∂2 c ∂c ∂c
g1 [T−1 ] and g2 [T−1 ] are decay parameters along x- = Dx0 2 + Dy0 2 − u0 − v0 (10)
f (mt) ∂t ∂x ∂y ∂x ∂y
and y-axes, respectively.
1906 R. K. SINGH ET AL.

Figure 1. Geometry of the 2D solute transport model problem.

To simplify Equation (10), introducing a new time into M and N number of equal sub-intervals with
variable as follows [33]: the sub-interval lengths DX and DY, respectively.
The governing solute transport Equation (12) together
t
with the conditions (13)–(17) are first approximated
T= f (mt)dt (11)
by CN and ADI methods. On approximations of the
0
governing equation by the CN and ADI methods, a
Now using Equation (11), Equation (10) is transformed pentadiagonal system of linear algebraic equations
into the new time domain as follows: and two set of tridiagonal system of linear algebraic
equations are obtained, respectively, at each time
∂c ∂2 c ∂2 c ∂c ∂c
R = Dx0 2 + Dy0 2 − u0 − v0 (12) level. Furthermore, the linear systems of the algebraic
∂T ∂x ∂y ∂x ∂y equations of each approximating method are solved
The geometric conditions are also transformed into graphically at fixed time period with the help of
the new time domain as follows: Matlab coding. The following two sub-sections
discuss the approximations of the governing transport
c = c0 ; 0 ≤ x ≤ a, 0 ≤ y ≤ b, T = 0 (13) equation for its conversion into the system of alge-
c = c1 exp (−g1 T); x ≥ 0, y = 0, T . 0 (14) braic equations.

c = c2 exp (−g2 T); x = 0, y . 0, T . 0 (15)

∂c Approximation by CN method
= 0; x = a, y . 0, T . 0 (16)
∂x
Using CN finite-difference scheme, the time derivative
∂c of the ADE (12) is approximated at the grid point
= 0; x . 0, y = b, T . 0 (17)
∂y (i, j, n + 1/2), and the space derivatives are replaced
by their means of central-difference approximations
at the n th and (n + 1) th time levels. In addition,
the retardation factor R of Equation (12) is approxi-
Solution of the 2D governing equation by
mated according to the previous time level (i.e. at
FDM
the time level n), so that the resultant finite-
Let T is divided into P number of sub-intervals with difference approximation would become linear-
each step length DT. Again, let a and b are divided ized [29]. Thus, Equation (12) is approximated as
 ENVIRONMENTAL TECHNOLOGY 1907

follows:
 n+1  n+1
ci,jn+1 − ci,jn Dx0 ci+1,j − 2ci,j + ci−1,j ci+1,j
n+1 n+1 n
− 2ci,jn + ci−1,j
n
Dy0 ci,j+1 − 2ci,j + ci,j−1 ci,j+1
n+1 n+1 n
− 2ci,jn + ci,j−1
n
Rni,j = + + +
DT 2 (Dx)2 (Dx)2 2 (Dy)2 (Dy)2
 n+1 
u0 ci+1,j − ci−1,j ci+1,j − ci−1,j v0 ci,j+1 − ci,j−1 ci,j+1 − ci,j−1
n+1 n n n+1 n+1 n n
− + − +
2 2Dx 2Dx 2 2Dy 2Dy

Using Equations (21) and (23) in Equation (19) gives

(18)
r1 r3 n+1 r1 r3 n+1
−r4 ci,j−1
n+1
− + ci−1,j +(Rni,j +r3 +r4 )ci,jn+1 + − ci+1,j
Rearrangement of Equation (18) takes the following 4 2 4 2
r1 r3 n r1 r3 n
form: =r4 ci,j−1 + + ci−1,j +(Ri,j −r3 −r4 )ci,j − − ci+1,j
n n n
4 2 4 2
r2 r4 n+1 r1 r3 n+1 (25)
− + c − + c + (Rni,j + r3 + r4 )ci,jn+1
4 2 i,j−1 4 2 i−1,j where, i =1,2,3, ...,M−1 and j =N.
r1 r3 n+1 r2 r4 n+1 Again, using Equations (20)–(23) in Equation (19)
+ − c + − c
4 2 i+1,j 4 2 i,j+1 yields
r2 r4 n r1 r3 n
= + c + + c −r4 ci,j−1
n+1
− r3 ci−1,j
n+1
+ (Rni,j + r3 + r4 )ci,jn+1
4 2 i,j−1 4 2 i−1,j
+ (Rni,j − r3 − r4 )ci,jn −
r1 r3 n
− c −
r2 r4 n
− c = r4 ci,j−1
n
+ r3 ci−1,j
n
+ (Rni,j − r3 − r4 )ci,jn (26)
4 2 i+1,j 4 2 i,j+1
(19) where, i = M and j = N.
The implicit Equations (19), (24–26) all together will
where, i = 1, 2, 3, . . . , M − 1, j = 1, 2, 3, . . . , N − 1, finally give an approximate solution of the solute
u0 DT v0 DT Dx DT Dy DT transport Equation (12). When M = N, the above
r1 = , r2 = , r3 = 0 2 , and r4 = 0 2 .
Dx Dy Dx Dy implicit equations result an M × M pentadiagonal
system of linear algebraic equations at each time
Again, approximating the derivative boundary con-
level. Solution of these equations is obtained in
ditions given in Equations (16) and (17) at the time
graphical form for a fixed time period with the help
level n for the extreme boundary points i = M and
of Matlab coding [34]. The obtained graphical
j = N, respectively as follows:
results are illustrated in the Results and Discussion
section.
n
ci+1,j = ci−1,j
n
, i=M (20)

n
andci,j+1 = ci,j−1
n
, j=N (21) Approximation by ADI method
The solution of the 2D ADE (12) by CN method
Also, at the (n + 1)th time level, Equations (20) and (21)
results a pentadiagonal system of linear algebraic
can be written respectively as
equations at each time level, whereas ADI technique
results a couple of two tridiagonal system of linear
n+1
ci+1,j = ci−1,j
n+1
, i=M (22)
algebraic equations in each time interval which are
more efficient to solve than the pentadiagonal
n+1
and ci,j+1 = ci,j−1
n+1
, j=N (23) linear system and hence reduces the computational
cost. Solution of the 2D ADE by ADI technique con-
Using Equations (20) and (22) in Equation (19) gives tains two stages of computation in each time inter-
val [35].
r2 r4 n+1 r2 r4 n+1 In the first stage, all the spatial and temporal deriva-
− + ci,j−1 −r3 ci−1,j
n+1
+(Rni,j +r3 +r4 )ci,jn+1 + − ci,j+1 tives of Equation (12) are replaced by central-difference
4 2 4 2
r2 r4 n r2 r4 n approximations of second order in which the
= + ci,j−1 +r3 ci−1,j
n
+(Rni,j −r3 −r4 )ci,jn − − ci,j+1 x-derivative entries are approximated at the (n + 1/2)
4 2 4 2
(24) time level having unknown concentration, whereas the
y-derivative entries are approximated at the time level
where, i =M and j = 1,2,3, ...,N−1. n having known concentration. Thus, Equation (12) is
1908 R. K. SINGH ET AL.

expressed as Equation (12) is expressed as

ci,jn+1/2 − ci,jn n+1/2

ci+1,j − 2ci,jn+1/2 + ci−1,j
n+1/2
−r4 ci,j−1
n+1
+ r7 ci,jn+1 + r6 ci,j+1
n+1
Rni,j = Dx0
DT/2 (Dx) 2
n+1/2
= r1 ci−1,j + r8 ci,jn+1/2 − r3 ci+1,j
n+1/2
(29)
n
ci,j+1 − n
2ci.j + ci,j−1
n
+ Dy0 2
(Dy) Dy0 DT Dx DT
where, r7 = Rni,j + , r8 = Rni,j − 0 2 , and j runs
n+1/2
ci+1,j n+1/2
− ci−1,j n
ci,j+1 − ci,j−1
n (Dy)2 (Dx)
− u0 − v0 from 1 to N − 1 for each value of i = 1, 2, 3, . . . , M − 1.
2Dx 2Dy
(27) Again, approximating the derivative boundary con-
ditions (16) and (17) at the extreme points i = M and
After rearrangement, Equation (27) takes the form j = N, respectively, as follows:
n+1/2
−r1 ci−1,j + r2 ci,jn+1/2 + r3 ci+1,j
n+1/2
n+1/2 n+1/2
ci+1,j = ci−1,j , i=M (30)
= r4 ci,j−1
n
+ r5 ci,jn − r6 ci,j+1
n
(28)

u0 DT Dx0 DT Dx DT and
where r1 =+ , r2 = Rni,j + 0 2 ,
4Dx 2(Dx)2 (Dx)
u0 DT Dx0 DT v0 DT Dy0 DT
n+1
ci,j+1 = ci,j−1
n+1
, j=N (31)
r3 = − , r4 = + ,
4Dx 2(Dx)2 4Dy 2(Dy)2
Dy DT v0 DT Dy0 DT Also, at the nth time level, Equation (17) can be approxi-
r5 = Rni,j − 0 2 , r6 = − , and i runs from 1 mated for the boundary point j = N as
(Dy) 4Dy 2(Dy)2
to M − 1 for each value of j = 1, 2, 3, . . . , N − 1. n
ci,j+1 = ci,j−1
n
, j=N (32)
In the second stage, the procedure is vice-versa of the
stage one for the next half time interval. Approximating
Using Equation (30) in Equations (28) and (31) in
the y-derivative entries of Equation (12) at the (n + 1)th
Equation (29) yield, respectively
time level having unknown concentration, while the
x-derivative entries approximated at the (n + 1/2) time n+1/2
level having known concentration. Thus, in this stage, (r3 − r1 )ci−1,j + r2 ci,jn+1/2 = r4 ci,j−1
n
+ r5 ci,jn − r6 ci,j+1
n
(33)

Figure 2. Two-dimensional contaminant concentration distribution profile obtained by CN method.

 ENVIRONMENTAL TECHNOLOGY 1909

Figure 3. Two-dimensional contaminant concentration distribution profile obtained by ADI method.

where, i = M, and j = 1, 2, 3, . . . , N − 1. where i = 1, 2, 3, . . . , M − 1 and j = N.

and (r6 − r4 )ci,j−1
n+1
+ r7 ci,jn+1 and − r4 ci,j−1
n+1
+ r7 ci,jn+1 + r6 ci,j+1
n+1

n+1/2
= r1 ci−1,j + r8 ci,jn+1/2 − r3 ci+1,j
n+1/2
(34) n+1/2
= (r1 − r3 )ci−1,j + r8 ci,jn+1/2 (36)

where i = 1, 2, 3, . . . , M − 1 and j = N. where, i = M and j = 1, 2, 3, . . . , N − 1.

Using Equation (32) in Equation (28) and (30) in Using Equations (30) and (32) in Equation (28)
Equation (29) yield, respectively yields
n+1/2
−r1 ci−1,j + r2 ci,jn+1/2 + r3 ci+1,j
n+1/2 n+1/2
(r3 − r1 )ci−1,j + r2 ci,jn+1/2 = (r4 − r6 )ci,j−1
n
+ r5 ci,jn ; i = M, j = N
= (r4 − r6 )ci,j−1
n
+ r5 ci,jn (35) (37)

Figure 4. Two-dimensional contaminant concentration distribution profiles obtained by (a) CN method and (b) ADI method.
1910 R. K. SINGH ET AL.

Using Equations (30) and (31) in Equation (29) yields obtained by CN and ADI methods, respectively. In each
n+1/2 figure, the solute concentration is maximum at the
(r6 −r4 )ci,j−1
n+1
+r7 ci,jn+1 = (r1 −r3 )ci−1,j +r8 ci,jn+1/2 ; i = M, j = N
origin from where it decreases with space and attains
(38) minimum level near the final boundaries. The distri-
For M = N, the two set of implicit Equations (28), (33), bution of solute concentration is symmetrical about
(35), (37) and (29), (34), (36), (38) separately results the the diagonal line due to same amount of input concen-
tridiagonal system of linear algebraic equations. All trations assigned along the coordinate axes. Both the
the implicit Equations (28), (29), (33–38) solved figures depict almost similar contaminant distribution
together with the help of Matlab coding and the profiles; however, a minor difference can be observed
graphical results obtained are illustrated in the next between them. The advancement of the contaminant
section. concentration in Figure 2 is slightly greater than
Figure 3.
The solute concentration distribution profiles
Results and discussion obtained at time t = 1 year by CN and ADI methods
In this study, the seepage velocity patterns are con- under linear sorption are also compared graphically by
sidered as u = u0 exp (−mt) and v = v0 exp (−mt). The contour plot as shown in Figure 4. The solute concen-
input values of advection and dispersion coefficients trations having maximum amount along the coordinate
and other transport parameters are taken same in hori- axes decrease with space towards the final boundaries in
zontal and transversal directions to obtain the symmetri- the clay medium. The advancement of the solute con-
cal graphical solutions. The following input data are centration in both the figures can be seen throughout
used to obtain the graphical solutions of the governing the medium. But still, it is not clear which one gives
equation [30,36,37]: more accurate result. However, the further discussion

M = 20, N = 20, P = 50, a = 1 km, b = 1 km, c0 = 0.03, c1 = 0.50, c2 = 0.50,

m = 0.002 year−1 , u0 = 0.015 km/year, v0 = 0.015 km/year, Dx0 = 0.15 km2 /year
Dy0 = 0.15 km2 /year, r = 2.19, f = 0.55(Clay), kd = 0.25, g1 = 0.001 year−1
g2 = 0.001 year−1 , a = 0.25, and b = 0.30.
Figures 2 and 3 depict the solute concentration distri- will be helpful to judge about the accuracy between
bution profiles in clay formation at time t = 1 year the two methods used in this study.

Figure 5. Contaminant concentration distribution profiles obtained by CN method (dotted lines) and PDEtool (solid lines) under
special case.
 ENVIRONMENTAL TECHNOLOGY 1911

Figure 6. Contaminant concentration distribution profiles obtained by ADI method (dotted lines) and PDEtool (solid lines) under
special case.

The accuracy of a numerical solution is generally them with another numerical solution obtained by
observed by comparing with analytical solution of the PDEtool available in Matlab software. The approximate
same model problem. In this study, it is difficult to solution of the solute transport equation is obtained
obtain the exact solution of the present governing by PDEtool, which is based on FEM [34], under special
model problem. Even though, the accuracies of the case (m = 0, g1 = g2 = 0, R = 1). The numerical sol-
two numerical solutions are examined by comparing utions obtained by CN and ADI methods under the

Figure 7. Contaminant concentration distribution profiles at different time periods obtained by CN method under linear sorption.
1912 R. K. SINGH ET AL.

Figure 8. Contaminant concentration distribution profiles in different porous formations obtained by CN method under linear
sorption.

special case are compared with the solution obtained by while Figure 6 depicts slightly heterologous. This
PDEtool as shown in Figures 5 and 6, respectively. implies that the CN method gives more accurate result
Figures 5 and 6 are obtained at time t = 1 year subject than the ADI method for the present model problem.
to time interval length DT = 0.02 year and the spatial Therefore, from Figure 7 and onwards all the graphical
interval lengths DX = DY = 0.05 km. The solute concen- solutions are obtained by CN method only.
tration distribution profiles obtained by CN method and Figure 7 depicts the solute concentration distribution
PDEtool in Figure 5 depict analogous to each other, profiles obtained by CN method under linear sorption at

Figure 9. Contaminant concentration distribution profiles obtained by CN method for different values of the non-linearity power.
 ENVIRONMENTAL TECHNOLOGY 1913

Figure 10. Contaminant concentration distribution profiles obtained by CN method under Langmuir non-linear sorption and no
sorption.

time t = 1, 2, 3 years in clay medium. The concentration with the inlet and outlet boundaries. In addition, the
of solute increases with time whereas decreases with contaminant concentration increases on increasing
space throughout the finite homogeneous medium. the value of non-linearity power under the consider-
Also, it can be observed that the increasing rate of ation of sorption.
the contaminant concentration slightly gets lower with Figure 10 depicts the solute concentration distri-
time. bution profiles under the consideration of Langmuir
Figure 8 depicts the concentration transport profiles non-linear sorption and no sorption. The solute concen-
in various soil media at time t = 1 year obtained by tration attains lower level under the Langmuir sorption
CN method under linear sorption. The contaminant con- condition as compared with the no sorption condition.
centration achieves highest level in silt formation as The variation between the concentration profiles
compared to the sand and gravel media. Whereas, the increases with space up to the extreme boundaries.
contaminant concentration attains lowest level in the
gravel formation. There may be observed larger vari-
ations among the three contaminant concentration
Conclusions
transport profiles at the intermediate location as com- The 2D ADE under linear and non-linear sorption subject
pared to the initial and final boundary areas. to the axial input sources has been solved numerically
Figure 9 shows the contaminant distribution by CN and ADI methods to examine the solute concen-
profiles at time t = 1year for different values of the tration distribution profiles in finite uniform soil media.
non-linearity power of the Freundlich sorption iso- The numerical solutions have been illustrated graphi-
therm. The solute concentration profiles are obtained cally with the help of Matlab coding. The ADI method
under the no sorption, linear sorption, and non- is found more efficient than the CN method. However,
linear sorption conditions. The solute concentration the CN method gives the more accurate result than
attains the highest level under no sorption condition the ADI method as compared with the solution obtained
as compared to the consideration of linear and non- by PDEtool. Hence, the CN method may be used effec-
linear sorption. There can be observed a larger gap tively to solve the present model problem subject to
between the solute concentration profiles obtained axial input sources in finite uniform porous media. The
for without sorption and with sorption. Larger vari- present study may further be extended to solve the
ations among the solute transport profiles are solute transport equation in heterogeneous semi-
observed at the intermediate location as compared infinite porous media.
1914 R. K. SINGH ET AL.

Acknowledgements [11] Ma Y, Sun CP, Haake DA, et al. A high-order alternating

direction implicit method for the unsteady convection-
The authors are thankful to the Indian Institute of Technology dominated diffusion problem. Int J Numer Methods
(Indian School of Mines), Dhanbad and Adamas University, Fluids. 2012;70:703–712.
Kolkata for providing facilities to complete the present work. [12] Savovic S, Djordjevich A. Finite difference solution of the
one-dimensional advection-diffusion equation with vari-
able coefficients in semi-infinite media. Int J Heat Mass
Disclosure statement Transf. 2012;55(15–16):4291–4294.
[13] Savovic S, Djordjevich A. Numerical solution for tem-
No potential conflict of interest was reported by the author(s). porally and spatially dependent solute dispersion of
pulse type input concentration in semi-infinite media.
Int J Heat Mass Transf. 2013;60:291–295.
Funding [14] Djordjevich A, Savovic S. Solute transport with longitudi-
nal and transverse diffusion in temporally and spatially
The author(s) reported there is no funding associated with the dependent flow from a pulse type source. Int J Heat
work featured in this article. Mass Transf. 2013;65:321–326.
[15] van Genuchten MT, Leij FJ, Skaggs TH, et al. Exact analyti-
cal solutions for contaminant transport in rivers
Data availability statement 2. Transient storage and decay chain solutions. J Hydrol
Hydromech. 2013;61:250–259.
Some or all data, models, or code that support the findings of
[16] Wadi AS, Dimian MF, Ibrahim FN. Analytical solutions for
this study are available from the corresponding author upon
one-dimensional advection–dispersion equation of the
reasonable request.
pollutant concentration. J Earth Syst Sci. 2014;123
(6):1317–1324.
[17] Singh MK, Das P. Scale dependent solute dispersion with
References linear isotherm in heterogeneous medium. J Hydrol.
2015;520:289–299.
[1] Schwartz FW, Zhang H. Fundamentals of groundwater.
[18] Djordjevich A, Savovic S, Janićijević A. Explicit finite-differ-
New York (NY): John Wiley & Sons; 2013.
ence solution of two-dimensional solute transport with
[2] Leij FJ, Skaggs TH, van Genuchten MT. Analytical sol-
periodic flow in homogenous porous media. JHydro
utions for solute transport in three dimensional semi-
Hydromech. 2017;65(4):426–432.
infinite porous media. Water Resour Res. 1991;27:2719–
[19] Guymon GL, Scott VH, Herrmann LR. A general numerical
2733.
solution of the two-dimensional diffusion-convection
[3] Batu V. Applied flow and solute transport modeling in
equation by the finite element method. Water Resour
aquifers: fundamental principles and analytical and
Res. 1970;6:1611–1617.
numerical methods. Boca Raton: CRC Press; 2005.
[20] Hossain MA, Yonge DR. On Galerkin models for transport
[4] Wang GT, Li BQ, Chen S. A semi analytical method for
in groundwater. Appl Math Comput. 1999;100:249–263.
solving equations describing the transport of contami-
[21] Povich TJ, Dawson CN, Farthing MW, et al. Finite element
nants in two-dimensional homogeneous porous media.
methods for variable density flow and solute transport.
Math Comput Model. 1999;30(9–10):63–74.
Comput Geosci. 2013;17:529–549.
[5] Dehghan M. Numerical solution of the three-dimensional
[22] Zhao Y, Bu W, Huang J, et al. Finite element method
advection-diffusion equation. Appl Math Comput.
for two-dimensional space- fractional advection-dis-
2004;150:5–19.
persion equations. Appl Math Comput. 2015;257:553–
[6] Ataie-Ashtiani B, Hosseini SA. Error analysis of finite
565.
difference methods for two-dimensional advection-dis-
[23] Chen JS, Hillman M, Chi SW. Meshfree methods: progress
persion-reaction equation. Adv Water Resour.
made after 20 years. J Eng Mech. 2017;143:04017001.
2005;28:793–806.
[24] Askari M, Adibi H. Numerical solution of advection-
[7] Savovic S, Djordjevich A, Tse PW. Explicit finite difference
diffusion equation using meshless method of lines. Iran
solution of the diffusion equation describing the flow of
J Sci Technol Trans Sci. 2017;41:457–464.
radon through soil. Appl Radiat Isot. 2011;69:237–240.
[25] Moranda A, Cianci R, Paladino O. Analytical solutions of
[8] Guerrero JP, Skaggs TH. Analytical solution for one-
one-dimensional contaminant transport in soils with
dimensional advection-dispersion transport equation
source production-decay. Soil Syst. 2018;2(3):40.
with distance-dependent coefficients. J Hydrol.
[26] Singh MK, Singh RK, Pasupuleti S. Study of forward–back-
2010;390:57–65.
ward solute dispersion profiles in a semi-infinite ground-
[9] Kumar A, Jaiswal DK, Kumar N. Analytical solutions to
water system. Hydrol Sci J. 2020;65(8):1416–1429.
one-dimensional advection-diffusion equation with vari-
[27] Banaei SMA, Javid AH, Hassani AH. Numerical simulation
able coefficients in semi-infinite media. J Hydrol.
of groundwater contaminant transport in porous media.
2010;380:330–337.
Int J Environ Sci Tech. 2021;18(1):151–162.
[10] Chen JS, Chen JT, Liu CW, et al. Analytical solutions to
[28] van Genuchten MT., Alves WJ. Analytical solutions of the
two-dimensional advection-dispersion equation in
one-dimensional convective-dispersive solute transport
cylindrical coordinates in finite domain subject to first-
equation (No. 1661). US Department of Agriculture,
and third-type inlet boundary conditions. J Hydrol.
Agricultural Research Service; 1982.
2011;405:522–531.
 ENVIRONMENTAL TECHNOLOGY 1915

[29] Zheng C, Bennett GD. Applied contaminant [33] Crank J. The mathematics of diffusion. Oxford: Oxford
transport modeling. Vol. 2. New York (NY): Wiley- University Press; 1975.
Interscience; 2002. [34] Yang WY, Cao W, Kim J, et al. Applied numerical methods
[30] Bosma WJP, Van Der Zee SE. Analytical approximation using MATLAB. River Street, Hoboken: John Wiley & Sons;
for nonlinear adsorbing solute transport and first- 2020.
order degradation. Trans Porous Media. 1993;11(1):33– [35] Hoffman JD. Numerical methods for engineering and
43. scientists. New York (NY): McGraw-Hill, Inc; 1992.
[31] Yadav SK, Kumar A, Kumar N. Horizontal solute transport [36] Singh MK, Kumari P. Contaminant concentration predic-
from a pulse type source along temporally and spatially tion along unsteady groundwater flow. Modelling and
dependent flow: analytical solution. J Hydrol (Amst). simulation of diffusive processes. Springer Cham:
2012;412–413:193–199. Springer. 2014. p. 257–275.
[32] Freeze RA, Cherry JA. Groundwater. Englewood Cliffs (NJ): [37] Serrano SE. Propagation of nonlinear reactive contaminants
Prentice-Hall, Inc.; 1979. in porous media. Water Resour Res. 2003;39(8):1288.