Journal of Hydrology 380 (2010) 330–337

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: [Link]/locate/jhydrol

Analytical solutions to one-dimensional advection–diffusion equation

with variable coefficients in semi-infinite media
Atul Kumar, Dilip Kumar Jaiswal, Naveen Kumar *
Department of Mathematics, Faculty of Science, Banaras Hindu University, Varanasi 221005, India

a r t i c l e i n f o s u m m a r y

Article history: In the present study one-dimensional advection–diffusion equation with variable coefficients is solved
Received 2 June 2009 for three dispersion problems: (i) solute dispersion along steady flow through an inhomogeneous med-
Received in revised form 5 November 2009 ium, (ii) temporally dependent solute dispersion along uniform flow through homogeneous medium and
Accepted 9 November 2009
(iii) solute dispersion along temporally dependent flow through inhomogeneous medium. Continuous
point sources of uniform and increasing nature are considered in an initially solute free semi-infinite
This manuscript was handled by P. Baveye,
Editor-in-Chief, with the assistance of medium. Analytical solutions are obtained using Laplace transformation technique. The inhomogeneity
Renduo Zhang, Associate Editor of the medium is expressed by spatially dependent flow. Its velocity is defined by a function interpolated
linearly in a finite domain in which concentration values are to be evaluated. The dispersion is considered
Keywords: proportional to square of the spatially dependent velocity. The solutions of the third problem may help
Advection understand the concentration dispersion pattern along a sinusoidally varying unsteady flow through an
Diffusion inhomogeneous medium. New independent variables are introduced through separate transformations,
Dispersion in terms of which the advection–diffusion equation in each problem is reduced into the one with the con-
Spatial and temporal dependence stant coefficients. The effects of spatial and temporal dependence on the concentration dispersion are
Aquifer studied with the help of respective parameters and are shown graphically.
Ó 2009 Elsevier B.V. All rights reserved.

Introduction advection–diffusion partial differential equations with constant

coefficients have been obtained subject to variety of initial and
With respect to the importance of hydro-dynamic dispersion boundary conditions, and incorporating the additional effects on
process studies in water quality management and pollution control dispersion due to adsorption, first-order decay and zero-order pro-
particularly in aquifers, the dispersion has been referred to as a duction. Deviating from the ideal conditions, and coming nearer to
hydraulic mixing process by which the waste concentrations are real dispersion problems, unsteady nature of solute transport as
attenuated while the waste pollutants are being transported well as of the flow field, and inhomogeneity of the medium have
downstream. The concentration distribution behavior with space been the key features. Analytical solutions of the former type of
and time is described by a partial differential equation of parabolic problems have been obtained by using a direct relationship be-
type known as advection–diffusion equation. This equation is tween D and u (Ebach and White, 1958; Rumer, 1962). Scheidegger
equally important in soil physics, bio-physics, petroleum engineer- (1957) summarized the analyses on relationships between the two
ing and chemical engineering for describing similar processes. A parameters and had proposed two possible relationships: (i) D is
one-dimensional linear advection–diffusion equation, derived on proportional to u and (ii) D is proportional to u2 . Inhomogeneity
the principle of conservation of mass, is of the medium was addressed by dividing the medium into strati-
  fied layers, transport properties being homogeneous in each layer
@C @ @C but differing from the other layer. But analytical solutions were
¼ Dðx; tÞ  uðx; tÞC
@t @x @x possible for two or three layers only. To overcome this difficulty,
If D and u are constants then the two are called dispersion coef- porosity-position variable continuous functions were defined but
ficient and uniform velocity of the flow field, respectively. Cðx; tÞ is with them the dispersion problems were solved only by numerical
the dispersing solute concentration at a position x along the longi- methods. Later, on the basis of the observation of Matheron and
tudinal direction at a time t. Many analytical solutions of the deMarsily (1980) that in large subsurface formations dispersivity
may vary with time or distance, analytical solutions of one-dimen-
sional problems were obtained using scale-dependent dispersion
* Corresponding author. Address: P-12, New Medical Enclave, BHU, Varanasi
221005, India. Tel.: +91 9450541328. and uniform flow. Research papers related to these problems have
E-mail addresses: nks_1953@[Link], naveen@[Link] (N. Kumar). been reviewed in our previous paper (Jaiswal et al., 2009). A

0022-1694/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/[Link].2009.11.008
 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337 331

concise review of the experiments of the dependence of the disper- ized by the linear variation of increasing nature but of small order,
sion coefficient on Peclet number, is contained in the paper by in velocity of the flow field. The second problem investigates the
Neung-Won et al. (1985). Neelz (2006) has analyzed the limita- temporally dependent solute dispersion along uniform flow. The
tions of analytical solution for advection–diffusion equation with third problem elucidates the solute transport behavior along un-
coefficients being function of space variable. steady flow domain through inhomogeneous medium. The first
The engineering background of one-dimensional hydro-dy- two problems are also solved analytically for the pulse type point
namic dispersion problems have been addressed in many more sources in our previous paper (Jaiswal et al., 2009). In the present
realistic works, some are cited here. Watson and Jones (1982) con- paper the point sources of uniform and of increasing nature are ap-
sidered velocity dependent hydro-dynamic dispersion. Fry et al. plied in a continuous time domain. The third problem is an addi-
(1993) derived a comprehensive set of solutions for one-dimen- tional contribution in which the solute transport along
sional non-equilibrium solute transport with first-order decay sinusoidally varying flow field has been found much slower than
and zero-order production in a semi-infinite soil system. Lin and that along steady flow field, both the flow fields being through
Ball (1998) presented analytical solution for solute diffusion in a the inhomogeneous medium. The effects of inhomogeneity and
two-layer porous medium for arbitrary boundary and initial condi- temporal dependence on the solute transport are also studied.
tions by using Green’s function approach in Laplace domain. The With the help of such studies, the uniform or increasing degrada-
underlying idea in the work of Chen and Sheu (2000) in the devel- tion of underground hydro-environment, that on the earth surface,
opment of steady discretization scheme was to incorporate the that of the atmospheric clean air, may be assessed with reasons.
analytical solution, obtained within a one-dimensional context, Such degradations are caused due to the advective diffusive trans-
into the formulation of scalar transport governed by the convec- port of pollutant tracers from the point sources. A location in the
tion–diffusion equation with linear production in two dimensions. river bed where drainage meets, emission point of smokes from a
Chen et al. (2003) solved the advection–dispersion equation in chimney, are examples of the point sources. On the basis of the
cylindrical coordinates in a radially convergent flow field by apply- data collected from different locations of a polluted region on daily,
ing a Laplace-transformed power series technique. Didierjean et al. or weekly or fortnightly basis, such studies may further be modi-
(2004) implemented the quadrupole method to simulate the ef- fied. Hydrological problems like salt water intrusion, evapotranspi-
fects of heterogeneities in one-dimensional advective and diffusive ration can also be understood in similar situations.
transport of a passive solute in two-layered infinite porous media.
Karahan (2006) proposed one-dimensional advection–diffusion Analytical solutions
equation with finite difference method using implicit speardsheet
simulation. The model presented by Smedt (2007) for solute trans- The linear advection–diffusion partial differential equation in
port in rivers consists of an advection–diffusion equation for trans- one-dimension is written at the out set of the previous section
port in the main channel with a sink term describing diffusive where symbols have also been defined. Let us write this equation
transport to the hyporheic zone. Jia et al. (2008) used the general- as
ized conditional symmetry approach to study the functional sepa-  
ration of variables of the non-linear diffusion equations. Huang @C @ @C
¼ D0 f1 ðx; tÞ  u0 f2 ðx; tÞC ð1Þ
et al. (2008) presented an unconditionally stable finite element ap- @t @x @x
proach to solve the one-dimensional fractional advection–diffusion In above equation D0 and u0 are constants whose dimensions
equation. Moreira et al. (2009) presented a review of the solutions will depend upon the expressions for f1 ðx; tÞ and f2 ðx; tÞ, respec-
(obtained by generalized integral Laplace transform technique) for tively. Let us introduce a new independent variable using a trans-
the one and two-dimensional, time-dependent advection–diffu- formation (its premise is explained in the Discussion section)
sion equations focusing the application to pollutant simulation in Z
atmosphere, assuming both Fickian and counter-gradient models dx dX 1
X¼ or ¼ ð2Þ
for a wide class of problems. Guerrero et al. (2009) presented a for- f1 ðx; tÞ dx f1 ðx; tÞ
mal exact solution of the linear advection–diffusion equation with
Applying this transformation on Eq. (1), it becomes
constant coefficients for both transient and steady-state regimes
by generalized integral Laplace transform technique. @C @2C @
Since analytical solutions of hydro-dynamic dispersion prob- f1 ðx; tÞ ¼ D0 2 þ u0 ðf2 ðx; tÞCÞ ð3Þ
@t @X @X
lems, even in one-dimension may be obtained in limited cases so
numerical solutions of the problems applicable to realistic engi- Analytical solutions of advection–diffusion equation (1) are ob-
neering have been obtained using finite difference and finite ele- tained in the following three hydro-dynamic dispersion problems
ment methods. In most of these problems either the media are of (discussed in the last of the earlier section). In each problem the
finite extent or the semi-infinite media are transformed into finite semi-infinite medium is considered initially solute free. The con-
ones. But Zhao and Valliappan (1994a,b,c) used time-dependent tinuous point sources of uniform and of increasing nature are
infinite element approach to simulate contaminant transport prob- introduced at the origin of the medium.
lems in infinite media. They considered one and two-dimensional
advection–diffusion equations with constant or variable coeffi- Dispersion through inhomogeneous medium
cients, addressing the anisotropy of the media as well as leakage
effects in porous fissured media. But such numerical solutions need The inhomogeneity of a medium causes variation in velocity of
analytical solution of a similar problem for the comparison to val- the flow through it. A variation of increasing nature is considered.
idate the numerical method used. In this context the analytical From the previous study (Jaiswal et al., 2009) the expression for the
solutions to a new class of one-dimensional spatially and tempo- velocity in a finite longitudinal region 0 6 x 6 L, is
rally dispersion problems through inhomogeneous semi-infinite
uðxÞ ¼ u0 ð1 þ axÞ;
media are obtained in the present paper. These may be used to
construct ‘the mass transport function’ for a new type of transient where a ¼ b=L is another parameter of dimension ðlengthÞ1 ,
infinite element (Zhao, 2009). Three such problems are dealt with accounting for the inhomogeneity of the medium. The velocity in-
analytically. In the first problem solute dispersion is studied creases from a value u0 at x ¼ 0 to a value u0 ð1 þ bÞ at x ¼ L. The in-
through inhomogeneous medium. The inhomogeneity is character- crease in velocity is of small order to insure that it satisfies the
332 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337

  
Darcy law in case the medium is porous or satisfies the laminar con- C0 lnð1 þ axÞ pffiffi
Cðx; tÞ ¼ ð1 þ axÞ1 erfc pffiffiffiffiffiffiffiffi  b t
ditions of the flow in a non-porous medium, a necessary condition 2 2a D0 t
for velocity parameter in the advection–diffusion equation. Further  
lnð1 þ axÞ pffiffi
dispersion parameter is considered proportional to square of the þð1 þ axÞd erfc pffiffiffiffiffiffiffiffi þ b t ð18Þ
velocity (Scheidegger, 1957). Thus in Eq. (3) we consider 2a D0 t

f1 ðx; tÞ ¼ ð1 þ axÞ2 and f 2 ðx; tÞ ¼ ð1 þ axÞ ð4Þ where

Then in Eq. (1) constants D0 and u0 may be referred to as dispersion
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
coefficient and velocity at x ¼ 0. The transformation in Eq. (2) gives
2 w20 u0 þ aD0 u0
a new space variable w0 ¼ ðau0  a D0 Þ; b¼ þ au0 ¼ pffiffiffiffiffiffi ; and d ¼ :
4a2 D0 2 D0 aD0
1
X¼ ð5Þ
að1 þ axÞ The partial differential equation (14) may be obtained directly by
putting the expressions from (4) in (1) where the variable Z will
Hence the partial differential equation (3) becomes
be lnð1 þ axÞ. Also for f1 ðx; tÞ ¼ 1 and f2 ðx; tÞ ¼ 1 in the advection–
@C @2C @C diffusion equation (1) the problem reduces to that in an ideal case
¼ a2 D0 X 2 2 þ au0 X  au0 C: ð6Þ
@t @X @X and its solution is well known. These facts validate the application
of the transformation (2).
To proceed further, initial and boundary conditions are defined.
The semi-infinite medium is supposed initially solute free, so we
Continuous input concentration of increasing nature
consider
Cðx; tÞ ¼ 0; x P 0; t ¼ 0 ð7Þ The input condition defined by Eq. (8) is of uniform nature. It
means that the input concentration at the origin of the medium
Further the continuous input concentration is introduced at the ori-
(water bodies on the surface, groundwater level) and hence its
gin of the domain. The concentration gradient at the infinity is as-
source of pollution on the earth surface remains uniform at all
sumed zero. The analytic solutions are obtained for uniform input
times. But it may not be a real situation. In fact, with increasing hu-
and input of increasing nature, respectively.
man activities on the surface, the pollution at the source and so the
input will increase with time. Such situation may be defined by a
Continuous input concentration of uniform nature
condition of mixed type. We consider

The two boundary conditions are @C

Dðx; tÞ þ auðx; tÞC ¼ u0 C 0 at x ¼ 0; t > 0; ð19Þ
@x
Cðx; tÞ ¼ C 0 ; x ¼ 0; t > 0; ð8Þ
where a is a real constant. For expressions (4) this condition
@C becomes
¼ 0; x ! 1; t P 0 ð9Þ
@x @C
D0 þ au0 C ¼ u0 C 0 at x ¼ 0; t > 0 ð20Þ
The initial and boundary conditions may be written in terms of new @x
space variable X as This condition may be written in terms of space variable Z defined
CðX; tÞ ¼ 0; X P 0; t ¼ 0; ð10Þ by Eq. (13) as

CðX; tÞ ¼ C 0 ; X ¼ 1=a; t > 0; ð11Þ @C

aD0 þ au0 C ¼ u0 C 0 at Z ¼ ln a; t > 0 ð21Þ
@Z
@C
¼ 0; X ¼ 0; t P 0; ð12Þ
@X Now the advection–diffusion equation (14), initial condition (15),
Using a transformation input condition (21) and second boundary condition (17) comprise
a dispersion problem in ðZ; tÞ domain similar to the problem C6 in
Z ¼  ln X; ð13Þ
ðx; tÞ of van Genuchten and Alves (1982) hence or otherwise (using
partial differential equation (6) reduces to the advection–diffusion Laplace transformation technique) the desired solution may be
coefficient with constant coefficients written, for a – 1, as

@C @2C @C   
¼ a2 D0 2  x0  au0 C; ð14Þ u0 C 0 1 lnð1 þ axÞ pffiffi
@t @Z @Z Cðx; tÞ ¼ pffiffiffiffiffiffi ð1 þ axÞ1 erfc pffiffiffiffiffiffiffiffi  b t
2 D0 b þ c 2a D0 t
where x0 ¼ ðau0  a2 D0 Þ and the initial and boundary conditions  
1 lnð1 þ axÞ pffiffi
become  ð1 þ axÞd erfc pffiffiffiffiffiffiffiffi þ b t
bc 2a D0 t
CðZ; tÞ ¼ 0; 0 6 Z < 1; t ¼ 0; ð15Þ
2c
þ ð1 þ axÞb1 expfðb2  c2 Þtg
CðZ; tÞ ¼ C 0 ; Z ¼ ln a; t > 0; ð16Þ b2  c2
 
lnð1 þ axÞ pffiffi
@C erfc pffiffiffiffiffiffiffiffi þ c t ð22aÞ
¼ 0; Z ! 1; t P 0; ð17Þ 2a D0 t
@Z
Thus the dispersion problem undertaken reduces to the initial auffiffiffiffi
where in addition to c ¼ p 0 w0
 p ffiffiffiffi and b1 ¼ pcffiffiffiffi þ 2aw20D , w0 , b and
and boundary value problem (14)–(17), similar to the problem D0 2a D0 a D0 0

C5 of van Genuchten and Alves (1982) hence or otherwise (using d are same as defined in solution (18). In condition (19) if a ¼ 1,
Laplace transformation technique) the desired solution may be then above solution will not occur as b ¼ c hence in this situation
written as the solution may be obtained independently as follows:
 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337 333

  
u0 C 0 1 lnð1 þ axÞ pffiffi CðZ; TÞ ¼ 0; Z P 0; T ¼ 0; ð32Þ
Cðx; tÞ ¼ pffiffiffiffiffiffi ð1 þ axÞ1 erfc pffiffiffiffiffiffiffiffi  b t
D0 4b 2a D0 t
  CðZ; TÞ ¼ C 0 ; Z ¼ 0; T > 0; ð33Þ
1 2b
 ð1 þ axÞ 1 þ pffiffiffiffiffiffi lnð1 þ axÞ þ 4b2 t
d
4b a D0
  @C
lnð1 þ axÞ pffiffi ¼ 0; Z ! 1; T P 0: ð34Þ
erfc pffiffiffiffiffiffiffiffi þ b t @Z
2a D0 t
rffiffiffiffi  # The initial and boundary value problem defined by Eqs. (31)–
t x0 1 2 2 (34) in ðZ; TÞ domain is similar to the problem A1 in ðx; tÞ of
þ exp lnð1 þ axÞ  flnð1 þ axÞg  b t
p 2a2 D0 4a2 D0 t van Genuchten and Alves (1982) hence or otherwise (using
ð22bÞ Laplace transformation technique) the desired solution may be
written as
where x0 , b and d are same as defined in solution (18).
Cðx; tÞ
      
Temporally dependent dispersion along uniform and steady C0 x=f ðmtÞ  u0 T u0 x x=f ðmtÞ þ u0 T
¼ erfc pffiffiffiffiffiffiffiffiffi þ exp erfc pffiffiffiffiffiffiffiffiffi ;
flow 2 2 D0 T D0 f ðmtÞ 2 D0 T
ð35Þ
Temporally dependent solute dispersion of continuous uniform
point source and that of increasing nature along a uniform and where T may be written in terms of t using transformation (26) for
steady flow in a longitudinal semi-infinite homogeneous and ini- an expression of f ðmtÞ.
tially solute free medium are considered. As in the previous study
(Jaiswal et al., 2009), we consider Continuous input concentration of increasing nature
f1 ðx; tÞ ¼ f ðmtÞ and f 2 ðx; tÞ ¼ 1 ð23Þ
The input concentration at x ¼ 0 of increasing nature may be
where m is a coefficient whose dimension is inverse of that of t. So defined by the condition
D0 and u0 in Eq. (1) may be referred to as the initial dispersion coef-
@C
ficient and uniform velocity, respectively. Thus m may be consid- Dðx; tÞ þ uðx; tÞC ¼ u0 C 0 at x ¼ 0; t > 0 ð36Þ

@x
ered proportional to u20 =D0 . From Eq. (2) we have another space
variable In view of expressions (23) and transformations (2), (24), (26), and
x (30), the condition reduces to
X¼ ; ð24Þ
f ðmtÞ @C
D0 þ u0 C ¼ u0 C 0 at Z ¼ 0; T > 0 ð37Þ
in terms of which partial differential equation (3) becomes
@Z
Thus advection–diffusion equation (31), initial condition (32), input
@C @2C @C condition (37) and second boundary condition (34) comprise a dis-
¼ D0 2 þ u 0 ; ð25Þ
@T @X @X persion problem in ðZ; TÞ domain similar to the problem A2 in ðx; tÞ
where of van Genuchten and Alves (1982) hence or otherwise (using La-
Z t
place transformation technique) the desired analytical solution
dt may be written as
T¼ ; ð26Þ
0 f ðmtÞ sffiffiffiffiffiffiffiffiffi
" ( )
is a new time variable. An expression of f ðmtÞ is chosen such that C0 T ðx=f ðmtÞÞ2 u20 T u0 x
Cðx; tÞ ¼ 2u0 exp   þ
f ðmtÞ ¼ 1 for m ¼ 0 (representing uniform dispersion) or for t ¼ 0 2 pD0 4D0 T 4D0 2D0 f ðmtÞ
(initial stage). Also we should get T ¼ 0 at t ¼ 0, so that the nature  
x=f ðmtÞ  u0 T
of the initial condition in the new time domain does not change and þerfc pffiffiffiffiffiffiffiffiffi
2 D0 T
Laplace transformation may be used. To proceed further same initial      
u0 x u2 T u0 x x=f ðmtÞ þ u0 T
and boundary conditions of the first problem, are considered.  1þ þ 0 exp erfc pffiffiffiffiffiffiffiffiffi
D0 f ðmtÞ D0 D0 f ðmtÞ 2 D0 T
Continuous input concentration of uniform nature ð38Þ
where T may be written in terms of t using the transformation (26)
The initial and boundary conditions (7)–(9), in the context of
for an expression of f ðmtÞ.
new independent variables become

CðX; TÞ ¼ 0; 1 < X P 0; T ¼ 0; ð27Þ

Dispersion along unsteady flow through inhomogeneous
medium
CðX; TÞ ¼ C 0 ; T > 0; X ¼ 0; ð28Þ
To predict the concentration distribution behavior along un-
@C
¼ 0; X ! 1; T P 0; ð29Þ steady flow (particularly a sinusoidally varying unsteady flow do-
@X
main) through an inhomogeneous medium, it is assumed that
Further using the inhomogeneity causes an increase in velocity of linear nature
Z ¼ X; ð30Þ defined by Eq. (4) and dispersion is proportional to square of the
velocity (Scheidegger, 1957). Also the flow is supposed to vary with
the initial and boundary value problem in the domain, 0 6 Z < 1, time hence dispersion is also supposed to vary temporally in the
becomes same proportion. We consider following expressions in Eq. (1) in
degenerate forms
@C @2C @C
¼ D0 2  u0 ; ð31Þ
@T @Z @Z f1 ðx; tÞ ¼ f ðmtÞð1 þ axÞ2 ; and f 2 ðx; tÞ ¼ f ðmtÞð1 þ axÞ ð39Þ
334 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337

So the constants D0 and u0 may be referred to as the initial dis- 0.7

persion coefficient and uniform velocity, respectively, at x ¼ 0. t = 0.1 (yr)
t = 0.4 (yr)
Introducing a new time variable T 0.6 t = 0.7 (yr)
Z
t = 1.0 (yr)
T¼ f ðmtÞdt; ð40Þ
0.5

and using the new space variable X defined by Eq. (5), advection–
diffusion Eq. (1) may be reduced into the one with constant coeffi- 0.4
cients as

C/C0
@C @2C @C 0.3
¼ a2 D0 X 2 2 þ au0 X  au0 C ð41Þ
@T @X @X
0.2
It may be noted that T obtained from Eqs. (26) and (40) will
have different expressions for an expression of f ðmtÞ. For the same
initial and boundary conditions, the desired solutions may be ob- 0.1
tained. In case of uniform input condition it will be the same as
that in Eq. (18). For the input condition of increasing nature, the 0.0
solution will be same as that in Eq. (22). The only change will be 0.0 0.2 0.4 0.6 0.8 1.0
that T defined by Eq. (40) will occur in place of old time variable x (km)
t in both the solutions. Fig. 2. Solute dispersion of input of increasing nature along a flow of velocity
u ¼ u0 ð1 þ axÞ and D / u2 , described by solution (22a), for a = 1.0 km1.
Results and discussions

The analytical solutions (18) and (22a) of the first problem ob- 1.0
tained for uniform input and increasing input, respectively, are a = 0.1
a = 1.0
plotted for a set of input data C0 = 1.0, D0 = 0.71 km2/year,
u0 = 0.60 km/year, at t (years) = 0.1, 0.4, 0.7, 1.0. The concentration 0.8
values are evaluated in a finite longitudinal region 0 6 x 6 1.0. The
inhomogeneity parameter a in Eq. (4) is assigned a value 1.0 km1.
It means velocity varies from u0 to 2u0 and dispersion coefficient 0.6
varies from a value D0 to 4D0 in 0 6 x ðkmÞ 6 1:0. Fig. 1 depicts
C/C0

the concentration distribution behavior of uniform input. Fig. 2

illustrates the concentration values of an input which increases 0.4
with time. The constant a, which occurs in input condition (19),
is assigned a value 0.1. It has been observed that the concentration
value at x ¼ 0, at a particular time, decreases if a increases. The 0.2
solution (22b) for a ¼ 1 is not traced. Fig. 3 compares the concen-
tration values ðC=C 0 Þ of solutions (18) and (22a) at t = 0.70 years,
for a = 1.0 km1 (shown by dashed curves) and a = 1.0 km1 0.0
0.0 0.2 0.4 0.6 0.8 1.0
(shown by full lines curves), respectively. In this way the effect
x (km)
of inhomogeneity of the medium on the solute dispersion through
it has been studied. It may be observed that for both the types of Fig. 3. Comparison of solutions (18) and (22a) at t = 0.70 years for two values of a,
point sources, the concentration value at a particular position in respectively.
a medium of higher inhomogeneity (causing larger variation in

1.0
1.0 t = 0.05 (yr)
t = 0.1 (yr) t = 0.20 (yr)
t = 0.4 (yr) t = 0.35 (yr)
t = 0.7 (yr) 0.8 t = 0.50 (yr)
0.8 t = 1.0 (yr)

0.6
0.6
C/C0
C/C0

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x (km) x (km)

Fig. 1. Solute dispersion of uniform input along a flow of velocity u ¼ u0 ð1 þ axÞ Fig. 4. Temporal dependent dispersion of uniform input along uniform flow for
and D / u2 , described by solution (18), for a = 1.0 km1. f ðmtÞ ¼ expðmtÞ described by solution (35).
 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337 335

0.7 1.0 t = 0.2 (yr)

t = 0.05 (yr)
t = 0.20 (yr) t = 0.5 (yr)
0.6 t = 0.35 (yr) t = 0.8 (yr)
0.8 t = 1.1 (yr)
t = 0.50 (yr)
t = 0.2 (yr)
0.5
0.6

0.4

C/C0
0.4
C/C0

0.3

0.2
0.2

0.1 0.0
0.0 0.2 0.4 0.6 0.8 1.0
x (km)
0.0
0.0 0.2 0.4 0.6 0.8 1.0 Fig. 7. Concentration distribution pattern of uniform continuous input along
x (km) sinusoidally varying unsteady velocity through inhomogeneous medium, discussed
in the third problem, shown by four dotted curves. The only solid curve represents
Fig. 5. Temporal dependent dispersion of increasing input along uniform flow for the solution (18).
f ðmtÞ ¼ expðmtÞ given by solution (38).

concentration values for the former expression are higher than

velocity along the domain 0 6 x 6 1:0) is smaller than that in a
those for the latter expression. It becomes more evident as we
medium of less inhomogeneity.
move away from the origin. In case of the point source of increas-
The curves in Figs. 4 and 5 represent the analytical solutions
ing nature the difference is evident at a larger time. In Figs. 3 and 6
(35) and (38) for uniform input and increasing input, respectively
the two curves originating from concentration values 1.0 at x = 0,
of the second problem. The concentration values are evaluated in
represent the case of uniform input concentration. Similarly the
the same domain 0 6 x ðkmÞ 6 1:0, but at different times
two curves originating from different input concentration values
t ðyearsÞ ¼ 0:05; 0:2; 0:35 and 0:5. The input data also differs, ex-
(other than 1.0), represent the case of increasing input.
cept C 0 ¼ 1:0. The other values are: D0 = 1.71 km2/year,
The solute concentration dispersion behavior of uniform input
u0 = 1.60 km/year, m = 0.1 year1. The expression of f ðmtÞ and that
concentration of the third problem is drawn in Fig. 7. A sinusoidal
of the new time variable T obtained from Eq. (26), considered in
expression f ðmtÞ ¼ 1  sin mt, is considered. For it, using Eq. (40),
both the solutions are
new time variable is T ¼ ½t  ð1  cos mtÞ=m. It replaces old time
variable t, in the solution given by Eq. (18), to trace the figure.
1 For mt ¼ 2; 5; 8; 11;    ; f ðmtÞ has the values 0.09, 1.96, 0.01,
f ðmtÞ ¼ expðmtÞ and T ¼ ½1  expðmtÞ; respectively:
m 1.99,. . ., respectively, where m = 10.0 year1 has been chosen. Thus
The concentration distribution behavior of uniform input at the velocity u ¼ u0 ð1  sin mtÞ shows the sinusoidal variation at a
t = 0.5 years and that of increasing input at t = 1.1 years for two fixed time interval of 0.3 years. This solution is shown by four dot-
expressions f ðmtÞ ¼ expðmtÞ and f ðmtÞ ¼ expðmtÞ, respectively, ted curves in Fig. 7 at t (years) = 0.2, 0.5, 0.8, 1.1. The other input
are compared with each other in Fig. 6. The former expression is values are the same as considered to draw Fig. 1. The only solid
of increasing nature while the latter one is of decreasing nature. curve in this figure depicts the concentration values obtained from
It may be observed that in case of the uniform point source, the solution (18) of the first problem at t = 0.2 years. Thus the concen-
tration distribution behavior of a uniform input concentration
along steady flow and that along sinusoidally varying unsteady
flow, are compared with each other. Both the flow fields occur in
1.0
f(mt) = exp(-mt) an inhomogeneous medium as described in this paper. This figure
f(mt) = exp(mt) reveals that concentration values in the latter case are much less
than those in the former case at the positions x > 0. That is, in a
0.9
certain time period, a position away from the source, in a sinusoi-
dally varying unsteady flow field, gets less polluted than that in a
steady flow field. The similar trend has been found (though not
0.8
traced) at the other values of the continuous time domain, though
C/C0

the difference decreases with time.

0.7
It is evident that for a ¼ 0, from Eq. (4) we get f1 ¼ f2 ¼ 1 hence
partial differential equation (1) reduces to one with constant coef-
ficients (which describes the uniform solute dispersion along uni-
0.6
form flow through a homogeneous medium). But the latter’s
solution cannot be obtained by putting a ¼ 0 in solution (18) be-
cause nature of both the partial differential equations are different.
0.5
It may be explained with an easier example. Let us consider an or-
0.0 0.2 0.4 0.6 0.8 1.0 dinary differential equation of second order with variable coeffi-
x (km) cients: ð1 þ axÞ2 y00  ð1 þ axÞy0  3y ¼ 0. For a ¼ 0 it reduces to
ordinary differential equation with constant coefficients:
Fig. 6. Comparison of temporal dispersion along uniform flow of uniform input
y00  y0  3 ¼ 0. The general solution of the platter ffiffiffiffiffiffi one is
described by solution (35), at t = 0.5 years and that of increasing input described by
solution (38), at t = 1.1 years, for two expressions of f ðmtÞ. y ¼ c1 expðb1 xÞ þ c2 expðb2 xÞ, where b1 ; b2 ¼ ð1  13Þ=2. But it
336 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337

cannot be obtained by putting a ¼ 0 in the general solution in case of temporal dependence of solute dispersion of increasing
y ¼ c1 exp a1 x þ c2 exp a2 x of the former ordinary differential equa- nature, and (iii) in case of solute dispersion along steady flow
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
than that along sinusoidally varying flow. The analytical solutions
tion, where a1 ; a2 ¼ ða þ a2 Þ  ða þ a2 Þ2 þ 12a2 2a2 .
in each case may be used for a variety of variations in dispersion
Like other analytical methods, Laplace transformation tech- and velocity due to spatial dependence and temporal dependence,
nique, also provides approximate solution. Among all these meth- respectively.
ods, it is the simplest to use and yields the minimum error of
approximation. So this technique is the most analytically viable. Acknowledgements
The solutions obtained by it are commonly used in validating the
numerical solutions of more realistic problems. In the present The authors express their deep gratitude to the esteemed scien-
study it has become possible to use this technique by using a tists for their valuable comments and references after reviewing
new transformation (2) which introduces another space variable. the submitted paper so thoroughly. This work is part of the re-
This transformation may be said to be the solution of one of the search program for the award of Ph.D. Degree to first two authors.
three ordinary differential equations of the auxiliary system (La- Financial supports to both in the form of Rajeev Gandhi National
grange system) equivalent to a first-order linear partial differential Fellowship and UGC Research Fellowship, respectively, from Uni-
equation versity Grants Commission, Government of India, are gratefully
@C @C acknowledged.
¼ f1 ðx; tÞ  f2 ðx; tÞC
@X @x
But the transformation (2) may be used independently and po- References
sitive sign or negative sign may be assigned according to the prob-
Chen, J.S., Liu, C.W., Liao, C.M., 2003. Two-dimensional Laplace-transformed power
lem. For example, in the second problem one may get same series solution for solute transport in a radially convergent flow field. Advances
solutions by taking positive sign preceding the integral. But as in Water Resources 26, 1113–1124.
the positive sign in the first problem will yield a value like Chen, Y.H., Sheu, T.W.H., 2000. Two-dimensional scheme for convection–diffusion
with linear production. Numerical Heat Transfer, Part B 37, 365–377.
Z ¼  lnð1=aÞ in condition (16) hence here the negative sign is Didierjean, S., Maillet, D., Moyne, C., 2004. Analytical solutions of one-dimensional
necessary. But proceeding with a function like f2 ¼ ð1  axÞ in place macrodispersion in stratified porous media by the quadrupole method:
of that in Eq. (4), one has to take positive sign in the transformation convergence to an equivalent homogeneous porous medium. Advances in
Water Resources 27, 657–667.
(2). Ebach, E.H., White, R., 1958. Mixing of fluids flowing through beds of packed solids.
Some admissible expressions for f ðmtÞ in addition to the two Journal of American Institute of Chemical Engineering 4, 161–164.
considered in the solutions of the second and third problems, Fry, V.A., Istok, J.D., Guenther, R.B., 1993. Analytical solutions to the solute transport
equation with rate-limited desorption and decay. Water Resources Research 29,
may be as follows:
3201–3208.
Guerrero, J.S.P., Pimentel, L.C.G., Skaggs, T.H., van Genuchten, [Link]., 2009. Analytical
1 þ mt; ð1  mtÞ1 ; ð1 þ mtÞ1 ; cos mt; 1  mt  ðmtÞ2 solution of the advection–diffusion transport equation using a change-of-
variable and integral transform technique. International Journal of Heat and
It may be verified that for each expression f ðmtÞ ¼ 1 and T ¼ 0 for Mass Transfer 52, 3297–3304.
t ¼ 0, where T is the corresponding new time variable either ob- Huang, Q., Huang, G., Zhan, H., 2008. A finite element solution for the fractional
advection–dispersion equation. Advances in Water Resources 31, 1578–1589.
tained by transformation (26) or transformation (40). Also Jaiswal, D.K., Kumar, A., Kumar, N., Yadav, R.R., 2009. Analytical solutions for
f ðmtÞ ¼ 1 and corresponding T ¼ t for m ¼ 0, which reduces the temporally and spatially dependent solute dispersion of pulse type input
problems in the second and third problems, to the temporally inde- concentration in one-dimensional semi-infinite media. Journal of Hydro-
environmental Research 2, 254–263.
pendent problems. The values of mt, considered to draw Fig. 7, for Jia, H., Xu, W., Zhao, X., Li, Z., 2008. Separation of variables and exact solutions to
another value m = 0.0165 day1, represents a sinusoidal variation nonlinear diffusion equations with x-dependent convection and absorption.
in the velocity defined by u ¼ u0 ð1  sin mtÞ, at the uniform interval Journal of Mathematical Analysis and Applications 339, 982–995.
Karahan, H., 2006. Implicit finite difference techniques for the advection–diffusion
of 180 days, a pattern in surface water bodies and ground water in equation using spreadsheets. Advances in Engineering Software 37, 601–608.
tropical regions (Kumar and Kumar, 1997). Kumar, N., Kumar, M., 1997. Solute dispersion along unsteady groundwater flow in
a semi-infinite aquifer. Hydrology and Earth System Sciences 2 (1), 93–100.
Lin, C., Ball, W.P., 1998. Analytical modeling of diffusion-limited contamination and
Conclusions decontamination in a two-layer porous medium. Advances in Water Resources
21, 297–313.
Matheron, G., deMarsily, G., 1980. Is transport in porous media always diffusive? A
Analytical solutions to one-dimensional advection–diffusion counter example. Water Resources Research 16, 901–917.
equations with variable coefficients along with initial and bound- Moreira, D.M., Vilhena, M.T., Buske, D., Tirabassi, T., 2009. The state-of-art of the
ary conditions in a semi-infinite domain have been obtained in GILTT method to simulate pollutant dispersion in the atmosphere. Atmospheric
Research 92, 1–17.
three cases: (i) solute dispersion being proportional to square of Neelz, S., 2006. Limitations of an analytical solution for advection–diffusion with
velocity, which increases linearly in small order, from one end variable coefficients. Communications in Numerical Methods in Engineering 22
to the other of a finite longitudinal region, in which concentration (5), 387–396.
Neung-Won, H., Bhakta, J., Carbonell, R.G., 1985. Longitudinal and lateral dispersion
values are evaluated, (ii) temporally dependent solute dispersion in packed beds: effect of column length and particle size distribution. AICHE
along uniform and steady flow field, and (iii) solute dispersion Journal 31 (2), 277–288.
along unsteady flow through inhomogeneous medium. The Rumer, R.R., 1962. Longitudinal dispersion in steady and unsteady flow. Journal
Hydraulics Division 88 (4), 147–172.
sources of the pollution are of uniform and of increasing nature.
Scheidegger, A.E., 1957. The Physics of Flow through Porous Media. University of
Laplace transformation technique has been used in getting the Toronto Press.
analytical solutions. Such solutions may serve as tools in validat- Smedt, F.D., 2007. Analytical solution and analysis of solute transport in rivers
affected by diffusive transfer in the hyporheic zone. Journal of Hydrology 339,
ing numerical solutions in more realistic cases. A transformation
29–38.
(2) has made possible to use this technique in the present study. van Genuchten, [Link]., Alves, W.J., 1982. Analytical solutions of the one-dimensional
In the process some more transformations introducing new inde- convective–dispersive solute transport equation. In: Technical Bulletin, vol.
pendent variables have been used. The observations of the three 1661. US Department of Agriculture.
Watson, K.K., Jones, M.J., 1982. Hydrodynamic dispersion during absorption in finite
problems undertaken in context of concentration attenuation sand: the constant concentration case. Water Resources Research 18 (1), 91–
are that it is faster (i) in a medium of higher inhomogeneity, (ii) 100.
 A. Kumar et al. / Journal of Hydrology 380 (2010) 330–337 337

Zhao, C., 2009. Dynamic and Transient Infinite Elements: Geological, Geotechnical element technique: theory. International Journal for Numerical and Analytical
and Geoenvironmental Applications. Springer, Berlin. Methods in Geomechanics 18, 523–541.
Zhao, C., Valliappan, S., 1994a. Transient infinite element for contaminant transport Zhao, C., Valliappan, S., 1994c. Numerical modelling of transient contaminant
problems. International Journal for Numerical Methods in Engineering 37, 113– migration problems in infinite porous fractured media using finite/infinite
1158. element technique: parametric study. International Journal for Numerical and
Zhao, C., Valliappan, S., 1994b. Numerical modelling of transient contaminant Analytical Methods in Geomechanics 18, 543–564.
migration problems in infinite porous fractured media using finite/infinite