You are on page 1of 12

Hydraulics of Seepage from Trapezoidal Channels

Rajesh Kumar Mahato 1 and Subhasish Dey, M.ASCE 2

Abstract: A tractable analytical solution for steady-state two-dimensional seepage from a trapezoidal channel in a homogeneous, isotropic
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

porous medium of considerable depth is presented. The analysis is performed by applying the method of inversion and the Schwarz-
Christoffel transformation accounting for the capillary action. Because the right half-seepage domain is a mirror image of the left half-seepage
domain about the vertical axis (on either side of the channel central plane) owing to the axisymmetric channel, a solution is sought for the right
half-seepage domain. The results show that an increase in channel bottom width boosts seepage flux. On the other hand, an increase in
channel side slope also amplifies seepage flux. In addition, capillarity plays a subtle role in augmenting the seepage flux from a channel.
The analysis suggests that the dynamic capillary rise (i.e., the vertical rise of seepage water along the side slope owing to capillary action) is
always less than the static capillary rise. The analysis also presents the relation for the seepage velocity distribution along a channel perimeter.
The equations of seepage line coordinates yield the seepage line initially curving downward with a significant lateral shift and eventually
becoming vertical at a great depth. Particular solutions for the triangular and rectangular channels and the vertical slit, which is a channel with
vertical sides and negligible top width, can be obtained from the generalized solution for the trapezoidal channel. Therefore, this study
provides insight into the hydraulics of seepage from a trapezoidal channel, including its particular cases, revealing some new features.
DOI: 10.1061/(ASCE)HY.1943-7900.0001825. © 2020 American Society of Civil Engineers.
Author keywords: Conformal mapping; Hydraulics; Method of inversion; Open channel; Schwarz-Christoffel transformation; Seepage
flow.

Introduction Because of the importance of water management, researchers


are very interested in the underlying hydraulics of seepage from
In the field, seepage from channels is one of the most apparent man- channels. Studies on steady-state seepage from channels and its im-
ifestations of gravity. Water is conveyed through channels for vari- pact on groundwater systems have been extensively documented
ous purposes, for example, irrigation, industrial, and domestic uses. in classical textbooks (Muskat and Wyckoff 1946; Polubarinova-
A substantial quantity of usable water is lost during conveyance, Kochina 1962; Harr 1962; Bear 1972; Hálek and Švec 1979;
and a sizeable fraction of this loss is attributed to seepage from Kovacs 1981). The analytical computation of seepage loss is related
channels (Worstell 1976; Swamee and Chahar 2015). The loss to the relevant hydraulic properties of the ground and the boundary
due to seepage not only leads to the depletion of surface water res- conditions of the flow system (Bouwer 1969). Because seepage
ervoirs but is also a factor in waterlogging and salinizing adjoining flow systems are subject to a multitude of complexities, their analy-
land (Chahar 2007). Therefore, the study of seepage from channels sis usually starts with numerous simplified assumptions. Using the
is indispensable for figuring out the extent of the adjacent land to be method of inversion and the Schwarz-Christoffel transformation,
waterlogged or salinized. Channels are lined to control seepage, but Vedernikov (1939) studied two-dimensional steady-state seepage
lining deteriorates over time. The magnitude of seepage from a from a trapezoidal channel in a homogeneous, isotropic porous
channel with deteriorated lining gradually approaches that from medium of significant depth. He also examined seepage from a tri-
an unlined channel (Wachyan and Rushton 1987). Consequently, angular channel, which is a particular case of a trapezoidal channel.
the study of seepage is essential even for deteriorating lined chan- Additionally, capillarity plays a vital role in augmenting seepage
nels. Seepage from channels can also be thought of as an incidental from channels. The rise of water within the ground above the
recharge to the subsurface reservoir (Choudhary and Chahar 2014). groundwater level stems from capillary action. Vedernikov (1940)
Therefore, a deep understanding of seepage from a channel is im- studied steady-state two-dimensional seepage from a trapezoidal
perative to improve the management of water resource systems. channel considering the capillary action, aided by the method of
inversion and the Schwarz-Christoffel transformation. He presented
1 computations for trapezoidal and triangular channels with a side
Graduate Research Fellow, Dept. of Civil Engineering, Indian Institute
of Technology Kharagpur, West Bengal 721302, India. Email: rkm22@ slope of unity and a vertical slit and found that the vertical rise
iitkgp.ac.in of seepage water along a channel side slope owing to capillary ac-
2 tion is significantly less than static capillary rise.
Professor, Dept. of Civil Engineering, Indian Institute of Technology
Kharagpur, West Bengal 721302, India; Distinguished Visiting Professor, Risenkampf (1940) applied a methodology similar to that used
Dept. of Hydraulic Engineering, State Key Laboratory of Hydro-Science by Vedernikov (1940) to study the steady-state two-dimensional
and Engineering, Tsinghua Univ., Beijing 100084, China (corresponding seepage from a channel with low water level considering the capil-
author). ORCID: https://orcid.org/0000-0001-9764-1346. Email: sdey@ lary action in a homogeneous and isotropic porous medium of great
iitkgp.ac.in
depth. Morel-Seytoux (1961) analyzed seepage from a rectangular
Note. This manuscript was submitted on April 7, 2020; approved on
July 13, 2020; published online on October 9, 2020. Discussion period channel using the Schwarz-Christoffel transformation. He also
open until March 9, 2021; separate discussions must be submitted for investigated seepage from a nearly rectangular channel using con-
individual papers. This paper is part of the Journal of Hydraulic En- formal mapping and the Green-Neumann function. Swamee et al.
gineering, © ASCE, ISSN 0733-9429. (2000) proposed simplified algebraic equations for the computation

© ASCE 04020083-1 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


of seepage flux from triangular, rectangular, and trapezoidal chan- solutions subject to conformal transformations (Harr 1962). Flow
nels based on the derivations of Vedernikov (1939) and Morel- systems primarily driven by gravity in a porous medium consist of a
Seytoux (1961). They also examined minimum seepage loss seepage line (Muskat and Wyckoff 1946). In seepage hydraulics,
channel sections using simplified algebraic equations of seepage the position and geometry of the seepage line are not known a pri-
flux and a resistance equation of open-channel flow. Using the ori, but its hodograph can be readily constructed (Harr 1962).
method of inversion and the Schwarz-Christoffel transformation, Therefore, hodographs are used categorically in unconfined seep-
Swamee et al. (2001) and Chahar (2007) provided analytical solu- age studies. A hodograph can be conceived as a mapping of the
tions for quantifying seepage flux from rectangular and trapezoidal actual seepage field onto the (u–iv) plane, where u and v are the
channels, respectively, within a ground layer underlain by a drain- seepage velocities in the horizontal and vertical directions, respec-
age layer at a shallow depth. Chahar (2007) presented relations for tively (Polubarinova-Kochina 1962). Because hodographs help in
seepage velocity variation along a channel perimeter and seepage obtaining valuable information about the velocity conditions along
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

line profiles. He also deduced the particular solutions for triangular, surfaces, which are actually unknown on the actual seepage plane,
rectangular, and slit-shaped channels. they are helpful in solving practical problems. The hodograph can
Kacimov (2004) studied steady two-dimensional seepage into a also provide an aid to verify the compatibility of the underlying
porous reservoir bank from a zero-depth reservoir. He presented the assumptions on flow kinematics at the boundaries and the boundary
extent of capillary rise along the vertical and horizontal banks. In conditions. When they are compatible, the transformed boundary
addition, following the methodology used by Swamee et al. (2001) on the hodograph plane appears to be a closed boundary, whose
and Chahar (2007), Choudhary and Chahar (2007, 2014) studied portions may be at infinity (Bear 1972). The Schwarz-Christoffel
the seepage/recharge flux from an array of rectangular and trapezoi- transformation that conformally maps a polygon composed of
dal channels, respectively. A solution was obtained for the drainage straight-line segments onto an auxiliary plane is frequently used
layer at a shallow depth with or without pressure. They also pre- in seepage studies. When the segments of the hodograph plane
sented the relations for the variation of seepage velocity along a are not rectilinear, but they or their extensions intersect at a point,
channel perimeter and seepage line profiles. From a generalized the method of inversion is commonly used in order to obtain the
case Choudhary and Chahar (2014) deduced particular solutions inverse hodograph plane that is composed of rectilinear segments
for an array of triangular and rectangular channels for a drainage (Polubarinova-Kochina 1962). A variety of unconfined flow prob-
layer at a shallow depth with or without pressure and at a great lems can be solved by linking the inversion process to the velocity
depth. Furthermore, Samal and Mishra (2017) analyzed the seepage hodograph. If the complex potential plane, w ¼ ϕ þ iψ, where ϕ
flux from a triangular channel considering capillary action, which and ψ are the velocity potential and stream function, respectively,
was a particular case of the solution given by Vedernikov (1940). consists of rectilinear segments, the solution can be obtained
As the foregoing brief overview reveals, the majority of studies by means of the Schwarz-Christoffel transformation. The solution
did not take into account capillary action to determine seepage flux methodology consists of the mapping of the inverse hodograph
from a channel. Water rises along the channel perimeter owing to plane and the complex potential plane onto an auxiliary plane in
the capillary action that augments the wetted area. Considering the such a way that the images of the corresponding points are mapped
effects of capillary rise, Vedernikov (1940) was the only researcher either from the inverse hodograph plane or from the complex poten-
to provide an analysis only for trapezoidal and triangular channels tial plane to be identical. Combining the mapping function of the
with a side slope of unity and a vertical slit in tabular form. There- inverse hodograph plane and the complex potential plane onto
fore, his results failed to elucidate the dependencies of seepage the auxiliary plane, the mapping function of the actual seepage do-
characteristics on channel geometrical parameters and the boun- main onto the auxiliary plane can be readily determined. Once the
dary conditions of the seepage domain. Moreover, he did not carry maps of the actual seepage domain, inverse hodograph plane and
out an analysis of the seepage line, also called the phreatic line, complex potential plane onto the auxiliary plane are available,
considering capillary action. The impact of capillarity on seepage the characteristics of seepage can be effectively determined.
from a trapezoidal channel in terms of various channel geometric
parameters has thus far received inadequate attention, and this
study sheds light on this issue. Mathematical Formulation
This study therefore revisits the pioneering study of Vedernikov
(1940). Using the method of inversion and the Schwarz-Christoffel The schematic of a trapezoidal channel containing water is shown
transformation, the analysis is carried out on a trapezoidal channel in Fig. 1, where the x- and y-axes represent the horizontal (right-
in a homogeneous, isotropic porous medium of great depth. The ward) and vertical (downward) directions, respectively. The coor-
present endeavor therefore focuses on the dependencies of seepage dinate system that follows the left-hand rule has an origin at the
characteristics on channel geometric parameters and the vertical half-width of the free surface (that is, the half-top width). The chan-
rise of seepage water along channel side slopes owing to capillary nel bottom width, the top width of a free surface, and the water
action. Considering the asymmetrical seepage domain about the depth in the channel are denoted by B, T, and H, respectively.
vertical central plane of the channel, a solution is obtained for the The side slope of the channel that forms an angle απ with the hori-
half-seepage domain. zontal is represented by m (1 vertical: m horizontal), where α is a
factor. The horizontal extent and the vertical rise (from the top of
the free surface) of seepage water along the channel side slope ow-
Theoretical Background ing to capillary action are represented by xs and ys , respectively.
Henceforth, for brevity’s sake, xs and ys are called the dynamic
Conformal mapping has been used extensively for solving steady- capillary shift and rise, respectively. The channel runs through a
state two-dimensional potential flow problems with a seepage line. homogeneous, isotropic, incompressible porous medium of great
Notably, the Laplace equation governs steady-state incompressible depth. The groundwater table is assumed to be at a great depth such
flow through a homogeneous, isotropic, porous medium. The appli- that the channel remains hydraulically unconnected with the aqui-
cability of conformal mapping in two-dimensional flow cases is fer. In addition, no drainage layer in the vicinity of the channel and
rooted in the fact that the solutions for Laplace’s equation remain no sources or sinks in the flow system are considered. Darcy’s law

© ASCE 04020083-2 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


T c
xs e c e
d
f
e ys b
x
H 1 d
m a g
b c f u
(0.5 – ) k
B/2 k (0.5 – )
u
Seepage layer Seepage line a g f

b
d
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

a g
c e e
y c
v v
Fig. 1. Schematic of trapezoidal channel containing water. (a) (b)

Fig. 2. (a) The (u þ iv) plane; and (b) hodograph plane.

is considered to be satisfied within the seepage domain, and the


solution for a steady-state, two-dimensional potential flow is
sought. The seepage domain is composed of geometrically well- Because the segment ef is conceived to be a streamline (Fig. 1),
defined boundaries, as shown in Fig. 1. The essential prerequisite the normal velocity component vanishes along the segment ef,
for the determination of seepage characteristics is the characteriza- while the tangential velocity component prevails. Along the boun-
tion of the seepage domain geometry and the flow condition within dary bcde (Fig. 1), the pressure distribution can be considered to be
it (Kovacs 1981). The seepage domain is asymmetrical about the hydrostatic. Along the wetted perimeter, the pressure at a given
y-axis; hence the solution is obtained for the half-seepage domain depth y is
that includes the points abcdefg. Seepage water rises along
p ¼ γwy ð4Þ
the channel segment ef owing to the capillary action (Fig. 1). The
gauge pressure in the zone to the right of segment ef is negative
Inserting Eq. (4) into Eq. (2), it can be inferred that the velocity
(i.e., less than atmospheric pressure) as a result of the capillary ac-
potential function ϕ remains constant along the boundary bcde.
tion. The segment ef serves as the external boundary of the capil-
Consequently, the tangential velocity component vanishes along
lary fringe, and the pressure along it varies from zero at point e to
the boundary bcde. For the present case, the constant C in
−pc at point f, where −pc is the average capillary negative pres-
Eq. (2) is considered to be zero. The seepage line is defined as the
sure (suction) along the seepage line. The presence of air at atmos-
uppermost streamline within the seepage domain that separates the
pheric pressure is assumed to be on the other side of segment ef.
saturated region from the part of the medium through which no
Because no flow can take place across the capillary exposed boun-
flow occurs. Because it is a streamline, the stream function ψ be-
dary ef, it is deemed to be a streamline (Polubarinova-Kochina
comes a constant along it. Further, the average pressure along the
1962). Therefore, the boundary conditions of an impervious boun-
seepage line remains invariant. Therefore, Eq. (2) produces
dary are applicable along the segment ef. The physical plane, in
Fig. 1, is defined as z ¼ x + i y. By definition, the equations of ϕ − ky ¼ constant ð5Þ
a steady-state two-dimensional potential flow in a porous medium
(i.e., ground) are as follows: Notably, the velocity potential function ϕ along the seepage line
varies linearly with the vertical distance y corresponding to the flow
∂ϕ ∂h ∂ϕ ∂h system. The line of symmetry ba is also assumed to be a streamline.
u¼ ¼ −k ; v¼ ¼ −k ð1Þ
∂x ∂x ∂y ∂y Along the streamlines efg and ba, the stream functions ψ are as-
signed to be ψ ¼ −qs =2 and 0, respectively, where qs is the seep-
where h = hydraulic head; and k = coefficient of hydraulic conduc- age flux per unit width. It is thought that the seepage velocity
tivity. For y being positive downward, one can write h ¼ ðp=γ w − becomes the coefficient of hydraulic conductivity at a great depth,
yÞ, where p is the pressure at a given depth y and γ w is the specific ignoring the effects of infiltration and evaporation. Having identi-
weight of water. Then integrating Eq. (1) yields fied the flow conditions along the channel boundary and seepage
  domain, the next section presents an analytical solution for steady-
p
ϕ ¼ −k −y þC ð2Þ state two-dimensional seepage from a trapezoidal channel.
γw

where C is an integration constant. In terms of stream function ψ for


a steady-state two-dimensional potential flow, one obtains
Analytical Solution

∂ψ ∂ψ The steady-state seepage flux qs per unit length complying with


u¼ ; v¼− ð3Þ Darcy’s law can be expressed as (Swamee et al. 2000; Chahar
∂y ∂x
2007)
By definition, stream function ψ is constant for a given stream- qs ¼ kHFs ð6Þ
line, and the fluid flux between two streamlines equals the differ-
ence in the stream functions of those streamlines. The derivative of where Fs is the seepage function, which is dependent on the
a complex potential (i.e., dw=dz ¼ u − iv) is called the complex channel geometric parameters and the boundary conditions.
velocity. Fig. 2(a) outlines the uv plane of the seepage domain as shown

© ASCE 04020083-3 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


in Fig. 1. The uv plane can be obtained following the standard steps f

(Polubarinova-Kochina 1962; Harr 1962). The origin of the coor-


dinate system lies at point f. The channel bottom bc is mapped on
the uv plane along the line normal to it passing through the origin. e e f
u/(u2 + v2 )
Consequently, it lies along the v axis on the uv plane. The velocity c
(0.5 – ) g
d
at point b is finite, and hence the point is displaced along the v axis. 1/k d –qs /2
b c
Because a theoretically infinite velocity exists at points c and e, the f a
channel boundary ce forms a cut on the uv plane. Along the chan- b
a g
nel boundary ce, the minimum seepage velocity transpires at point
d, which is an intermediate point along the segment ce. The
so-called impervious segment ef gives rise to a straight segment
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

on the uv plane parallel to it passing through the origin. The point


f is a stagnation point and hence falls at the origin on the uv plane. v/(u2 + v2)
The impermeable line of symmetry ba falls along the v axis on the (a) (b)
uv plane. Finally, the mapping of the seepage line fg as a semi-
circle completes the diagram on the uv plane. The velocity hodo- Fig. 3. (a) Inverse hodograph plane; and (b) complex potential plane.
graph of the seepage domain is shown in Fig. 2(b). It is merely a
mirror image of the uv plane about the u axis.
Taking the origin of coordinates of the hodograph plane as the
center of inversion and keeping the radius of inversion equal to
unity, the inverse transformation on the hodograph plane can be
applied. In an inversion, the center of inversion maps onto a point
at infinity, and the point at infinity maps onto the center of inver-
sion. The semicircle passing through the center of inversion on the
hodograph plane transforms into a straight line on the inverse hodo-
graph plane that does not pass through the origin of coordinates of Fig. 4. Lower half-plane (ζ ¼ ξ þ iη).
the inverse hodograph plane. Furthermore, the straight lines or their
continuations passing through the center of inversion on the hodo-
graph plane transforms into straight lines passing through the origin
of coordinates of the inverse hodograph plane. Hence, it is evident the ζ plane relative to the known abscissae b and e. Thus, the region
that all the boundaries of the hodograph plane transform into recti- interior to the polygons, shown in Figs. 3(a and b), are mapped con-
linear segments. Consequently, employing the inverse transforma- formally onto the entire lower half-plane through the Schwarz-
tion, a rectilinear polygon can be obtained. The inverse hodograph Christoffel transformation (Fig. 4).
for the seepage domain in Fig. 1 is shown in Fig. 3(a). Both points c The conformal mapping of the inverse hodograph plane onto the
and e lie at the origin of coordinates of the inverse hodograph plane. lower half-plane is as follows (see Appendix):
The flow pattern on the physical plane, which is dependent on the Z
relative position and elevation of the drainage layer, characterizes the dz i ζ
¼− ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5
layout of the complex potential plane (Polubarinova-Kochina 1962). dw kðJ 1 þ J 2 Þ 0
Because the drainage layer is assumed to be at a great depth, the iJ 1
complex potential plane appears to be a semi-infinite strip. × ðγ − ζÞ−ð1þαÞ dζ þ ð7Þ
Fig. 3(b) shows the complex potential plane for the seepage do- kðJ 1 þ J 2 Þ
main in Fig. 1.
Notably, the inverse hodograph and the complex potential plane where β, λ, and γðγ > λ > βÞ are the points on the real axis of the
are composed of straight line segments. In other words, they are ζ plane corresponding to the vertices c, d, and f of the channel
polygons with a finite number of vertices, one of which is at infin- boundary, respectively; and J 1 and J 2 are the improper integrals
ity. Hence, the conformal mapping of the inverse hodograph and defined in the Appendix. The mapping parameters β, λ, and γ must
the complex potential plane onto the lower half-plane ζð¼ ξ þ iη, be determined. At point e (dz=dw ¼ 0; ζ ¼ 1), Eq. (7) reduces to
where ξ and η are the real and imaginary axes of the lower half- R1 α−1
plane, respectively) is carried out using the Schwarz-Christoffel β ζðζ − βÞ ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ
λ ¼ R1 α−1
ð8Þ
transformation. The lower half-plane is displayed in Fig. 4. The
β ðζ − βÞ ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ
transformation can be conceived as the mapping of polygons,
shown in Figs. 3(a and b), onto the lower half-plane (Fig. 4) in such
a way that the sides of the polygons extend through the real axis of Note that Eq. (8) makes it possible to evaluate the transforma-
the lower half-plane. The mapping is transferred by opening the tion parameter λ when the other transformation parameters β and γ
polygons at ag and extending a to ζ → −∞ and g to ζ → ∞. are known.
The mapping of vertices at infinity is advantageous, because the Employing the Schwarz-Christoffel transformation, the com-
factors corresponding to the vertices at infinity in the ζ plane do plex potential plane can be conformally mapped onto the lower
not emerge in the transformation, and the number of arbitrary val- half-plane as (see Appendix)
ues is reduced by one (Harr 1962). The Schwarz-Christoffel trans- Z ζ
iqs
formation makes it possible to arbitrarily map three points on the w¼− ζ −0.5 ð1 − ζÞ−0.5 dζ ð9Þ
real line of the ζ plane, which corresponds to three vertices of the 2π 0
given polygon (Harr 1962). Consequently, the vertices b and e are
mapped onto the real line of the ζ plane at ζ ¼ 0 and 1, respectively. Along the streamline efg (∞ > ζ ≥ 1), the complex potential
The other vertices of the polygons are mapped onto the real line of takes the form

© ASCE 04020083-4 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


qs pffiffiffi pffiffiffiffiffiffiffiffiffiffiffi iq Relationship between Static and Dynamic Capillary
wðζÞ ¼ lnð ζ þ ζ − 1Þ − s ð10Þ Rise
π 2
From Eq. (10), the velocity potential along the streamline efg is
Using Eqs. (7) and (9), the conformal mapping of the physical given by
plane (z ¼ x þ iy) onto the lower half-plane can be produced as
q pffiffiffi pffiffiffiffiffiffiffiffiffiffiffi
(see Appendix) ϕðζÞ ¼ s lnð ζ þ ζ − 1Þ ð18Þ
Z ζ Z ζ π
H
z¼− ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 Therefore, the velocity potential at point f (ζ ¼ γ) is as follows:
J 3 sinðαπÞ β β
 q pffiffiffi pffiffiffiffiffiffiffiffiffiffiffi
B ϕðγÞ ¼ s lnð γ þ γ − 1Þ ð19Þ
× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ þ þ iH ð11Þ π
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

2
As point f lies on the seepage line fg, the velocity potential can
where J 3 is an improper integral defined in the Appendix. Using also be written
z ¼ F1 ðζÞ from Eq. (11) and w ¼ F2 ðζÞ from Eq. (9), the charac-
 
teristics of seepage can be evaluated (Polubarinova-Kochina 1962). pa − pc
At point bðz ¼ iH; ζ ¼ 0Þ, the mapping of the physical plane onto ϕðγÞ ¼ −k − ys ð20Þ
γw
the lower half-plane yields
where pa is the atmospheric pressure. Taking atmospheric pressure
B 2J 4 to be equal to zero (gauge pressure) and using pc =γ w ¼ hc (where
¼ ð12Þ
H J 3 sinðαπÞ hc is the static capillary rise), Eq. (20) reduces to
where J 4 is an improper integral defined in the Appendix. ϕðγÞ ¼ kðhc þ ys Þ ð21Þ

Using Eqs. (19) and (21), a relation between the static capillary
Computation of Dynamic Capillary Shift and Rise rise and the dynamic capillary rise can be obtained as
The mapping of the impervious segment ef onto the lower half- hc q pffiffiffi pffiffiffiffiffiffiffiffiffiffiffi y
plane makes it possible to obtain the dynamic capillary shift and ¼ s lnð γ þ γ − 1Þ − s ð22Þ
rise, xs and ys , as follows (see Appendix): H πkH H
Eqs. (12) and (22) are obtained with a view to determining the
xs T J cotðαπÞ
¼ þ 5 ð13Þ mapping parameters.
H 2H J3

ys J Variation of Seepage Velocity


¼− 5 ð14Þ
H J3 The conformal mapping of the inverse hodograph plane (dz=dw)
onto the lower half-plane can be used to determine the seepage
where J 5 is an improper integral defined in the Appendix. velocity distribution along the channel perimeter. From Eq. (7)
one obtains
Determination of Seepage Line Profiles dz 1 u þ iv
¼ ¼
Using the mapping of seepage line fg onto the lower half-plane, the dw u − iv u2 þ v2
Z ζ
parametric equations of the seepage line fg can be obtained as i
¼− ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ
follows (see Appendix): kðJ 1 þ J 2 Þ 0
iJ 1
xðζÞ
¼−
qs J 6 x
þ s ð15Þ þ ð23Þ
H 2πkHðJ 1 þ J 2 Þ H kðJ 1 þ J 2 Þ

Z ζ Separating the real and imaginary parts of Eq. (23) and then
yðζÞ
¼
qs J 1
ζ −0.5 ðζ − 1Þ−0.5 dζ squaring and adding them along the channel bottom bc (β ≥ ζ ≥ 0)
H 2πkHðJ 1 þ J 2 Þ γ yields
Z ζ
q y J1 þ J2
þ s ζ −0.5 ðζ − 1Þ−0.5 dζ þ s ð16Þ V
¼ R ð24Þ
2πkH γ H k J 1 − 0ζ ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ
where J 6 is an improper integral defined in the Appendix. where V is the resultant velocity of seepage flow along the channel
perimeter. At ζ ¼ β, the denominator of Eq. (24) becomes zero.
Computation of Seepage Flux Hence, at vertex c of the channel, theoretically infinite velocity pre-
vails. It can be inferred from Eq. (24) that the minimum velocity of
The seepage function Fs provides the dimensionless seepage flux seepage flow prevails at the midpoint of the channel bottom.
per unit length of the channel. The seepage function Fs is obtained Similarly, the resultant velocity of seepage flow along the side
as (see Appendix) slope ce (1 ≥ ζ ≥ β) of the channel is given by

2πðJ 1 þ J 2 Þ V J1 þ J2
Fs ¼ ð17Þ ¼ Rζ ð25Þ
α−1
J 3 sinðαπÞ k β ðλ − ζÞðζ − βÞ ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ

© ASCE 04020083-5 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


The denominator of the preceding equation becomes zero at 1. For a given channel configuration and static capillary rise, the
ζ ¼ β and ζ ¼ 1. Hence, theoretically infinite velocity persists at mapping parameters β and γ were assumed.
points c and e on the channel perimeter. 2. With these assumed values, the mapping parameter λ, the dy-
Likewise, the relationship for the resultant velocity of seepage namic capillary rise ys , and the seepage function Fs were com-
flow along the segment ef (γ ≥ ζ ≥ 1) of the channel is given by puted using Eqs. (8), (14), and (17) for determining g1 and g2 .
3. Once g1 and g2 were obtained, the objective function f1 was
V J1 þ J2
¼ Rζ ð26Þ determined from Eq. (27).
k ðζ − λÞðζ − βÞα−1 ðζ − 1Þ−0.5 ðγ − ζÞ−ð1þαÞ dζ
1 4. The mapping parameters β and γ for the minimum value of the
objective function f 1 were evaluated aided by an iterative
The denominator of Eq. (26) becomes zero and infinity at ζ ¼ 1 method using the MATLAB version 8.5 routine fminsearch.
and ζ ¼ γ, respectively. Consequently, the resultant velocity of seep- 5. Once the parameters β and γ were obtained using Eq. (8), the
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

age flow becomes infinity and zero at points e and f, respectively. parameter λ was determined.
Note that the computed seepage velocities at points c, e, and f 6. Then the seepage function Fs was determined from Eq. (17).
along the channel perimeter from Eqs. (24)–(26) are in agreement 7. The dynamic capillary shift and rise, xs and ys , were calculated
with the seepage velocities considered at these points in the velocity using Eqs. (13) and (14), respectively.
hodograph. 8. Finally, the coordinates of the seepage line were obtained from
Eqs. (15) and (16).
The improper integrals involved in the derived relations were
Particular Cases computed numerically using Gaussian quadrature after removing
the singularities by changing the variables (Press et al. 1992). The
Rectangular Channel computations were performed for various channel geometric
parameters and static capillary rises in order to determine the
A rectangular channel is a particular case of a trapezoidal channel dependencies of seepage characteristics on them.
with vertical sides (m ¼ 0). The relations for a rectangular channel
can be obtained by using m ¼ 0 and α ¼ 0.5 in the relations for a
trapezoidal channel derived in the preceding section. Results and Discussion
Vedernikov (1940) presented the numerical results for trapezoidal
Triangular Channel channels with a side slope of unity (m ¼ 1) and variable bottom
A triangular channel is also a particular case of a trapezoidal chan- widths. Table 1 furnishes a comparison of the results obtained from
nel with zero bottom width. For a triangular channel, point c in the present study with those obtained by Vedernikov (1940). It is
different mapping planes coincides with point b, and the transfor- evident that the results obtained from the two studies correspond
mation parameter β becomes zero. However, the other transforma- satisfactorily. Note that the integrals involved in the present analy-
tion parameters λ and γ must be determined. Using β ¼ 0, one can sis and those in Vedernikov’s analysis appear in different forms, as
derive the relations for a triangular channel from the derived rela- only the half-seepage domain was analyzed in the present study.
tions for a trapezoidal channel. However, the departure of the present results from Vedernikov’s
results, although minimal, may be ascribed to differences in the
forms of the integrals and their numerical evaluations.
Vertical Slit The variations in the seepage function Fs with aspect ratio
A deep, narrow channel having vertical sides (m ¼ 0) and a neg- B=H for different relative static capillary rises hc =H are shown
ligible top width (T=H → 0) is regarded as a vertical slit. A vertical in Figs. 5(a and b) for channel side slopes m ¼ 1.5 and 0, respec-
slit is therefore a particular case of a trapezoidal channel. Thus, its tively. Importantly, the values of Fs ðB=H ¼ 0; m ¼ 1.5) and
relations can also be deduced from the derived relations for a trap- Fs ðB=H ¼ 0; m ¼ 0) (representing points on the ordinate scale)
ezoidal channel using m ¼ 0 and T=H → 0.

Table 1. Comparison of results obtained from present study with those


Computational Procedure obtained by Vedernikov (1940) for trapezoidal channels with side slope
of unity (m ¼ 1)
The equations developed are used to determine the seepage flux, qs =kH qs =kH ys =hc ys =hc
dynamic capillary rise, and seepage line profiles for various chan- T=H hc =H (Vedernikov) (present study) (Vedernikov) (present study)
nel configurations and static capillary rises. To begin with, for a 2 7.3 18 19.2 0.246 0.226
given channel geometry and static capillary rise, the parameters in- 4.1 12.6 31.2 31.9 0.24 0.235
volved in the mapping, β, λ, and γ, must be evaluated. Therefore, to 8.3 22 55.1 55.5 0.233 0.231
calculate the mapping parameters, an objective function f 1 can be 22.8 51.2 129.6 130.3 0.225 0.225
defined as 2 3 10.8 11.7 0.285 0.26
 2  2 3.8 4.8 17.7 18.1 0.273 0.268
B h 7.2 8.1 30.1 30 0.254 0.262
f1 ðβ; λ; γÞ ¼ − g1 ðβ; λ; γÞ þ c − g2 ðβ; λ; γÞ ð27Þ
H H 19.7 18.6 67.9 69.8 0.248 0.25
2 0.97 6.7 7 0.32 0.306
where g1 and g2 are the right-hand sides of Eqs. (12) and (22), 3.4 1.46 10 10.3 0.315 0.31
respectively. Because the objective function f 1 is the sum of 5.8 2.2 15.3 15.5 0.306 0.303
two quadratic parts, the minimum value of f 1 must be zero, which 13.2 4.1 29.7 29.9 0.287 0.29
can be achieved only when both the parts are zero, and hence, Note: The absolute values of ys =hc for this study are presented in Table 1,
Eqs. (12) and (22) are satisfied. The steps involved in the compu- because they are originally negative because the y-axis is positive in the
tation are as follows: downward direction (Fig. 1).

© ASCE 04020083-6 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Variations of seepage function Fs with aspect ratio B=H for different relative static capillary rises hc =H: (a) m ¼ 1.5; and (b) m ¼ 0. Broken
lines represent the variations of Fs with B=H without capillary rise, computed from the simplified algebraic equation of Swamee et al. (2000).

50 50

40 40

=6
B/H 4 =6
30 30 B/H 4
2
2
Fs

Fs

0
20 20 0

10 10

0 0
0 2 4 6 8 10 0 2 4 6 8 10
(a) hc /H (b) hc /H

Fig. 6. Variations of seepage function Fs with relative static capillary rise hc =H for different aspect ratios B=H: (a) m ¼ 1.5; and (b) m ¼ 0.

for different hc =H in Figs. 5(a and b) correspond to the triangular


channels and a vertical slit, respectively, while the Fs ðB=H ≠ 0;
m ¼ 0) curves in Fig. 5(b) signify rectangular channels. It is evi-
dent that for a given hc =H, Fs increases monotonically as B=H
increases. Further, for a given B=H, Fs increases with an increase
in hc =H. In addition, a comparison of the sets of curves for m ¼ 1.5
and 0 in Figs. 5(a and b), respectively, suggests that Fs ðm ¼ 1.5Þ is
greater than Fs ðm ¼ 0Þ for the given values of B=H and hc =H. The
broken lines represent the Fs ðB=HÞ curves for hc =H ¼ 0 (zero
capillary rise) in Figs. 5(a and b) for m ¼ 1.5 and 0, respectively,
computed from the simplified algebraic equation of Swamee et al.
(2000). It is apparent that the results of Swamee et al. (2000) match
perfectly with the present results.
It is indeed interesting to explicitly observe the dependencies
of seepage function Fs (i.e., trend of Fs curves) on relative static
capillary rise hc =H and channel side slope m for different
aspect ratios B=H, because Fig. 5 clearly illustrates these findings.
From this perspective, Figs. 6 and 7, which supplement the
conclusions drawn from Fig. 5 in terms of the explicit variations
Fig. 7. Variations of seepage function Fs with channel side slope m for
of Fs with hc =H and m, are presented. In essence, for a given
different aspect ratios B=H and for a given relative static capillary rise
B=H, Fs increases monotonically as hc =H and m increase
hc =H ¼ 1.
(Figs. 6 and 7).

© ASCE 04020083-7 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Variations of relative dynamic capillary rise ys =H with relative static capillary rise hc =H for different channel side slopes m: (a) B=H ¼ 0; and
(b) B=H ¼ 2.

-1 -0.6

-0.5
-0.8

-0.4
m=0
-0.6

ys /H
ys /H

-0.3
0.5
-0.4
1 2
-0.2
1.5 B/H = 0
-0.2
-0.1

0 0
0 4 8 12 16 20 0 2 4 6 8 10
B/H m

Fig. 9. Variations of relative dynamic capillary rise ys =H with aspect Fig. 10. Variations of relative dynamic capillary rise ys =H with chan-
ratio B=H for different channel side slopes m and for a given relative nel side slope m for aspect ratios B=H ¼ 2 and 0 and for a given re-
static capillary rise hc =H ¼ 1. lative static capillary rise hc =H ¼ 1.

The variations in relative dynamic capillary rise ys =H with rel- owing to the small values of bottom width B. On the other hand,
ative static capillary rise hc =H for different channel side slopes m ys =H diminishes sharply with an increase in m, tending to be even-
are depicted in Figs. 8(a and b) for aspect ratios B=H ¼ 0 and 2, tually independent of m for m > 10, which is not shown in Fig. 10,
respectively. Notably, the ys =HðB=H ¼ 0; m ≠ 0Þ curves and because the channel tends to have a flat base at large values of m.
ys =HðB=H ¼ 0; m ¼ 0) curve in Fig. 8(a) refer to the triangular Importantly, ys is at its maximum for channels with vertical sides
channels and a vertical slit, respectively, while the ys =HðB=H ¼ (i.e., m ¼ 0).
2; m ¼ 0) curve in Fig. 8(b) indicates a rectangular channel. It The seepage lines (x=H; y=H) relative to the ground level
is found that for a given m, ys =H increases continuously with for different relative static capillary rises hc =H for a trapezoidal
an increase in hc =H. Further, for a given hc =H, ys =H increases channel (B=H ¼ 2; m ¼ 1.5), a rectangular channel (B=H ¼ 2;
as m decreases. In addition, the curves suggest that the dynamic m ¼ 0), a triangular channel (B=H ¼ 0; m ¼ 1.5), and a vertical
capillary rise ys is always less than the static capillary rise hc , which slit (B=H → 0; m ¼ 0) are plotted in Figs. 11(a–d), respectively.
is in agreement with the findings of Vedernikov (1940). It is obvious that the seepage domain significantly increases with
To supplement the results obtained from Fig. 8 in terms of the an increase in hc =H. The seepage lines that emerge from the chan-
explicit variations in relative dynamic capillary rise ys =H with as- nel boundary initially curve downward with a significant lateral
pect ratio B=H and channel side slope m, ys =H as a function of shift, eventually becoming vertical at a great depth. Examination
B=H and m are presented in Figs. 9 and 10, respectively. It is evi- of seepage lines for different channels reveals that the width of
dent that ys =H remains invariant with respect to B=H for B=H > 3 the seepage domain at great depth nearly equals the seepage func-
(Fig. 9). Further, a close examination of Fig. 9 reveals that the ys =H tion Fs . This observation is in agreement with the assumption that
curves increase with B=H for 3 ≥ B=H > 0, indicating that the dy- the seepage velocity equals the coefficient of hydraulic conduc-
namic capillary rise ys briefly overcomes the channel constriction tivity (V ≈ k) at great depth.

© ASCE 04020083-8 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Seepage lines relative to ground level for different relative static capillary rises hc =H: (a) trapezoidal channel (B=H ¼ 2 and m ¼ 1.5);
(b) rectangular channel (B=H ¼ 2 and m ¼ 0); (c) triangular channel (B=H ¼ 0 and m ¼ 1.5); and (d) vertical slit (T=H → 0 and m ¼ 0).

Conclusions recognized that the seepage lines for different static capillary rises
become parallel to each other at great depth, and the width of the
Applying the method of inversion and the Schwarz-Christoffel seepage zone at that depth nearly equals the seepage function.
transformation, the analytical solution for seepage from a trapezoi- In essence, this study provides insight into the characteristics of
dal channel in a homogeneous, isotropic, porous medium at great seepage from trapezoidal channels and its particular cases, high-
depth is obtained. The key conclusions of the study are as follows. lighting the effects of channel geometric parameters and static
The results reveal that the seepage function depends on the capillary rise on seepage characteristics. For instance, the quanti-
channel configuration and the static capillary rise. It increases fication of seepage from a channel provides an understanding of the
monotonically with an increase in channel bottom width and static interaction between surface and subsurface water. In addition, the
capillary rise. Moreover, a close look at the results shows that the outcomes concerning the dynamic capillary rise make it possible to
seepage function attains its minimum for channels with vertical infer the minimum freeboard required for a given channel. To be
sides and increases considerably with an increase in channel side specific, if the existing freeboard is less than the dynamic capillary
slope. This observation can be attributed to the fact that an increase rise, then the adjacent ground becomes saturated, and the seepage
in channel bottom width, the static capillary rise, and the channel line emerges from the ground level instead of the channel
side slope expand the wetted zone from where the seepage emerges. side slope.
Regarding the dynamic capillary rise, it was observed to
continuously increases with increases in static capillary rise for a
given channel side slope. Analysis of the dependency of channel Appendix. Mapping Details
side slope on dynamic capillary rise reveals that it is at its maximum
for channels with vertical sides and decreases sharply with an in-
crease in channel side slope. Further, the dynamic capillary rise
Mapping of Inverse Hodograph Plane onto Lower
does not vary much with the channel bottom width. Importantly,
Half-Plane
the dynamic capillary rise is always smaller than the static capil-
lary rise. The inverse hodograph plane, shown in Fig. 3(a), is a rectilinear
The seepage lines demonstrate that the width of the seepage polygon with a vertex at infinity. The vertices of the polygon a,
zone widens with an increase in the static capillary rise. It is b, c, d, e, f, and g are mapped onto the real line of the ζ plane

© ASCE 04020083-9 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


at points −∞, 0, β, λ, 1, γ, and ∞, respectively. Hence, the interior lower half-plane. With interior angles equaling π, the factors corre-
region of the inverse hodograph is mapped onto the entire ζ plane. sponding to vertices c, d, and f of the semi-infinite strip becoming
When the vertices are mapped at −∞ and ∞, the factors corre- unity do not appear in the mapping. When the vertices are mapped at
sponding to vertices a and g becoming unity do not take part in −∞ and ∞, the factors corresponding to vertices a and g becoming
the mapping. Because the interior angle at vertex b equals π, its unity also do not appear in the mapping. The mapping of the com-
corresponding factor, which becomes unity, also does not take part plex potential plane onto the lower half-plane is
in the mapping. The mapping of the inverse hodograph plane Z ζ
(dz=dw) onto the lower half-plane is given by
w ¼ M2 ζ −0.5 ð1 − ζÞ−0.5 dζ þ N 2 ð36Þ
Z ζ 0
dz
¼ M1 ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ þ N 1
dw 0 where M 2 and N 2 are complex constants. Making use of the known
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

ð28Þ points in both planes, the complex constants are evaluated. For point
b (w ¼ 0 and ζ ¼ 0), from Eq. (36), the mapping yields
where M 1 and N 1 are complex constants. Using the corresponding
points between the inverse hodograph [Fig. 3(a)] and the ζ plane N2 ¼ 0 ð37Þ
(Fig. 4), the complex constants M 1 and N 1 are determined. At ver-
tex c (dz=dw ¼ 0; ζ ¼ β), from Eq. (28), the mapping produces At point e (w ¼ −iqs =2; ζ ¼ 1), from Eq. (36), the mapping
N 1 ¼ −M 1 J 1 ð29Þ yields

where iqs
M2 ¼ − ð38Þ
Z 2π
β
J1 ¼ ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ ð30Þ
0 Eq. (9) can be obtained substituting M 2 and N 2 into Eq. (36).
Similarly, for vertex a (dz=dw ¼ i=k; ζ → −∞), from Eq. (28),
the mapping yields Mapping of Physical Plane onto Lower Half-Plane
Z∞
− Using the mapping of the inverse hodograph plane and the complex
i α−1 −0.5 −ð1þαÞ
¼ M1 ðλ − ζÞðβ − ζÞ ð1 − ζÞ ðγ − ζÞ dζ þ N 1 potential plane onto the lower half-plane, the physical plane is
k mapped onto the lower half-plane. The mapping of the inverse
0
hodograph and the complex potential plane onto the lower half-
ð31Þ plane can be written as follows:
Setting −ζ ¼ τ , where τ is a variable, and using Eq. (29),

Eq. (31) becomes dz
Z ∞ ¼ M1 ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 ðγ − ζÞ−ð1þαÞ dζ ð39Þ
i dw
¼ −M 1 ðλ þ τ Þðβ þ τ Þα−1 ð1 þ τ Þ−0.5 β
k 0
× ðγ þ τ Þ−ð1þαÞ dτ − M 1 J 1 ð32Þ Z ζ
w ¼ M2 ζ −0.5 ð1 − ζÞ−0.5 dζ ð40Þ
The preceding equation can be rearranged as 0

i It is known that
M1 ¼ − ð33Þ
kðJ 1 þ J 2 Þ
dz dz dw
where ¼ ð41Þ
dζ dw dζ
Z ∞
J2 ¼ ðλ þ τ Þðβ þ τ Þα−1 ð1 þ τ Þ−0.5 ðγ þ τ Þ−ð1þαÞ dτ ð34Þ Inserting Eqs. (39) and (40) into Eq. (41) and then integrating
0
with respect to ζ, the physical plane onto the lower half-plane is
Substituting Eq. (29) into Eq. (33), one obtains mapped as
Z ζ Z ζ
iJ 1
N1 ¼ ð35Þ z ¼ M3 ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5
kðJ 1 þ J 2 Þ β β
ζ −0.5
Substituting M 1 and N 1 into Eq. (28), one obtains Eq. (7). × ðγ − ζÞ−ð1þαÞ dζ ð1 − ζÞ−0.5 dζ þ N 3 ð42Þ

Mapping of Complex Potential Plane onto Lower


where M 3 ð¼ M 1 M2 Þ and N 3 are complex constants that can be
Half-Plane
evaluated using the corresponding known points on the physical
The complex potential plane is sketched in Fig. 3(b). Because the plane and the lower half-plane.
seepage domain is assumed to be of significant depth, the complex For point c (z ¼ B=2 þ iH, ζ ¼ β), Eq. (42) yields
potential plane emerges in the form of a semi-infinite strip. Like the
inverse hodograph plane, the vertices of the semi-infinite strip a, b, B
N3 ¼ þ iH ð43Þ
c, d, e, f, and g are mapped onto the real line of the ζ plane at 2
points −∞, 0, β, λ, 1, γ, and ∞, respectively. Consequently,
the interior region of the semi-infinite strip maps onto the entire Then, for point e½z ¼ B=2 þ H cotðαπÞ, ζ ¼ 1], Eq. (42) yields

© ASCE 04020083-10 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


Z 1 Z
B ζ H½cosðαπÞ − i sinðαπÞJ 5 T
þ HcotðαπÞ ¼ M 3 ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 xs þ iys ¼ þ ð52Þ
2 β β J 3 sinðαπÞ 2

× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ where
Z γ Z ζ
B J5 ¼ ðλ − ζÞðζ − βÞα−1 ðζ − 1Þ−0.5
þ þ iH ð44Þ
2 1 1

Separating the real and imaginary parts of Eq. (44), the real part × ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ðζ − 1Þ−0.5 dζ ð53Þ
is given by
Z 1 Z ζ
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

H Separating the real and imaginary parts of Eq. (52), Eqs. (13)
− ¼ M3 ðλ − ζÞðζ − βÞα−1 ð1 − ζÞ−0.5 and (14) are obtained. They represent the dynamic capillary shift
sinðαπÞ β β
 and rise, xs and ys , respectively.
× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ ð45Þ Mapping of the streamline fg onto the lower half-plane makes it
possible to obtain parametric equations for the seepage lines. The
mapping of the inverse hodograph plane and the complex potential
Rearranging and using notation for the integral, the preceding plane can be written
equation reduces to Z ζ
dz 1
H ¼− ðζ − λÞðζ − βÞα−1 ðζ − 1Þ−0.5
M3 ¼ − ð46Þ dw kðJ 1 þ J 2 Þ ∞
J 3 sinðαπÞ
iJ 1 i
× ðζ − γÞ−ð1þαÞ dζ þ þ ð54Þ
where kðJ 1 þ J 2 Þ k
Z 1 Z ζ Z ζ
J3 ¼ ðλ − ζÞðζ − βÞα−1 ð1 − ζÞ−0.5 w¼
qs
ζ −0.5 ðζ − 1Þ−0.5 dζ ð55Þ
β β
 2π 0

× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ ð47Þ Inserting Eqs. (54) and (55) into Eq. (41) and then integrating
with respect to ζ, the seepage line fg is mapped onto the lower
Substitutions of N 3 from Eq. (43) and M 3 from Eq. (46) into half-plane as
Eq. (42) yield Eq. (11). Now, for each segment of the seepage do-
Z ζ Zζ
main the physical plane is mapped onto the lower half-plane. Along qs
the channel bottom bc (β ≥ ζ ≥ 0), from Eq. (42) one obtains z¼− ðζ − λÞðζ − βÞα−1 ðζ − 1Þ−0.5
2πkðJ 1 þ J 2 Þ γ
Z ζ Z ζ 

H
z¼− ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5
J 3 sinðαπÞ β β × ðζ − γÞ−ð1þαÞ dζ ζ −0.5 ðζ − 1Þ−0.5 dζ
 Z ζ
B
× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ þ þ iH ð48Þ þ
iqs J 1
ζ −0.5 ðζ − 1Þ−0.5 dζ
2 2πkðJ 1 þ J 2 Þ γ
Z
For vertex b (z ¼ iH; ζ ¼ 0), Eq. (48) yields iqs ζ −0.5
þ ζ ðζ − 1Þ−0.5 dζ þ zs ð56Þ
Z 0 Z ζ 2πk γ
B H
¼ ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5
2 J 3 sinðαπÞ β β Inserting z ¼ xðζÞ þ iyðζÞ and zs ¼ xs þ iys into Eq. (56)
 yields
× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ ð49Þ
qs J 6
xðζÞ þ iyðζÞ ¼ −
2πkðJ 1 þ J 2 Þ
Then Eq. (12) is obtained introducing the notation for the Z ζ
iqs J 1
integral as þ ζ −0.5 ðζ − 1Þ−0.5 dζ
Z 0 Z ζ 2πkðJ 1 þ J 2 Þ γ
Z ζ
J4 ¼ ðλ − ζÞðβ − ζÞα−1 ð1 − ζÞ−0.5 iq
β β þ s ζ −0.5 ðζ − 1Þ−0.5 dζ þ xs þ iys ð57Þ
 2πk γ
× ðγ − ζÞ−ð1þαÞ dζ ζ −0.5 ð1 − ζÞ−0.5 dζ ð50Þ where
Z ζ Z ζ
Along the segment ef, the mapping of the physical plane has the J6 ¼ ðζ − λÞðζ − βÞα−1 ðζ − 1Þ−0.5
form γ ∞

Z ζ Z ζ
He−iπα × ðζ − γÞ−ð1þαÞ dζ ζ −0.5 ðζ − 1Þ−0.5 dζ ð58Þ
z¼ ðλ − ζÞðζ − βÞα−1 ðζ − 1Þ−0.5
J 3 sinðαπÞ 1 1

−ð1þαÞ T Separating the real and imaginary parts of Eq. (57), Eqs. (15)
× ðγ − ζÞ dζ ζ −0.5 ðζ − 1Þ−0.5 dζ þ ð51Þ
2 and (16), which provide the coordinates of the seepage line, are
obtained. Finally, the equality M 3 ¼ M 1 M 2 is used to determine
For point f (z ¼ zs ¼ xs þ iys ; ζ ¼ γ), Eq. (51) produces a relation for the seepage function Fs . Substituting M 1 , M 2 , and

© ASCE 04020083-11 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083


M 3 into the equality yields Eq. (17), which shows a relation for the ζ = lower half-plane;
seepage function Fs . η = imaginary axis of lower half-plane;
ξ = real axis of lower half-plane;
τ = a variable;
Data Availability Statement ϕ = velocity potential (L2 T−1 ); and
Some or all data, models, or code generated or used during the ψ = stream function (L2 T−1 ).
study are available from the corresponding author by request.
• They include the tabular data in an Excel file corresponding to
the data presented in Figs. 5–11. References
Downloaded from ascelibrary.org by Indian Institute of Technology Kharagpur on 10/09/20. Copyright ASCE. For personal use only; all rights reserved.

Bear, J. 1972. Dynamics of fluids in porous media. New York: Dover.


Acknowledgments Bouwer, H. 1969. “Theory of seepage from open channel.” In Vol. 5 of
Advances in hydroscience, edited by V. T. Chow, 121–172. New York:
The authors acknowledge the JC Bose Fellowship (JBD) in Academic Press.
pursuing this work. Chahar, B. R. 2007. “Analysis of seepage from polygon channels.” J. Hy-
draul. Eng. 133 (4): 451–460. https://doi.org/10.1061/(ASCE)0733
-9429(2007)133:4(451).
Notation Choudhary, M., and B. R. Chahar. 2007. “Recharge/seepage from an array
of rectangular channels.” J. Hydrol. 343 (1–2): 71–79. https://doi.org/10
The following symbols are used in this paper: .1016/j.jhydrol.2007.06.009.
Choudhary, M., and B. R. Chahar. 2014. “Recharge from an array of
a, b, c, : : : g = points in seepage domain;
polygonal channels.” J. Hydrol. Eng. 19 (5): 918–930. https://doi.org
B = channel bottom width (L); /10.1061/(ASCE)HE.1943-5584.0000872.
C = an integration constant; Hálek, V., and J. Švec. 1979. Groundwater hydraulics. New York: Elsevier.
Fs = seepage function; Harr, M. E. 1962. Groundwater and seepage. New York: Dover.
F1 = right-hand side of Eq. (11); Kacimov, A. R. 2004. “Capillary fringe and unsaturated flow in a porous
reservoir bank.” J. Irrig. Drain. Eng. 130 (5): 403–409. https://doi.org
F2 = right-hand side of Eq. (9);
/10.1061/(ASCE)0733-9437(2004)130:5(403).
f 1 = objective function defined by Eq. (27); Kovacs, G. 1981. Seepage hydraulics. New York: Elsevier.
g1 = right-hand side of Eq. (12); Morel-Seytoux, H. J. 1961. Effect of boundary shape on channel seepage.
g2 = right-hand side of Eq. (22); Technical Rep. No. 7. Stanford, CA: Dept. of Civil Engineering,
H = water depth in channel (L); Stanford Univ.
h = hydraulic head (L); Muskat, M., and R. D. Wyckoff. 1946. The flow of homogeneous fluid
through porous media. Ann Arbour, MI: Edwards Brothers.
hc = static capillary rise (L); Polubarinova-Kochina, P. Y. 1962. Theory of ground water movement.
i = imaginary number; Princeton, NJ: Princeton University Press.
J 1 ; J 2 ; : : : J 6 = improper integrals; Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. 1992.
k = coefficient of hydraulic conductivity (LT−1 ); Numerical recipes in C. Cambridge, UK: Cambridge University Press.
M 1 , M 2 , M 3 = complex constants; Risenkampf, B. K. 1940. “Hydraulics of groundwater.” Proc. State Univ.
Saratovsky 15 (25).
m = channel side slope; Samal, K. P., and G. C. Mishra. 2017. “Analysis of seepage from a triangu-
N 1 , N 2 , N 3 = complex constants; lar furrow considering soil capillarity using inverse hodograph and con-
p = pressure at a given depth (ML−1 T−2 ); formal mapping technique.” ISH J. Hydraul. Eng. 23 (1): 1–12. https://
pa = atmosphere pressure (ML−1 T−2 ); doi.org/10.1080/09715010.2016.1213144.
pc = average capillary pressure (ML−1 T−2 ); Swamee, P. K., and B. R. Chahar. 2015. Design of canals. New Delhi,
India: Springer.
qs = seepage flux per unit length (L2 T−1 );
Swamee, P. K., G. C. Mishra, and B. R. Chahar. 2000. “Design of minimum
T = top width of free surface in channel (L); seepage loss canal sections.” J. Irrig. Drain. Eng. 126 (1): 28–32.
u = seepage velocity in horizontal direction (LT−1 ); https://doi.org/10.1061/(ASCE)0733-9437(2000)126:1(28).
V = resultant seepage velocity (LT−1 ); Swamee, P. K., G. C. Mishra, and B. R. Chahar. 2001. “Design of minimum
v = seepage velocity in vertical direction (LT−1 ); seepage loss canal sections with drainage layer at shallow depth.” J.
Irrig. Drain. Eng. 127 (5): 287–294. https://doi.org/10.1061/(ASCE)
w = complex potential (L2 T−1 );
0733-9437(2001)127:5(287).
x = horizontal axis (L); Vedernikov, V. V. 1939. Seepage theory and its applications in the field of
xs = dynamic capillary shift (L); irrigation and drainage. [In Russian.] Moscow: Gosstroiizdat.
y = vertical axis (L); Vedernikov, V. V. 1940. “Account of soil capillarity on seepage from a
ys = dynamic capillary rise (L); canal.” [In Russian.] Dokl. Akad. Nauk SSSR 28 (5).
z = physical plane (x þ iy) (L); Wachyan, E., and K. R. Rushton. 1987. “Water losses from irrigation ca-
nals.” J. Hydrol. 92 (3–4): 275–288. https://doi.org/10.1016/0022-1694
α = a factor; (87)90018-7.
β, γ, λ = mapping parameters; Worstell, R. V. 1976. “Estimating seepage losses from canal systems.”
γ w = specific weight of water (ML−2 T−2 ); J. Irrig. Drain. Div. 102 (1): 137–147.

© ASCE 04020083-12 J. Hydraul. Eng.

J. Hydraul. Eng., 2020, 146(12): 04020083

You might also like