Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Environ Fluid Mech (2011) 11:427–435

DOI 10.1007/s10652-010-9196-6

ORIGINAL ARTICLE

A new model for soil–water drainage problems

Oscar Castro-Orgaz

Received: 17 June 2010 / Accepted: 23 September 2010 / Published online: 6 October 2010
© Springer Science+Business Media B.V. 2010

Abstract Seepage flow is an agent related to the transport and dispersion of contam-
ination in groundwater. Steady two-dimensional seepage flow is governed by Laplace’s
equation, for which several solution techniques are available. Because computations are
complex from a practical point of view, simplified models encompass the Dupuit-Forchhei-
mer approach assuming a horizontal flow. However this approach is inaccurate in seepage
problems involving steep drawdowns. In this research, a new theoretical model for 2D seep-
age flow is proposed based on Fawer’s theory for curved flows Castro-Orgaz (Environ Fluid
Mech 10(3):2971–2310, 2010), from which a second-order equation results describing the
seepage surface. From this development, a numerical solution for the rectangular dam prob-
lem based on the second-order model is presented, whereas a simple first-order equation is
found to describe flow to drains under a uniform rainfall. The results of this new model are
compared with the full 2D solution of Laplace’s equation for typical test cases, resulting in
an excellent agreement.

Keywords Dam · Drains · Flow net · Seepage · Two-dimensional flow

List of symbols
h Seepage height (m)
K Curvature distribution parameter (–)
K H Hydraulic conductivity (m/s)
n Curvilinear coordinate along equipotential (m)
N Length of equipotential (m)

O. Castro-Orgaz (B)
Instituto de Agricultura Sostenible, Consejo Superior de Investigaciones Cientificas,
Finca Alameda del Obispo, 14080 Cordoba, Spain
e-mail: oscarcastro@ias.csic.es

123
428 Environ Fluid Mech (2011) 11:427–435

 
q Unit discharge m2 /s
L Length of rectangular dam; also half of drain separation (m)
R Radius of streamline curvature (m)
r Rainfall (m/s)
s Curvilinear coordinate along seepage surface (m)
t Travel time (s)
V Local velocity (m/s)
x Horizontal distance (m)
y Vertical distance (–)
ν Dimensionless curvilinear coordinate (–)
θ Angle of seepage surface with horizontal (rad)
φ Potential function (m/s)

Subscripts
1 Relative to inflow section
2 Relative to outflow section
s Relative to free surface streamline
b Relative to bottom streamline

1 Introduction

Seepage groundwater flow is a transport agent for the dispersion of pollutants in aquifers,
thereby indicating the importance of seepage modelling in environmental processes of sub-
surface water. Steady two-dimensional (2D) flow in porous media is described by Laplace’s
equation in the Cartesian x- and y-directions as [1–3]

∂ 2φ ∂ 2φ
+ =0 (1)
∂x2 ∂ y2

where φ = potential function. Several methods for solving Eq. 1 are available, including
analytical methods based on the complex potential and numerical solutions based on finite
difference or finite element methods [4–6]. 2D flow computations may be avoided if paral-
lel streamlines are assumed, for which the Dupuit-Forchheimer approach becomes simple.
However, this approach is inaccurate for 2D groundwater flow problems involving a curved
and sloped seepage surface [3]. An intermediate degree of refinement in the seepage flow
problem relates to a quasi-2D approach, where the kinematic variables are approached to
reduce Eq. 1 from the x- and y-directions to a model in the x-direction only. Both Kashef
[7] and Knight [3] obtained approximate models for seepage problems based on parabolic
assumptions on φ.
In this research, an alternative method is proposed. Steady 2D free surface flow may
be approximately treated by incorporating the 2D flow characteristics in the 1D potential
flow equations, using the so-called Fawer theory [8,9]. By analogy to free surface flows,
the pseudo-2D approach is used for free seepage flow in porous media, thereby improving
the classical Dupuit-Forchheimer approach and avoiding a full solution of Laplace’s equa-
tion. The new development is applied to seepage flow in earth dams and to free water flow
into drains. The novel model yields a second-order differential equation for the former case,
whereas a first-order differential equation describes the latter.

123
Environ Fluid Mech (2011) 11:427–435 429

Fig. 1 Definition sketch of 2D


seepage flow

2 Governing equations

Consider steady flow in porous media over a horizontal impermeable stratus (Fig. 1). Potential
flow with curvilinear streamlines is governed by the Euler equations, reducing in a stream-
line-centered s − n reference system to [10,11]
∂V V
= (2)
∂n R
with V = velocity, R = radius of streamline curvature and n = coordinate along a normal
(Fig. 1). Integration of Eq. 2 along an equipotential curve yields [10,12,13]
⎛ N ⎞

dn ⎠
V = Vs exp ⎝− (3)
R
n

with Vs = free surface velocity and N = length of equipotential curve. The curvature radius
may be related to its corresponding value at the free surface (subscript s) as [8]
1 1 K
= ν (4)
R Rs
with ν = n/N = dimensionless curvilinear coordinate along a normal and K = curvature
exponent. Using Eq. 4, Eq. 3 is integrated as
 

N ν K +1 − 1
V = Vs exp (5)
Rs K + 1

The discharge q through an equipotential is, using Eq. 5, [9]


1
N
q=N V dv = N Vs 1 − (6)
Rs (K + 2)
0

which was obtained using a Taylor series development of Eq. 5. The relation between N and
the seepage height h can be approximately prescribed assuming circular-shaped equipoten-
tials [12–14], resulting in

123
430 Environ Fluid Mech (2011) 11:427–435

 
N θ arctan h   1/2
= = 
1 + h 2 (7)
h sin θ h

where θ = seepage surface inclination relative to horizontal and prime denoting ordinary
differentiation with respect to x. Further, the free surface curvature is given by

h  −3/2
= hh  1 + h 2 (8)
Rs

Substituting Eqs. 7 and 8 into Eq. 6 gives

 −1/2 ⎛  −1 arctan(h  ) ⎞−1


q h  1 + h 2 hh  1 + h 2 h
Vs = ⎝1 − ⎠ (9)
h arctan (h  ) K +2

for the free surface velocity as a function of streamline curvature and inclination of the
seepage surface. Further, Darcy’s law states [1,2]


V = −K H (10)
ds

where V = flow velocity, K H = hydraulic conductivity and s = curvilinear coordinate in


seepage direction. Equation 10 at the seepage surface reads

dh  −1/2
Vs = −K H = −K H sin θ = −K H h  1 + h 2 (11)
ds

from which results by equalling Eqs. 9 and 11

⎛  −1 arctan(h  ) ⎞−1


q h hh  1 + h 2 h
⎝1 − ⎠ + h = 0 (12)
K H h arctan (h  ) K +2

which is novel second-order non-linear differential equation describing 2D seepage flow in


porous media. Below, Eq. 12 is applied to typical cases of practical interest in soil–water
drainage problems. Once the seepage surface h = h(x) is determined from Eq. 12 the free
surface velocity Vs is available from Eq. 9, and the bottom velocity follows from Eq. 5 for
ν = 0. Thus, the travel time of a pollutant particle
may be estimated for the free surface and
bottom streamlines as ts = ds/Vs and tb = dx/Vb , respectively.

123
Environ Fluid Mech (2011) 11:427–435 431

Fig. 2 The rectangular dam


problem

3 Application to seepage flow in rectangular earth dam

Consider 2D flow in a rectangular earth dam (Fig. 2).


Using a Taylor series development
h h 2

≈1+ (13)
arctan (h ) 3
from which Eq. 12 simplifies to by retaining first order terms

q hh  h 2
1+ + + h = 0 (14)
KHh K +2 3
Further, for the basic case K = 1 involving a linear variation of streamline curvature, Eq. 14
reads

q hh  + h 2
1+ + h = 0 (15)
hKH 3
According to [7], the discharge q is exactly given by Dupuit’s equation
q 1  2 
= h − h 22 (16)
KH 2L 1
which is used in Eq. 15, where h 1 = upstream flow depth, h 2 = downstream flow depth and
L = length of dam.
The results obtained from Eq. 15 using a fourth-order Runge–Kutta method with
the boundary conditions h(x = 0) = h 1 and h  (x = 0) = 0 are compared in Fig. 3 with the

Fig. 3 Flow across rectangular


dam for h 1 /L = 1 and
h 2 / h 1 = 0.25 from (open circle)
Baiocchi’s 2D results, Knight [3],
(solid line) present approach,
Eq. 15

123
432 Environ Fluid Mech (2011) 11:427–435

Fig. 4 Flow across rectangular


dam for L/ h 1 = 1 and h 2 / h 1 = 0
from (open circle) 2D results by
Hornung and Krueger [15],
(solid line) present approach,
Eq. 15

Fig. 5 Seepage flow to drains

numerical solution of Eq. 1 by Knight [3], resulting in an excellent agreement. Note that the
boundary condition h  (x = 0) = 0 means no-flow at x = 0, whereas the actual flow is q.
Thus, this boundary condition means that the seepage slope is small there.
The numerical results of the Polubarinova-Kochina equations [2] are shown in Fig. 4, as
presented by Hornung and Krueger [15] for h 1 /L = 1 and h 2 / h 1 = 0. The agreement of
Eq. 15 with Polubarinova-Kochina’s solution by Hornung and Krueger [15] is again excellent,
supporting the accuracy of Eq. 15.
The present model was further applied considering only streamline inclination effects (e.g.
K → ∞) resulting in a poor agreement. Thus, in the present case the seepage curvature is
important.

4 Application of seepage flow to drains

Consider steady 2D flow to drains under a uniform rainfall of intensity r (Fig. 5).
The distance between drains is 2L, and, by symmetry, only half of the problem is consid-
ered. For this problem q = q(x) = r x, from which Eq. 12 reads
⎛  −1 arctan(h  ) ⎞−1
rx h hh  1 + h 2 h
⎝1 − ⎠ + h = 0 (17)

K H h arctan (h ) K +2

123
Environ Fluid Mech (2011) 11:427–435 433

Fig. 6 Flow to drains for


r/K H = 0.3 from (open circle)
2D results by Knight [3],
(solid line) Eq. 17 with K = 15,
(dashed line) K = 1

Fig. 7 Flow to drain for


r/K H = 0.3 from (open circle)
2D results by Knight [3],
(solid line) Eq. 18

The test case presented by Knight [3] for r/K H = 0.3 is considered in Fig. 6, where
his 2D flow results are included. Equation 17 was solved numerically using a fourth-order
Runge–Kutta method for this case assuming K = 1. To avoid a singularity at x = 0, where
q = h  = 0, a point in the vicinity of x = 0 was considered. The results are included in Fig. 6
indicating good agreement for 0 < x/L < 0.8, whereas results are poor near the drain.
A second computation was done using a higher K values. From a trial-and-error resulted
K = 15. It can be seen that this value improved the results, yielding excellent agreement
with Laplace’s equation (Fig. 6).
Inspecting Eq. 17 reveals that K is a weighting factor. Thus, the larger K , the smaller
the effect of streamline curvature. The large K value fitted in this analysis indicates that the
streamline curvature effect is irrelevant for flow to drains, whereas streamline inclination
needs consideration for an accurate solution. The extreme case K → ∞ gives for Eq. 17

rx
h  = − tan (18)
KHh

where only streamline inclination effects are considered, given by N / h. Equation 18 was
originally proposed by Jaeger [12] as a particular case of the more general model presented
herein. The numerical solution of Eq. 18 using a fourth-order Runge–Kutta method is included
in Fig. 7, resulting in excellent agreement with the 2D flow results by Knight [3].

123
434 Environ Fluid Mech (2011) 11:427–435

Fig. 8 Flow to drain for


r/K H = 0.1 from (open circle)
2D results by Gurehian and
Youngs [5], (solid line) Eq. 18

Another test case for r/K H = 0.1 is considered in Fig. 8, where the 2D flow results
obtained by solving Eq. 1 by the finite element method are included [5]. The results obtained
from Eq. 18 are further included, again resulting in a good prediction.

5 Conclusions

A new model for soil–water drainage problems is presented based on an approximate theory
for 2D potential flows. The model involves a second-order, non-linear differential equation
including the effects of both seepage curvature and inclination. The new model was applied
to the rectangular dam problem considering a simplified version where non-linear terms are
dropped. Predictions were compared with exact 2D flow solutions, resulting in an excellent
agreement of the new model. The model was further applied to steady 2D flow to drains,
resulting in irrelevant effects of seepage surface curvature, whereas consideration of seepage
inclination is necessary. A first-order equation describing the seepage surface was then solved
and successfully compared to the full 2D result from Laplace’s equation. The developments
presented in this research can be applied to seepage flow problems related to environmental
hydraulics with pollutants.

Acknowledgements The Author was supported by a contract of modality JAE-DOC of the program “Junta
para la Ampliación de Estudios”, CSIC, National Research Council of Spain, co-financed by the FSE.

References

1. Muskat M (1942) Flow of homogeneous fluids trough porous media. McGraw-Hill, New York
2. Polubarinova-Kochina PY (1962) Theory of groundwater movement. Princeton University Press, Prince-
ton, NJ
3. Knight JH (2005) Improving the Dupuit-Forchheimer groundwater free surface approximation. Adv
Water Resour 28:1048–1056
4. Chapman TG (1957) Two-dimensional ground-water flow through a bank with vertical faces. Géotech-
nique 7:35–40
5. Gurehian AB, Youngs EG (1975) The calculation of steady-state water-table heights in drained soils by
means of the finite element method. J Hydrol 27:15–32
6. Raudkivi AJ, Callander RA (1976) Analysis of groundwater flow. Arnold, London
7. Kashef AA (1965) Seepage through earth dams. J Geophys Res 70(24):6121–6128
8. Castro-Orgaz O (2010) Steady open channel flows with curved streamlines: the Fawer’s approach revised.
Environ Fluid Mech 10(3):297–310

123
Environ Fluid Mech (2011) 11:427–435 435

9. Castro-Orgaz O (2010) Equations for plane, highly curved open channels flows. J Hydraul Res 48(3):405–
408
10. Rouse H (1938) Fluid mechanics for hydraulic engineers. McGraw-Hill, New York
11. Montes JS (1998) Hydraulics of open channel flow. ASCE Press, Reston, VA
12. Jaeger C (1956) Engineering fluid mechanics. Blackie and Son, Edinburgh
13. Montes JS, Chanson H (1998) Characteristics of undular hydraulic jumps: experiments and analysis.
J Hydraul Eng 124(2):192–205
14. Hager WH (1985) Equations for plane, moderately curved open channel flows. J Hydraul Eng
111(3):541–546
15. Hornung U, Krueger T (1985) Evaluation of the Polubarinova-Kochina formula for the dam problem.
Water Resour Res 21:395–398

123

You might also like