Calculation Methods

for
Two-dimensional Groundwater Flow

y/o>o

p. van der Veer

 llllillll.iMIIII
'II' i n
liilii'lli'liiillirHlii
imiiiMifiiim,'
r
•o
o -^
u> o

Calculation Methods
for
Two-dimensional Groundwater Flow

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN DOCTOR IN DE

TECHNISCHE WETENSCHAPPEN AAN DE TECHNISCHE
HOGESCHOOL DELFT, OP GEZAG VAN DE RECTOR
MAGNIFICUS PROF. DR. IR. F. J. KIEVITS,
VOOR EEN COMMISSIE AANGEWEZEN DOOR HET
COLLEGE VAN DEKANEN, TE VERDEDIGEN OP ^^>^ SaSd
WOENSDAG 1 NOVEMBER 1978 TE 16.00 UUR

DOOR

PETER VAN DER VEER

CIVIEL-INGENIEUR
GEBOREN TE APELDOORN

BIBLIOTHEEK TU Delft
P 1780 5033

C 526816
 DIT PROEFSCHRIFT IS GOEDGEKEURD DOOR DE PROMOTOR

PROF. DR. IR. J. C. VAN DAM

.
To my wife Puck
and
my daughter Nienke
r

Acknowledgements

The author is indebted to Mrs. O. C. M. Erkelens-Post for typing the manuscript and
to Mr. M. A. J. van Bijsterveld for preparing the figures.
Contents

Page

7 Introduction

Part I: Basic Theory

10 1 Basic Laws
12 2 Definitions
14 3 Anisotropy
17 4 Boundary Conditions

Part II: Existing Solution Methods

Analytical Methods

24 5 Direct Methods
24 5.1 Direct Integration
25 5.2 Separation of Variables
27 5.3 Integral Transforms

33 6 Indirect Methods
33 6.1 Superposition
36 6.2 Green's Functions

39 7 Methods based on Complex Function Theory

39 7.1 Method of Pavlovskii
44 7.2 Method of Vedernikov-Pavlovskii
48 7.3 Hodograph Method

Approximative Methods

54 8 Method of Fragments
58 9 Graphical Flow Net Method
59 10 Finite Difference Method
62 11 Finite Element Method
Page Part III: Analytical Function Method

70 12 General Description
72 13 Composition of the Solution

74 14 Boundary Conditions
74 14.1 General
76 14.2 Specific Expressions to be used

87 15 The Approximative Function 02(2)

87 15.1 General
87 15.2 Algebraic Functions
89 15.3 Transcendental Functions

94 16 The Calculation of Flow Problems

113 17 Special Cases of Flow in a Half Plane
122 18 Multi-Valued Character and Singularity of the used Functions
127 19 General Comparison with Other Methods
129 20 Examples

145 Samenvatting
145 Summary
146 Appendix 1: Definitions with respect to Complex Denotations
148 Appendix 2: The Sine Transform
150 Appendix 3: Part of Computer Output of Example 7
157 Appendix 4: Comparison of Exact ^nd Approximative Solution of
Example 9
160 Appendix 5: Listing of Computer Program
169 Principal Notations
170 References

6
Introduction

The motion of groundwater plays an important role in many civil engineering

projects. Low-situated polders receive groundwater from the surroundings, while
from water reservoirs there may be a groundwater flow towards the surroundings if
the water table in the reservoir is higher than the groundwater head in the surround-
ings. In cases of groundwater recovery for public water supply or at a trench, it is
important to know the consequences of the abstraction of groundwater; a sharp fall
of the groundwater head may cause damage to structures and to the vegetation. If it
is desired to confine the adverse effects of groundwater recovery, it is important to be
able to predict the consequences of various actions. Knowledge of the groundwater
motion plays a role in problems of groundwater pollution and groundwater manage-
ment too; the stability of dikes and shore constructions also depends on the motion
of groundwater (e.g., the behaviour at a seepage surface and overpressures under an
impermeable dike revetment). It is thus seen that it is important to have a good
knowledge of the motion of groundwater.
Although in many cases it is possible to find mathematical solutions for groundwater
flow problems, it is often difficult to define exactly the relevant parameters in a
flow problem. Generally the average coefficient of permeability can be determined
rather well in situ, but it is much more complicated to determine the anisotropy,
especially when the direction of the anisotropy is not known previously. It is almost
impossible to determine in situ the exact location and permeability of thin layers of
clay or silt; in addition, these layers may have sharply differing properties. The same
problems are encountered in determining the properties of resistance layers on the
talus and bottom of a canal. Generally in a calculation these lines are not equipotential
lines; in other words, the effect of the resistance layers (clay or silt) may not be neg-
lected in the calculation. An extra complication in these problems is the alteration of
the resistance properties of the layer when it bursts off as a result of overpressures.
All this shows how unsure a description of groundwater flow may be. This does
not mean, however, that there is no need for reliable calculation techniques. The
availability of elTective mathematical tools can give a good insight into the effects of
any alteration of parameters of the problem. In many cases, therefore, geohydrologic
calculations consist of some calculations of the same problem with varying para-
meters. In that way an insight is obtained into the consequences of the lack of know-
ledge of parameters of the problem. That experience can be useful for the decision
whether additional information has to be compiled, for example, by field measure-
ments.
The performance of reliable geohydrologic calculations may result in a lower safety

7
coefficient for the relevant work because more knowledge has been obtained about
the phenomenon

This treatise on calculation methods has been restricted to two-dimensional ground-

water flow Although in reality groundwater flow is always three-dimensional, in
many cases the flow can be assumed to be two-dimensional
Such cases are, for example, a horizontal flow region without replenishment of water
from above or below in vertical flow regions the flow can be assumed to be two-
dimensional if the relevant section is present over a great length (dikes, canals, rivers
and long structures)
Some basic assumptions have been made
— The porous medium is incompressible
— The coefficient of permeability is sectionally constant, in other words, the coeffe-
cient of permeability may vary if the variation is not continuous but step-wise There
may be resistance layers between different sand layers The soil may be anisotropic
— The fluids are incompressible and there is an abrupt alteration of density going
from one fluid to another
— There is only groundwater in a saturated zone
— Anywhere and at any moment the flow is laminar This means that Darcy's
relationship between groundwater head-gradient and specific discharge always holds

8
I Basic Theory
1 Basic Laws

The groundwater head cj) is defined by

(1.1)
pg

where

p- pressure with respect to a reference pressure (atmospheric pressure)

p - density
g - acceleration of gravity
y - elevation with respect to a reference level.

The specific discharge i\ in a direction s is given by Darcy's Law;

where k is the coefficient of permeability.

The condition of continuity for an element dxdy is given by the following expression
(see figure I.l): d{v^dy + v^dx) + P(x,y,t)dxdy = 0.

positive d(v,dy) is outflow

positive d(vydx) IS outflow
dy
positive p(x,y,t)is outflow

Figure I 1 Flow through an element

10
In this expression P(x, y, f) is the amount of water abstraction per unit of time and
per unit of area.
From the previous expression the continuity equation is derived;

dv^ dv..
^ + ^=-P(x,y,r) (1.2)

If Darcy's Law is combined with the continuity equation (1.2) a differential equation
is obtained that defines the flow problem together with the boundary conditions. For
a homogeneous isotropic porous medium the differential equation becomes;

k ^ + k—} = P{x,y,t) (L3)

'dx^^ dy^

In the case that P{x, y, t) = 0, for each x, y, t (1.3) reduces to the well-known differential
equation of Laplace;

2 + ^ = 0 (1-4)
dx^ dy

In that case the function <^(x, y) is apparently a harmonic function (see Appendix 1).

11
2 Deflnitions

Potential
The relationship between specific discharge and groundwater head is given by Darcy's
Law, for a homogeneous isotropic porous medium the following expressions hold

d(l> d(j)
Vy. = —k — and v, = —k —
dx ' dy

The potential <P is defined as the product of the groundwater head and the coefficient
of permeability <P = kcj) With this variable Darcy's Law can be written in a form
where the specific discharges are derived from the potential

(^(p d0
and V. = (15)
dx dy

Because the potential 4> is equal to kcp with constant k, also (? satisfies the differential
equation of Laplace (in the case that P(x, y, t) = 0) So the potential is a harmonic
function (see appendix 1)

d^<p d^<P _
'd^^'d/^^

Stieam function
Another important function is the stream function f This function has the property
that the amount of water that per unit of time passes a line between two points is
equal to the stream function difference between those points
This follows from the definition of the stream function

^f flV
and 1-, (16)
tiy dx

(For example the amount of water that passes a vertical line between two points at a
mutal distance A\ is equal to i^Ay According to (I 6) this is equal to —AW)
If the expressions (I 6) are substituted in the continuity equation (I 2) with P(\,), f) — 0
It IS seen that the stream function also satisfies the Laplace differential equation, so "f
IS a harmonic function too

12
 ~d^^W^^
The relationship between the potential and the stream function is found by combining
the expressions (1.5) and (L6). This yields;

d0 df d(P d'P
dx dy dy dx

These are just the Cauchy-Riemann relationships (see Appendix 1), which means that
the functions ^ and f are conjugate harmonic functions. So in a homogeneous
isotropic porous medium, equipotential lines (<P = constant) and stream lines
( f = constant) are perpendicular.

Complex potential
Using the potential <P and the stream function f, another important function can be
defined:
The complex potential Q is defined by;

Q = (p + iW

Because <P and f are conjugate harmonic functions of x and y, the complex potential
Q is an analytical function of the complex variable z = x -I- iy.

13
3 Anisotropy

If the coefficient of permeability has not the same magnitude in any direction, the
porous medium is anisotropic By a simple geometric transformation the anisotropic
region (coordinates x and >. z = x -I- ly) can be transformed into another region
(coordinates x and y^^ = x + iy) that can be considered as an isotropic region

Therefore a region with coordinates x and y is considered in which the maximum and
the minimum coefficients of permeability are k^„ and A.^,„ The maximum coefficient
of permeability occurs in a direction x* that has an angle a with the x-axis, as
shown in figure I 2

Figure I 2 Main directions of permeabihty

The relationship between z = x + i\ and z* = \* -I- ; ) * is given by

z* = : exp(-ia)

In other notation

X* -I- (V * = (\ -I- (v) (cos a - i sin a)

From this it follows

X* = V cos Oi + \ sin a

}* = \ cos a — \ sin a

14
According to Darcy's Law the following expressions hold:

del) d<t>
V.' = - ^ m a x ^ and V = -'^•nin^ (1-7)

The stream function f is defined by;

d'P dW
V = - ^ a n d v = ^ (1.8)

By eliminating of i\* and t-y from (L7) and (1.8) it follows:

3«F d4> 8^ d<p

mm ^ ^ ^ — k (1.9)
ax* dy* "^^ dx*

Now the coordinates z* = x* -I- iy* are transformed into z = x + iy according to:

x*=xJkZ. y* = lJk~ (LIO)

Substitution of (1.10) in (L9) yields;

dW I d^ dW r d^ , ^
- ^ = -V'^min'fmax^ and — = V'f„i„/c„,ax^ (II 0

A Active coefficient of permeability for the z - plane is defined by A: = yjk^;„k^^^.

Then (1.11) can be written in the following form;

S'V _ d^ d'f _ d(f>

dx dy dy dx

If the potential 4> is defined as <P = kcp for the z - plane there come the following
expressions:

d^P d<P d'P d<P

dx dy dy- + a^
or:
d<P d^P d0 dv
(L12)
dx dy Ty~ ~ 5x

The relationships (1.12) between the stream function *? and the potential fp in the

15
z - plane are just the Cauchy-Riemann relationships. So <P and P are conjugate
harmonic functions in the z - plane. So the functions <P and P satisfy the Laplace
differential equation and Q = <P + i'P is an analytical function of z = x + iy.
So the flow in an anisotropic region can be calculated in the same way as for an iso-
tropic region if the following, transformations are applied (where the direction of the
maximum coefficient of permeability has an angle a with the positive x-axis);

z* = z exp( —la)

X =' X 1^ f^max

I = y*ljKmm
In other notation, where z is expressed directly in z, this is:

z = —j=^ Re{zexp(-(a)} H— Im{zexp(-ia)}

The fictive coefficient of permeability for the z-plane is:

"•max "min

16
4 Boundary Conditions

A groundwater flow problem is defined by:

— continuity condition,
— darcy's Law, and
— boundary conditions (of space and time).
In addition, factors as anisotropy and inhomogeneities have to be taken into account
and also the variation of the density of the fluids. In this Chapter a schematic survey
of the most important boundary conditions is given, as they are relevant for the
treatise of the existing solution methods in Part II. A more comprehensive formu-
lation is given in Chapter 14 in a form that is convenient for the Analytical Function
Method of Part III.

Equipotential line
An equipotential line is a line of constant potential <P. For a homogeneous porous
medium, also the groundwater head <j) is constant along such a line.

Stream line
A stream line is a line along which a fictive water particle moves (where the physical
porestructure is neglected). No water passes a stream line. From the definition of P
(see (1.6)) it follows that the stream function is constant along a stream line.

Seepage line
A seepage line is a line along which the water leaves the soil and freely streams off
The thickness of the seepage layer is generally very small. Therefore at the seepage
line the pressure may be assumed to be equal to the atmospheric pressure (which is
the reference pressure that is set to zero). From the definition of the groundwater
head cj) (see (LI)), it follows;

p = 9ip<P - py)-

Substitution of p = 0 in this expression gives the boundary condition at the seepage

line:

(j) = y .

17
So a seepage line is a line along which the groundwater head is equal to the known
height y.

Phreatic line
In a steady state a phreatic line without precipitation or evaporation is a stream line
Then the stream function is constant If there is precipitation, the stream function "/'
increases per unit of length with an amount that is equal to the precipitation on that
unit of length In any case (steady or non-steady), the condition holds that the
pressure along the phreatic line is equal to the atmospheric pressure Then the same
condition is arrived at as for the seepage line;

<t> = y-

So a phreatic line is also a line for which the groundwater head is equal to the height y
(although generally y is not previously known).

Interface
An interface is a separation line between two fluids with different densities In a steady
state an interface is a stream line, then the stream function P is constant In any
case (steady or non-steady), the condition holds that the pressure on both sides of
the interface is equal
From the definition of the groundwater head </> (see (I 1)), it follows

p = 9ip<t> - py)-

If this expression is used to define the pressure on one side of the interface and the
pressure on the other side is defined by

Pc = d(Pc(t>c - Pcyl

it follows from p = p^ that

ptf, - py = p^(t)^ - pj,

from which.

, P - Pc ^ PA ,r . ,x
(t> = y+ (113)
P P
So an interface is a line along which the groundwater head 0 varies in a prescribed
way with y (although > generally is not previously known).

li
If p^ = 0 is substituted in (1.13) or, in other words, on one side of the interface the
density is zero, the expression reduces to;

<t> = y,

which is just the condition for the phreatic line.

So a phreatic line may be considered as the special case of an interface when one of the
fluids has no density.

19
 I

II Existing Solution Methods

In this Part a general survey is given of the most important solution methods. The
principles of each method are discussed briefly and without a complete derivation,
and the methods are illustrated with examples. The following sub-division is used:
— Analytical methods. In these methods a solution in a closed form is obtained.
- Approximative methods. In some of these methods an approximative closed
form is obtained (e.g., in the method of fragments, see Chapter 8), while in other
approximative methods a solution is found where the relationship between the
parameters that define the problem is not directly visible (e.g., the finite element
method, see Chapter 11).

22
 ^

Analytical Methods

.
5 Direct Methods

Direct methods are methods where the solution is obtained by direct operations of
the differential equation that defines the flow.

5.1 Direct Integration

Generally direct integration of a differential equation is only possible for one-dimen-

sional flow. A very simple example with one-dimensional flow in a confined aquifer is
given in figure II. 1:

x = x, </)=(/)i

Figure 11 1 Confined aquifer

The differential equation for this flow is (see (1.4));

d^<t)
= 0
dx'

From direct integration it follows:

^ = C'lX -I- Co

Substitution of the boundary conditions yields the solution:

, ('^i - 0o) ^,

24
5.2 Separation of Variables

This method has been comprehensively treated in literature (see, for example, Wylie,
1966, Arflcen, 1970).
The solution of the differential equation;

2 + ^ = 0 (III)
dx' dy

is assumed to be the sum or the product of two functions of respectively x and y

<P = F(x) + G{y), (II.2)

or: 0 = f(x) G{y). (II.3)

If the form (II.2) is used, then from substitution in the differential equation (II. 1) it
follows:

Because F(.x) and G{y) are functions of respectively x and y, (II-4) can only be satisfied
if f "(.v) as well as G"{y) are constant. Then it follows from direct integration;

F(x) = a^x' + b^x + c'l

(n.5)

G(y) = a^y' + b^y + Cj

By substitution of (n.5) in (11.4) it follows;

«1 = -«2

If «, = a and c, -I- f2 = ^' ^ then the solution has the following form:

(j) = a{x' - y') + hix + /)2>' + c

The four degrees of freedom in this expression depend on the boundary conditions of
the problem.

If the product of the two functions F(x) and G(y) is used, see (11.3), the following
expression is found by substitution of (IL3) in (ILl):

25
 m ^ Gjy)
F"(x) G"(y)

This expression can only be satisfied if both members of the equation are constant:

nx) . G(y)
F"(x) = c and G"(y)
—-- = —c

Examples of functions that satisfy these equations are sin(x), cos(x), sinh(x), cosh(x),
exp(x), etc.
For example, consider the following functions;

F(x) = (^1 exp(ax) and

G{y) = cos(fly)

The flow problem is defined by;

(^ = </), exp(flx) cos(ay)

This function generates the groundwater flow through a dam of variable width (see
figure II.2). The width of the dam is seen from 0i = (/)j exp(ax) cos(ay) or

y = - arccos(exp( —fl.x)). Then the width of the dam is:

a
n 1
h= arccos(exp( — ax))
la a

(/) = 0

2a

Figure II 2 Horizontal flow through a dam of variable width

26
5.3 Integral Transforms

An integral transform is an operation of the differential equation that generally

results in an equation or differential equation that is easier to solve. By an inverse
transformation of the solution of that equation, the solution of the flow problem is
found. For the transformations useful tables are available (see, e.g., Erdeiyi, 1954). In
the literature the theory of integral transforms has been given (Churchill, 1958,
Sneddon, 1972, Ditkin and Prudnikov, 1965). Applications are given by Bruggeman,
1972.
In this section some important integral transforms are mentioned and some applica-
tions are shown. The integral transforms that will be discussed are those of Laplace
and Fourier.
Generally the Laplace transform is used to transform the time variable and the Fourier
transform is used for one of the space variables x, y.
Generally a linear integral transform has the following form;

T{F{t)}=i'^K(t,s)F(t)dt

The various integral transforms differ in the type of K(f, .s) and the integration bounda-
ries a and b.

LAPLACE TRANSFORM

The Laplace transform of a function F(t) results in a function f{s) and is defined by:

L{f(f)}=jg^exp(-6t)F(t)rft=/(s)

A more convenient form is;

L{h(x,t)} = jg- exp{-st)h(x,t)dt = h{x,s)

Some properties are;

4^f^'-j = ^^(^'^)-M-^'0)
jd'hix,t)\_d'h(x,s)
^V^^\~^x^ "-^

27
As an example, the problem of Edelman is chosen (see Bruggeman, 1972) horizontal
groundwater flow towards a canal in which the water table is lowered suddenly

1
j^-"^'^
M D

X
^ ^ ^ ^ ^ ^
Figure II 3 The problem of Edelman

The differential equation that defines two-dimensional groundwater flow is (see (I 3))

kj^ + k-^ = P(x,y,t)

dx dyl

As the flow IS assumed to be horizontal, the variation of <^ in v-direction may be neg-
lected Then the differential equation becomes

d'(h

Along the phreatic line (^ = \ (see Chapter 4) The differential equation will be written
in a form where the original water table is the reference level and h denotes the
lowering of the phreatic line (then (p = -h at the phreatic line)
The term P(x, t) was defined as the abstraction of water in an element at location x at
time t (see figure I 1) here it is the storage increase that is caused by the increase of
the elevation of the phreatic line
The term P(x, t) is a quantity per unit of area and per unit of time (see Chapter 1)

dh\
Here the storage increase -u— occurs along the phreatic line (where p is the

( ^h\ ,
storage coefficient) To obtain the storage increase per unit of area, \—l^~^] has to
be divided by the height of the water column It is assumed that h <^ D, and so D may
be used for the height of the water column

28
According to this assumption, the term F(x, t) may be written as;

/' Sh
p X, t)=
^ ' D dt
(negative dh/dt means outflow from a vertical element hdx: so negative dh/dt conforms
to positive P(x, t)).

The differential equation now becomes (where cj) = —h):

d'h _ p dh
'dx'~l) dt

or:
^2
d^h p dh
a? = To Vt ("-^^

The boundary conditions are;

h{x, 0) •= 0
/i(0,0 =/io
/i(oo,t)= 0.

Application of (II.6) to the differential equation (II.7) yields the transformed equation;

g = A„5 _,(,„), ,,,.8)

In this equation the first boundary condition {h{x, 0) = 0) can be substituted; and the
two other boundary conditions can be written in transformed form if (II.6) is used:

^(0, .s) = hjs and

(11.9)
/j(oo, s) = 0

If ^(x, 0) = 0 is substituted in (II.8), the following form is found:

d'h p -
^^-i~sh =0 (ILIO)
dx' kD ^ '

A general solution for this differential equation is;

h = A exp(ax) -I- B exp( - ax) (II. 11)

29
By substitution of this solution in the differential equation (11.10) and inserting the
boundary conditions (II.9) in (II. 11), the following expression is found;

r K ( PS

This is a well-known form. The inverse transformation can be found in tables. The
solution of the problem is then found to be;

X / p \
h = h^ erfc
^2V kDtJ

The function erfc(x) is defined by l-erf](x) and:

erf(x) = —= |5 exp( — u')du

Jn

FOURIER TRANSFORM
Depending on the type of problem, some kinds of Fourier transforms can be applied.
In the following example the infinite Fourier-sine transform is used:
The infinite Fourier-sine transform of a function F(x) results in a function/(r); it is
defined by;

S{F(x)] = \^ F{x)sm[rx)dx = f{r)

Written in a more convenient form:

S{h(x, r)} = jo Kx, t) sin (rx)dx = h(r, t)

Some properties are;

\d'h(x, t)
= -r'h(r,t) + rh(0,t)
dx'

(if;/!(x, t) = — / ! ( x , r ) = 0)
dx
(11-12)

S{0} = 0

30
The inverse transformation is given by;

;,(x, f) = ^ j J ;i(r, t) sin {rx)dr (IL13)

n

The same example as for the Laplace transform is shown here to illustrate this Fourier
transform (see Bruggeman, 1972). The differential equation that generates the flow is
(see(n.7)):

d'h p dh , ^

The boundary conditions are:

^(x,0) = 0
h{0,t) =/lo
^(oo, t) = 0

The boundary condition ^(oo, f) = 0 has already to be satisfied for application of

the Fourier-sine transform (it is a condition in 11.12; our problem satisfies it).
Application of (11.12) to the differential equation (11.14) yields the transformed
equation: _
, > p dh
-r'h -hr/iO,t = — —
^ ' kD dt

The boundary condition /i(0, t) = /lo can be substituted in this equation.

The remaining condition ^(x, 0) = 0 is given in transformed form by;

/!(x,0) = 0 (ILl 5)

By inserting the boundary condition for h{0, t) in the transformed differential equation,
the following expression is found;

dh ,kD- kD , ^
+ r' — h = rho— n.l6)

of p p

A general solution for this equation is;

h = A exp(af) + B exp(-at) + C (IL 17)

By substitution of (1117) in (IL 16) and inserting the boundary condition (IL15) in the
general solution (11.17), the following expression for h is found;

31
 1
h=ho - exp( - kDtr'/p) /r

Using (11.13) the inverse transform h can be found:

h=-\tK exp( - kDtr'/p)/r sm{rx)dr

or;
^ ^ 2/10 ^ . sirHrx) ^^ _ 2/.0 ^^ exp( -/cZ>r.V,) ^^^^^^^^^
n r n r

The first integral is equal to n/2, and the second can be reduced to a well-known
function. Then the solution of the problem is found:

h = h,
^'^[l^jkDt)^
or;

h = /in erfc I - /
2\/ kDt

32
6 Indirect Methods

6.1 Superposition

The superposition principle holds for the Laplace differential equation

If (pi and (f)2 are solutions of V^t/) = 0, then 03 = fl<^i -I- b4>2 is also a solution
This can be shown simply

V'(j)3 = W'(a(pi + b(j>2) = aV'4)i + bS/'cjy^

V^^i = 0 V'<i)2 = 0 then ^'(i)^ = 0

Examples of application of the superposition method are flow in an infinite plane

with sources and sinks, and the method of images

Example 1 The flow pattern due to a sink and a source in a lineai flow field in an
infinite plane is found by a superposition of three separate flow patterns (see figure
114)

Linear flow field (/), = ax + b

Source in (xj^yi) with discharge Q 4>2 = — r \ri[^J{x - Xj)^ + (\ - \2f'\

Ink

+Q
Sink in (\3 >3) with discharge Q ^3 = ln[v'^(x - X3)' + (y - Vi)^]
Ink

Example 2 The method of images is used for problems with sources and sinks where
some special boundary conditions can be satisfied if extra fictive sinks or sources
are located outside the region of interest
In the case of a straight boundary, the property is used that in the middle between a
sink and a source the straight line perpendicular to the connecting line is an equipoten-
tial line In the case of two sinks or two sources, that line is a stream line, (see figure
115)

33
 -•-T 1
01
1 1 [ [-»—I 1

• > *

-* *
.»_ »
*- *—
•- • (XaYg)
•—i 1 ' ' ' l-^-J 1
(X2y2)

0=(/)^ + (/)2+(/)3

sink
m{x3y3)

Figure II 4 Superposition of solutions

Another application of the method of images is the flow field of an excentncally

located sink in a circular island, see figure II 6 Here the property is used that in the
case of a source and a sink in an infinite plane, the equipotential lines are circles This
can be shown by a simple deiivation
Consider a sink and a source both with discharge Q. in respectively (\i, i J and
(\2, \2) The groundwater head in the plane is given by

l(x-x,)' + {i-^,)'Y
0= — In (II IS)
2nk

34
a)

two sinks

b)

sink and source

Figure II 5 a Flow according to two sinks

b Flow according to a sink and a source

For an equipotential line <>

/ = (/>! is constant. From (11.18) it then follows directly:

[(x - xi)'+ (y - y,)'^' /2nkcf>,

-^ \ = exp ~ ] = constant
[{x - X2)'+ [y - y2)'f ^ Q
or: /AnkcPi
(x - Xi)^ + (y - y^)' = exp (x - x^)^ + (y - y2

This can be written as;

^_^^ZL^2y^l,_y^-ay2 X, - ax- >'i - ay 2

1 -a 1 -a +
yy^ - ay2^
(11-19)
I -a 1 -a

35
 Figure II 6 Excentncally located sink m a circular island

where a = e\p(4nk (j>i/Q)

The expression (II 19) is the equation of a circle with its centre at

>i - a\2
Xn = and \o (II 20)
1 I - a

and with radius R according to

,2
Xi — a x ; 31 - ayi ax-, >l^ - a)2^
R' = +
1 -a I -a 1 -a 1- a

Generally only (xj, \ J is known and the location of the centre of the island From
(II 20) the location of the fictive source can be calculated yielding

^2 = (^1 - ^o(l - a))/a and ^2 = (>i - >o(l - "))/«

With these known values of (xj, Va)- ^he groundwater head distribution over the
island can be calculated using (II 18)

6.2 Green's Function

The solution method that uses Green's function has been described in literature (see
for example. Bear Zaslavski, Irmay, 1968 and De Wiest, 1969) The method will be
illustrated here with a simple example

36
The flow IS considered in a region D that has a smooth boundary C (see Bear,
Zaslavski, Irmay, 1968) The values of the head in the points A(x, \) in D are (f){A).
(j) satisfies V'(f> = 0. so <^ is a harmonic function Suppose that the values of (j) are
known in all points of C as a function of the location s on the boundary according to
a function y(s) Then it is possible to express (f)(A) in another harmonic function
G(/4, A') that IS called a Green's function The function G(A, A') is a function of the
coordinates of two points (A{x, v) and A'{x\ v')) such that for each point A{x,y) the
value (t)(A) can be expressed as a function of the values of/"(s) at the points A' on C
and the partial denvate of G with respect to n (n is the inner normal on C)

dG{A, A')
<t>{A) = \J[s) ds(A') A' on C (11.21)
dn{A')"

In literature (see, eg Bear, Zaslavski, Irmay, 1968) for several rather simply shaped
regions the Green's functions can be found For example, for the right half plane
Green's function is given by

1 "(x - x')' + (y - y'f

G{x,y, x',y') = In (II22)
An' {x + x')' + (y - y f

As an illustration, this method is used to describe the flow in a half plane with the
following conditions

<ti{x\yX <o = (^1/2 4>{x\y\ >o = -<^i/2

Figure 11 7 gives a sketch of the flow pattern

Figure II 7 Vortex flow

The solution is the flow according to a vortex in the origin

4> - ' arctgl- (II 23)

n \x

37
This solution can also be found by means of the Green's function method, using
(II 21) and (II 22)
The combination of (II 21) and (II 22) yields an expression for the groundwater head
in the right half plane

(II 24)

[x - xf + (v - I )'
where F=
(x + x')' +(y- yf

After differentiation and substitution of x = 0 , (11.24) becomes

'^1 . f o dy' </>! dy'

^ In""^' -''x'^ (y - y'f In "" ^^' = <' x' + [y - V)'

or
dy'
6 = ^x{-y J^JlA__ti
^ 2n
^_ ny -y)-<x. X'
„2 +, (y
/„- - y)'
„\2 In ''^~' x' + (y-v')'

arctgl
y - y' 0. arctg y - y\
2n (y - ) . ) - . » 2n (> - > ) - -y

(\>i , y
0 arctg -
n \x
With this expression the result (II23) is obtained

38
7 Methods based on Complex Function Theory

7.1 Method of Pavlovskii

This method is well-known in literature (Bear, Zaslavski, Irmay, 1968, Verruyt 1970).
A region in the z-plane can be conformally mapped upon a region in another plane,
e.g., by elementary functions such as sin(z), ln(z), z", etc. A region with a polygon
boundary can be mapped on the upper half plane Im {C] > , 0 by the well-known
Schwarz-Christoffel integral:

z = afo(^ - Ci)"'' (^ - G'" ••• (^- - Q-'-d^- + P (11-25)

In this expression ^j are the images of the edge points of the polygon on the real axis c.
The factors k^ correspond to the alteration of direction in each edge point according
to k = a/n, see figure II 8; a en ^ are complex constants.

Figure II 8 Conformal mapping by the Schwarz-ChristotTel integral

A large number of examples of conformal representations is given by Kober (1957).

In literature (see, for example, Verruyt 1970) a more comprehensive treatise of the
Schwarz-Christoffel integral can be found.
In the method of Pavlovskii some problems are solved by conformal mapping of Q
on z, while other problems are solved by mapping the z-plane on another plane and
solving the problem in that plane. For both kinds of problems, an example is given:

39
Example 1 Flow in a confined aquifer to a low-situated polder, see figure II 9.

0 =0

Figure II y Polder aquifer

The problem is symmetrical, therefore it is sufficient to consider half the flow region
Figure II 10 gives the z-plane and the fl-plane

A D (

y Z cj
B X

Figure II 10 z plane and ff-plane

\t Q

The application of the sine transform (see Appendix 2) for both the Q- and z-regions
yields the mappings as represented in the upper half planes C and C*. see figure II 11

-1 +1
Figure II 11 Upper half planes "* and

40
For this mapping the z-plane and the 0-plane are first rotated over n/2. Then they
are linearly transformed, and translated over n/2. So they are first mapped upon a
half infinite strip with width n (figure 11.12).

Figure II 12 Mapping upon a half infinite strip.

The mapping functions are;

[ n n\
C* = sin !z — h -
\ D 2)
n n
L = sin iQ—I—
' e 2
This can be reduced to:

cos ( D cosh (IL26)

fiQn (Qn
= cos — cosh — (11.27)

The relationship between the C*- and the C-plane is found by a transformation, so
that the points D and B of the C*-plane are mapped upon the points — I and +1;

41
 2(\_-^
C=i
1 - cs
From figure 11.10 it follows; ZD = L + ID

Then according to (11.26):

L
CB = cos ( (• — (L -f iD)) = cos -n + i—n cos 71 - (L— =
D D
(iLn Ln
— cos — cosh —

Then ( is given by;

2(1
C = 1 - (11.28)
1 -f cosh
(f)
A combination of the expressions (0.26), (0.27) and (0.28) yields the relationship
between Q and z;

1 - coshi
fi = — arccosh (11.29)
n 1 4- coshI — 71

\D J!

This expression is the solution of the flow problem. Because only half the flow region
is considered, Q represents half the seepage discharge per m'. Q can be calculated if at
some distance from the polder the groundwater head with respect to the polder level
is known:
From (11.29) it follows using <P = kcp:

1 - cosh — n
0 = Re — arccosh
kn
1 -I- coshi — n
D

42
or
nk(j)
e=
1 — cosh — n
Re arccosh 1 - 2
1 -I- cosh - n
\D

With this formula Q can be calculated, cj) is the difference between the groundwater
head at the point z and the polder level

Example 2 A number of sinks in the vicinity of a contraction of a canal, see figure

n 13:

number of sinks n
sink discharge Qj
location of sink Z.

canal (/)=0

Figure II n Sinks near a canal

The boundary ABC A is assumed to be an equipotential line

The z-plane is mapped on the upper half plane using the Schwarz-Christoffel integral
(II 25), see figure II 14

Figure II 14 Upper half plane ;

43
The point A is mapped onto infinity and does not appear in the integral (see Verruyt
1970)
The alterations of direction in the points B and C of the polygon are — 7r/2 and 7r/2

Then kg = - i kc = + i

The points B and C are chosen to be mapped on C = 0 and C = 1 Then the Schwarz-
Christoffel integral gets the following form for this problem

z "f'L-^ii'"*'
After integration the constants a and /? are determined in such a way that the points B
and C correspond with the points z = ; and z = 0 in the z-plane, see figure II13
That leads to

z = i { 7 a r ^ + in[Vc + 7(c-i)]} (II 30)

In the C-plane the problem can be solved simply by use of the method of images

^^-'27r C - L,
(ini)

The expressions (II 30) and (II 31) form together the solution of the flow problem
Therefore c is eliminated from (II 30) with known z Then from (II 31) the complex
potential Q can be evaluated

7.2 Method of Vedernikov-Pavlovskii

This method can be found in literature (Bear, Zaslavski, Irmay, 1968. Verruyt, 1970)
The function Z is defined by
Z ^ Q + ikz

From this definition it follows by separating real and imaginary parts (using
Z = Zi + 1Z2)
Zi = 0 - ky
Z2 = P + kx

Using the function Z (Zhukovski s function) problems can be solved with boundaries

44
that consist solely of horizontal equipotential lines, vertical stream lines and a
phreatic line.

For a horizontal equipotential fine Z^ is constant:

Z[ = <Pi — ky^ = constant

For a vertical stream line Z2 is constant;

Z2 = 'Pi + kxi = constant

For a phreatic line without precipitation Zi = 0. This can be shown by substituting
the boundary condition (j) = y and thus <P = ky in the expression for Z , ;

Zi = 0 - ky = ky - ky = 0

The properties just mentioned, can be used in the problem of Nelson-Skornakov (see
Verruyt. 1970). The problem deals with a flow under a dike that separates a low-lying
polder from higher surroundings, see figure 11.15;

Fiaure II 15 Problem of Nelson-Skornakov

In figure 11.16 the Z-plane and the fi-plane are given for this problem. The position
of the origin is shown in the figure. The potential in the polder and that in the surround-
ings just correspond with the elevation of the relevant ground level. The image of the
flow region correspond with the right half plane of Z. In the fi-plane, the image of the
flow region is a half infinite strip.

45
 m

c ^0

Fiaure 11 16 7 plane and Q plane

The relationship between Z and Q can be found by application of a sine transform

(see Appendix 2)
First the half infinite strip in the i3-plane is mapped on a half infinite strip with width
n
n in the f3*-plane The Z-plane is rotated over and transformed linearly so that the

points C and B are mapped in the C-plane on the points C = and -I-1, see
figure II17

1^
^ kL
kH 2

Figure 1117 plane and Q* plane

Therefore it is considered that for point C Z2 is given by

Z2=^ P + kx = kL

By application of a sine transform to the i3*-plane, the relationship between Z and Q

IS found (using the relationship between Q and Q* see figure II 17)

46
 2iZ (Qn
-—-h 1 = sin ( -
kL \kH
2iZ Qn
JL cos kH

or
ikL Qn
Z = 2cos^
2kH

Substitution o{ Z - Q + ik z leads to

, , nQ\ iQ
z = LCOS — H (II 32)
2kH J k

In this expression L is unknown (the position where the phreatic surface meets the
drain is not known) The parameter L can be calculated by using the condition that
for point D the v-coordinate is the smallest on the traject CDA
From expression (II 32) it follows

X = LRe<!cos^( ^—][
\2kHj\ k

Along CDA <P = 0 then

nQ\
cos^( _y^ 1 = cos^l — 1 = cosh^
akHj "' \2kH \2kHJ

Then at CDA the following condition holds

nP\ P
X = Lcosh' (II 33)
2kH

The condition that x has its minimum at point D for all points on the traject CDA can
be formulated in the following way

dx ^ /7t'^\ /7r"^\ TT 1
-— = 0 or 2L cosh — sinh — — = 0
d'P \2kHj \2kHj2kH k
For point D then follows

47
 sinh
kH J nL
and thus
kH (2H\
Po = —arcsinh — (1134)
n \T^LJ

The relationship between f ^ and L* is found from (II 33)

L* = L cosh- -" - ^ (1135)

\2kH / k ^ '

By elementary operations one term in (II 35) can be rewritten

1 -I- cosh I _ 1
— T
1 -I- I 1 -I- sinh21 '^'^DW
\2kHj \kH kH

Substitution of this expression and (II 34) in (II 35) yields

or in dimensionless form

|-2s{'-('*Cr)TM»-"''(i
From this expression the parameter Lean be calculated if L* and H are given, and
from (II 32) the complex potential can then be evaluated for an> point

7.3 Hodograph Method

This method can be found in literature (see e g Verruyt 1970) The complex function Q
IS a function of the complex variable z = \ 4- n The partial derivative with respect
to X can be written as follows

dQ _dQ dz _ dQ
dx dz dx dz

The function w is defined by

48
 dQ dQ d<P d'Fl
I —
1^ dx + dx]
According to the Cauchy-Riemann relationships (see Chapter 2), the following
condition holds:

dW d4>
5^

Then:

d<P d<P
W = hi
dx dy

After substitution of Darcy's Law, expressed in derivatives of the potential, it follows:

w = v^ — iv„ (0.36)

dQ
The inverse form of w is given by:
dz

-Jv. dQ (11.37)

From the linear appearance of i-^ and r, in (11.36), it is seen that straight equipotential
lines and straight stream lines in the flow region are represented by straight lines in
the u-plane. Because for those lines i\./i\ = constant they are going through the
origin. A phreatic stream line is represented in the hodograph-plane (the plane of
specific discharges), and thus in the M-plane too, by a circle which can easily be
derived in the following way;

hodograph-plane w-plane

Figure 1118 Representation of phreatic surface in the hodograph plane and

in the w-plane

49
 dip dy v.,
^ ^ = — k — = - ksmoi = —Ze-
es ds i\

v' + ki\. = 0 -» v' + i' + ki\, = 0^vj + { L\, 4-

The latter expression is the equation of a circle with radius k/2. and the centre located
a t ( - ( V 2 ) , seefigure0.18.
In the vv" '-plane the phreatic line is represented by a straight line parallel to the real
axis at a distance — \/k. That can easily be shown (using v' = - kVy. from the pre-
vious derivation);

1 V:C + iVy _ V^ + iVy _ V^ + iVy

Vx - iVy v/ + Vy' -kv.. kvy k

So the imaginary part is constant and equal to — \/k.

The function w is a function of fi. The solution of a problem is found by expressing

w~' inQ. Therefore the conformal mapping technique is used and (11.37) is integrated.
This procedure is illustrated by an example given in figure 11.19. It is the problem of
Vreedenburgh (see De Vos, 1929, Kozeny, 1931 and Verruyt, 1970).

J
^ ^ - ^ ^ ^ ^ ^ ^
Z
J^ . \
^^^^^ /A D X C
" • ' • ' ' '---•-*i

'•••-' —- ' ' ' ««S55^>>W>>>^>565565^<->(y%*565655d««»«*56

Figure II49 Problem of Vreedenburgh.

In figure 11.20 the hodograph plane; the w-plane, the w '-plane and the fl-plane are
given:

50
 D

i: PI t"- A

A i ^V
W [n\ Q
)
D 0 '' cf
Figure II 20 Conformal mappings for the problem of Vreedenburgh

The relationship between w"' and Q is found directly from figure IL20:

w"' = -Q/kQ

Substitution of this expression in (0.37) yields the solution of the problem:

1 r , ^'
z = — \QdQ = hc
kQ •" 2kQ
The integration constant c is determined by the condition that for point D : z = 0, and
Q — 0. From that it follows that c = 0. Then the solution is:

Q^
z = ov: Q = ^/2/cQz
2kQ ^ ^

51
 1

Approximative Methods
8 Method of Fragments

In the method of fragments the flow region is sub-divided into a number of sub-
regions This sub-division is made in such a way that for each sub-region a more
simple solution can be found For example, such a sub-division can be made if one or
several known straight lines may be assumed to be an equipotential line
The method is illustrated with an example
In the problem of figure II 21 the discharge is calculated.

Figure 1121 Seepage towards a basin

The solution of the problem can be given m closed form using conformal mapping
of the region AEFCD upon the upper half plane Im{C} > 0 (method Pavlovskii)
However, the numerical evaluation of this solution may give numerical problems
for some values of the quotients h/Dj and Dj D2
A more simple and approximative solution is obtained if the line segment BC is
assumed to be an equipotential line (with unknown potential 0,) The flow problem
IS solved for the sub-regions on both sides of the line segment BC. with the unknown
potential 0^ being eliminated for both Then the approximative solution of the flow
IS found In figure II 22 the conformal mapping process is given schematically

54
 sub-region1:
0

D1 B c ^Q n
I B I fl>
-1 +1 ;2wc

sub-region 2:

C
R
y
'x
S
i
i 0 +1
^n*^
3v
Figure II 22 Conformal mappings for the basin problem

Sub-region 1 (ABDA) :
Conformal mapping of the z-plane on the C-plane and of the C-plane on the fi-plane
yields;
ZTT
2 1 — cos I
C=
f^
1 - COS — n
\
\Di J
l(Q-0i) 1
COS

A combination of these expressions leads to:

2 1 — cos iz —
Q
'2 [ (11.38)
= 0,4— arccoss( / h
1 — cos - - 7 1

.Sub-region 2 (EFBE):
The relationship between C* and z is found by means of a sine transform;

^*=^n/^J + ^ (0.39)

55
The relationship between Q and Z* is given for subregion 2 by the following Schwarz-
Christoffel integral.

Q = afo*^-*(;. - c ) - i ( / - \)-'d}. + j?

The constants a and ji are determined with the conditions that the points F and C are
represented correctly in the z-plane.
The solution is an elliptic integral of the first kind (see Abramowitz, Stegun, 1968)

Q - (p, F{w\c)
^1 He) (II 40)

where;

w = arccos^_/j^"^-^i' ("41)

Jn{D2 -h)
c = sin^
V 2D,

f(wlc) = jS(l - csm'9)~'-dQ

K(c) = Fhc

The discharge quantity Q is found by calculation of Im[f2J in the point B (see figure
II 22), using (0.39), (11.40) and (11.41).
This yields;

^ ( 1 - <•)
^Q = -^ ^ II 42
0, K(c)

By eliminating <f>i from (1138) and (II 42), the relationship between Q. Q and z in
subregion 1 is found.

,2( 1 - cos I iz —
Q =Q ,
m ^i
-4- arccos<
\ ;
\ D,^
; I (II 43)
LK(1 - C ) 71 [ //,
— COS — n
D,

If the potential is supposed to be known at a point ZQ, the discharge Q can be calculated
from (0.43). Then the following expression is found-

'' \
 *0
Q= -J 7
r ^2 1 - COS iZo —
- 4- Re <\n- arccos I
K{\ - c) ( h
1 — cos --n

where 0^ is the potential at the point ZQ.

57
9 Graphical Flow Net Method

The flow net method is a graphical method based on the property that stream lines
and equipotential lines are perpendicular if the flow region is homogeneous and
isotropic Generally A0 = AP is chosen then a figure is found where the stream
lines and the equipotential lines form elementary squares
Starting from a first sketch in which the expected flow pattern is drawn, a process of
improving the drawn pattern leads after some iterations to a convenient flow pattern
Then the quantity z l ^ is the same over each square also AP is the same over each
square With that knowledge the discharge quantity Q and the potential at any point
in the flow region can be found from the drawn flow pattern
As an example, a flow net is roughly sketched for the groundwater flow in a confined
aquifer, see figure II 23 (The analytical solution for this problem can be found using
the method of Pavlovskii and has a form with an elliptic integral of the first kind)

a).|=2.0m^day .^ 40 m ^ ^^ =0

Fiaure II 23 Flow net under a dike

The number of squares between two stream lines going from $ , = 20 m^/day to
02 = 0 IS 10 So each square represents a decrease of the potential of 0 20 m^/day
To calculate the discharge Q the discharge per square (AP = A0 = 0 20 m^/day) is
multiplied by the number of squares between two equipotential lines

e = 5 X 020 = 1 m^/day

Generally the discharge is expressed in the coefticient of permeability Then 0 , may

be given by for example, 0^ =2k (using the definition 0 = k(j)) Then in the same
way Q = \ k m' day is found

58
10 Finite Difference Method

In the finite difference method the flow region is sub-divided into a number of rec-
tangular or square elements. According to a basic assumption about the variation of
the groundwater head in the elements, the solution can be calculated. Generally the
function 4>(x, y, f) is calculated for a moment f = f,. If the flow is non-steady, a dis-
cretisation in time is used and the-position of the boundary (e.g., a phreatic line)
after a step in time is calculated from the flow pattern. Then the function f/;(x, y, t) is
calculated for the moment f = f, 4- At, etc.
The variation of the groundwater head is assumed to be linear within each element.
In figure 0.24 square elements are considered;

02

03 00 01

a X

04

Figure II 24 Square elements.

The function 4>(x, y, t) is continuous everywhere but not dilferentiable everywhere

because at the sides of each element it is generally nodded.
The partial second derivatives of 0 with respect to x, and y can be expressed as
follows:

d'(t> 1 01 - 00 00 - 03
= -2 [01 + 0 , . - 2 0 o ]
dx' a a a

and
d\(P_ I 02 - 00 00 - 04
[02 + 04 - 20o]
Ty'~a

59
Substitution of these expressions in the differential equation (I 3) that generates
two-dimensional groundwater flow, yields

1
[01 + 02 4- 03 -h 04 - 40o] = P(\, \, r,)

or
00 = i[01 + 02 + 03 + 04 - "' •P(^. \, fl)] (II 44)

^^0 f^^0
This expression can also be derived from —=- and - ^ from a Taylor series
dx' dy

The Taylor series is defined by

J(x) = f(a) + f'(a)(x -a) + f"(a)''—^^~ + / » ^ - ^ - ^ +

When this formula is applied to a grid with sides Ax and A\ (see figure II 25)

Ax

Ay

00 01

Ay

Figure IT 2S Rectangular elements

It yields

0 d'cj) (A\)' f > (Ax)"" d-'cf) (Ax)^

01 = 00 + :; ^^ + 4-
dx' 2 ' dx^ 3' "*" dx'^ 4'
(> c'(l) (A\)' r^0 (Ax)^ d^cj) (Ax)^
rx dx 2 dx^ 3' ax* 4'

60
Addition of these expressions leads to;

d'd) , c'*0 (Ax)"^

0, + 03 = 200 4- ~{Ax)' + 2 - 4 4 f + ... 0.45)
ox dx 4!

a*0 (Ax)"
If all terms from 2 — are neglected, then (11.45) can be written in the following
form;
d'(p 01 -h 03 - 200
(0.46)
dx' (Ax)'
d'4>
In an analogous way an expression for ^-y is obtained:
dp
g 0 _ 02 + 04 - 200
(0.47)
dy' ~ ' (Ay)'

Substitution of (0.46) and (0.47) in the differential equation (L3) leads to:
01 + 03 - 200 , 02 + 04 - 200
= ^(-^,.v,f,)
(Ax)' (Ay)'

From this expression it follows that:

i^x)' (Ay)' | 0 i + 03 02 + 04
00 = ^.;...;2
2l(Ax)' + ;;7V2i
(Ay)']l - I(A.x)'
T T ^ + '(Ay)
h : ! -nx,y.t,)\ (ii.48)

If the sides Ax and ziy are chosen of the same magnitude zlx = zly = a, then (0.48)
reduces to the previously-found formula (11.44):

00 = i [ 0 i + 02 + 03 + 04 - a'P(x,y,t^)^

a*0 a*
The error made by this approximation is of the order 2 --^ • - -
dx 4!

Expression (0.44) is the basic expression that is used for this finite difference method.
All terms appear linearly in that expression. A set of linear equations will result
when the groundwater heads in the nodal points are related to each other and
to the boundary conditions. That set of equations can be solved by well-known
techniques, resulting in known heads in the nodal points. For points within the ele-
ments, the head is calculated by linear interpolation according to the basic assump-
tion.

61
11 Finite Element Method

In the finite element method generally the flow region is sub-divided into a number of
triangular elements Using an assumption about the variation of the groundwater
head in the elements, the solution can be calculated Generally the function 0(\, i, f)
IS calculated for a moment f = f i If the flow is non-steady, a discretisation in time
IS used and the position of the boundary (e g . a phreatic line) after a step in time is
calculated using the specific discharges at the moment f = f. Then the function
0(Y, J , f) IS calculated for f = ri 4- At, etc
The finite element method is based on a variational principle The basic formulas can
be derived using a mathematical variational analysis Here it is shown that a somewhat
alternative derivation using an energy concept leads to the same result
For groundwater flow an energy flux function can be defined by e = in*q(f> The
dimension of e for two-dimensional flow is energy per unit of length and per unit of
time The parameter m* is given by m* = pi Thus

e = pqicj) (1149)

The dimension of m* is mass per unit of length and per unit of time Note that the
density p has the dimension of mass per unit of area in our two-dimensional case
From the appearance of the specific discharge v in the definition of e it is seen that e
IS a vector function The energy flux per unit of time in a direction n (through a unit
of length with direction / perpendicular to n) is given by e„ where e„ is the component
of e in the direction n Consequently the energy transport per unit of time, dE, through
a line segment dl is given by

dE = e„dl = pgi„(t>dl

,ev+ ^ y dy

Figure II 26 Energj flux through an elementar\ rett ingle

62
The concept used in this Chapter says that in a flow region there will be such a ground-
water head distribution that the total loss of energy transport per unit of time due to
the flow is a minimum. (Of course this is also a variational principle).

In figure 0.26 the energy flux e per unit of time through an elementary rectangle is
considered; e is composed of the following components:

in x-direction: e^ = pi\g(j)
in y-direction: Cy = pVy.gcj)

The loss of energy transport dE, in the rectangle is given by (see figure 0.26):

~ dx dv 4 dydx
dx ' dy

or, after substitution of (11.49) and Darcy's Law and considering the case with constant
density p, constant coefficient of permeability k and P(x, y, f) = 0 (see Chapter 1):

'd'4> d'cj)
py^ dx dy 4- pg<l>k dx dy
dx) \dy) J

The second term of this expression is zero (see (1.4)). Thus the expression becomes:

(10 Y /a0^2"
pgk dx dy

For the whole flow region R the loss of energy transport per unit of time E is then
given by:

50 V fdcj)^''
-4* = pgkj\R dx dv (11.50)
(LX dy

If a quantity A is defined according to

A*
A=
2pg
then (11.50) becomes:

dx dy (0.51)

63
This expression is well-known in literature (see, for example, Verruyt, 1970)
Expression (0 51) is used to solve groundwater flow problems According to the
mentioned energy concept, there will be such a groundwater head distribution that
the total loss of £ is a minimum which occurs if in each point 0 has such a value that

dA
= 0 (II 52)

Approximative solutions are obtained by sub-dividing the flow region in a number

of elements An assumption is made about the variation of the groundwater head in
an element and the unknown heads in the nodal points are calculated by application
of (II 52) It will be shown that this leads to a set of linear equations
The solution of this set of equations gives an approximative solution of the problem
Generally triangular elements are used (see figure II 27) and the variation of the
groundwater head in an element is assumed to be linear in \ and \, see figure II 28

hj^'r^s^

2 iy
X

^ 1 (xi-y^)

Figure II 27 Triangular clement

The groundwater head within an element is given by

0 = «! V 4- «2V + ^3 (II 53)

The constants Oj, 02, «3 are chosen in sucha way that in(\,, ii), (xj, i2)and(x3, v-,)
the heads just are respectively 0,. 02 and 03
The constants a are given by

«1 = (^101 + ''202 + ''303)/^

ai = (Cl01 + ^202 + C303)//l (II 54)
a, = (^101 4- d2(t>2 + di^i)/A
where

64
 ^1 = >'2 - ^3
''2 = ^3 - .Vi
^3 = y i - yi
Ci = .X3 — X2
C2 = X, - X3
(0.55)
C3 = X2 — .Xi
dl = X j y j - X3y2
^2 = ^3>'i - x^y^
ds = ^ 1 ^ 2 - ^2^1
^ = ^i(>'2 - >'3) + ^2(^3 - yi) + X3(yi - y2)

ixgyz)
0 y ("ivi)

Figure 11.28 Linear variation of groundwater head m an element

To determine the contribution to the quantity A (see (11.51)) of an element, the partial
derivatives of 0 with respect to x and y are obtained from (0.53):

50 c50
(0.56)
ox

After substitution of (0.53) and (0.56) in (0.51), the following expression is obtained
for the contribution of the element with number /:

Aj = \ki\R[ai + a^dxdy (0.57)

where R^ denotes the area of this element.

As ai, f/j, c/3 are constant within an element, (11.57) becomes;

Aj = lKa\ + aDW^dxdy

65
 The integral ^^^dxdy is equal to the area of the triangular element;

lzll/2.

Where A is given in (11.55).

Then the mentioned contribution of the element y is given by:

k\A\
(a,' + a2') (0.58)

According to the energy concept, the correct head distribution is such that for each
head 0, (in nodal point with number i) the condition holds that A has its minimum.
According to (0.52):

50,. ^^50,

where ^ denotes summation over all elements of which 0,. is an angular point.
Considering that o,, 02- "3 are functions of the nodal heads (see (11.54)) this expression
can be written in the following form, using (0.58):

da I da2
= 0
4

or, considering that /c is a constant:

ca. 4- a. ca-i (0.59)

dcj^, ^ 50J
The derivatives that appear in this expression are obtained from (0.54). For the
nodal point /' = 1 that is used in the following as an example they are:

5ai bi 5^2 ft
501 A 501 A

Then (0.59) becomes:

foi f,
(11.60)
I A A

Substitution of (0.54) in (O.60) leads to;

66
 (b^'cpi + b,b24>2 + ''1^303 + ' • i ' 0 i +
I
+ CiC2 02 4- CiC3 03)i = 0 (11.61)

In other form (0.6I) is given by;

f ; [ F i 0 i + F202 + ^303] = 0 (11.62)

where;

P,=p^|(V+Ci^

Pi = r , | ( ^ i ' ' 2 + '^'I'^z)

P 3 = ^ - | ( ' ' l ^ 3 + qC3)

and b^, b2, b^, c,, C2, Cj, being given by (11.55).
The expression (0.62) is a linear equation in the head of the nodal point 1 and the
heads of the surrounding nodal points, as shown in figure 0.29.

Figure II 29 Elements arond a nodal point *

The condition (0.62) for all nodal points leads to a set of linear equations where the
unknowns are the nodal groundwater heads. The boundary conditions have to be
taken into account. The set of equations leads to an approximative solution of
the problem. For points within the elements, the head is calculated by linear inter-
polation according to the basic assumption (0.53).

67
Ill Analytical Function Method
12 General Description

In part III a calculation method is described for two-dimensional groundwater flow

problems, using complex functions Anisotropy is taken into account according to
the theory that is mentioned in Chapter 3 (simple geometric transformation and
defining a fictive coefficient of permeability) Inhomogeneities with respect to the
coefficient of permeability or the density of the fluid are accounted by sub-division
of the flow region in corresponding sub-regions that all have a constant coefficient
of permeability and density
The sub-regions are coupled in the calculation (see Chapters 14 and 16) by means of
special connecting conditions at the separation lines

In the following text the flow is considered in regions or sub-regions that already
have been transformed for their anisotropy, and are then further considered as if they
were isotropic So in Part III anisotropy is not mentioned further in formulas In the
computer program at the end of Part III it is accounted according the theory of
Chapter 3
Only pure two-dimensional flow is considered here Multi-layer systems where in
each aquifer the flow is assumed to be two-dimensional and between the aquifers
there are semi-pervious layers, are semi-three-dimensional They are not dealt with
here
According to the theory mentioned in Chapter 1 two-dimensional groundwater
flow in an isotropic region is given by

d'(t) d'cp P(x, y, r)

^'^Jy'^ k

The analytical function method (A F M ) is based on the calculation of flow patterns,

which means that non-steady flow is calculated as the non-steady behaviour for a
sequence of steps in time So there is a discretisation in time At a particular moment
the flow pattern is calculated and then from the specific discharge distribution a new
position of the boundary is calculated after a step in time (for example the changed
position of an interface) Then the flow pattern is calculated at the following moment
and so on (see Chapter 16) The non-steady character of the flow is accounted in the
boundary conditions The differential equation for the flow pattern is

5^0 5^0 _ P(x, v)

'd^^'dj'~ k

70
The term P(x. v) represents the abstraction of water per unit of time and per unit of
area in the interior of the region at the considered moment If such an abstraction
area reduces to zero where the product of P(\, y) and the area remains constant, then
there is a sink in the region
In two-dimensional flow problems supply or abstraction is generally concentrated in
sinks or sources For cases with a line abstraction (drain) a better generation of the
flow pattern is usually made by assuming that line as an equipotential line rather
than as a line where the specific discharge per unit of length is constant (although
that can be accounted by the formula of a distribution of sinks)
It is shown in Chapter 14 (dealing with the boundary conditions) and in Chapter 16
(concerning the procedure used in solving groundwater flow problems) that the storage
alteration along the boundaries need not be accounted in a supply or abstraction term
in the differential equation when flow patterns are calculated The storage alteration
is accounted when the position of the boundary after a step in time is calculated
If only abstraction or supply at sinks and sources is present, then for the whole
region (except these singular points) the flow is described by the Laplace differential
equation

5x'' dy

Complex functions are very suitable for the solution of flow problems that satisfy the
Laplace equation (III 1) The real and imaginary parts of any analytical function
satisfy the Laplace equation (see Appendix 1) This means that an analytical function
that satisfies the boundary conditions at all points of the boundary is the exact
solution of the flow problem A calculation method that uses the complex potential
Q = 0 + iP (see Chapter 2) has the advantage that the solution also contains the
stream function.
For the analytical function method (A F M) analytical functions with degrees of
freedom are used so that a flow pattern is generated that satisfies the boundary
conditions at a number of points of the boundary Because generally such a solution
does not satisfy the boundary conditions at all points of the boundary, the solution
IS not exact However, because the complex functions are analytical, the solution
satisfies the Laplace equation everywhere except at singular points Therefore the
solution IS exact within an approximative boundary The solutions obtained in this
way are different from those of the numerical methods like the finite difference method
and the finite element method (see respectively Chapters 10 and 11). In those methods
a solution is found that is approximative over the whole region

71
13 Composition of the Solution

For each sub-region there is an analytical function that generates the flow in it, and
IS defined in the sub-region and on its boundary
The presence of sinks and sources can be accounted by addition of the well-known
logarithmic expressions for sinks and sources (according to the principle of super-
position (see Chapter 6))

The general solution for the flow in a sub-region has the following form

Q(z) = Q2(=) -H Qi(z) + Qo (III 2)

where

Q2(z) approximative part,

i3,(z) exact part for sinks and sources, and
QQ reference constant

The expression Q(z) = Q2(:) + ^i(-) is the basic solution of the flow problem The
complex constant QQ IS a reference constant that produces a translation of the values
of the potential and the stream function The use of a reference constant is allowed
because the flow pattern does not depend on absolute values of the potential and the
stream function but on potential and stream function gradients The complex reference
constant can be used to define the potential and the stream function at a point, which
then becomes a reference point and the complex potential is defined with respect to
that point
It IS not neccessary that there is one and the same reference point for the stream
function and the potential There may be two different points The reference con-
ditions with respect to 0 and P in two points then yield the values of the complex
reference constant if the flow problem is otherwise defined
The complex function Qi(z) generates the flow due to sinks and sources in the flow
region
By the principle of superposition the complex potential due to sinks and sources is
given by

Q,(z) = l^Un(z-z) (10 3)

j-i 2n

where

72
 m : number of sources and sinks
Qj-. discharge of sink (Q > 0) or source (Q < 0)
Zj ; Zj = Xj 4- iyj'. position of sink or source

The complex function Q2(-) is an approximative function that is superimposed on the

flow according to sinks and sources.
The function Q2(:) is defined in such a way that at a number of points of the boundary
the boundary conditions are satisfied (or the connecting conditions with other sub-
regions). To achieve this, the function Q2(-) contains a number of degrees of freedom
that are so defined that the boundary conditions are satisfied by the solution
Q = Q2 + Qi + QQ. For the choice of an appropriate function Q2 there are many
possibilities.
It is essential that the function Q2 is an analytical function because only then does it
satisfy the Laplace equation. In addition, the appearance of the degrees of freedom
in the formula has to be such that the process of calculation is as simple as possible.
The most simple suitable form is used when the degrees of freedom appear linearly
in the solution;

^2(r) = i y, F,(z) (01.4)

j=l

where:
n : number of terms of iJj
jj : constant
Fj-. analytical function of z

The degrees of freedom 7^- in fi2(-) "^^y be real, imaginary or complex. The use of
complex yj has the advantage that there are two degrees of freedom per term yjFj(z)
This reduces the amount of calculation work because a maximum number of degrees
of freedom is used with a minimum number of complex functions Fj(z). Therefore
here the choice is made using degrees of freedom of complex type.

73
14 Boundary Conditions

14.1 General

In this Chapter the boundary conditions are drawn up in a form that is convenient
for use in the analytical function method (A F M ) Expressions are derived from
formulas for a 'general boundary' by varying some of the parameters
The 'general boundary' is shown in figure III I There is a thin sflt layer between two
sub-regions, and the density of the fluids in both sub-regions may be different Para-
meters for one of the two sub-regions are denoted by the subscript c The resistance
of the silt layer is c^, and its thickness is set to zero

Figure III I General boundary

The position of a point of the silt layer is denoted by the real variable / and by l^ for
the other sub-region The positive directions for / and /^ are opposite, having been
chosen in such a way that an anti-clockwise rotation is carried out when the whole
boundary of a sub-region is followed
In the situation shown in figure III I there will be a flow through the silt layer
Application of Darcy's Law to an element of the silt layer yields

dP -4> + 4>*

or
0-0,* + ./,^ = O (10 5)
cl

74
The fictive groundwater head 0^.* is a groundwater head that is derived from the
actual 0^. The relationship between (p* and 0^, is the following well-known ex-
pression (which expresses the relationship between two groundwater heads that
hold the same pressure at a point with height y);

, * Pc, Pc - P
4>c = - 0 . y
p p

Then (111.5) becomes;

, Pc , ^ S'P P - Pc ,,„..
0 0c + Cs ^ 7 = y I0.6
P dl p

In addition, the continuity condition has to be satisfied;

dP /dP\

Expression (10.7) is only relevant at the separation line between two sub-regions.
The conditions (01.6) and (III.7), if relevant, arc sufficient to arrive at an exact solution
if they are posed at all points of the boundaries. Here, however, they are posed at
selected points of the boundary ('boundary points'). The accuracy of the solution can
be improved by increasing the number of these boundary points, as well as by using
the condition that the expressions (10.6) and (I0.7) remain valid in the close neigh-
bourhood of the boundary points (then the variation of parameters in the direction /
and resp. /^ are considered). Here the choice is made to use also these 'derivative
conditions' for the boundary points. The expressions with respect to the variations
in the direction / are given by (where it is mentioned that the positive directions for
/and I,, are opposite):

d^_p^d^ d'W _p - p, dy
dl p dl "^ ^' dl' ~ p dl
or:
^A++'H-^]
^' fit]^c:-^
+ ^ = = fLU^ ^ ''^--^ ("1-8)
dl p[di). ^' dl' p ~ dl
and:
d'P 5 /dP
'W^TiydT
or:
d'P /d'P
2 t .,2 , = 0 (I"-9)
WVdi
75
14.2 Specific Expressions to be used

Equipotential line with silt layer

In Chapter 4 it was mentioned that along an equipotential line the groundwater head
is constant. Here the situation is considered where there is a thin resistance layer
between the equipotential line and the flow region, as shown in figure I0.2;

Figure III 2 Equipotential line with silt layer

This situation corresponds to many practical situations where thin silt or clay layers
are present at the bottom and talus of canals, lakes, etc.
The expressions to be derived here are found from (III.6) and (10.8) by putting /) = p^
and 0^ = 01 (which is a known constant). Then the result is:

dP
0 + c, 0. (OLIO)
5r
50 d'P
and: (10.11)

dQ
In calculations Q and — are used instead of 0. P and their derivatives with respect
dz

to /. Therefore (III. 10) and (III. 11) are written in complex form:
dQ dz\
Re {Q (10.12)

76
 idQ 1 dz d'Q
Re< — k Tl ic, iC = 0 (0LI3)
dl' Iz'
[dz

Stream line boundary

In Chapter 4 it is mentioned that a stream line boundary is impervious, see figure IIL3:

OO

Figure III.3 Stream line boundary.

The expressions are found from (III.6) and (III.8) by substitution of the condition that
the resistance of the silt layer is infinitely large:

00

Division by c^ in (I0.6) and (I0.8) yields:

0 Pc 0c , f""^ P - Pc y
P c, dl P Cs

1 50 p, (dcjA d'f p-p, \ dy

and 1 — 4 ,=
Cj dl pc,\dlj^ dr p c^ dl

For Cj -> 00 these expressions become:

dT
0
dl

d"p
and = 0
'df'

77
In complex denotation they are given by:

idQ dz]
Im < ^ =0 (10.14)
^dz dl

dQ d'z d'QfdzV]
and Im<^-- - y + - - , - .,,1 > = 0 (III. 15)
[d: dl' dz' \dL

Seepage line with silt layer

In Chapter 4 it is mentioned that at a seepage line the groundwater can freely leave
the soil and run ofT. Generally the thickness of a seepage layer is very small, and then
the pressure at the seepage line is equal to the atmospheric pressure, which is a
reference pressure, set to zero. Using the definition of groundwater head, the condition
for a seepage line is:

0 =y

Figure 111.4 Seepage line with silt layer

So at a seepage line the groundwater head varies in a prescribed way. In this study
the general case of a seepage line with a silt layer was considered, see figure 01.4.
If the resistance of the silt layer is set to zero in the expressions that will be given, the
conditions are obtained for a common seepage line.
The seepage line expressions can be found from (III.6) and (III.8) by the substitution
oSp, = 0.
This yields;

dP
0 + c, ~dl

78
and
50 d"F dy
Ti^^'~di' ^Ji
These expressions are given in complex denotation by

ff2 dQ dz}
H'k-''^i:.-dl\ = ''^^'^ ("^1^^
and
dQ 1 dz d'z d'Q
Re } = lm{|} (01,7)
k dl dz' J,
It should be noted here that generally the position of a seepage line is known, as it
follows from the geometry of the flow region
However generally the upper point of the seepage is not known this point is also a
point of a phreatic line, the position of which is usually not known previously So the
problem of the unknown upper point of a seepage line has to be considered as a
phreatic line problem

Phreatic line
In Chapter 4 it has been mentioned that at a phreatic line the pressure is equal to the
atmospheric pressure (that was set to zero) Using the definition of groundwater head,
the following condition was derived

0 =v

So a phreatic line can be considered as a line at which the head varies in a prescribed
way
If the position of the phreatic line in steady flow is known and there is no precipitation,
the stream line conditions (III 14) and (III 15) may be used If there is precipitation
the condition should be used that at the phreatic line P varies so that its variation per
unit of length in horizontal direction corresponds to the precipitation In A F M
50 d\
calculations the condition 0 = \ and the derivative condition— = are used be-
5/ dl
cause those conditions may be used for steady flow as well as for non-steady flow and
are independent of the presence of precipitation
Generally the position of the phreatic line is not known previously The steady
position of the phreatic line is calculated from the non-steady behaviour where
other boundary conditions are invariable After a number of steps in time it is seen

79
that there is practically no further movement of the phreatic line In the solution
It IS seen then that the specific discharge components normal to the phreatic line have
become very small and the stream function variation along the phreatic line is very
small too, if there is no precipitation In the presence of precipitation the stream
function vanes so that this variation just corresponds to the precipitation
The precipitation is taken into account in the following way
At the moment f = 0, the flow pattern is calculated, using a known or assumed
position of the phreatic line Displacement of the phreatic line in a step in time is
calculated using the specific di-^charges along the phreatic line The rise according
to the precipitation has to be superimposed The rise (or fall) of the phreatic line
also depends on the storage coefficient p of the soil The effect of the precipitation N
in a step in time At is

NAt
Ay =
P
A combination of the effects of the precipitation and specific discharges leads to the
position of the phreatic line after one step in time (see Chapter 16)
The new position of the phreatic line is then used to calculate the problem again, and
so on

Figuie HIS Phreatic line

The phreatic line expre-^sions to be used in flow pattern calculations can be found
from (III 6) and (III 8) by the substitution of p^ = 0 and c^ = 0, which yields

0 =3
and
50 dy
~di^~di

80
In complex form these expressions are given by.

Re<;|j. = Im{z} (III 18)

and
Re = Im (III 19)
dz k dl

Interface
It IS mentioned in Chapter 4 that an interface is a separation line between two fluids
with dilTerent densities In steady state an interface is a stream line If the steady
position of the interface is known, the expressions (III 14) and (III 15) can be used
In any case (steady or non-steady), the condition holds that the pressure is equal on
both sides of the interface, it has already been shown that this can be expressed by the
following condition

P ^ P C ^ . ^ P A (III 20)
P P
So an interface can be considered as a line along which the groundwater head varies
in a prescribed way with the height y Here (III 20) is chosen to be used because it
holds for steady as well as for non-steady flow Generally the position of the interface
IS not known previously The steady position of the interface is calculated from the
non-steady behaviour where other boundary conditions are invariable After a
number of steps in time, the interface will be practically at rest Then it is seen in the
solution that the normal components of the specific discharges at the interface have
become very small The calculation procedure starts from a known or assumed posi-
tion of the interface, and then the flow pattern is calculated The specific discharges
at the interface are used to calculate the interface position after one step in time Then
the problem is calculated again, and so on

Figure III 6 Interface

81
Two interface expresions are found from (III 6) and (III 8) using the condition that
c, = 0, yielding

A. Pc A P - Pc
0 - - 0c = y
p p
and
^ Pc^ / ^ 0 \ _ P - Pc SV
dl p \ dl J, p dl

In addition, the continuity condition (III 7) and its derivative expression (III 9) hold

cP (cP\

and
d'^V (d'P\

-^' - b" j< ^'

(10 21)

In complex form, these expressions are given (after multiplication by p) by (10 22)

(III 23)

(III 24)

\dQ dz (dQ cz\ 1

\dQ d'z (dQ r'z\ d'Q(dzy (d'Q(dzy\]

In the special case that iheic i> a stationai> fltiid on one side of the inleiface, then
50
01 = 01 = constant and -^ = 0. the conditions (III 23) and (111 24) are not relevant
dl
as these formulas are continuity conditions for the case with flow on both sides of
a sepaiation line between sub-regions In that special case the interface is not a
separation line between sub-regions
The expressions (III 21) and (111 23) then become

82
 Re jp-^> = (/; - / y J Im{z} 4-p,0i

and:
(p dQ dz] , ^ \dz]

''H*57 4^<''-''<""'y
Inhomogeneity line with silt layer
An inhomogeneity line separates two sub-regions with dilTerent coefficients of
permeability or anisotropy (direction or magnitude). The general case discussed
here is that in which there is a silt layer at the inhomogeneity line. This agrees with
many natural situations.

Figure III 7 Inhomogeneity line with silt layer

Two boundary condition expressions are found from (III.6) and (III.8) using the
condition that p = p^.. This yields:

d'P
0 - 0, -h c, — = 0

and:
50 /50\ d'W

In addition, the continuity condition (III.7) and its derivative expression (III.9) hold;

83
 dT (dP

and:
d"P /d'P
= 0
dl' dl'

These four expressions are given in complex form by:

Q /Q dQ dz]
Re
k \k

dQ 1 dz . d'z 1 dQ dz d'Q
Re
k^~"'W k~d^ ~dl dz' V-
-- 0(111.26)

(dQ dz (dQ dz\ ,

(IIL23)
^"te57n^Wci='
(dQ d'z (dQ d'z\ d'Q(dz\' (.
'""{i^w-y-d^wj/i^Ksij -['1(|riH« im.n

SUMMARY OF BOUNDARY CONDITION EXPRESSIONS

Equipotential line with silt layer

Q dQ dz]
(10.12)
• ' • ' • ' ^ 5 / 1 = '^' i

„ f'>^ [1 dz d'zl d'Q r f^A'i

"'h [k dl '''di'j dz' hUJJ (10.13)

Stream line boundary

Im dQ dz (10.14)
dz dl

(dQ d'z d'Q(dz\'

Im< :r H -\ = 0 (01.15)
[dz dl' dz' \dl

84
Seepage line with silt layer

\Q dQ dz]
(0L16)

\dQ "1 5z . d'z~ d'Q

Re ic, Im (10.17)
\dz [k dl "^ dl'] "d? dl

Phreatic line

Re
{^-« (I0.18)

(dQ 1 dz\ (dz

Re (nL19)
1^7

Interjace

Reif-fUH*--»-,)•".!-') (IIL21)

(p dQ dz (p dQ 5z\ 1 , ^ f5z
(IIL22)
"' i ;i7 aJ n ^ ^ 7 i , h ' " - " • " " ' Ti
(dQ dz /dQ dz
= 0 (I0.23)
^"^(d7 57"^ Vd7 5/

Im
dQ d'z
dz dl'
dQ d'z\
\dz dl'J,
d'Q
m
dz' \dlj
d'Q
'd^Vdi
/dzY
= 0 (III.24)

Inhomogeneity line with silt layer

Q dQ dz
Re ic. = 0 (I0.25)
J7 57

(dQ '1 5z . d'z I dQ dz d'Q

Re = 0(nL26)
k Ji'^-'^W ^^Jlh di 1^'

85
 (dQ dz (dQ dz\ ]

, (dQ d'z (dQ d'z\ d'Q(dz\' (d'Q(dz\'\]

15 The Approximative Function Q i (z)

15.1 General

It IS mentioned m Chapter 13 that the following form was chosen for the approxi-
mative pait of the solution

fl2(^-)=Z).^» ( I " 4)
j=i

where
n number of terms in Q2(:)
,1 complex constant
Fj(z) analytic function of z

Theie aic many possibilities for the choice of the function Fj(z) Analogous to a
sub-division for functions of a real variable (see Rektorys, 1969), a difference is made
between algebraic functions and transcendental functions Algebraic functions are
polynomials (rational functions) or quotients of polynomials (fractional rational
functions) Functions that are not algebraic are transcendental functions in which
distinction is made between elementary and higher transcendental functions Exam-
ples of elementary tianscendental functions are z°, tugononietric h^peibolic and
exponential functions and the related inveise functions Highci tianscendental func-
tions are defined by dillcrential equations (foi example Bessel functions) oi integials
(e g , elliptic integrals)

15.2 Algebraic Functions

It was mentioned in Section 15 1 that algebraic functions are polynomials or quotients

of polynomials A polynomial contains its degrees of freedom in the form of con-
stants If the quotient of two polynomials were used, the degiees of freedom would
not appear linearly in the expressions for the boundary conditions thus making the
calculation more complicated For example the function

fo,z 4- O2-

87
has the first derivative

(fc.z + b2:')(ai + 2^2^) - (aiz + a2z')(b, + 2b2z)

^^''" (biz + b2z')'

As constants Oi, a2, b^ 62 do not all appear linearly in these expressions, the use of
this kind of function therefore, would result in complicated non-linear boundary
condition expressions So the polynomial is a better choice

Q2(:) = a,z 4- a2z' + a^z' + + a„z'' (HI 27)

In this case there is not a constant in the polynomial because in the solution there is
already a constant (Q^)
Expression (III 27) agrees with the previously chosen form (III 4)

Q2(z) = t ,/,(z)
J 1

This IS seen by writing (III 27) in another form

^2(=)=ia,z^ (0128)

Generally, (III 28) may be suitable the function Fj(z) = z^ is analytic

When (III 28) is combined with the general solution (III 2) and the boundary condi-
tions (see Chapter 14) a set of linear equations is obtained in the unknowns aj and
QQ After solving this set of equations, the solution is known
In general it may be a disadvantage of polynomials that there may arise numerical
problems when there are many terms (then Izl^ becomes very large)

A special kind of polynomial is the well-known Lagrange interpolation polynomial,

which in complex form is given by the following expression

" " (~ - z)
02(z) = I y. n ^r-^.
j - l , - l ( Z j - Z , )

This Lagrange polynomial has the special property that the complex constants ,j are
just the values of the function in the points z = Zj

^2{=j) = 1,

In principle the interpolation polynomial of Lagrange may be used for the solution

88
of flow problems. There is, however, an important practical disadvantage; Kosten
(see Kuipers en Timman, 1969 chapter XII) notes that the Lagrange polynomial is
not convenient for practical calculations because of the considerable amount of
calculation work necessary for evaluating values of the function.

15.3 Transcendental Functions

It is mentioned in Section 15.1 that there are elementary and higher transcendental
functions, the latter being defined by integrals or differential equations. Because the
amount of calculation work required for evaluating values of functions is generally
much greater for higher transcendental functions than for elementary transcendental
functions, only the latter functions are discussed.
In principle, there are many possibilities that may be used for Fj(z) in (III.4);

Q2(=)-tyjFj(z)
.i=i

Some examples of the use of elementary transcendental functions in the function

^2(2) are;

Q2(z) = t yj sin (z - z)
j=i

^2(2) = I yj[sin(z - Zj) + cos(z - z^.)]

j=i

Q2(:) = t yjH= - Zj) (10.29)

where Zj are the boundary points (points where the boundary conditions are posed,
see Chapter 14). It is not neccessary that the positions of the boundary points (z^)
appear in the expression for Q2(z). The form (III.29) with (z — Zj) was chosen for its
analogy with the appearance of (z — z/) in the interpolation polynomial of Lagrange.
Although in principle the expressions mentioned for Q2(z) and many others might be
useful, a further selection is made between elementary transcendental functions:
When the computer time needed for the evaluation of values of some elementary
transcendental functions are compared, it is seen that the time needed for complex
logarithms is about 25",, less than for complex sines and complex cosines (see IBM
Systems Reference Library 360S-LM-501). This is shown in the following table,
where the time needed for a complex sine evaluation is set to 100%;

89
 Function Name Time ('%)

sin(z) CSIN(Z) 100

cos(z) CCOS(Z) 100
exp(z) CEXP(Z) 87
ln(z) CLOG(Z) 75

On account of this fact the use of complex logarithms is preferable For example,
there could be used (III 29)

Q2(z) = i,^\n(z-z) (10 29)

j = i

For the case that ,, is complex, this expression represents a flow due to sources and
sinks with strength Xj and vortices with strength fij (where ij = oij + ifi^ in the
points z, In principle, (III 29) can be used to approximate flow patterns, but this
has the disadvantage that the logarithmic function is singular at the boundary
points Zj That difficulty can be avoided by choosing the points z^ outside the flow
region and posing the boundary conditions at other points on the boundary Although
this can be done it is not efficient because points have to be defined at the boundary as
well as points outside the region In addition, when the points z^ of (III 29) are chosen
close to the boundary, many terms would be needed to prevent inaccuracy caused by
the singular behaviour of the complex logarithms Then the calculation would require
a considerable amount of work
The situation with many sinks sources and vortices can be replaced by a system of
continuous distributions of sinks sources and vortices over the boundaries The use
of those distributions has the advantage that the expression for Q2(z) is not singular
in the boundary points and a flow pattern is generated that is more smooth in the
surroundings of a boundary point
The boundary points are chosen at the middles of the boundary segments over which
there is a distribution of sinks sources and vortices
To ensure the smoothness of the flow pattern at the boundar> and for efficiency the
boundary segments are chosen in sequence without gaps Then only the end points
and the properties of these boundary segments form the boundaiv input for a
calculation program
In the following the distribution strength has been chosen to be constant for each
boundary segment
The expression for the complex potential due to a distribution of complex strength
IS derived first The potential due to a sink (or source) and a vortex in the point z^ is
given by

90
 Q(z) =2^- ln(z - zJ

where Q is the complex strength.

When the quantity Q is distributed over a line segment v, \-2 on the .v-axis according to;

dQ
" =dx

where q = r -\- is, then the complex potential is given by:

After integration, and neglecting the integration constant, it can be written as:

Q(z) = - ^ [(z - X2)Hz - X2) - (z - .vi)ln(z - .v,)) (III.30)

Zn

When the end points of the segment are chosen at arbitrary positions in the complex
plane, the complex potential can be simply derived from (III.30) by a translation b
and a rotation 0 of the coordinates as shown in figure 10.8.

z* = ze'" + b

where h is complex.
The points z^ = .v, and Zj = .Xj have become the following positions in the trans-
formed plane:

Zi* = x^e"' 4- b and

Z2* = X2e'" + b

where;

0 = arg(z2* - z,*)

A constant c* per line segment z*z2 is defined by c* = exp( — (0) or:

c* = e x p ( - i arg(z2* - Zi*))

91
 Im b

•^-x

Re b

Figure III.8 Rotation and translation.

Otherwise written, using (z2* - z,*) = Izj* - z,*l exp(iarg(z2* - Zi*)) (10.31)

|. * _ *|
* 1-2 -1 '
(-2* - -.*)

When the inverse expressions (for z, Zi = .Vi and Z2 = .Vj)

z = c*(z* - b)
x,=c*(z,*-b)
X2 = r*(z2* — b)

are substituted in (III.30), there comes:

Q(:*) = - ^- {(c*z* - c*z,*)ln[c*z* - c*Z2*] +

2n
- (c*z* - c*zi*)ln[c*z* - c*Zi*]}

So the complex potential due to a distribution of complex strength q over a line

segment ZiZj is given by;

Q(z) = -^{(z- Z2)\n[c(z - Z2)] - (z - zi)ln[c(z - z J ] ] (10.32)

271

where c is defined by c = - ^
(-2 ~ '^i)

The use of (10.32) in the formula for the approximative part Q2(z) of the solution

92
yields (where each distribution is denoted by the subscript j)

^2{z) = i - ^ ^ {(z - Z2,)ln[c/z - Z2,)] - (z - z i > [ c , ( z - Zi,)]}

(III 33)

, Iz2, - ZiJ
where ''J = 7"^ \ (III 34)
(Z2j - -ij)

The constant — l/27r is not relevant in (III 33) because it can be accounted in the
constants qj Because of the visibility that qj represents a distribution strength, the
form (III 33) with —qJ2n is chosen to be used in the general solution (III 2)

In the literature the use of singularities and distributions of singularities is well-

known Lamb (1932) showed that a flow that satisfies the Laplace differential equation
can be generated by an appropriate distribution of sinks and sources or vortices over
the boundaries
In aerodynamics, applications are given by Von Karman (1927) and Mc Nown and
Hsu (1949) Von Karman calculated the pressure distribution on the surface of air
ships Therefore he located distributions of sinks and sources over a number of line
segments at the axis of symmetry At a corresponding number of points on the air
ship surface, he posed the condition that the normal velocity is zero, which resulted
in linear equations with the unknown distribution strengths The solution was used
to calculate the pressure distribution on the air ship surface
Mc Nown and Hsu generated an inlet flow between two walls, using separate vortices
outside the flow region The way of solving the problem was analogous to that of
Von Karman
De Josselin de Jong (I960, 1969) used singularity distributions for generating flow
through a porous medium with varying properties of fluid or porous medium He
located singularity distributions at interfaces These distributions were chosen in
such a way that the problem reduced to a homogeneous problem that could be solved
by usual methods The solution of a flow problem then consists of two parts one part
that IS the flow due to the singularity distributions and a second part that makes the
solution (including the first part) satisfy the boundary conditions

93
16 The Calculation of Flow Problems

It \Vas mentioned in Chapter 13 that the general solution, generating the flow pattern
in a sub-region, is given by:

Q(z) = Q2(z) + Q,(z) + Qo (111.2)

where QQ is a reference constant, fii(z) is the exact part that generates the flow due
to the sinks and sources, and Q2(z) is the approximative part that contains para-
meters that are used to meet the boundary conditions at a number of points.
Substitution of (I0.33) and (III.3) in (I0.2) yields:

«(--) = 1-^4' «-• - =^Mcj(= - Z2j)] - (-- - M,)ln[o(z - Zi,)]} +

j = i

m Q
+ ^ f i | n ( z - z , ) + f2o (III-35)
j=i 2n

where c^ is given by (III.34):

I22,- — z, .-I
- - " " (III.34)

The constant q^ = r^ 4- is^ is the complex distribution strength over the fine segment
from z,, = .Vij + (Vij to Z2, = X2, + (}'2r

The first and the second derivatives of 0(z) with respect to z also appear in the bound-
ary condition expressions mentioned in Chapter 14.
These are found by differentiation of (10.35);

dQ(z) _" q,c . n, , V 2,

dz -,^1 2^i..L.,v- -2,, ..^LS^- -^'J'-^4^,2;r(z-z,)
(10.36)

d'Q(z) "
'^'"'^ ^ 1 I ,y -2^ mM\
dz' ->, 2. |(z-z2.) (z-zijj '.?.2.(z-z/ <"•-'')

The complex reference constant QQ in (10.35) permits a translation of the values of

the potential and the stream function.

94
The value ofQo can be defined by two equations The equation with respect to Im{i2o}
may be given by the condition that at a point z^ the stream function is set to zero

Im{Q(zo,)}=0

The equation with respect to Re{i2o) 's a similar equation for the potential in a point
Zo2 however, it is not permitted to choose an arbitrary value of the potential at the
point Zo2 because values of the potential are involved in the boundary conditions
Therefore it is convenient to choose the point Zo2 Jt the boundary Then for the equa-
tion with respect to Re{fio) th^ potential at the point ZQ2 is related to the potential
at the boundary or just outside the boundary (eg, if there is an equipotential line
with a silt layer) This equation has the same form as that of the previously discussed
boundary condition equations (e g, for an equipotential line with a silt layer) Because
boundary conditions are posed at the middles of the boundary segments, the point
Zo2 may not be chosen at the middle of a boundary segment (then the equation with
respect to Re[(2oj would be the same as a boundary condition equation consequent-
ly, the set of equations would be undetermined)
The calculation procedure for solving flow problems will now be outlined The general
solution IS given by (III 35) and a combination of (III 35) (III 36) and (III 37) with the
boundary condition expressions (see Chapter 14) leads to a set of linear equations
The unknowns are the complex distribution strengths i/,(c/j = 'j + '^j) and the com-
plex reference constant QQ SO when there are n boundary segments the result of posing
boundary conditions and reference conditions is a set of (2n 4- 2) linear equations for
each sub-region After solving this set, the complex potential can be evaluated at any
point by substitution of ^^ and QQ in (III 35)
The specific discharges are given by

d0 (dQ] (dQ dz] (dQ]

d0 (dQ] (dQ dz] (idQ] \dQ]

(III 38)

If the flow problem is non-steady where one or more parts of the boundary are
moving (phreatic line or interface), then that behaviour is calculated in the following
way (The expressions hold for a phieatic line Similai expressions hold for an inter-
face but then the storage coefficient p has to be replaced by the effective porosity )/)
Starting from a known or assumed position of the boundary, the flow pattern is cal-
culated in the way described, followed by the calculation of the specific discharges at
the moving boundaries These specific discharges are used to calculate the position
of the boundary after one step in time The alteration of the position of a point in

95
a step in time is given by (using (III 38))

(dQ] At
Ax = Re<-
l dz \ p
(III 39)
(dQ] At

where p is the storage coefficient of the soil and At is the time step size If the moving
boundary is a phreatic line with precipitation, the rise due to the piecipitation has
also to be accounted In Chapter 14 it was mentioned that the rise according to
precipitation A^ is given by

At
Ay = N (III 40)
P

where N is the precipitation per unit of length and time Then the position alteration
of a point of the boundary (phreatic line) in a step in time is given by the sum of
(III 40) and (III 39)

(dQ] At
Ax =
^^\d7] 7
(0141)
(dQ] At
^} Im-^ -} + N
P

After correction of the boundary position the flow pattern after a step in time is
calculated and then the procedure is repeated
Generallv the position of a phreatic line is not known previously, but it is pointed
out in Chapter 14 that the steady position of that line can be found by calculation of
the non-steady behaviour of the flow After some steps in time the boundary will
reach a position that is practically at rest In the calculation the normal component
of the specific discharge at the phreatic line will then tend to zero
If the normal direction is denoted by n (see figure III 9), the specific discharge in
rt'-direction is given by

d0 dP
(III 42)
dn^ 'df

or
dQ
Im (III 43)

96
 . y

Figure III 9 Tangential and normal direction at the boundary

When large steps in time are chosen, the specific discharges at the end of the step in
time may differ considerably from those at the beginning of the step in time. In such
cases the average velocity of the specific discharge over the step in time should have
been used; then some iterations might be necessary. In most cases, however, it will
be more convenient to choose the time step size so that these effects are small.
The time step size that is convenient depends on the kind of problem.

The equations that are used for generating flow patterns and the expressions for
correcting the position of a moving boundary will now be detailed, with the ex-
pressions (10.35), (01.36) and (III.37) being written in a more convenient form:
The following notations are used:

FO,(z) = - -^ {(z - Z2;in[c/z - Z2;] - (z - z.^lnCc-(z h)T>

Fl,(z) = - | i {ln[c,(z - Z2,)] ~ In[c,(z - Zi,)]}

c, 1 1 1
F2,(z) =
^^ ((Z - Z2,) {Z - Z,) (10.44)

GO(z) = X ^ l n ( z - z ;
j = i ^''^

Gl(z) = I Q.
j=i 2n(z - z)

QJ
G2(z) = I
,t^i 2rr(z - z/

97
Then the expressions (III.35), (10.36) and (IIL37) may be given by.

Q(z) = t q,FOiz) + GO(z) + QQ (10.45)

j=i

dQ(z
X q,Fliz) + Gl(z) (10.46)
dz ,

d'Q(z)
= X 9 , F 2 » + G2(z) (I0.47)
j= i

The specific discharges can be expressed by the following formulas (using (10.38) and
(10.46)):

tx= - R e I .//l,(z) + Gl(z)

j = i

(10.48)
Vy = i<^]l qjF\(z) + G\(z)

The expressions (10.41) for the alteration of the position of a point of a phreatic
line can be expressed using (10.46).

Ax= - R e | X <?,fl» + Gl(z)J^

(III 49)
At
Ay I m i l q,F\,(z) + G\(z)\ + N
j = i P

The reference equation Im Q(ZQI) = 0 for the stream function can be written in the
following form, using (10.45);

Im < X qjFOj(zQ) + GO(ZQ) + QQ> = 0. or otherwise arranged (using

gj = ''j + >\)

I
j = i
r^lmlFO^Q)] +s,Re|fO,(zo)l Vo = I m ; - G O ( z j ; (IO.50)

This is a linear equation in r^. s^ and PQ.

It has already been mentioned in this Chapter that the equation for Re [fig] has the

98
same form as a boundary condition equation.
The boundary condition equations are based on the boundary condition expressions
dz d'z
of Chapter 14 in which the terms'— and --^ appear, representing the form of the
dl dl
boundary. For a straight line, the following conditions hold:
dz d'z
— = constant —^ = 0
dl dl'
d'z
In this text only straight line segments of the boundary are used, thus -^ = 0-

This is permitted when the boundary is not sharply curved. A sharply-curved bound-
ary can be approximated by a number of straight line segments.
The following boundary condition equations refer to boundary conditions for the
middle of a boundary segment with number;*, the end points of which are Zi^. and
dz
z,,.. The term — can be discretisized for that segment by:
" dl
^ ~ (^2/ - Zl/)

dl - \Z2, - zj\

Analogous to (IIL34), this can be written with c^., thus;

^ ~ ( ^ 2 / - Z l / ) ^ _1_
dl
or:
1^2/ - Zi/
Cf = j ^ ^T (01.51)
(Z2/ - Zi,*)

U'
Im some of the following expressions Im <--> appears

In a similar way this can be discretised by:

5 z | ^ lm{z2f - Zi^.}
Im-
dl

Further, this is denoted by e^.:

Im{z2/ - zi .
(111.52)
Iz2/ - Zi,J

99
Equipotential line with silt layer
The first boundary condition expression is (III. 12)-

(Q dQ dz,

Muhiplying by k and combining it with (III.45), (III.46) and (01.51) yields;

Re\Y.qjFOj(z) + GO(z) + QQ-,^'''' lq,F^z) + G\(z) = k<P,.

After some rearranging, it can be written in the following form:

I
j= i
r^Re{/fl»} + ..,Im{-Hl/z)} + 0Q =

kc.
= Re{/c(/)i - GO(z) + 1 —Gl(z)} (10.53)
c,.
where;

HI^-) = F O ^ i - / — ' F I j z ) (I0.54)

The second boundary expression is (10.13);

dQ 1 dz d'zA d'Q
r /^z\n
Re
7/7 I Ji~-'''-dl'\ dz' L'HWJ = 0

By substitution of —^ = 0 and multiplication by k, and after combining with (01.46),

(I0.47) and (10.51), there is obtained:

Re X g / l j z ) + Gl(z) X ^ / 2 ; z ) + G2(z) = 0
/ U=i

Some rearranging yields;

t r,Re{H2,(z)}+s;m{-//2/z)] = Re^ Gl(z)-H i - ^ G 2 ( z )

j=i L
(10.55)
where;

100
 //2» = - ^ F l » - , ' ^ F 2 » (IIL56)
Cj. Cj.

Stream line boundary

The first boundary condition expression is (01.14);

(dQ dz]
Im< }=0
^.dz dl

Substitution of (I0.46) and (10.51) yields;

Im I q/ip) + Gl(z) = 0
.j=i

Rearranged, this becomes:

I
j = i
r ; m < ; i F I , ( z ) l + s ^ R e l ' Fl,(z) = Im<| G\(z)\ (10.57)

The second boundary condition expression is (III. 15):

(dQ d'z d'Q(dz\']

.d^z
Substitution of—z- = 0 yields:
dl'

-{S©'1-
Substitution of (I0.47) and (01.51) gives:

1
Im X ^/2,(z) + G2(z)

In rearranged form, this is

n

r , I m ^ - ^ F 2 , ( z ) i + 5,Re|-^F2,(z) = Im G2(z)
1
(10.58)

101
Seepaqe hue with silt la\ei
The first boundar> condition expression is (III 16)

\Q dQ dz
Rsi k, \m\z\
k \lz (I

This expression is almost identical with the expression for an equipotential line with
silt layer (III 12) The term (/>! here is replaced by Im{z} By such a replacement in
(III 53). the first equation for a seepage line with silt layer is obtained, where it is
noted that lm[z} = Re{-iz}

I
j = i
,,Re{Hl^-)} + s ; m { - H l ^ ' ) } •Pn =

ikc,
= Re{-;z - GO(z) + -—Gl(z)} (III 59)

where Wlj(z) is given by (III 54)

The second boundary condition expression is (III 17)

(dQ ri dz d'zl d'Q r (d-Y^1 , 1^'

" ^ 7 7 ~k Ti~ ''UY\ dz' H-^)\ r'-^u
d'-
By substituting ^ ^ = 0 and multiplying by k, there results
di'

^ \dQ dz , d'Q dz\'] , , f5z

R e i - , - ^, - ikc,-rT\ — ] \- = klm< —
dz dl ' dz' \d J I ]dl
The left-hand part of this equation is the same as the expression for an equipotential
d'z
line with sih layer (III 13) with j = 0 and muhiplied by k By putting k Im
dl

dz
in the right-hand side of the equation and noting that Im -^ > is given by c^. (see

(III 52)), the resuU is

r,Rc{H2,(z)} + sJm{-H2p)}
j=i

= Re^kej.- ^ G\(z) + -%G2(z) (III 60)

102
where H2j(z) is given by (10 56)

Phreatic Ime
The first boundary condition expression is (III 18)

R e ^ ^ ^ = Im{2}

Muhiplying by k and combining with (III 45) yields

Re-^ Z g / 0 » + GO(z) + QQ\ = k Im{z}

After some rearranging and noting that k Im{z} = Rs{-ikz} there is obtained

r,Re{FO/z)} + s,Im{-fO,(z)} + 00 = Re{ - ikz - GO(z)}

j=i (III 61)
The second boundary condition expression is (III 19)

IdQ 1 dz] \dz

^'h-z-kTi\-H-di
After muhiplying by k and combining with (III 46) and (0151) and noting that
(• z
Im < , > IS given by e^* according to (III 52), there comes

Re X g/l,(z) + Gl(z) = ke,

-;=i

In rearranged form, this becomes

z
J=I
r^Re<j-Fl»| + s ; m j - - F l » = Rt\ke^. Gl(z)
(III 62)

Interface
The first boundary condition expression is (III 21).

pQ (pQ
Re (p - p,)Im{z}

103
Substitution of (01.45), and using:

(p-p,)lm{z] =Re{-i(p-p^)z}
yields;

Re Z q/O^z) + GO(z) + QQ 4-
.j=i

X q/0,(z) + G0(z) + QQ = Re{-i(p-p^)z}

-j=i

In rearranged form, this becomes

z
J=l
r,Re<j^FO,(z) +i,lm -^FO/z; + -%o +

r,Re<;-|FO,(z)[-f .,Im^FO,(z) -^^o) =

^ j = i

R e < j - i ( p - / ; , ) z - ^ G O ( z ) + f^GO(z) (III.63)

If one of the fluids is stationary, the term

Re •( I — I > is constant

R^<(f]} = Pc<^.

In that case this term is known, and can be placed in the right-hand side of the
equation;

z
J=l
r,Re^J^V0,(z) +s,Im -^FO,(z) ^0

Rc{p,4>c-i(p-P.)=-lG0(z) (III.64)

The second boundary condition expression is (10.22):

JpdQdj^ /pdQdAl Jd^

[k dz dl \k dz dljA ^^ ^'' [dl

104
Substitution of (OL46), (OL5I) and (10.52) yields;

Re^^ - - Z q/l,(z) 4- Gl(z)

\kCf j=i

P i r X q/l,(z) + Gl(z)
+ k CJ*
(P - Pc)ef
Lj=l

In rearranged form this is;

z
J=l
'•.Re<;^f fMz)[ + ^ I m ] - ^ - i F l »
k c,.
4-

z ,,Re^^FMz)}4-Mm{--^FMz)
J=l

= R^\(p- p,)e,. - ^ ^Gl(z) + (^ -i-Gl(z) (IIL65)

If one of the fluids is stationary, one term in (111.22) vanishes:

Re^/'^^^r^^O
\k dz dl

In that case the second equation becomes:

z
J=l
.,Re^^AFl,(z)j + . , I m | - ; ; l F l »

P 1
= Re{(/;-/;,)e,.-^--GI(z) (IIL66)

The third boundary condition expression is (01.23):

(dQ dz (dQ dz'^

^[dz dl'^Kdz dl),]

The first term of the left-hand side of (10.23) is the same as the left-hand part of the
first stream line condition. The second term has the same form, here for the adjoining
sub-region. Using the first stream line equation (111.57) one finds;

J=l
z Im<^ ^F\j,z)\ + s^Rt\-F\lz^ 4-

105
 + (Z r,Im<!-Fl,(z)j + s , R e j - - F l l ' )
vj=l

= Im^--^Gl(z)-(-iGl(z; (10.67)

The fourth boundary condition expression is (10.25):

jdQ d'z /dQ d'z\ d'Q/dzV /d'Q/d^

= 0
^"^{dz 'W'Kdz W),^l?\di) ~ Uz'Xdi

Substitution of—^ = 0 yields:

W'

-^$(f-(S(r-o (01.68)

The first term of the left-hand side of (01.68) is the same as the left-hand part of the
second stream line condition (III.58). The second term has the same form, here for
the adjoining sub-region. Using the second stream line equation gives:

z
J=i L
'•^ImiAf2/z)l + s^.Rej-i^F2,.(z

rjlm ~ F 2 / z ) l + .s-^.Re - - ^ F 2 , ( z ;
\J=i J J

= Ini^_ J^G2(z) + (-^G2(z) (10.69)

Inhomogeneity line with silt layer

The first boundary condition expression is (III.25):

(Q (Q\ dQ dz\
•"^7-7)-'•'•• 77 71^ = "
Substitution of (10.45), (I0.46) and (01.51) yields:

R e j ! ( I ^//O/z) + GO(z) -f floV Q ( Z gjFOj(z) + GO(z) + QQ]] +

k\M j=i

1
' • ^ s - ^ Z ^/l,(z) + Gl(z))|^ = 0

106
In rearranged form, this is:

z
J=l
r,Re<jiHl»| + 5 , I m | - i w i »
^ ^ ^

+ 11 r,Re<j-^FO»j + .v,Im|iFO»
k
•.j=i

(111.70)

where H\j(z) is given by (10.54).

The second boundary condition expression is (10.26);

(dQ "1 dz d'zl /I dQ dz\ d'Q r f^^y]

Uz _k dl ''^dl'j ^ [k "dl Jl)^ ' dz' hU/JJ
Substitution of—. = 0 yields:
dl'
(dQ 1 dz (\ dQ dz\ d'Q(dz\']
^'\Tz k Ji + U 7/7 5/1 - '"^-dAji) I = '
By substitution of (01.46), (10.47) and (III.51) there comes:

Rearranged, this becomes

n

z
1=1
./e^jHy.-) +»,lm ~1H2,(--)

H,?,b''n77;^'''--r'''"'{-fc7;'"''-''
(10.71)

where H2j^z) is given by (10.56).

107
The third and the fourth boundary condition expressions are (III.23) and (10.24).
They are the same as for the interface, and so the corresponding equations are (01.67)
and (10.69).

SUMMARY OF EXPRESSIONS AND EQUATIONS

Complex potential and derivatives:

m = i g/H') + GO(z) + QQ (01.45)

j=i

dQ(z)
Z ^//l,(z) + Gl(z) (10.46)
"dT j = i

cl'Qi-)
-rf = S '//2,(z) + G2(z) (10.47)
dz j= 1

Definitions:

FO,(z) = - -U^(z - Z2,)ln[c> - Z2,)] - (z - zi;in[cjz - z,)-\

Fl/z)=--^^ln[c/z-zj]-ln[cl--zi;]

F2,(z) = --'- ^
2n [(z - Z2j) (z - Zi j

GO(z)= X f - ^ l n ( z - z ; (10.44)
j=i 2n

Gl(z) = Z e.
j=i2;r(z - z)

-%
G2(z) = Z
j = i 2n(z - z)

ike,
H l » = FO,(z)--^Fl> (10.54)

1 ike
H2» = - F l » - - ^ / 2 » (01.56)

108
 _ 1^2; - Zljl
C; (III 34)
i^2, - Zu)

\z2f - Zi/I
CJ. (10 51)
(^2/ - Z,;.)

Im{z2/ - Zij.}
^ 1*
(III 52)
IZ2/ - Zi,.l

Specific discharges:

''x= - R e < ! Z ^//1,(--) + Gl(z

j - i
(10 48)
i;, = Im<^X ^ / M z ) + Gl(z;
U=l

Components of alteration of the position of a point of a phreatic line in a step in time:

At
Ax= - Re<j Z ^ / l , ( z ) + Gl(z)
^; = 1 ) P
(III 49)
At
Av = \m\ j:^q/\p) + Gl(z)\ + N

Equations with respect to QQ

Potential
One of the expressions (III 53), (III 59), (III 61), (III 63) or (III 70)

Stream function

r;m{FO/zo)} + s,Re{FO^'o)} + f'o = Im[-GO(zo)) (0150)

j = i

Boundary condition equations:

Equipotential line u/f/j silt la\er

t <,Re{Hl»} + s ; m { - H l » } + <Pa

109
 R e U 0 , - G0(z)4- - ^ G l ( z ) (0153)

ikc
Z ,,Rc{H2j(z)\ +s;m[-//2^-)] Rei--G\(z) + -^G2(z)
j - i c ,. c ,.
(11155)

Stream line boundai \

J=i
z ';"ij--:^iA-)j + ^.R^|--:fM Im<!--~Gl(z)^ (III 57)

J=l
z '•;"'i:V2.(4 + ^.Rej-^F2,(z) Im _ G2(z) (I0 58)

Seepage line with silt la\er

z
J - l
r^Rs{H\j(z)} +Sjlm{-H\j(z) + 0Q =

ikc,
Re - i z - G O ( z ) + - ^ G l ( z ) (III 59)
^f

z
J = l
,jRc{H2j(z)}+Sjlm{-H2^(z)\

= R e | / c e , . - - G l ( z ) + -'^-^G2(z) (III 60)

Phreatic line

J=l
z ^,Re{FO^-)) + 5 ; m { - F 0 ^ - ) } + 0Q = Re{-ikz- GO(z)] (10 61)

z
J = l
;,Re<;- F\Y.)\ + Sjlm]-^^F\j(z)

ReUt' Gl (0162)

110
Interface

z
J=l
r^Re<|jFO,(z)[ + . y m | - ^ F O » n^« +
+ (Z '•M-i''n=)\ + -\^^\j.FOiz: --<?.) =

Rei-,(.-.Jz.-^^*^.(''^ (10.63)

z '•^'^'^^7;^'^*4^'^'"'{"L-7^^^^-'^
J=l
-h

+ (Z ,Rel?ln,w| + ,i™{-^ln,H
Vj=l

- Re-j(„ ~ p j i y - t' J- Cl(z) + ft' J_ Gl(;) (10.65)

z
J=l
r,Im<|--Flj(z)[ + 5 j R e | l F l / z
c, J

•;m^--Flj(z);> + SjRe<'--Fl,(z)
MZ
^j=i

Im - Gl(z)-(--Gl(z: (I0.67)

'-;'"1;'-^^2;z)j + SjRej-LF2j(z) +

+ z r,Im :-2^2/z)j + .SjRe|--VF2j(z)

Vj=l

= Im^--~G2(z) + I^G2(z) (10.69)

If one of the fluids is stationary (p^ and cj)^ constant) then (10.63) becomes;

z
J=l
r,Re^[Vo,(z)[ + , s ; m | - ^ ' F O » . ^ . =

= Re^p,0,-,(p-p,)z-^GO(z; (III 64)

111
and (10.65) becomes;

z
J=l L
'•jReif-J:f!.(--)}+ ^ ; ' " { - ^ - ; fiAi

= Rc][p-p,)e,,-^^^^G\(z) (0166)

Inhomogeneity line with silt laver

j=l L
rjRe^^Hlj(z) +5jlm -^Hlj(z)
^h
0,
rjRe^--FOj(z)f + SjImj^FO,(z;
\j=i

-^-'^^•^m-:;-'-) (10 70)

z
J=l L
rjRe^iH2j(z)j + 5 j I m | - l / / 2 j ( z )

r,ReiJ-n,(.-)} + y,.{--Lf,j.-:
Vj=l

= Re^--Ul(,)-(-^U,y)^,,_f^C2H (11171)

/-;m<'-^Flj(z)j + SjRe|--Flj(z;
I 4-

1
+ (Z ;m --FI» +5jRe --Fl/z
w=i

Im<^ Gl(z) (III 67)

z
1=1
r,lm<;-^F2(z)| + S j R e i ^ F2(z)
+

r Im , F2j(z)^+s,Re<j-^--,F2j(z;
Vj=i ' c/

Im^-J^G2(z) + (-^G2(z) (III 69)

112
17 Special Cases of Flow in a Half Plane

In the foregoing Chapter distributions of sources, sinks and vortices were used to
generate flow patterns in regions that do not extend to infinity. In this Chapter it is
shown that those distributions are also useful for two classes of half-plane flow,
having the advantage that then it is not necessary to confine the region by an assumed
fictive boundary. When the flow is generated in the whole half-plane, the influence of
the local boundary conditions can be seen from the calculation results (generally this
property is only found in analytical methods (see Chapters 5, 6, 7)). Using numerical
methods, such as the finite element method and the finite difference method, always
a schematised fictive boundary has to be taken into account to cut off a region of
interest from the rest of the half plane.
Both classes discussed here deal with half-plane flow with each class being defined by
special boundary conditions at the real axis. Of course, in principle also problems
can be calculated that can be reduced to these classes by a conformal mapping
technique. Regions that have a closed boundary in the form of a polygon can also
be mapped upon a half plane; however, that is not relevant here as flow problems
in that kind of regions can be solved by the method of the foregoing Chapters.

Two classes of half plane flow will now be defined. The half plane is assumed to be
homogeneous and isotropic. The flow in the lower half-plane y ^ 0 is considered
because this has a good connection with many problems in reality.
The two classes of half-plane flow are defined by;

Class I: On the real axis the following conditions hold (where x^, > x„):

-x ^ .Xf, 0 = 0^ (constant)
(10.72)
x ^ x^ 0 = 02 (constant)

Class II: On the real axis the following conditions hold (where .x^ > x^:

x^ Xf, f = f 1 (constant)
(10.73)
X ^ .x„ f = f 2 (constant)

For many practical problems one of the conditions (III.72) and (10.73) can be sup-
posed to hold. Between the points .x^ and x^ several boundary conditions may be met.
(In practice there only the boundary types of stream line and equipotential line

113
(with silt layer) can be used (in arbitrary number and combinations) because the
boundary (the x-axis) is a straight line) For both classes the lower half-plane may
include an arbitrary number of sources and sinks in its interior Using the method of
images (see Chapter 6), fictive sinks and sources are accounted in such a way that
the relevant condition (III 72) or (III 73) remains satisfied

First, some properties of the complex potential due to a distribution of sinks, sources
or vortices over the real axis are shown

A distribution of constant strength over the line segment XjXj is given by (III 30)

«(z) = - ^ {{z - X2)\n(z - X2) - (z - xi)ln(z - x,)}

On the real axis Q is given by

^{x) = - ^Jix - X2)ln(x - X2) - (x - Xi)ln(x - Xi)} (III 74)

let /(x) = (x - X2)ln(x - Xj) - (x - Xi)ln(x - x J }

then (III 74) is Q(x) = - — f(x)

2n

or Q(x) = - ^ {Re{y(x)} + ,Im{y(x)} (III 75)

Zn

The function Im{y(x)] has a special property at the x-axis

Using the definition ln(z) = InLI 4- ( arg(z), Im{/(x)} is given by

Im{/(x)] == (x - X2)arg(x - X2) - (x - x,)arg(x - x,)

For X > V2 arg(x - Xi) = 0
arg(x - X2) = 0
then for x > V2 Im(/(x)) =0
For \ < Xi arg(x - X,) = -n
arg(x - X2) = -n
then for \ < \ 1 lm{f(x)] = n(x2 - xi)
For \ i < \ < \2 arg(x - Xi) = 0
arg(x - \ 2 ) = -TT
then for Xi < \ < X2 Im{/(z)j = ^7r(x - X2)

In figure III 10 the variation of Im{/'(x)j at the x-axis is shown

li4
 7r(X2-Xi)|
. Im {f(x)t

X2

Figure HI 10 Imaginary part of/(v)

The real part of/(x) does not have this special property The function Re{f(x)} is
given by

Re{/(x)} = (x — X2)lnlx — Xjl - (x — xjinix — xJ

The special behaviour of Im{/"(x)} is used for problems where the stream function P
of the potential 0 only varies between two points of the x-axis and has constant values
outside these points, according to the defined Classes I and II This will be shown
now When there is a distribution of sinks and sources only, q is real q = r + is = r
Then from (III 75)

'P(x) = lm{f(x)}
2n

So for distributions of sinks and sources over a part of the real axis, the stream
function only varies over that part (This satisfies Class II) When there is a distribution
of vortices only q = r + is = is and it is seen from (III 75) that

^(x) = -~ Im{/(x)}
271

So for distributions of vortices over a part of the real axis, the potential only varies
over that part (Thus it satisfies Class I)

By a combination of a number of these distributions (of only real or only imaginary

strength) a flow pattern can be generated that satisfies (III 72) or (III 73), whereas be-
tween the points x„ and x^ an arbitrary variation of ^ or f is approximated by a
sequence of straight line segments (Generally that arbitrary variation of 0 or f
is not known previously as it is the result of meeting the boundary conditions between
x„ and Xfc) Figure III 11 is a schematic representation The figure shows (according
to (III 72) or (III 73)) constant values of 0 or f for x ^ x„ and x ^ x^, and at 5 line
segments between x^ and x^ a linear variation of 0 or P

115
 02or %

(PorV' 1 ; ' 1 - - ^

_ v
L-J
\ 1 ! 2
i
^ 3
!__-IX-?L^
' 4 Is:
xa Xb
Figure III 11 Combination of distributions at the v-axis

The complex potential due to a number of distributions of sinks and sources or

vortices is found from (III.30):

Q(z Z - i-g {(2 - ^2,)ln(z - X2; - (z - xijln(z - Xi,)} + QQ.

1=1In
where qj is real or imaginary.

Sources and sinks can be accounted by also using extra fictive sources and sinks
in the upper half-plane so that the condition (10.72) or (01.73) remains valid. For
Class I, where for x ^ .x„ and x > x^ the potential is constant, fictive sources in
the upper half-plane are accounted if there are sinks in the lower half-plane For
Class II, where for v ^ x„ and x ^ x,, the stream function is constant, fictive sinks
in the upper half-plane are accounted when there are sinks in the lower half-plane.
The expressions for the complex potential and its derivatives are given for both
Classes by the following expressions:
various boundary conditions

Figure III 12 C'lass I of half plane flow

116
Class 1: seefigure111.12

f2(z)= Z - - M ( z - X 2 , ) l n ( z - X 2 , ) - ( z - x J l n ( z - x J } +
j = i ^'^

,= i 2 7 r (z - Zj)

^ = Z - ? {ln(z - X2,) - ln(z - X,)} +

dz j=i 2n

" 0
+ Z --iz-z) (z--zy
j=i2n

d^m ^ f _ L^j ( 1
,= 1 2n[(z-X2) (z-x,)

-Qi
+ z\ 2n _(z-z)' (z--zf

Class II: seefigure10.13

various boundary conditions
, f'^,

Figure III n Class II of half plane flow

^(2) = 1 - ^ { ^ - ^2j)'n(z - X2) - (z - Xijln(z - x,)} +

i=i 2n

m Q
+ Z~H(z-zJ(z-z,)]+f2o
j=i2n

117
 dQ(z) " r
az j=i 2;r

m Q
j=i zn iz-^j) (z-z.)

d'i2(z) ^ ^ _ 0 ( 1
,2 Z. -,. 4-
dz^ J=l 2;r [(z - X2j) (z - x,^)

m Q 1 1
+ z - ^ i^-^f 4-
(^^-'^)
1=1 2n
A combination of the general solution with the boundary conditions between the
points x^ and x^ at the x-axis leads to a set of linear equations, in which the unknowns
are the distribution strengths. It is noted that here there is one degree of freedom per
boundary segment. In the middle of each boundary segment the first of each pair of
boundary conditions (see Chapter 14) is posed. It has already been mentioned that
relevant boundary conditions at the boundary segments may be a stream line and
an equipotential line with silt layer (in arbitrary number and combinations). The
resulting equations can be derived directly from those of Chapter 16 by putting r^ = 0
for Class I problems and .s^ = 0 for Class II problems;
The following survey gives the boundary condition equations and relevant expres-
sions;

Definitions:

/Oj(z) = - ^ { i ^ - ^2j)Mz - X2j) - (z - x , ; i n ( z - x,j)}

/l;(z) = - ^ {Mz - X2j) - ln(z - Xi j }

1 i 1 1
/2.(z) =
2n [(z - X2) (z - X i ;
z —z
gOl(z) = Z ? ^ l n
j=i 2n \_z - Zj_

Qi^ 1
I
gn(z) = i=i2nl(z-Zj) (z - Zj]

.2i(z) = z ? ^ ] - - L ^ . ^ (III 76)

'r',2nl (z-z)' (z-'zjf

118
 " 0 (111.76)
gO\\(z)= Z f H(z-z,)(z-I.)]
j= 1 zn

tflll(z) = X ej 1
j=i27r[(z - Zj.) (z - z^

e, f 1 1
^'"(^) = ,?i2.l ( z - z / (z-z,)
/il/z) =/0/z)-//ccJI,(z)
/i2,(z) =f\.(z)-ikcJ2j(z)

Complex potential and derivatives:

Class I;
Q(z) = t iSjfOjiz) + gOl(z) + QQ (01.77)
j = i

dQ(z) "
(10.78)
dz- - = Z ' V i / ^ ) + 9n(z)
j = i

d'Q(z)
(10.79)
"7z^ = J, ''^''^^^'^ + ^^^^'^
Class II;
Q(z) = X r / 0 » + gOII(z) + QQ (III.80)
J=i

dQ(z
= Z ' • ; / l » + 0lII(^) (10.81)
dz j=

d'Q(z)
= Z '•jf2j(z) + g2ll(z) (10.82)
"dz'' i=i

Specific discharges:

Class I;

t>, = - R e - X isjflj(z) + gn(z

7=1
(OL83)
,^ = Im^ X /i-,/l,(z) + 0ll(z:
U=l

119
Class II;

v,= - R e | z ^ r ^ / l , ( z ) + .ylII(z)|
(10 84)
tV = I m | f ^ r ^ / i ; . - ) + .ylII(z)|

Equations with respect to QQ

Pcnential:
Class I;
Z s,lm{-/0^-o)) +0Q = Re{/c0, - 30I(zo)) (10.85)
j = i

Class 0 ;

Z /-.Rel-ZO/zo)! +0Q = Re[^0, - 301I(ro)} (III.86)

j = i

Stream junction:
Class I:

t 5/eJ/O^-o)) + ' f o = Im[-yOI(zo)} (10 87)

1=1

Class 11;

Z r;m(/-0,(zo)} + PQ = Im[-tyOII(zo)) (IIL88)

1=1

Boundary condition equations:

Equipotenticd line with silt layer

Class I-

Z i,Im(-/7l/z)! +0Q = Rc\kcP, - gOl(z) + ikc s^/ii(z)! (111.89)

1=1

120
Class II;

Z 0Re{/il,(z)l +0Q = Re;/c0i - yOII(z) + ikc^qm(z)] (III.90)

j = i

Stream Ime
Class I:

Z .v,Re{/Mz)) = I m ( - ^ l I ( z ) } (10.91)
j = i

Class II:

Z rjlm{l \j(z)] = Im{ -glll(z)} (10.92)

121
18 Multi-Valued Character and Singularity of the used Functions

Multi-valued character
The general solution (0.35) and its derivative (01.36) contain logarithms of a complex
variable. Such logarithms are defined by;

ln(z) = lnlzl -f /arg(z)

This function is multi-valued. In calculations generally a single-valued function is

used that is defined by;
ln(z) = Inlzl 4- larg(z)
(10.93)
-n < arg(z) ^ n

This is the definition of the logarithm of a complex variable that is currently available
in computers. From (01.93) it is seen that the imaginary part of In(z) has a discon-
tinuity 2n at (x < 0, v = 0). This discontinuity is encountered in the stream function
due to the flow of a sink at the point z^,:

Q = ^\n(z-zJ
2n
or;
Q Q
0 + IT = -- Iniz - zJ + i— arg(z - z„)
2n 2n

The argument discontinuity causes a stream function discontinuity at Re'z — z^j < 0,
Im{z — Zp} = 0, as shown in figure 01.14:

Figure III.14 Flow to a sink in ; = r

122
The complex potential due to distributions of complex strength also involves loga-
rithms, and therefore it has discontinuities. The expression for the complex potential
due to a distribution of complex strength between the points z, and Z2 is (III.32):

IjC^
Q = {(z - Z2)ln[c(z - Z2)] - (z - zi)ln[c(z - z,)]}
2n

The argument discontinuity causes a discontinuity in the real and imaginary parts
of Q, and appears when the imaginary part of the complex logarithm is zero and
the real part is negative.
For the first logarithm in (10.32) if;

Im{c(z Z2) 0
(10.94)
Re{c(z ^2)} < 0

For the second logarithm in (III.32) if;

Im{c(z - zi)} = 0
(10.95)
Re{c(z - zi)} < 0

This means that generally the complex potential according to (01.32) has discontin-
uities in the real and imaginary part, at the line through the points z^ and Z2 over the
part that is shown in figure 10.15:

discontinuity in one term

((III 94) holds)

discontinuity in two terms

((III 94) and (11195)hold)

Figure III 15 Location of discontinuities

Because the argument step is 2?!, the discontinuity between z, and Z2 (see (01.32) and
figure 10.15)) is given by:

123
 A,Q = f (z - Z2)/27r = uic(z - Z2) (III 96)
2n

At the extended part of ZjZi it is given by (see (III 32) and figure (III 15))

A2Q = f ((z - Z2)i27r - (z - Zi)i2;r) = iqc(z, - z.} (Ill 97)

271

It IS easily seen from (III 96), which is a linear expression that the complex potential
step A^Q (between z, and Z2 has a linear variation going from z = Zj (at z = Zj it is
zero) to z = Zi (there it is ; ^ c (zi - Zj)) In the extended part of Z2Zi, the step remains
constant at this value which is seen from (III 97)

Using the mentioned single-valued definition of the logarithm in some cases where the
extended parts of the boundary segments are located in the flow region the solution
would have inadmissible discontinuities (see figure III 16) Therefore in the computer
program mentioned in Chapter 20 the values of the complex logarithms are corrected
so that the discontinuities are shifted to lines outside the flow region, see figure
III 16

Figure III 16 Shift of discontinuities from line a to I line b

It is noted that for multiply connected regions the discontinuities cannot be avoided
in this way In such cases the flow region is cut into two or more sub-regions m such
a way that each sub-region is simply connected and the discontinuity of the mentioned
function within the region can be avoided see figure III 17
Then at the cuts there is an inhomogeneity line where the properties of soil and fluid
are the same at both sides

124
 Figure III.17 Subdivision of multiply-connected regions.

Singularity
The general expression for the complex potential in the analytical function method is
(I0.35):

m = Z - ^ ' {{= - Z2,)ln[c,(z - Z2,)] - (z - zi,)ln[c,(z - z,,)]} +

j= 1 2n
"• 0
-f Z -,-•<" - z , ) + «o
j=\ Zn

The function Q(z) is singular at the sinks and the sources, but not at the end points of
the boundary segments (z,, and Z2j). This follows from the fact that the limit:

hm {(z - Z2)ln[c(z - z^)]}

• "2

is equal to zero. This can be shown simply by application of L'Hopitals' rule:

ln[c(z - Z2)]
lim ((z - Z2)ln[c(z - Z2)] = lim 1
(z - =2)

lim {z - Z2} = 0
(Z-Z2)
(=^=2)'

Of course, this holds for the point z = Zi too. So the complex potential only has
singularities at sinks and sources.

The derivatives of i3(z) with respect to z, given by (III.36) and (10.37);

125
 ^ ^ = Z - ' ^ ' { I n ^ z - Z2,)] - ln[c,(z - z,,)]} +
dz j=i In

Qj
j=l 27t(z -

d'o{z)^ y _gj^ -Qj

dz' jh 2n .{=-=2j) (z
+ .Z
fi -12
27r(z - Zj)

are singular at the points Zij, Z2j, and Zj.

126
19 General Comparison with Other Methods

The following general comparison with other methods deals with aspects of pure
two-dimensional flow because the analytical function method is restricted to two-
dimensional flow
The analytical function method has the properties that it is an approximative numer-
ical method and it is rather general for two-dimensional flow
In part II a review has been given of the most important existing solution methods
Only the finite difference method and the finite element method have these proper-
ties too, and in each of them just like in the analytical function method, the solution
IS found by generating and solving a set of linear equations However, there are some
essential diflerences
- There is a difference in theoretical background In the finite difference method the
solution IS found using a discretisation of the differential equation whereas in the
finite element method it is found by using a variational principle The analytical
function method uses analytic complex functions
In the finite element method as well as the finite difference method the flow region is
sub-divided into a large number of elements Corresponding to this number of ele-
ments the number of equations that is generated increases In many cases where
there are sharp alterations of the potential, such as may occur in the presence of sinks
and sources it may be necessary to increase the number of elements greatly to
ensure accuracy Then the number of equations becomes very large and consequently,
the required computer storage too For the analytical function method this does not
apply because the number of equations is relatively small as the number of equations
IS defined by the number of boundary segments and in addition, sinks and sources
aic simply accounted by extra terms in de general solution The number of equations
is proportional to the number of boundar> segments For the finite element method
and the finite difTerence method the number of equations is proportional to the
number of elements
- For the finite element method and the finite difference method the array of coeffi-
cients IS a diagonal band matrix The array of coefficients for the analytical function
method is complete in the case of one region and consists of blocs when there are
more sub-regions, however, these arrays are generally smaller than the arrays of
the finite element method and the finite difTerence method, because generally the
number of boundary segments is relatively small
- In the hnite element method and the finite difference method the unknowns are
the groundwater heads in a large number of points within the flow region from which
discharges can be calculated For the analytical function method the unknowns are

127
the complex distribution strengths So for the calculation of groundwater heads in
points of the region, values of an analytical function have to be evaluated Then
discharges are also known (because the analytical function contains the potential as
well as the stream function)
In the case of non-steady flow in the finite element method the problem may be
encountered that due to a sharp rise of a phreatic line the upper elements become very
oblong That may be disadvantageous for accuracy because the basic assumptions
about the variation of the groundwater head within an element might become a bad
approximation for these upper elements For such cases there has to be generated a
new element grid or a part of it Analogous problems may be encountered when there
IS a sharp fall in time of a phreatic line These kinds of problems are not encountered
in the analytical function method because it does not use an element discretisation of
the region
- When the coefficient of permeability varies considerably a sub-division in many
sub-regions must be applied in the analytical function method In such cases the
method becomes disadvantageous with respect to the finite element method and the
finite difference method because then the number of sub-regions might become the
same as the number of elements in the finite element method oi the finite difference
method Then the computer time needed for calculations will be larger for the ana-
lytical function method because the unknowns in the set of equations are not ground-
water heads as in the two other methods
- The amount of input data for computer calculations is generally less for the
analytical function method because a computer program based on that method needs
only general information (number of sub-regions, coefficients of permeability etc)
and boundary information (position and properties) Consequently the amount of
work for checking data is less Howevci, it u noted that when e g , a finite element
program is used, for many problems a mesh generator can be used to reduce the
amount of input work
For some classes of half-plane flow the analytical function method provides a
solution for the whole half plane For those classes it is not necessary to cut off
the part of interest from the rest of the half plane In the finite element method and
the finite difference method such a cut always has to be made for half-plane flow
problems
For the finite element method as well as the finite difference method the solution is
approximative over the whole region The analytical function method provides a
solution that is exact within an approximative boundary So the analytical function
method has another character than the finite element method and the finite diflerence
method Because its solutions are exact within an approximative boundary the
method is semi-analvtic

128
20 Examples

To illustrate the analytical function method a computer program was made that is
rather general for two-dimensional groundwater flow It can be used for the calcu-
lation of steady and non-steady flow patterns in arbitrary shaped regions that may
contain several fluids and inhomogeneities The anisotropy may have arbitrary direc-
tions and magnitudes and there may be sources and sinks The input of the program
consists only of general information (number of sub-regions, coefficients of permea-
bility, etc) and boundary information (position and properties) The computer pro-
gram comprises a main programme, two subroutines and three small function sub-
programs The main program has an overall controlling task and it provides the
input and output The subroutine OPS Tgenerates the equations, using the boundary
condition formulation of the foregoing Chapters The set of equations is solved by the
subroutine SIMQ That subroutine, or a corresponding one, is usually available on
computers in a scientific subroutine package The function sub-program CAF
calculates some complex expressions that appear in many equations the function
sub-program CPF calculates the part Qi(z) of the complex potential that is due to
the presence of sources and sinks and the function sub-program COF evaluates the
complex potential using the distribution strengths that are known from the solution
of the set equations
The output of the program consists of coordinates groundwater heads, stream func-

ZU3

DY

ZU2

ZU1

Figure I I I IS Output end

129
 MAIN (l)PST

READ
input data next region

I X
call (|)PST region information

call SIMQ
next boundary point |

^. t
next region |
I next region boundary information
X
region-information
[step in x-directlon
z
print information right memljerof
two equations
I step in y-dlrection~
X
coefficients for:
no (Poand Wo

next coefficient numtjen

calculation of
fl,v^,Vy X
coefficients
print result
in two equations

calculation of
".Vx,Vy,Zn
print results

no

no

STOP RETURN

Figure III 19 Fssential parts of computer program

130
tion values, and x- and \-components of the specific discharge These values are
printed for the boundary points and one or more output grids, that are specified in
the input data These output grids are each defined by four complex parameters
starting point ZLl, end points ZU2 and ZL/3, and the complex step parameter
(DX + iDY) as shown m figure III 18

In addition, for points of the boundary the position is printed after one step in time
Before calculating the complex potential at a point of an output grid, it is first checked
whether the point is within the region or not If not, no calculation of the complex
potential is carried out and the program passes to the next point of the output grid
This procedure has the advantage that output grids may be defined very roughly, so
that all required information can be given to the program by a minimum of input
data Figure III 19 gives schematically the two relevant parts of the computer pro-
gram The MAIN program and the subroutine OPST A listing of the computer
program is given in Appendix 5 It is noted that for the computer program the terms
d'z
—J have been set to zero (see Chapter 16)
In the following some examples are discussed that illustrate the power of the calcula-
tion method outlined The figures are roughly sketched by hand from numerical
output data that were obtained using an early version of the computer program In
practise a plot program will be used toaether with the calculation program of Appen-
dix 5 (The listed program conforms figure III 19) Then accurate plots can be pro-
duced automatically and the development using these programs is very easily done
If not defined otherwise, for all examples the following dimensions are used Length
meters time days mass tons (1000 kg)

EXAMPLE 1

Stead] flow in a dike homogeneous isottopic soil

Figure III 20 shows the flow region There is an impermeable revetment on the talus
The open part of the talus is clean there is no silt layer, so it is an equipotential line
At some distance from the canal the flow is nearly horizontal, there a vertical equi-
potential line IS used as boundary of the region For this homogeneous isotropic soil
the coefficient of permeability is 1 m/day and the groundwater head difference between
the canal and the vertical equipotential line is 1 m Figure III 20 shows the flow
pattern At the boundary the sub-division in boundary segments that was used is
indicated
From the groundwater heads at the talus it is seen that the outflow of groundwater is
concentrated mainly at the upper zone of the open part of the talus

131
 stream line
line of constant head
end point of boundary segment

Figure HI 20 Flow pattern in a dike

EXAMPLE 2

Steady flow in a dike, homogeneous isotropic sod with a dtain

In figure III 21 the same dike is shown as in the first example, but here there is a drain
in the vicinity of the lower end of the revetment, possibly having been left behind
from the building phase of the dike It can now be used to reduce the overpressures
at the revetment, for example, to make it possible for some repairs to be carried out
on It In another situation, it would be possible that the water flowing from the vertical
line of constant head is polluted and the dram could be used to intercept the water
in order to prevent it reaching the canal

stream line
line of constant head
end point of boundary segment

Figure III 21 Flow pattern in a dike with a sink

The same boundary conditions are used as in the first example The discharge of the

132
drain is 1.0 m^/m'day (for a length of 500 m that is about 21 m'/hr)
From figure III 21 it is seen that the polluted water is almost completely intercepted
by the drain

EXAMPLE 3

Steadv fhw m a dike, homogeneous anisotropic soil

Figure III 22 shows the same dike as in the first example, but in this case the soil is
anisotropic The direction of the maximum coefficient of permeability has an
angle of 15*" with the positive x-axis (that is, about parallel to the 1 4 talus) The ratio
of the maximum to the minimum coefficient of permeability (the anisotropy factor) is
10 0.

stream line
Ime of constant head
end point of boundary segment

Figure III 22 Flow pattern in anisotropic soil

EXAMPLE 4

Moment oj non-stead\ flow with a phreatic Ime, homogeneous anisotropic soil

This example deals with the groundwater flow in a region that was rectangular in its
initial state (see figure III 23) At the left, right and lower parts of the region the bound-
ary IS impermeable The upper boundary is a phreatic line In the region there is a
sink and a source
Figure III 23 shows the flow region and the flow pattern The soil is anisotropic, the
maximum coefficient of permeability is 1 4 m/day, the anisotropy factor is 2 0, and
the direction of the maximum coefficient of permeability has an angle of 15 with
the positive x-axis. The storage coefficient is 0 2
This discharge of the sink is 2 5 m^/m'day. The same discharge is infiltrated at the

133
 stream line
line of constant head
end point of boundary segment

phreatic line at t =0.1 day

phreatic line at t = 0

x(m)
Figure III 23 Flow pattern in anisotropic soil with sink and source

source From the specific discharge distribution along the phreatic line the alteration
of the position of the phreatic line in a step in time is found After a period of
0 I day, the phreatic line is raised about 6 cm above the source and the fall above
the sink is also about 6 cm

EXAMPLE 5

Moment oJnon-stead] flow with an inteijace homogeneous anisotropic sod w ith a sink
and a source m the heai \ fluid and a sink m the lighter fluid
Figure III 24 shows the flow region There are two fluids Left, right and lower bounda-
ries are impermeable The groundwater head at the upper boundary is constant
The flow pattern is given in figure III 25 The soil is anisotropic, the maximum coeffi-
cient of permeability is 1 4 m/day, the anisotropy factor is 2 0, and the direction of the
maximum coefficient of permeability has an angle of - 15° with the positive x-axis
The density of the lower fluid is 1,025 kg/m^. that fluid contains a sink and a source,
both with the same discharge (2 5 m^/m'day) The density of the upper fluid is 1,000
kg/m^. in this fluid there is a sink with a discharge of 3 5 m^/m day
The upper boundary of the region is an equipotential line without silt layer
It IS seen from figure III 25 that there are groundwater head discontinuities at the
interface, caused by the density difference between both fluids The displacement of
the interface in a step in time was calculated from the specific discharges at the inter-
face At the end of the step in time (2 days) the rise of the interface above the source is
about 40 cm and the fall of the interface between the two sinks is also about 40 cm
(»; = 0 2) It IS seen in figure III 25 that the upper sink mainly receives water origi-
nating from the upper equipotential line boundary

134
 14
(j) = 20.0

12

10

8
Interface position at t = 2 days
6
Jl'l
PsW25 Vinterface position at \.=Q
k=1.4
an=2D ia=-2.5 Q=2.5

y(m)

0 2 4 6 8 10 12 14 16
— * - x(m)
Figure III.24 Two fluids in anisotropic aquifer.
interface
stream line
14t line of constant head
• end point of boundary segment
12

x(m)
Figure III.25 Flow pattern for the problem of Figure 111.24. 135
EXAMPLE 6

Moment of non-steady flow with cm interface, a seepage line and a phreatic line, inhoi7w-
geneous anisotropic soil with a sink and a source in the heavy fluid and a sink in the lighter
fluid
A fictive non-steady flow problem that involves together all boundary conditions
that have been discussed in Chapter 14 is given in figure 01.26. The flow region
consists of two sand layers that are both anisotropic. The soil properties are different
for both sand layers.
miiiiiittiiiiinntiuu
precipitation 0.05

Figure III 26 Two fluids in inhomogcncous anisotropic aquifer

136
 interface
stream line
— line of constant head
• end point of boundary segment
20 m n silt layer

18

16

14

12

10

2
y{m)^

*- x(m)

Figure III 27 How pattern for the problem of Figure III 26

At the separation line between the sand layers there is a thin layer of clay In the second
sand layer there are two fluids with different densities In the heavy fluid there are a
sink and a source of equal discharge In the lighter fluid there is a sink m the second
sand layer The upper boundary of the region involves a phreatic line with precipita-
tion that ends at the talus of a canal Perfect unimpeded streaming out of the soil
solid IS not possible because there is a silt layer on the talus
This silt layer is present at the bottom of the canal too, but there it has a greater

137
resistance then at the talus In the canal there is a fictive structure by which the water
table in the canal is higher than the lowest point of the seepage line
The flow pattern is given in figure III 27 It is clear from the figure that there are
groundwater head discontinuities at the interface and at the clay and silt layers From
the specific discharges at the moving boundaries the displacements in a time interval
were evaluated, and the positions after one step in time are given in figure III 26 by
the dotted lines (p = /; = 0 2) From the figure it is clear that there is only a small
movement of the interface On the other hand, there is a sharp drop of the phreatic
line This is caused by the abstraction of water at the upper sink m the second sand
layer

EXAMPLE 7

Moment of non-steadv flow with two interfaces, a seepage line and a phreatic line
inhomoqeneous anisotropic soil with a sink and a source in the salt water and a sink in
the fresh watei
The regional geometry here is almost the same as in the previous example, except for
one essential difference Instead of an impermeable lowest boundary here there is an
interface with a third fluid That fluid is assumed to have a constant groundwater
head The problem definition is given in figure III 28, and the flow pattern is given in
figure III 29 The displacements of the moving boundaries in a time interval are indi-
cated in figure 01 28 by the dotted lines
It IS remarkable that now there is a sharp rise of a part of the lower interface and the
whole upper interface
These rises are the consequence of the abstraction of water at the upper sink in the
second sand layer Because there is a thin layer of clay between the sand layers, and
now there is a possibility of water supply from underneath, this sink now receives
considerably less water from above Consequently the fall of the phreatic line is less
than in Example 6
As an illustration a part of the computer output is guen in Appendix 3 The complex
distribution strengths (RCQ + i ICQ) are given in the first pages, followed in later
pages by the groundwater head (PHI), the stream function (PSI) and the components
of the specific discharge (VX and VY) The parameters KO, FI and WR are not rele-
vant in this context This calculation was made using the sub-division of the boundary
in the segments that are indicated in figure III 29 To illustrate the influence of the
number of boundary segments, the calculation was repeated where the number of
boundary segments was doubled The corresponding part of the computer output
IS given in the next part of Appendix 3 It is seen that for this extreme case (high
groundwater head gradients that are caused by sink and source), the diff"erences
between the results of the two calculations are comparatively small Apparently the
first choice of the number of boundary segments was reasonable

138
 Hinuiiiniiiuntiiiiiu
precipitation 0i05

y(m)

-x(m)

Figure III 28 Three fluids in inhomogeneous anisotropic aquifer

139
 -. _ Interface
stream line
— line of constant head
• end point of boundary segment

y(m)

x(m)

Figure III 29 Flow pattern for the problem of Figure III 28

140
EXAMPLE 8

The behaviour in time oj an interface homogeneous isotropic sod

In figure 111 30 the starting position of an interface between fresh and salt ground-
water IS given by the horizontal line indicated by f ^ 0 This example deals with
groundwater flow in a polder aquifer The upper boundary of the flow region consists
of semi-pervious layers which locally have a smaller resistance (c^ = 20 days instead of
Cj = 100 days) The sub-soil consists of sand, with a coefficient of permeability
of 10 m/day To a depth of 400 m the soil is homogeneous and isotropic at that depth
there is a very course layer (gravel or shells) As the supply of sah groundwater through
that layer is quite possible, the horizontal line at 400 m depth is assumed to be an
equipotential line At the beginning (t = 0) the interface is horizontal at a depth of
200 m This is a steady position if the groundwater head at the entire upper boundary
IS -1-40 m higher than the head in the salt groundwater (0m)

_ 800 400 800

(/)=4,Cs=100 Cs=20|t$8 |:-i(Poider) Cs = 100,(/)=4

FRESH /y = 100
interface at t < 0
SALT /^g = 1 0 2

gravel or shells,(p = 0
Figure III •^O Interface in polder -tquifci

At f = 0 a sudden fall of the groundwater head is brought about above the part
of the upper boundary where c, = 20 days (see figure III 30), (reclamation the
groundwater head is lowered from -I- 4 to — 1 m) The consequence of the reclamation
IS a flow of the fresh water as well as the sah water Figure III 31 shows the position of
the interface between fresh and sah groundwater as a function of time
In the calculation only half the profile was used because the problem is symmetrical
From the figure it is clear that after some years there is sah water seepage into the
new polder but after about 3^ years the steady state has practically been reached
The relatively high speed of the phenomenon was caused by the great groundwater
head differences (5 m between the polder and its surroundings and 1 m between the
polder and the groundwater head in the salt water at 400 m depth) To indicate the
influence of the size of the step m time, the problem was calculated twice, where the
time step size was respectively 04 and 0 2 year Figure III 32 gives a comparison for
the top of the interface and a point at 100 m from the top that does not reach the polder

141
level. From the figure it is seen that there are differences, although these are relatively
small. So the choice of a time step size of 0.4 year led to a reasonable result (for that
time step size the polder received salt water seepage after 2.7 years instead of 2.9 years
for the calculation with a time step size of 0.2 years).

800 400 800

-*+<- -*i
£^=J09'fei Cs^20^a)=-1, Cs = 100.(f)=4

gravel or sne1ls,(p=0
Figure III 31 Position of interface m time

^t(year) 10 20 3 0
I
y(m)
,->>^'

^ ; ^ ^ ^ ^ ^ 0 0 m FROM TOP
-100
J^
//^ j t = 04year
^ —— j t = 0 2year
•200

Figure III 32 influence of time step size

EXAMPLE 9

Simple half-plane flow

A simple example of half-plane flow is given in figure I0.33: end points of
boundary
segments
in numerical
calculation .-

y=0, x=-1.00
x=-0.99
x=-0.97
x=-0.93
X=-a85
X=-069
x= - 0 3 7
X=+Ou37
X= +0£9
X = *0S5
X= +a93
X=-i-a97
13 boundary segments in numerical calculation X= +099
Figure I I I 33 Simple half plane flow X= +100

142
The problem deals with the groundwater flow from the surroundings towards a low-
situated polder The analytic solution of the flow problem can simply be obtained by
means of the method of Pavlovskii (conformal mapping, see Chapter 7) Figure III 34
shows the i2-plane

i cl >^ I
Figure III 34 Q plane

Using the sine transform (see Appendix 2), the relationship between z = \ + ly
and Q — 0 + iP can easily be found The result is

Q = — arccos( — z)
n

The flow was also calculated using the calculation method outlined in Chapter 17
A comparison of the results of both calculations is given in Appendix 4 From the
tables of that Appendix it is seen that the sub-division in 13 boundaiy segments of
the 'polder boundary' in the numerical calculation yields very accurate results
Including the reference equations, there were only 15 linear equations to be solved

EXAMPLE 10

Half-plane flow
A somewhat more complicated problem is the groundwater flow from the surround-
ings towards three low-situated canals of which the water tables mutually differ One
of the canals has a bottom on which a silt layer is present In the half plane there are
two sinks of equal discharge, see figure III 35 The flow pattern is given in figure III 36
It is seen from the figure that the water abstracted from the right sink originates merely
from the right canal and infinity The left sink receives water from the middle canal
and infinity From the middle and the right canals water is abstracted while the left
canal receives water, which comes from the middle canal and infinity

143
 Figure III.35 Half plane flow.

r^- x(m)
-6. -5. -4. -3. -2. - •1. 9 1. 2.
C^=1.0 (/)=5D,Cs=5.0 <p=3.0

(m)
-1.

-2.

-5.

stream fine
-6.
line of constant hiead
resistance layer under
line of constant head
end point of boundary -7.
segment

Figure III 36 Flow pattern for the problem of Figure III ^5

144
SAMENVATTING

Dit proefschnft over berekeningsmethoden voor tweedimensionale grondwaterstrommg bestaat uit dne
dclcn In het eerste deel wordt de basistheone vermeld In het tweede deel wordt een overzicht gegeven van
de belangrijkste bestaande oplossingsmethoden In het derde deel wordt een numerieke rekentechniek
beschreven die is gebaseerd op het gebruik van analytischc functies \oor het benaderen van tweedimcnsio-
nalc stromingsbeelden (anal>tische functie methode)
Een formulering van randvoorwaarden wordt gegeven waarin in de praktijk voorkomende randen zoals
equipotentiaalliin met sliblaag etc voorkomen Bij de berekening wordt het stromingsgebied onder-
verdeeld in deelgebieden die elk constante vloeistof- en grondeigenschappen hebben De deelgebieden
worden verbonden met behulp van aansluitvoorwaarden voor punten op de scheidmgslijnen tussen de
deelgebieden
Twee klassen van stroming in een halfvlak kunnen ook worden berekend metde/e melhodemet als voordecl
dat het met nodig is om op zekere afstand een schematische begrenzmg aan te brengen
Fen computcrprogramma werd geschreven dat gebaseerd is op dc analytischc functie methode Dit
programma kan worden gebruikt voor de berekening van beelden van stationaire en niet-stationaire
stroming in gebieden van willekeurige vorm die putlcn en bronncn mcerdcie vloeistoffen en anisotrope
inhomogene grond mogen bevatten De invoer van het computerprogramma bestaat loutcr uit algcmcnc
inlormatic (gegevens met betrekking tot doorlatendheid anisotropic dichtheid etc) en randmformatie
(plaats en eigcnschappen van de rand) De oplossingen die worden verkrcgen met behulp van de analvtische
functie methode zijn binnen een benaderde rand exact

SUMMARY

This thesis on calculation methods for two-dimensional groundwater flow is subdivided into three Parts
In the first Part the basic theory is outlined In the second Part a review is given of the most important
existing solution methods In the third Part a numerical calculation technique is developed that is based
on the use of analytic functions for approximating two-dimensional flow pitterns (analytical function
method)
A boundary condition formulation has been given that involves piactital boundaries like 'equipotential
line with resistance', etc
In the calculation a flow region is divided into sub-regions that all have constant properties of fluid and
soil The sub-regions are coupled by connecting conditions for points of the separation lines between the
sub regions
Two classes of half-plane flow can also be calculated by this method giving the advantage that for those
problems no schematic boundary at some distance is necessary
A computer program was written, based on the analytical function method This program can be used for
the calculation of steady and non-steady flow patterns in icgions of arbitrary shape that may include sinks
and sources several fluids and soil that nia\ be inhomogeneous and anisotropic The input of the computer
program consists onlv of general information (data with respect to permeabilitv anisotropv density etc )
and boundary information (position and properties of the boundary)
Solutions that have been found by the analytical function method are exact withm an approximative
boundary

145
Appendix 1: Definitions with respect to Complex Denotations

Complex numbers
The complex number z is defined by:

r = X
. - -I- i y

where .x and i' are real numbers and / is the so-called imaginary unit that has the
following property:

i^ = - 1

The real numbers .x and r are respectively the real and imaginary part of the complex
number z. This is denoted by:

.X = Re!.-}
y = lm{.-}

Complex functions
The complex function /(r) = u(x.y) + /r(.v,3) is analytic when u and v satisfy the
so-called Cauchy-Riemann relationships:

du dv du dv
dx dy dy dx

In the mathematical literature (e.g., Wylie, I960) this can be found more compre-
hensively. From the Cauchy-Riemann expressions by dilTerentiation one finds:

d'u _ d'v d'u _ d'v

dx' dxdy dy' dydx

Additions yields:

d'u d'u

146
In a similar way one finds:

d'v d'v _
^'^~dy'~^

So the functions u and i that satisfy the Cauchy-Riemann relationships satisfy the
Laplace differential equation. Functions that satisfy the Laplace differential equa-
tion are called harmonic functions, and when these satisfy the Cauchy-Riemann
relationships they are called conjugate harmonic functions. So an analytic function
j(z) = M(.Y, r) -I- iv(x, y) consists of two conjugate harmonic functions. Conjugate
harmonic functions (u and v) have the property that i/(.\-, y) = Cj (constant) and
(•(.V, v) = C J (constant) are perpendicular lines.

147
Appendix 2: The Sine Transform

The sine transform has the property that a half-infinite strip is conformally mapped
upon the upper half plane (see figure A 1) accoiding to

H = sin(r)

Figure A 1 Sine transform

This IS seen simply by

w = sin(x -I- M) = sin(x) cos(/\) + cos(x) sm(iy)

Using

cos(n) = cosh(()
sin(n) = (sinh(i)

It follows w = sin(\) cosh(i) -I- / cos(\) sinh(i),

for \ = + - u = ± cosh(t)

so lm(n j = 0

for I = 0 w = sin(\)

so Imju j = 0

148
For a point within the half-infinite strip:

n n
< .X < - and so : cos(.x) > 0
2 2 ^ '

>• > 0 and so : sinh(y) > 0.

Then : lm[u') = cos(.x) sinh(y) > 0.

( n n \
So a point of the half-infinite strip I < .x ^ - , y > 0 I is mapped upon a point
of the upper half plane.

149
Appendix 3: Part of Computer Output of Example 7

Region I is the lower region; region 2 is the middle region and region 3 is the upper
region (see figure 10.28).

Recjion I

X Y KO FI WR RCQ ICQ

000 - 100 23 .000 .000 .482

3 000 - 300 23 .000 .000 483
7 000 - 100 23 .000 .000 .448
9 000 200 23 .000 .000 .338
13 000 400 23 .000 .000 .393
16000 (X)0 21 .000 .000 .523
16000 3 000 21 .000 .000 .509
16 000 5S00 23 .000 .000 .472
13 000 .S600 23 .000 .000 .531
9 000 5 800 23 .000 .000 .538
7 000 6 000 23 .000 .000 .492
3 000 6 300 23 .000 .000 .508
000 6 100 21 .000 .000 .412
000 3 000 21 000 000 425

Initial complex potential 6.933 - 7 822

150
Reilion 2

X Y KO FI WR RCQ ICQ

000 6 100 23 000 000 389 289

3 000 6 300 23 000 000 377 - om
7 000 6 000 23 000 000 465 - 146
9 000 5 800 23 000 000 524 - 128
1 3 000 5 600 23 000 000 485 - 185
IA 000 5 800 21 000 000 587 006
16 000 9 000 21 000 000 600 -033
16 000 1 3 000 31 000 5 000 550 192
1 3 000 12 900 31 000 5 000 509 - 007
9 000 12 200 31 000 5 000 465 101
7 000 11 700 31 000 5 000 384 - 034
3 000 11 200 31 000 5 000 376 - 209
000 11000 21 000 000 .585 090
000 9 000 21 000 000 .552 - 0 45

Initial complex potential 3 9S1 - 6 148

Riquin ^

X Y KO FI WR RCQ ICQ

000 11000 000 5 000 281 026

3 000 11 200 000 5 000 282 069
7 000 11 700 000 5 000 .336 058
9 000 12 200 000 5 000 .368 - 031
13 000 12 900 000 5 000 .324 - 150
16 000 13000 000 000 347 105
16 000 16 000 000 000 343 - 03S
16 000 19 000 -)-) 000 000 365 064
12 000 19 000 22 000 000 388 - 206
S 000 18 500 12 000 2 000 586 - 470
6 000 17 500 11 18 500 2 000 075 - 272
3 000 16 000 11 18 500 4 000 -066 - 255
000 16 000 21 000 000 225 - 092
000 14 000 21 000 000 213 139

Initial complex potential 1 296 - 451

151
Reijuin 1

X Y PHI PSI VX VY

1000 1000 17991 - 852 008 - 063

3 000 1000 18 020 -1045 001 - 077
5 000 1000 IS 036 -1 309 041 - 106
7000 1000 17918 -1 553 095 - 048
9 000 1 000 17 742 - 1 536 061 065
11000 1000 17612 - 1 192 -010 172
nooo 1000 17617 - 643 -049 192
15 000 1000 17615 - 197 - 031 138
1000 3 000 18 185 - 873 Oil - 021
3 000 3 000 18 277 -916 - 105 - 009
5 000 3 000 22 064 - 2 027 3838 164 -1028 405
7 000 3 000 17 925 - 2 124 279 028
9 000 3 000 17 368 - 2 008 268 054
11000 3 000 13 249 - 714 -3838 036 1028 489
13 000 3 000 17 064 - 333 - 143 0S7
15 000 3 000 17 199 -089 -045 083
1000 5 000 18 142 - 3 382 016 024
3 000 5 000 18 092 -3251 015 071
5 000 5 000 17 934 -487 102 102
7 000 5 000 17 5S9 -257 169 056
9 000 5 000 17 204 - 162 162 012
11000 5 000 16957 - 171 056 - 005
1 3 000 5 000 16 934 124 -023 037
15 000 5 000 16 974 007 032 046

152
Example 7:
The number of boundary segments is now doubled.

Ri'i/ion 1

X" Y KO FI WR RCQ ICQ

000 - 100 23 000 000 747 593

1 500 - 200 23 000 000 792 .401
3 000 - 300 23 .000 000 733 .353
5 000 - 200 23 000 000 752 307
7 000 - 100 23 000 000 .730 349
8 000 050 23 000 .000 .720 315
9 000 200 23 000 000 681 163
11000 300 23 000 000 .655 -009
1 3 OCX) 400 23 000 .000 .650 -4.39
14 500 200 23 000 .000 .60S -635
16 000 000 21 000 .000 .826 083
16 000 1 500 21 .000 .000 .852 -074
16000 3 000 21 .000 .000 850 - 196
16000 4 400 21 .000 .000 825 - 381
16 000 5 800 23 .000 .000 727 593
14 500 5 700 23 .000 .000 .758 457
nooo 5 600 23 .000 .000 .814 271
1 1 000 5 700 23 .000 .000 815 234
9 000 5 800 23 .000 .000 843 138
8 000 5 900 23 .000 .000 842 088
7 000 6 000 23 .000 .000 809 039
5 000 6 150 23 .000 .000 801 - 089
3 000 6.300 23 .000 .000 832 - 099
1 500 6 200 23 000 .000 825 262
000 6 100 21 .000 .000 743 323
000 4 550 21 .000 .000 .748 105
000 3 000 21 .000 .000 752 - 101
000 1450 21 .000 000 741 - 308

Initial complex potential - 602 11851

Reqion 2

X Y KO FI WR RCK ICQ

000 6 100 23 000 000 520 479

1 500 6 200 23 000 000 .538 329
3 000 6 300 23 000 000 .560 101
5 000 6 150 23 000 000 565 - 053
7 000 6 000 23 000 000 601 151
8 000 5 900 23 000 000 .608 - 194
9 000 5 800 23 000 000 630 177
11 000 5 700 23 000 000 617 - 171
13 000 5 600 23 000 000 632 - 173
14 500 5 700 23 000 000 .624 - 312
16 000 5 800 21 000 000 .751 131
16 000 7 400 21 000 000 776 002
16 000 9 000 21 000 000 .777 058
16 000 11 000 21 000 000 .738 193
16 000 13 000 31 000 5 000 668 409
14 500 12 950 31 000 5 000 683 209
13 000 12 900 31 000 5 000 .610 142
11000 12 550 31 000 5 000 .648 105
9 000 12 200 31 000 5 000 636 163
8 000 11950 31 000 5 000 630 138
7 000 11 700 31 000 5 000 .573 - 001
5.000 11450 31 000 5000 .573 - 099
3.000 11 200 31 000 5 000 .561 - 244
1.500 It 100 31 000 5 000 .549 - 373
.000 11000 21 000 000 .775 186
.000 10 000 21 000 000 .783 078
.000 9 000 21 000 000 .761 -027
.000 7 550 21 000 000 .739 - 188

Initial complex potential -196 -8 373

154
Reqion 3

X Y KO FI WR RCQ ICQ

.000 1 1.000 31 .000 5.000 325 071

1.500 11.100 31 .000 5 000 .333 - 008
3.000 11 200 31 .000 5.000 .328 - 052
5.000 11.450 31 .000 5.000 ,337 - 109
7 000 11.700 31 .000 5 000 .367 -.082
8 000 11.950 31 .000 5 000 .378 -.079
9 000 12.200 31 000 5.000 .388 -.090
11 000 12.550 31 .000 5.000 .373 -.080
13.000 12.900 31 .000 5.000 .338 - 170
14500 12.950 31 .000 5.000 .327 - 226
16.000 13.000 21 .000 .000 .392 153
16 000 14 500 21 000 000 .398 065
16 000 16.000 21 .000 .000 .396 -016
16.000 17.500 21 000 .000 .385 -- 120
16.000 19 000 22 000 .000 .412 133
14000 19 000 22 000 .000 .431 - 020
12 000 19.000 22 000 .000 .449 - 186
10 000 18 750 22 000 .000 .445 - 320
8 000 18 500 12 .000 2.000 .688 .. 445
7.000 18 000 12 .000 2.000 .696 - 628
6 000 17 500 18.500 2.000 .030 - 433
4 500 16 750 18 500 2 000 .004 -- 31S
3.000 16 000 18 500 4 000 -.128 - 283
1 500 16 000 18 500 4 000 - 125 260
000 16 000 21 .000 000 .237 - 112
.000 15 000 21 .000 .000 .237 - 131
.000 14000 21 .000 .000 .231 - 154
.000 12.500 21 .000 000 .225 - 193

Initial complex potential 199 061

155

u
Rcilion 1

X V PHI PSI VX VY

1 000 1 000 17 968 -1041 Oil - 056

3 000 I 000 17 993 -1 207 -003 067
5 000 1 000 18013 - 1443 037 - 096
7 000 1000 17 898 -1659 092 - 039
9 000 1 000 17721 -1616 059 072
11 000 1 000 17 596 - 1 244 - 010 186
13 000 1 000 1 7 SS9 - 678 -053 194
15 000 1 000 17 591 -211 -028 142
1 000 3 000 18 110 -1 048 -005 - Oil
3 000 3 000 18215 1064 - 110 000
5 000 3 000 22 006 - 2 150 3838 159 -1028 396
7 000 3 000 17871 -2221 275 037
9 000 3 000 17 M 7 - 2 082 265 063
11 000 3 000 n 198 -765 -3838 037 1028 497
13 000 3 000 17012 -363 - 143 095
15 000 3 000 17 143 -096 - 043 092
1000 5 000 18 044 -3 549 014 038
3 000 5 000 17 996 - 3 387 Oil 082
5 000 5 000 17 846 - 596 097 110
7 000 5 000 17 504 - 343 165 065
•^000 5 000 17 126 - 225 158 018
11000 5 000 16 880 - 218 055 001
13 (KK) 5 000 16 853 - 152 -023 045
15 000 *: 000 16 889 003 - 030 059

156
Appendix 4: Comparison of Exact and Approximative Solution of
Example 9

-X _x
approximated exact approximated exact

0 0 0000 0
0 05 00001 1 80 0 3806 0 3797
0 10 0 0002 1 85 0 3911 0 3901
0 15 0 0004 190 0 4011 0 4002
0 20 0 0006 195 04108 0 4099
0 25 0 0003 200 0 4201 04192

0 30 0 0003 2 05 0 4291 0 4282

0 ^5 0 0003 2 10 0 4379 04370
0 40 -00014 2 15 0 4464 0 4455
0 45 -00014 2 20 0 4546 0 4537
0 50 - 0 0006 2 25 0 4626 04617

0 55 0 0005 2 30 0 4704 0 4695

0 60 0 0015 2 35 0 4780 04771
0 65 0 0021 2 40 0 4854 0 4845
0 70 - 0 0005 2 45 0 4926 04917
0 75 0 0006 2 50 0 4996 0 4987

0 80 0 0010 2 55 0 5065 0 5056

0 85 0 0003 2 60 05132 05123
0 90 0 0006 2 65 0 5197 05189
0 95 - 0 0000 2 70 0 5261 0 5253
100 0 0174 0 2 75 0 5324 05316

105 0 1023 0 1002 2 80 0 5386 0 5377

1 10 0 1428 0 1412 2 85 0 5446 0 5437
1 15 0 1736 0 1722 2 90 0 5505 0 5496
1 20 0 1994 0 1981 2 95 0 5563 0 5554
1 25 02218 0 2206 3 00 0 5620 0 5611

1 30 02419 0 2408 3 05 0 5667 0 5667

1 35 0 2602 0 2591 3 10 0 5730 0 5721
1 40 0 2770 0 2760 3 15 0 5784 0 5775
145 0 2927 02917 3 20 0 5837 0 5828
1 50 0 3074 0 3063 3 25 0 5888 0 5830

1 55 03212 0 3202 3 30 0 5939 0 5931

1 60 0 3342 0 3333 3 35 0 5990 0 5981
165 0 3467 0 3457 3 40 0 6039 0 6031
1 70 0 3585 0 3575 3 45 0 6088 0 6079
1 75 0 3698 0 3689 3 50 06135 06127

157
 —x
approximated exact approximated exact

3 55 06182 06174 4 80 07173 07164

3 60 0 6229 0 6220 4 85 0 7206 0 7198
3 65 0 6275 0 6266 4 90 0 7240 0 7231
3 70 0 6320 06311 4 95 0 7273 0 7264
3 75 0 6364 0 6355 5 00 0 7305 0 7297

3 80 0 6408 0 6399 5 05 0 7338 0 7329

3 85 0 6451 0 6442 5 10 0 7370 0 7361
3 90 0 6493 0 6485 5 15 0 7401 0 7393
3 95 0 6535 0 6527 5 20 0 7433 0 7424
4 00 0 6576 0 6568 5 25 0 7464 0 7455

4 05 06617 0 6609 5 30 0 7494 0 7486

410 0 6658 0 6649 5 35 0 7525 07517
415 0 6697 0 6689 5 40 0 7555 0 7547
4 20 0 6737 0 6728 5 45 0 7585 0 7577
4 25 0 6775 0 6767 5 50 0 7614 0 7606

4 30 06814 0 6805 5 55 0 7644 0 7635

4 35 0 6851 0 6843 5 60 0 7673 0 7664
4 40 0 6889 0 6881 5 65 0 7701 0 7693
4 45 0 6926 06917 5 70 0 7730 0 7722
4 50 0 6962 0 6954 5 75 0 7758 0 7750

4 55 0 6998 0 6990 5 80 0 7786 0 7778

4 60 0 7034 0 7026 5 85 07814 0 7806
4 65 0 7069 0 7061 5 90 0 7841 0 7833
4 70 0 7104 0 7096 5 95 0 7869 0 7860
4 75 07139 0 7130 6 00 0 7896 0 7887

158
—x —X

approximated exact approximated exact

0 0 5000 0 5000 0 80 0 2033

0 05 0 4839 0 4841 0 85 0 1777
010 04678 0 4681 0 90 0I4I7
015 04517 0 4521 0 95 0 0988
0 20 0 4356 0 4359 100 0 0000
0 25 04195 04196
105 0 0000
0 30 0 4034 0 4030 1 10 0 0000
0 35 0 3873 0 3862 1 15 0 0000
0 40 0 3695 0 3690 120 0 0000
0 45 0 3505 03514 125 00000
0 50 03316 03333
0 55 03126 0 3146
0 60 0 2936 0 2952
0 65 0 2747 0 2748
0 70 0 2544 0 2532
0 75 0 2288 0 2301

159
Appendix 5: Listing of Computer Program

1 I M P L I C I T C O M P L E X (C,Z)
2 D I M E N S I O N T E X T I S D ,AA( 1)
3 I N T E G E R N S 3 ( 5 I ,NS1(5I
« COMMON /Cni/ZPA(5,10),ZUA(5,5,1),ZQA(5),ZRA(5,10O,ll,DT,ROZ,FIZ,
S CFI«(b).GAIS|7l,CPAIS,in).WRAIS,laO,ll,NG,PORI5l,
6 CNA(5,3,l),NGAt5,5,l),IFA(5,iaO,ll,KRA(5,100,l),IRA(5,10a,ll,NFI
7 COMMON /CnZ/BP(5J0I,NV
a EQUIVALENCE (INO.AAdll
9 C A N I F I C V , A N , A H | : C E X P I I tl. 0, I . 0 ) • AH ) tCMPL X ( ( R E A L ( C V ) * S O R T t A N ) ) ,
in CAIMAGICVI(
11 c
12 c»
13 c
1« EPS Cl*«(-'t I
IS CI = (o. J .1 .01
16 PI: 3.11 15926535
IT W E A 0 1 ? , 115 ) T E X T
1» a R I T L ( O ,116 )TrXT
19 U R I Tt (b ,112 1
20 R E A D ( 5 , 10 1IVG,NFA,NFI,DT,')0Z,FIZ
21 N S l :Nr
22 N G = I APS (NG)
23 N S 2 :hFA
21 N F A :IA-3 5 t N F A )
25 IF( Nf I. N E . J I R F A n ( 5 , l J 2 ) ( F I A ( I ) , I = l , N F I )
26 D O 1..S I G : 1 , N G
27 REA 0(5, l - 1 5 l ( N A ( I G , J , l ) , J : l , 3 ) , ( N G A ( I G , J , l ) , j : l , N G »
28 >.S3 ( i&)= N H ( It , 1 , 1 I
29 MA ( Ic,l , 1 ) : I A = ' S ( N A ( I G , 1 , 1 ) 1
RFA 0(5, 1 1 7 ) ( G « ( I & , J 1 , J : 1 , 1 ) , P 0 R ( I G ) , Z J A ( I G )
il •JSM ( IT)= 1
?2 IF I GA( IG , 1 ) . L T . n . I N S I t 1 6 l r - 1
SJ G A ( Ic,I ) : A B S ( e A ( I G , l l )
31 N P : NA(I G , 3 , l l
SS I F ( NH.G T . M R E A C I ( 5 , 1 1 8 ) ( ? P A ( I & , J ) , 0 P A ( I G , J ) , J : l , N P I
36 \u:NO (I S , 3 , 1 1
37 IF( Nb.F r . T ) G O T O 1-15
38 RLA n (5, l J 1 ) ( ( 7 U « ( I 3 , J , l l , I : l , M l , j r l , N U )
39 CON Tl^•u F
-JFA A-1
m UO 1 ,9 IGrl.'IG
12 NA(1 r-, 1 , 1 )
13

11 READ FOUNOARY AND INSERT EXTRA BOUNDARY SEGMENTS

15
16 or li IR^l ,N';
17 jT = n
IR I S u n ^ ^ l - F A t t l R - l 1*1
19 ''EAn(5,111)ZRA(I&,ISUPS,l),KrR , «A. ( I G , I S U B S , 1 ) , I F A ( I 6 , I S U B S , 1 ) >
51 r,i(PA(lC,ISll8S,l),IRA(IG,I5UBS,n
51 I F ( I K . F C . 1 ) E O T O 113
52 IFlIR.KT.llISIIBMzISIIBS-NFA
5J I F C J F A . E O . 1 )GCTO 113
31 11U C n Z : ( Z ' ^ A ( I & , I S U B S , l > - Z R A ( I G , I S U B M , l ) l / N F A
iS JO 119 J : l , N F A M
56 ISU3nJrISUB''«J
57 <.RA(IG,ISlJgMJ,l):ZRA(IG,ISUBM,l)«J*CDZ
58 KRA(lr,IS'JB''J,l)=KRAlIG,ISUBM,l)

160
 59 IFA(iE,ISUPI'J,ll:lFA(IG,ISUBM,ll
bC l,PA(lG,ISUet'J,lt:WRA(IG,ISUBM,ll
Ll IPA(IO,ISIJB»'J,ll:Ir!A(IG,ISURM,l)
62 119 C O N T I N U E
b3 IF(.NOT.(IR,El.NR.ANO.JT.EQ.ul)GOTO 113
Dl ISJBHrNFA*(NR-1)•!
65 ISU8S:l
66 JT:l
o7 bOTO 1 1 "
6" 113 CONTlNc'F
69 NA(IG,1,1);NFA»NR
7" I F t N F A . F U . 1 ) G O T 0 1J9
71 I G A : I C » 1,1 J
72 IGA^t.Or ( I G A , 2 )
DO 111 I G B = 1 , N G
71 IH=NuA(IG,IFB,1)
75 I F ( I H . r C . n ) G O T O 111
76 IHC=NFA*(IH-1)•!
77 IF(IoA.fE.U>IHC=IHC*NFA-l
78 NGA(1G,IGB,1l=IHC
79 111 CONTINUE
tT 199 CONTINUE
61

82 .RITt rcFSULT OF INPUT DATA (INCL. GENERATED EXTRA SEGMENTS)

6f
31 I F ( N i l . F T . J IGCTC 138
85 WRITLta,li:)-jn,NFA,NFI,riT,ROZ,FIZ
66 IF(NfI.NE.I)HRITE(6,l''2)(FIA(I),Irl,NFI)
87 1^0 IZf IC = l,Nr,
88 yRITL(6,ll6)(NA(IG,J,!),j:l,3l,(NCAtIG,J,ll,j:l,NG)
69 «RITL(fc,li7)(CA(IG,JI,J=l,1),P0P(IGI,ZaA(IG)
9n NP:NA(ir,2,1)
91 IFlNP.CT.i)WRITE(5,ir)8)(ZPA(IG,J),0PA(IG,JI,J=l,NP)
92 NU^NA(IC, 3 , 1 )
93 IFI N u . E C . I.IGOTO 13P
91 y P I T L t O ,1'11)((ZIJA(IG,J,I),I-1,1),J:1,NU)
9": CONTIIUE
96
97 DO l.,9 I G : 1 , N ' -
98 uA(IG,'J)rGA(IR,«)«PI/18".ri
99 tj R = N A ( I f;, 1 , 1 )
luO NV=NV«,•NR«2
lul I F ( N J 1 . L T . O ) » » I T E ( 6 , 1 1 C I ( Z R A ( I G , I R , 1 ) , K P A ( I G , I R , 1 I ,

loZ tIFA(lC,:R,ll,wRA(IG,IP,l),IRA(I6,IR,l),IR:l,NR)
1 .3 J CONTINUE
lul URITL(6,112)
1 5
1 J6 r»V2=tvV»''iV
1U7 CALL RLODA (AA ( 1 ) ,NV2,IE'')
1J8 IF(lLP.KE.a)«RITE(6,9r!l)IEP
1-9 IF(lEP.NF.T)STQP
11" CALL OPST I A A ( I N D ) )
111
112 CALL SIMO lAA (IND ) , 8 B , N V , K S )
113
111

115 WRITE REGION AND BOUNDARY DATA AND SOLUTION OF EQUATIONS

116
117 I F ( N S ? . G T . O ) G O T O 121
118 0 0 1<;1 I G : 1 , N G
119 NTrn
120 NR:NA(IG,1,I)
121 NP:NA(IG,2,1)
122 AH:GA(IC,1)*10J.J/PI
123 l«l:l
121 I F d G . C C . 1 )G0T0 127

161
125 J:IG-1
126 DO 128 I:l,J
127 IM1=I>'1»2*NA t l , 1 , 1 )»2
128 128 CONTINUE
129 127 yRITL(b,l?U)I6,GA(IG,l),GA(IG,Z),GA(IG,3),AH,NP
ijn NT:NT»1J
131 IF(NP.EC.TIGOTO 122
132 WRITE(6,123lt7PA(IG,J),0PA(IG,J),J:l,NP)
133 I . R I T L (6 , 1 31 )
131 NT:NT»6
135 122 liRITL(6,125)
136 NT=NT»3
137 00 121 I R : 1 , N R
138 IM:IK1->IR-1
139 IFO:IFA(IG,IR,1)
11" IF (IFO.Flj . 3 ) F I r n , 3
111 IF( IFP.ME.J)FI:FIA(IFO)
112 isues = lf->NR
113 kRITL(f,126)ZPA(IG,IR,l),KRA(IG,IP,l),FI,URA(IG,IR,l),BB(IH),BB(
111 ri su^.,)
115 NT:NT«1
116 Tr/hiT r n c i i u D T T c i f c . i T n i T C
IF(NT.C0.5J)URITEI6,130)IG
117 IFtNI.'_C.5U)NT = 3
118 121 CONTIKUF
119 ISUPS:IMi»2tNR
15 I b R I T C ( b , l ? 9 ) B B ( I S U B S ) , 8 " ( I SUSS*1)
151 121 CONTIM/E
152

153 BOUNDARY OU T P U r
151
155 DO 1 1 1 T G r l . N G
156 NT = "J
157 nN jRK =- NNAAI (l tI >G, , 11 , 11 )1
158 AawN r= fG; aA r( TI Gn ,. 3 )i
159 Afln-'jAiii.-,^;/iyKi
K:GA(IG,2)/S0RT(A l ANN)i
160 AMzGAirn.ui
AH=GA(IG,1)
161 JA:3
162 IF(Ni1(IG).LT.U)GOTO 132
163 DO 132 IRrl.NP
161 IF(Ni3( IG ) .LT. DGOTO 111
165 K 0 : K K A ( IG , IR, 1 I
166 IF(.NOT.(K0.EO.?2.OR.K0.E0.23))GOTO 132
167 111 IF ( JA.EC,'1)WRITE(6, 133) IG
168 JAr 1
169 ZO:ZKA(IG,IP,1)
170 IFtIR.LT.NR)ZP:ZRA(IG,IR«l,l)
171 IF(IR.fO.NR)ZP:ZRA(IG,l,l)
172 Z:0.5*(ZO»ZP)
173 COZCUFIIG.Z.Jl
171
175
176
177
178
179
180
181
182
183
181

185 6RI0 OUTPUT

186
187 NU = N A ( I G , 3 , 1 )
188 IF(NU.FC.OIGOT0 111
189 0 0 116 j : l , N U
190 NT;0
191 WRITE(6,112IIG

162
192 ZUl:ZUA(IG,J,l)
193 ZU2:ZUA(IG,J,2)
191 ZU3=ZUA(IG,J,3)
195 ZU1=<.UA ( I G , J,1)
196 I F t C A P S t Z U 2 - Z U l ) .G T . E P S ) C R : ( Z U 2 - 2 U 1 ) / C A B S ( Z U 2 - Z U 1 )
197 IF(CAPS(ZUZ-ZUl).L T.EPS)CR=(2U3-ZU2)/CARS(ZU3-ZU2)
198 RECR^REAL(CR)
199 AICRZAIMAG(CR)
2U0 DXZREAL(ZU1)
2J1 0Y:AIHAG(ZU1)
2J2 I F ( A a S ( R E C R ) . L T . E P S)CS T = C T » 0 Y
203 I F ( A B S ( P E C R ) . G E . E P S)CS T r C M P L X ( D X , < D X » A I C R / R E C R ) )
231 I F ( A b S ( D X ) . L T . E P S ) NX:l
205 IF ( A b S ( 0 Y ) . L T . E P S ) NY=1
2 J6 I F ( A B S ( D X ) . G T . E P S . AND C A B S ( Z U Z - Z U l ) . G E . E P S l
207 tNX= ( ( R E A L ( Z U Z - Z U l ) • EPS )/0X)^1
208 IF(AbS(DX).GT.EPS. AND. CARS(ZU7-ZU1I.LT.EPSI
209 tUXz((REAL(ZU3-ZU2) •EPS ) / D X ) ^ 1
210 I F ( A b S t P Y ) . G T . E P S . A N D . C A B S ( Z U 3 - Z U 2 ) .GT.''PS )
211 t N Y : ( ( A I M A G ( Z U 3 - Z U Z )*EP S ) / D Y ) • 1
212 IFtABStDY).CT.EPS. AND. CA8StZU3-ZU2).LT.EPS I
213 C N Y = ( ( A I M A G ( Z U 2 - Z U 1 )^EP S ) / 0 Y ) ^ 1
211 ZU:ZU1-CST-CI»DY
215 DO 117 I Y = 1 , N Y
216 ZU=ZU^CI«OY
217 DO IIP I X : l , N X
218 ZU^ZU^CST
219 Hro. J
220 DO 135 IR=1,NR
221 ZO:ZftA(IG,IR,I)
222 IFI I h . N E . N R ) Z P = ZRA I I G , I R * 1 , 1 )
223 IFtIK.EC.NR)ZP:ZRA ( I G , 1 , 1 )
221 JA = n
225 IF ( C A P S ( 2 U - Z 0 ) . L E . EPS) J A r l
226 I F ( J A . L C . 1 )ZU:ZU*( Z P - Z 0 ) / C A B S ( Z P - Z 0 ) * 2 . I 1 » E P S
227 I F ( C A B S ( Z U - Z P ) . L E . EPS ) J A : 2
228 IF(JA.EO.2)ZU=ZU*( Z O - ? P ) / C A B S ( Z 0 - Z P ) » 2 . 3 * E P S
229 IF(AtS(AIMAG(CABS( Z P - Z 0 ) / ( Z P - Z 0 ) » ( Z U - Z 0 ) ) ) . L T . E P S
230 t . A N D . C A P S ( Z U - Z P ) .L T . C A B S ( Z P - Z O )
231 C . A N D . C A B S ( Z U - Z O ) . L T . C A B S ( Z P - Z O ) ) 6 0 T 0 113
232 HrH^nIHAG(CLOG(CAB S ( Z O - Z U ) / ( Z 0 - Z U ) » ( Z P - Z U ) ) )
233 135 CCNTINUF
231 I F ( A D S ( H - - ' . I 1 4 P I ) .G T . ' l . U D G O T O 118
235 113 COiCUF(IG,ZU,M)
236 CO:C)'PLX( (REAL (CO) /AK ) , A I 1 A G ( C 0 ) )
237 CV:CUF(IG,ZU,1)
238 C W = C A M F ( C V , A N , AH)
23° CV=-C0\J6(CV)
21" NTzNT^l
211 WRITE(6,111)ZU,C0, CV
212 IFtJA.EO.1)ZUrZU-( Z P - Z 0 ) / C A n S ( Z P - Z 0 ) » 2 . ' l * F P S
213 IFtJA.CC.2)ZU:ZU-( Z 0 - Z P ) / C A B S ( Z 0 - Z P ) » 2 . n » E P S
211 IF(NT.EC.?5)WRITE (6 , 1 1 2 ) I G
215 IF(NT.LC.?5)NTn
216 118 CONTi^UF
217 ZU:21J-NX*CST
218 117 CONTIMLF
219 116 CONTINUE
25n 111 CONTINUE
251 C**
252 STOP

2S3 101 F0RMAT(I5,2I9,F13.3,2F9.3)

251 102 FORMAT(eF9.3)
2S« 101 FORMAT(8F9.3)
256 106 FORMATdS , 7 1 9 )
257 107 FORMAT(5F9.3/2F9.3)
258 108 FORMAT!3F9.3)
259 no F0RMAT(2F9.3,I5,I9,F13.3,I5)
260 112 F0RMAT('1',5X,'REGION',I1///6K,'X',8X,'Y',11X,"PHI",6X,'PSI',9X,
261 t'VX' ,7X,'VY'//)

163
262 UI FORMATC ',
263 120 FORMAK '1',
261 f,F8.3/23X,
265 123 F0RMAT(5X,'
266 131 FORMAT(///)
267 125 FORMAT(5X,'
268 126 F0RMAT(2F9.
269 130 F O R M A K '1 • ,
270 t, 'FI ' ,7X, 'w
271 129 F0RMAT(/2IX
272 133 FORMATI ' I ' ,
273 t'VX' ,7X,'VY
271 131 F O R M A T ( • ',
275 HZ FORMAT( ' 1 ' )
276 115 FORMAT(18A1
277 116 F O R M A K '1'/ //2(5X,71(1H*)/),3(5X,'»',72X,•*•/),
278 t5X, • f , 5 X , •M 0 T 6 R 0 : MODEL FOR TWO-DIMENSIONAL GROUNDWATER'
279 t' F L 0 h ' , 9 X ,'*•/
280 [ 5 X , • » ' , 1 7 X , 'BASED ON ANALYTICAL FUNCTION METHOD ' , 1 7X,'» ' /
281 t3(5X,'«',72 X,'••/),2(5X,71(1H»)/),2(5X,'«',72X,'••/),
282 C3 (5X , •»',18 A I , • * • / ) , 2 I 5 X , ' » ' , 7 2 X , ' » ' / ) , 5 X , 7 1 ( 1 H » ) )
283 90 F O R M A K ' IE R = ' , I 3 , 7 X , ' F O U T eiJ HET D Y N A M I S C H D E C L A R E R E N '
281 C/15X,'2IL D O C U M E N T A T I E VAN REODA'//)
285 END
286 C
287 C
288 C

280 SUBROUTINE OPST(AA)

290 I M P L I C I T COMPLEX (C,Z)
291 L O G I C A L LJO
292 DIM'ENSION A A ( 1 )
293 COMMON /CJl/ZPA(5,in),ZUA(5,5,1),Z0A(5),2RA(5,100,11,OT,R02,FIZ,
291 f.FIA(o),GA(5,7),CPA(5,lU),WRA(5,109,l),NG,P0R(5l,
295 CNA(5,3,l),NGA(S,5,l),IFA(5,19Q,l),KRA(5,laO,ll.IRA(5,100,l),NFI
296 COMMON / C 3 2 / B P ( 5 JO),NV
297 CANF(Z,AN,AH)rCMPLX(REAL{Z*CEXP(-(0.0,1.0l*AH)),(S0RT(AN)»AIHA6«
298 f.Z*CF*P(-( '1.i),l.3)«AH) ) ) )
299 CANIF(CV,AN,AH)=CEXP((n.3,1.0)»AH)»CMPLX(I REAL(CV)*SORT(AN))>
3.3 CAIMAG(CV) )
3J1
CIZ (<..>,!. J)
3J2
00 1J1 I;I,NV
3 J3
6B( I ) : 1.0
3 J1
DO lo2 J:l,NV
3,5
ISUBS-(J-l)*NV*I
3J6
3 ,7 AA( I S U F S ) = U . J
3.,8 laz CONTINUE
3..9 1 11 CONTINUE
310
311 00 lo3 IG=1,NG
712
313 REGIuN
311
315 J i; 1
316 IHB : i G
317 N P z NA ( i r , i , l )
31S N R B rNR
319 N P : NA ( I G , ' , 1 )
32" R O r 5 A ( I G, 1)
3^1 AK = G A ( I G , 2 ) / S C R T ( G A ( I G , 3 ) )
322 AH- C A ( in, 3)
323 A H - GA( I G , 1 )
321 IMl
325 IF ( I & . E . C . 1 ) G 0 T 0 11
326 J = IG - 1
3^7 D O lol I;l,J
3ZP I M l :ll' (•2»NA(I,1,1)*2
3;>o ICJI C O N TIN UE

164
330 118 on 1^5 IRrl,NP
331 C
332 C* BOUNuAf^Y-SECTION
333 r
331 I>"=IM1'2* I IP-1 )
335 IGC=iRA(IG,IR,1)
336 IF ( I G C E C .u)GOTO 106
3J7 ROC=oA(IGC,1)
33e AKC;bA(IGC,2)/SQRT(GA(lr,c,3))
339 ANC=&A(IGC,3)
31" AHC=uA(IGC,1)
311 NPC^NA(IGC,2,1)
312 W R : W . . A ( IG , IR, 1 )
313 IFO^IFA(IG,IR,1)
311 I F ( I t O . N E . j ) F I O : F I A ( I F O )
315 KOrKKA(IG,IR,1)
316 ZO;ZKA(IG,IR,1)
317 IF(Ir(.LT.NR)ZP:ZRA(IG,IR»l,l)
318 I F ( I n . F O . N R ) Z P = Z R A ( I G , l , l )
319
Z=J.5*(Z0^ZP)
3i,n
CDZ=iP-Z0
351
DX;PLAL(cnz)
352
OY=AIMAC(CDZ)
353
DL^CABS(CDZ)
351
UPZP.O'DX'OY
355
DV=DX**2-DY*«2
356

357 C»* EOUATIONS

358 C
359
la? r
331 C* RIGHT-HAND SNO OMEGA-ZERO
3t>2 C
363 L J i j r (KC . N E.21.AND.JO.EQ.O)
361 I F ( L J ' ) 7F 1= 7 0 * U . 7 5 * ( Z P - Z 0 )
3b5 107 F I l - J..
3c6 FI2::>|.
367 IF ( K u . f C. 11 ) E I l r R O « F I O
368 IF(Ko.E C. 12.0R.KO.E0.22.0P.KO.EC.23.AND.J*(IGB-IGC>.GT,n)FIlrj*RO*
369
CAIMAC(Z )
373
ir(KO.: C . 2 3 . A N D . I G C . E 0 . 0 ) F 1 1 - F I 1 • R O Z * ( F I Z - A I M A G ( 7 ) )
371
IF IHU.L C . 1 2 . 0 R . K 0 . E 0 . 2 2 . 0 R . K 0 . E C . 2 3 . A N D . J » ( I G B - I G C ) . G T . D ) F I 2 = J « R O «
372
r A IM A c ( ZP - 7 0 ) / C A B S ( 2 P - Z 0 )
373
I F ( K 0 . fC . 2 3 . A N D . I G C . E 0 . " ) F I 2 : F I 2 - R 0 2 * A I M A G ( Z P - Z 0 ) / C A B S ( Z P - Z 0 )
371
A L 1 - .J.'
3 !5
AL2=..
376
AL3= J.'
377
ALI;,. J
37P
IF t .NOT ( K U . E 0 . ? l . O R . J * ( I G B - I G C ) . L T . n ) ) A L 1 - J * R 0 / A X
37°
IF(KO.' 0 . 1 1 . 0 r . K O . E Q . 1 2 . O R . K O . E 0 . 3 1 . A N D . J » ( I G B - I G C ) . G T . O I A L 2 : ( l J » l
3bO
381 £ ) / 2 I ' R O »wR
3o2
IF{Ku.L 0 . 2 1 ) A L 2 - 1 . I
3d3 IF(J»(I G B - I C C ) . L T . 1 ) A L 2 = J * J
3i;1 AL3:o«« L l
365
AL1=AL^
3u6 IF(J«(I G B - I G C ) . L T . 1 ) 4 L 1 : J
3b7 CP ) = cpr( I G8 , 2 , J I
3t>8 CPI=CPF (IGB,Z,1)
389 CP2=CPF (I RB , Z , 2 )
390 CP1=CA^ IF ( C P l , A N , A H )
391 CP2rCA^ IF(CP2,AN,AH)
392 IF(.NOT .LJOtbOrn 121
393 CP IdzCP F ( I G P , 7 F 1 , 3 )
391 CPl1:CP F ( I G ° , Z F J , 1 )
395 CPl0-CA f l " ^ ( C P l 1 , A N , AH )
396 r5 P ( I h ) Br B I 1 M ) * F I 1 - A L 1 « R F A L ( C P I ) - A L 2 / D L « ( D Y » R E A L ( C P 1 ) * 0 X * A I M A G ( C P 1 )
397 f )
398 3 B ( I t • ! )zoe(I^.•l|^pI2-/^|_3/oL*(OX*REAL(CPl)-
f J Y • A 11- A G ( C P 1 ) ) - S L 1 / ( D L * * 2 ) » ( D P * R E A L ( C P 2 ) ^ D V * A I M A G ( C P 2 ) )

165
399 IF(LJ'>.ANn.J.E0.1 )IM-niMl*2*NOB
1 in IF(LJ' )PB(IM l)=BB(IM))*FIl-AL 1 « R E AL ( C P 3 0 ) - A L 2 / D L » ( D Y * R E AL ( C P U ) '
1 ll f.DXJAi'-AGICPl i) )
1o2 I F ( L j O | ! S U S S : ( I M l * 2 . N P e - l ) * N V * I M J
1 J 3 IF(LJ")AA(ISU^S)=AL1
1 JI ISU=>-:(lMl^2*NRB-l)«NV»IM
1 J5 AA ( I S t i r S ) = A L 1
1.'6 IF(.NCT.1J.EQ.1.A-.|D.IP.FC.NR))GOTO 110
1 7 2 3 = Z ,A ( I G )
1 iR D B ( U * J ) - - A I M A G ( C P F ( I R , Z - ; , J ) )
1 ..9 I S U B S : | I M M : » N R ) * N V * I M * 3
11" AA(ISU' 5 ) ^ 1 . I
111 11 1 CONTI'-UF
112 C*

113 00 112 IRB=1,NRB

111 ZOVr^ZRA I I S B , I R 6 , 1 )
115 I F ( I R B . L T . N R B ) Z P V = Z R A ( I 6 B , I R B « l , l l
116 I F ( I R B . E 0 . N R B ) Z P V = Z R A ( I G B , 1 , 1 I
117 C
118 c» LEFT-HAND
119 c
120 Aa-CAF(Z V,ZP V, A N , A H , 0 )
,Z0
121
122
Al^CAF(2 ,20 v,zp V , A N , A H , I )
A2TCAF(Z ,zo V,ZP V , A N , A H , 2 )
123 AIR^CANI F ( C Al,A N,AH)
121 AIS-CANI F ( ( CI*C AI ) , A N , A H )
125 A 2 R = C A N I F ( C A2,A N, AH)
126 A 2 S = C A N I F ( ( CI*C A2) ,AN,AH)
127 F ( . N O T . L J Q ) GOTO 122
128 AOr) = C A F ( Z F O
,70V ,ZPV,AN,AH,ni
129 A n = CAF( ZFO
,ZOV ,ZPV,AN,AH,1>
130 A l O R r C A N I F I CAIT
131
,AN,AH)
A l O S r C A N I F ( (CI*
132
CAID),AN,AH)
S U B S = ( I M 1 ^ 1 RP-2
133 122 A ( I S U E S ) = AL
)*NV^IM
1*RF A L ( C A n ) ^ A L 2 / 0 L » ( D Y » R E A L ( C A l R ) ^
131 X»AIMAG( CAl
R) )
135 SUBS=(IM M I
136 Re-2 «NR8)*NV^IM
A ( I S U f - S ) = AL
137 1*RE A L t C I * C A n ) ^ A L 2 / D L * ( D Y « R E A L ( C A l S ) ^
X*AI>'AG( CAl
1 JS S) )
S U R i : : ( IM 1^1
139 A ( I SUES) -AL
R P - 2 )*NV^IM^1
110 D L * » 2 ) * ( DP*
3/UL » ( D X * P E A L ( C A 1 R ) - D Y * A I M A G ( C A 1 R ) ) ^ A L 1 /
111 S U B S = ( I M 1^1
REAL ( C A 2 R ) ^ D V * A I M A G ( C A 2 R ) )
112 A ( I S t l P S ) = AL
RP-2 NRB)*NV^IM+1
113 L l / (PL»» 2 ) •
3/OL » ( D X » R E A L ( C A 1 S ) - D Y * A I M A G ( C A 1 S ) ) ^
111 CA F ( . K O I . L (DP* R E A L ( C A 2 S ) ^ D V * A I M A G ( C A 2 S ) )
J j )
115 I SUBS=(IM GOTO 123
1^1
116 I A( TSIJ^S) RB-2 )*NV^IM"
= AL
117 AX*AI"'AC( 1*RE AL(CA'll.l)*AL2/DL*(DY*REAL(CA10R)^
CAl
IIP f 0 S U K ^ r t I M 1 ^ 1 JP ) )
119 I A( I S U . >;) = AL P - 2 • N R F ) * N V ^ I M O
R
ISO A X * A i H A R tC A l 1*RF A L ( C I * C A 1 0 ) » A L 2 / D L * t D Y * R E A L ( C A 1 0 S ) ^
l 2 l JS) )
CU F ( . N O T . ( J . E
1[.2 123 I A., = CAF ( 2 " ' , Z
C. 1. AND.IR.'^O.NR ) )G0T0 113
153 CS b E i - ( I M l + I
0 V,7 PV,AN,AH,0)
151 I A( I S I I l < ) - A I
RP-2
)*NV*IM^3
lt.5 ASUPs:(IM 1«I
MAG(
CA 1)
156 I A ( l o U - 5) : A I
WL-2
• NRl«NV^IM» 3
157 A ONT I M J F M«i:( CI'CAO)
158 11' L

159 112 CONTINUE

160 IF(J.Et.1)60T0 111
161 J=l
162 GOTO 115
163 111 I F ( .t«OT. ( K O . E a . 3 1 . O R . K O . E C . 2 3 . A N 0 . I 6 C . N E . ' 3 l >60TO 116
161 J = - l
165 115 IH^IbP

166
166 IGB-IGC
167 IGC=iH
168 RH=RC
1b9 RO-ROC
170 ROC-HH
171 HH;Ah
172 AKrAKC
173 AKC^KH
1 71 RHrAK
175 AN^ANC
176 ANC^kH
1 77 RHzAH
178 AHrAHC
179 A H C - K H
16" I H = NP
181 NPrNPr
1o2 NPC-IH
163 NR6 = i » A ( I G P , l , l )
1o1 CDZ=-(ZP-ZO)
185 UX=RLAL(COZ)
1b6 DYrAIM(iC(CDZ)
187 DL=CAPS(CDZ)
1o8 CPT2.otrX»0Y
169 ov:;nx**?-nY**2
19r IMl:l
191 IFdtE.EC.DGOTO 120
192 JH:ICP-1
193 DO 117 I : l , J H
191 IMi:IM1^2*NA(I,1,1)^2
195 117 CONTiNUE
196 120 I F ( J . N E . 1 )GOTO 1U7
197 116 I F ( L J O ) Ja::l
198 C**
199 105 CONTINLE
5..n 1 J3 CONTINUE
5 .1 RETUKN
5 J 2 END
5 i3 C
5 ll C
5 J5 C

5„6 COMPLF X F U N C T T U N C A F ( Z , Z O V , Z P V , A N , A H , J )
5 ,7 IMPLIC IT C O M P L E X ( C Z )
5 .8 CANF(Z ,AN,AH)=C"PLX(REAL(Z*CEXP(-(0.0,1.0)'AH)),(SORT(AN)*AIMA6I
5 9 r Z t C E x P ( - ( i) . 0 , 1 . 0 ) * A H ) ) ) )
510 ; P S : i " , rl»«^-^ I
511 P T 2 : 2 . 1*3.1115926535
512 C I = ( J . , 1 . .'))
513 N1 = 0
511 N2= J
515 ZOw;CA ' . F ( 7 U V , A N , A H )
516 ZPW;CA t F ( Z P V , A N , A H )
517 ZWzCit F ( Z , A N , A H )
518 I F t J . ^ :.J.AND.(CABS(ZW-70W).LE.EPS.OR.CABS(ZW-ZPH).LE.EPSl)ZW=Ztf+
510 TLPS
52" C=CAi.? (ZP w - z o w : / (ZPW-ZOW)
•^Jl C Z l : c * (ZW - z o w )
5 22 c z 2 ; c « (7M -ZPW)
523 ( 7P w-zoi.) * 0 . 5
5 21 IFt AIM AG( C Z l ) . L T .EP S . A N D . ( A B S ( R E A L ( C Z l ) ) - D ) . L T . E P S ) C Z l : C Z l ^ C I * E P S
5 25 I F ( A I K AC( C Z 2 ) . L T .EP S . A N n . ( A B S ( R E A L ( C Z 2 ) ) - n ) . L T . E P S ) C Z 2 : : C Z 2 ^ C I » E P S
526 IF( AI G( C Z D . L T . 3 0 . A N D . P E A L ( C Z l ) . L T . D ) N l : l
52 7 I F ( A 11- AG( C Z 2 ) . L T . 1 o . A N D . R E A L ( C Z 2 ) . L T . ( - 0 ) ) N Z r l
5^8 I F ( J.N c . -I )GOT0 1 J l
529 I F ( C A P S (2 W - Z O W ) . LE E P S ) C A F = - C Z 2 * C L 0 G ( C Z 2 ) - N 2 * P I 2 * C I * C 7 2
5io IF( CAf- S (Z W - Z P W )LF
551 . EPS)CAF=CZ1*CL0G(CZ1)^N1*PI2*CI*CZ1
I F t C A P S IZ W - z o w ) . GT F P S . A N n . C A B S ( Z W - Z P W ) . G T . E P S ) C A F - - C Z 2 * C L 0 G ( C Z 2 ) »
532 C C 2 1 * C L or,( C Z l ) ^ P I
5J3 2*C I * ( N 1 * C Z 1 - N 2 * C Z 2 )
GOTO 1
531 IF(J.r . 1 ) C A F : - C * ( C L 0 D ( C Z 2 ) - C L 0 G ( C Z 1 ) ) » P I 2 * C I » C » ( N 1 - N 2 )

167
535 I F ( J . F T . 2 )CAF = - C » ( 1 . l / ( Z W - Z P W 1 - 1 . 0 / ( 2 W - Z 0 W ) )
5 36 1.J2 CAFZCAF/PI2
5 37 RETUK^-
538 END
539
5ir
511

512 0MOLF> NCTI ON CPF(TG,Z,J)

517 "'PLiCI OMPL LX (C,Z )
511 OMMUN 1/ZP A(5,n),ZUA(5,5,1),Z3A(5),ZRA(5,lDd,l),DT,R0Z,FIZ,
515 lA(t), 5,7) ,ePA(5,lJ),WHA(5,lJ0,l),NG,P0R(5),
516 A(5,3, f.GA ( 5 , 5 , l ) , I F A ( 5 , l J ' l , l ) , K P A ( 5 , n O , l ) , I R A ( 5 , 1 0 3 , l ) , N F I
5l7 ANF(Z, AH)r C M P L X ( R E A L ( 7 * C t X P ( - ( 0 . 0 , l . Q ) * A H ) ) , ( S O R T ( A N ) * A I M A G (
518 *crxp( .•"1,1 . I ) * A H ) ) ) )
5lo 1 2 - ^ . ., 1115 9 2 6 5 3 5
5 6' PS^iO. (-1)
5D1
557
PF=('.. .":)
PrNA(I ,1 >
553 F ( N1 ,t ir.oT 0 101
551 NrPi(i
555
)
H=GA(I
5 56 WrCAMF- AN ,AH )
557 0 K2 ,NP
556 P; 0KA ( 11
559 P:?,"i ( I )
53" P-C A»'f , AN , AH )
Sol F (CA"^ -ZP).LE.EPS)ZU=ZW^EPS
5o? F ( J.ri CPFrCPF+0P/PI7*CL0G(2W-ZP)
5o3 F( J . E l . CPFrCPF^3P/(FI2*(ZW-ZP))
5o1 F(J.F^
565 CPF=C''F-0P/(PI2*(Zw-ZF)**2)
ONT.tU
566 1U2
FTU.iN
5b7 lOl
NO
5oP
569
5 7-"

571 COMPLEX FUNCTION C 0 F ( I 6 , Z , J l )

572 IMPLICIT COMPLEX (C,Z)
573 COMMON /C n / Z P A ( 5 , 1 0 ) , Z U A ( 5 , 5 , 1 ) , Z O A ( 5 ) , Z R A t 5 , 1 0 0 , 1 1 , D T , R 0 Z , F I Z ,
571 f.FIA(b),GA(5,7),0PA(5,10),WRA(5,10O,l),NG,P0P(5),
575 fNA(5,3,l),NGA(5,5,l),IFA(5,10n,l),KRA(5,100,l),IRA(5,100,ll,NFI
576 COMMON / C O Z / B " ( 5 J 0 ) , N V
577 CI = ( u . 1 , 1 .C)
578 AN;GA(IG, 3)
579 AH=GA(IG,1)
5 SCI NR^NAdG, 1,1)
561 IMirl
5o2 I F ( I G . E O . 1 ) G O T O 103
5B3 J-IG-1
5t)1 DO 1 J l 1=1,J
565 IMlrIMl42*NAlI,1,1)•2
566 1 Jl CONTIMiF
Sb7 1"3 COF = ( I . 1, " . , ; )
588 JO 1 .2 I R ^ l ,NB
580 ZOV^iPA(IG,IR,1)
590 I F ( l H . L T . N R ) Z P V =ZRA(ir, , I R 4 l , l )
591 IF(IK.fC.NR)2PV-ZRA(IG,l,l)
592 ISUBSl^IMl+IR-l
5V3 ISUBb2=IMl*IR«NR-l
591
CC = C M P L X ( B B d S U B S l ) , P P ( I S U B S 2 ) )
595
C0F:cnF4C0*CAF(Z,Z0V,ZPV,AN,AH,Jl)
596
CONTiNUE
59-'
ISURi=IMli2*\P
598
COF=l,rF.(l-Jl)»C«''LX(Fb(ISU°S),EBdSUBS^l))»CPF(IG,Z,Jl)
599
Ri^TU.(N
6 "\ END

168
Principal Notations

Q = 0 + iP complex potential (i?(~')

0 = kcj) potential (L't~^)
P stream function (L^f ')
<i> groundwater head (L)
Z = X + iy complex variable (L)
Z = X - iy complex variable (L)
x,y cartesian coordinates (L)
t'x' I', components of specific discharge (Lf ~')
At step in time (f)
k coefficient of permeability (Lf ~')
/' storage coefficient (dimensionless)
'/ effective porosity (dimensionless)
/' density (ML" ^)
P pressure (ML~f~')
Q discharge of source or sink (I?f~')
'll = '•j + '-l complex variable (Lf"')
distribution strength of sources (or sinks), resp. vortices (Lt~ ')
/ boundary variable (L)
dz d'z
-^^ complex variable for boundary segment (dimensionless)
d( dl'
Cs boundary resistance (f)
Ax, Ay components of displacement of a point of moving boundary in a
step in time ( L)
N precipitation (Lt~')
c subscript that refers to an adjoining sub-region
m number of sources and sinks
n number of boundary segments
complex constant
analytic function
Re{ } real part
Im{ j imaginary part

169
References

ABRAMOwtTZ M AND I A SiE'.bN 1968 Handbook of mathematical functions Dover Publications

New York
ARFKEN, G , 1970 Mathematical methods for physisists 2nd ed Academic Press, New York
BEAR, J , D ZASLAVSKI AND S IRMAY, 1968 Physical principles of water percolation and seepage,
UNESCO, Pans
BRUI EMAN G A 1972 Wiskunde in de geohydrologie R 1 D Den Haag
CHURCHILL, R V 158 Operational mathematics 2nd ed Mc Graw-Hill New York
DE JOSSELIN DE JONG, G , 1960 Singularity distributions for the analysis of multiple-fluid flow through
porous media. Journal of Geophysical Research, Vol 65 No 11
DE JOSSELIN DE JONO, G , 1969 Generating functions in the theory of flow through porous media
De Wiest (editor). Flow through porous media. Academic Press, New York
DITKIN, P A AND A P PRUDNIKOV, 1965 Integral trdnsforms and operational calculus Pergamon
Press Oxford
DE WIEST R J M , 1969 Green's functions m the flow through porous media,
De Wiest (editor), Flow through porous media Academic Press, New York
ERDELYI, A , W MAI.NUS, F OBFRHETTINGER, F TRICOMI, 1954 Tables of integral transforms, Mc Graw-
Hill, New York
K.OBER, H , 1957 Dictionary of conformal representations Dover Publications New York
KOZENY, J , 1931 Grundwasserbewegung bei freiem Spiegel, Fluss und Kanalversickerung Wasserkraft
und Wasserwirtschaft 3
KUIPERS, L AND R TIMMAN, 1969 Handbook of mathematics, Pergamon Press, Oxford
LAMB, H , 1932 Hydrodynamics 6nd ed , Cambridge University Press London
Mc NOWN, J S AND E Y Hsu 1949 Pressure distributions from theoretical approximations of the flow
pattern. Heat transfer and Fluid Mechanics Institute (Paper reprints by ASME pp 65 76)
SNEDDON, I H , 1972 The use of integral transforms, Mc Graw-Hill, New York
VERRUIJT A , 1970 Theory of groundwater flow, Mac Millan, London
VON KARMAN T , 1927 Berechnung der druckverteilung an LuftschifTkorpern Abh Aero Inst Aachen
No 6, pp 3-17
Vos, H C P DE, 1929 Enige beschouwmgen omtrent de verweekingslijn in aarden damnien De Water-
staatsingenieur 17(1929)335 354
WYLIE C R 1960 Complex variables and applications Mc Graw-HiU, New York

170
In the series of Rijkswaterstaat Communications the following numbers have been published before

No 1 * Tidal Computations in Shallon Watei

Dr J J Dronkerst and prof dr ir J C Schonfeld
Report on Hxdrostatic Leiellinq accros the Westerschelde
Ir A Waalewqn 1959

No 2 * Computation of the Deaa Pattern for the Netherlands Delta Works

Ir H Ph van der Schaaft and P Vetterli, Ing Dipl E T H , 1960

No 3 * The Aqinq of Asphaltic Bitumen

Ir A J P van der Burgh J P Bouwman and G M A Steffelaar 1962

No 4 Mud Disliihution and Land Reclamation in the Fastern Wadden Shallows

Dr L F Kampst, 1962

No 5 Modem Constiuction of Wing-Gates

Ir J C le Nobel 1964

No 6 A Structure Plan for the Southern IJsselmeerpolders

Board of the Zuyder Zee Works 1964

No 7 The Use of Explosiies foi Cleaiinq Ice

Ir J vander Kley, 1965

No 8 Vu Design and Constimlion of the Van Bnenenoord Budge actoss tin Riiei Nieuwe Maas
Ir W J vander Ebt 1968

No 9 Electronu Computation of Water Lei els in Rneis duiinq Hiqh Discharqes

Section River Studies Directie Bovenrivieren of Rijkswaterstaat 1969

No 10* The Canalization of the Lowei Rhine

Ir A C de Gaa> and ir P Bloklandt 1970

No 11 * The Haiinqiliet Sluices

Ir H A Ferguson ir P Bloklandt and ir drs H Kuiper 1970

No 12 The Application of Piecewise Pohnomuils to Piohlems of Cune and Suiface Appioximation

Dr Kurt Kubik 1971

No 13 Si stems foi Automatic Computation and Plottinq of Position Fixinq Patterns

Ir H Ph van der Schaaft 1972

No 14 The Realization and Function of the Notliern Basin of the Delta Project
Deltadienst of Rijkswaterstaat 1973

No 15 /•) sical-Enqineeiinq Model of Reinforced Conciete Frames in Coinpiession

Ir J Blaauwendraad 1973

No 16 Naiiqation Locks foi Push Tons

Ir C Kooman 1973

No 17 Pneumatic Ban leis to reduce Salt Intrusion through Lacks

Dr ir G Abraham, ir P van der Burg and ir P de Vos 1973

* out of print
In the series of Rijkswaterstaat Communications the follovMng numbers have been published before
(continuation)

No 18 E\peiiences with Mathematical Models used foi Water Qualit\ and Haiei Piohlems
Ir J Voogtanddr ir C B Vreugdenhil 1974

No 19* Sand Stahilizalion and Dune Buildinq

Dr M J Adnanianddr J H J Terwindl 1974

No 20 The Road-Picture as a Touchstone for the thiee dimensional Desiqn of Roads

Ir J F Springer and ir K E Hjizinga (also in German) 1975

No 21 Push ToHs m Canals

Ir J Koster 1975

No 22 Lock Capacit] and Traffic Resistance of Locks

Ir C Kooman and P A de Bruqn 1975

No 23 Computei Calculations of a Complex Steel Biidqe lenfied h\ Model Inusliqalions

Ir Th H Kayser and ir J Binkhorst 1975

No 24 The Kreekrak Locks on the Scheldt-Rhine Connection

Ir P A Kolkman and ir J C Slagter 1976

No 25 Motornni Tunnels built h\ the Imnicised Tube Method

Ir A Glerum ir B P Rigter ir W D Evsink and W F Hcins 1976

No 26 Sah Distribution in Estuaiies

Ri|kswaterstaat Delft Universitv of Technology Delft Hydraulics Laboratorv 1976

No 27 Asphalt Reietment of D\ke Slopes

C ommittee on the Compaction of Asphalt Revetments of Dyke Slopes 1977

* out of print

172
Stellingen

De door Verruyt in 1968 gepubliceerde exacte oplossing van het probleem van Badon
Ghijben kan worden uitgebreid tot een oplossing waarin de aanwezigheid van nuttige
neersiag op de freatische lijn is verwerkt
Verruijt A 1968 A note on the Ghijben Herzberg formula (Bull Int Ass Sci Hydrol
XIII 4-12 )
Van der Veer P 1977 Analytical solution for steady interface flow in a coastal aquifer
involving a phreatic surface with precipitation (J Hydrol 34 1-11)

II

Bij een geschikte keuze van twee constanten in de op een eendimensionale beschou-
wing gebaseerde formule voor de hgging van de grenshjn tussen zoet en zout grond-
water in het probleem van Badon Ghijben is deze formule gelijk aan die welke volgt
uit de exacte oplossing van het tweedimensionale probleem

Van Dam J t. 1976 Fresh Water Salt Water Relationships (college geohydrologic
TH Delft afd Civielc Techniek)
Van der Veer P 1977 Analytical solution for steady interface flow m a coastal aquiter
involving a phreatic surface with precipitation (J Hydrol 34 1 11)

III

De sinds 1929 bekende, door Vreedenburgh gevonden, exacte oplossing voor de

grondwaterstrommg naar een drainage, waarvan m 1959 door Glover een gewijzigde
vorm werd gepubliceerd die de strommg van zoet grondwater boven stilstaand zout
grondwater beschrijft kan worden uitgebreid tot een vorm die een gehjktijdige stro-
mmg van zoet en zout grondwater beschrijft
De Vos H C P 1929 Enigc beschouwmgen omtrent de verweekingslijn in aarden
dammen (De Waterstaatsingenieur 17 335 354)
Glover R E 1959 The pattern of fresh watei flow in a coastal aquifci (J Gcophys Res
69 no 8 457-459 |
Van der Veer P 1977 The pattern ol fresh and salt groundwater flow in a coastal aquiler
(Deltl Progr Rep 2 137 142)

IV

De oplossing voor een gehjktijdige stroming van zoet en zout grondwater kan met
met behulp van de hodograafmethode worden gevonden de exacte oplossing voor
het geval waarbij behalve het grensvlak ook een freatische hjn aanwezig is, dus het
probleem van Badon Ghijben uitgebreid met de stroming van zout grondwater, kan
echter toch worden gegeven.
Van der Veer. P . 1978 Analytical solution for a two-fluid flow in a coastal aquifer
involving a phreatic surface with precipitation (J Hydrol . 34 271-278)

In een bijzonder geval van grondwaterstroming met een freatische lijn en een grens-
lijn tussen zoet en (al dan niet stromend) zout grondwater is zowel de freatische lijn
als de grenslijn een rechte lijn; de grenshjn is ook een rechte als er in plaats van de
freatische hjn een rechte kwellijn aanwezig is
Van der Veer. P . 1977 Analytical solution for steady interface flow in a coastal aquifer
involving a phreatic surface with precipitation (J, Hydrol 34 1 11)
Van der Veer. P . 1978 Analytical solution for a two-fluid flow in a coastal aquifer
involving a phreatic surface with precipitation (J Hydrol . 34 271-278)

VI

Door de definiering van een geschikte variabele kunnen analytischc oplossingen wor-
den gevonden voor problemen van tweedimensionale grondwaterstroming in gebie-
den waarvan de rand bestaat uit een honzontale semi-doorlatende hjn en (eventucel)
verticale stroom- en/of potentiaallijnen. In deze gebieden mogen putten en bronnen
voorkomen.
Van der Veer. P . 1978 Exact solutions for two-dimensional groundwater flow problems
involving a semi-pervious boundary (J Hydrol . 37- 159-168)

VII

De in de vorige stelling bedoelde variabele is niet geschikt voor het berekenen van de
grondwaterstroming tussen twee honzontale semi-doorlatende lagen. ook al zijn de
weerstandseigenschappen van die lagen gelijk.

VIII

Naast het verschil tussen de hoogtehgging van de freatische lijn legen een gesloten
taludbekleding en de buitenwaterstand tegen het talud kan het karakter van de
grondwaterstroming. (stationair of niet-stationair), een grote invloed hebben op
de overdrukken onder die taludbekleding.
Van der Veer. P . 1976 Overdrukken onder gesloten dijkbeklcdingen Geohydrologischc
aspecten van de waterbouwkunde 1 (Pt-b. 31. nr 9. 547 550)
 IX

Het zoeken naar de analytischc oplossing van een stromingsprobleem geeft veelal
meer inzicht dan het vinden van een oplossing met behulp van een numeriek model.

Het verdient aanbeveling om de actieve muziekbeoefening op scholen sterk te bevor-

deren.

XI

De Nederlandse Spoorwegen verdienen een compliment van de minister van Weten-

schappen omdat het sinds dc invoering van zgn. werkcoupe's in treinen voor foren-
zen mogelijk is om wctenschappelijk onderzoek te doen tijdens treinreizen in het
woon-werkverkeer.

XII

Het vervangen van auto's door trapauto's kan een gunstige invloed hebben op de
verkeersveiligheid, het energieverbruik en de volksgezondheid.

Stellingen behorende bij het proefschrift Calculation Methods for Two-dimensional Groundwater Flow
van P. van der Veer, Delft, 1978.