GEOLOGY AND WATER

DEVELOPMENTS IN APPLIED EARTH SCIENCES

VOLUME 1
GEOLOGY AND WATER
An introduction to fluid mechanics
for geologists

by

RICHARD E. CHAPMAN
University of Queensland,
St. Lucia, Brisbane, Australia

MARTINUS NIJHOFF / DR. W. JUNK PUBLISHERS /1981

THE HAGUE / BOSTON / LONDON
Distributors:

for the United States and Canada

Kluwer Boston, Inc.

190 Old Derby Street
Hingham, MA 02043
USA

for all other countries

Kluwer Academic Publishers Group

Distribution Center
P.O. Box 322
3300 AH Dordrecht
The Netherlands

This volume is listed in the Library of Congress Cataloging in Publication Data

ISBN-13: 978-94-009-8246-8 e-ISBN-13: 978-94-009-8244-4

DOl: 10.1007/978-94-009-8244-4

Copyright © 1981 by Martinus Nijhoffl Dr. W. Junk Publishers bY, The Hague.
Softcover reprint of the hardcover 1st edition 1981
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system,
or transmitted in any form or by any means, mechanical, photocopying, recording, or otherwise,
without the prior written permission of the publisher,
Martinus NijhoffiDr. W. Junk Publishers by, P.O. Box 566,2501 CN The Hague, The Netherlands.
 THIS BOOK
is dedicated
to my wife
JUNE
with love and affection

Where is the Life we have lost in living?

Where is the wisdom we have lost in knowledge?
Where is the knowledge we have lost in information?
T.S. Eliot, 1934, The Rock
 CONTENTS

Preface XI

Symbols XV

1. Introduction: Liquids at rest 1

Introduction 1
Properties of fluids 1
Statics 5
Buoyancy 7
Liquid-filled porous solids in static equilibrium 11
Bulk density 13
Pressure and stress 14
Field confirmation of Terzaghi's relationship 18
Dimensional analysis 22

2. Liquids in motion 26
Bernouilli's theorem 30
Pipe flow 34
Dimensionless numbers 39
Pressure 39
Stokes' law 40
Hydraulic equivalence 43
Note on Poiseuille's experiments 46

3. Liquid flow through porous sands 49

Henry Darcy's experiments 49
The coefficient of permeability 52
Tortuosity 55
The limits of Darcy's law 64
Upper limit of Darcy's law 65
VIII

4. The aquifer and fields of flow 73

Fields of flow 76
Producing water well 82
Unconfined aquifers 88
Natural sinks 89

5. Aquifers: Springs, rivers, and man-made drainage 93

Water from the Danube to the Rhine 98
Oceanic islands and coastal aquifers 105
Arid regions: The qanat ofIran 110
The Great Artesian Basin 114

6. Movement of pore water, and abnormally high pore pressures 124

Compaction of sediment 124
Compaction of sands 125
Compaction of muds, clays, and 'shales' 126
Pore pressures 131
Other possible causes of or contributors to abnormally-high
pore pressures 140
Pore-water migration in sedimentary basins 145

7. Role of pore water in deformation of sedimentary basins 149

Pre-orogenic deformation 149
Growth structures 149
Causes of growth structures 154
Orogenic deformation 161
Lubricated sliding 161

8. Pore water and sliding 173

Gravitational sliding 175
Un lubricated 175
Lubricated sliding 180
Effect of sea-level 187

9. Conclusion 191
Faults and water movement 191
Flow of two immiscible liquids in porous sediments 194
Capillary pressure and surface tension 202
 IX

Appendix. Review of commonly-quoted works on Darcy's law 205

Discussion 210

Conversions 214

Glossary 218

Postscript 223

Subject index 224

Author index 227

 PREFACE

Water is one of the world's threatened resources: it is also a substance of importance

in Geology. For some years I have felt the need for a book that sets out the
fundamentals of fluid mechanics, written for geologists rather than engineers. The
efforts to repair my own deficiencies in this respect led me along various unfamiliar
paths, few of which were unrewarding. This book is the result of my journeys
through the literature and as a geologist in several parts of the world.
It has been written for students of geology of all ages, in the simplest terms
possible, and it has one objective: to provide a basis for an understanding of the
mechanical role of water in geology. It has not been written for experts in ground-
water hydrology, or specialists in the fluid aspects of structural geology: it has been
written for geologists like me who are not very good mathematicians, so that we can
take water better into account in our normal geological work, whatever it might be.
The fundamentals apply equally to mineralization, geochemistry, and vulcanology
although they have not been specifically mentioned. It has also been written for the
university student of geology so that he or she may start a career with some
appreciation of the importance of water, and understanding of its movement.
A casual glance through the references at the end of each chapter (which include
works of interest in addition to those specifically cited) will reveal many works
published over 50 years ago, some over 100, and a few over 200 years ago. This
should not come as a surprise, because there has been little new at an elementary
level in the last 40 years. The very old references have been included because many of
them reveal deep insight into the phenomena they discuss, and I have found second-
hand reports of their contents frequently misleading. I hope the reader will find
them as rewarding as I have. In any case, let the reader not underestimate the value
of the older work. For example, Poiseuille's expression for the kinematic viscosity of
water at temperatures to 45°C (the range of his experiments), published in 1841,
could be used today with little error.
Stokes would doubtless be appalled at some of the uses to which the law that bears
his name is put, because he realized its limitations. Perrault is sometimes called 'the
father of ground-water hydrology', but Mariotte is more worthy of that title. There
is little doubt that Perrault failed to understand most of what he discussed, while
Mariotte had a very good understanding of his subjects.
From around the turn of the Century, the works of King and of Veatch are full of
XII

interest; and I suspect that the pace of modern development would preclude repeti-
tion of many of their studies of the fluctuations of water-levels in wells.
Perhaps the greatest surprise of my journeys through the literature concerns the
origin of what is now known as the Ghijben-Herzberg relationship, in coastal
aquifers. Badon Ghijben, to use his proper name, was the junior author ofthe note
that has so often been cited (one wonders how many of those who cited him had
actually seen the note). Drabbe and Badon Ghijben clearly understood the move-
ment of ground water under their dunes, but they are only remembered for the 1: 42
ratio of the Ghijben-Herzberg lens. And I find it hard to believe that Herzberg,
writing 12 years later, had not heard of Drabbe and Badon Ghijben from just along
the coast into Holland. Soon afterwards, another outstanding Dutchman appeared
- Jan Versluys - whose work has been almost entirely neglected. True, he wrote
much of his early work in Dutch (but so did Drabbe and Badon Ghijben), and his
paper on the dune-water theory, so full of understanding and perception, was
incompletely translated into English and published without his diagrams.
In drawing attention to these early workers, it has not been my intention to
denigrate later workers - far from it. Outstanding amongst later workers in the
context of geology is M. King Hubbert, alone and with various collaborators. His
contributions to the understanding of fluid flow, and his applications of mechanics
and fluid mechanics to structural geology, have been central to my studies of the last
decade or so. Forty years ago he pointed out a number of common misunderstand-
ings in the matter of fluid flow. But the misunderstandings have persisted, and are
certainly no less common now than they were then.
I have sought understanding at a simpler level, and in so doing make no apol-
ogies. The simple approach has positive merits, not the least ofwhich is that it can be
understood by people with few mathematical skills. A simple approach yields an
expression for intrinsic permeability that takes porosity and the tortuosity of the
pore passages into account - at least to the point where an understanding of their
roles is achieved. It shows Darcy's law to be applicable to porous glass as well as to
glass beads, and that the dimensionless coefficients in these two extremes differ only
by a factor of about two, whereas the permeabilities differ by a factor of at least 109 .
This in turn suggests that fluid movement in very small pores is no different in
principle from that in large pores (in spite of many published statements to the
contrary). An extension ofthis simple approach to the vexed question of the flow of
two immiscible liquids through porous material yields results that are consistent
with experimental data without calling on anisotropy or any other feature to
account for the shape of relative permeability curves, and the well-known fact that
the sum of the relative permeabilities of two immiscible liquids is always less than
unity.
In structural geology, the simple approach leads to results of the correct order of
magnitude for such things as the proportion of a block that can be overthrust by
sliding on a gentle slope.
 XIII

It is my hope that readers will not only find these and the other topics of
interest, but will use them as a stepping-stone to better, more rigorous, studies.
The first two chapters outline the fundamentals offluid statics and dynamics, and
develop simple equations for flow in open channels and in pipes. The third chapter
builds on this base, and examines Darcy's law and the flow of liquids through
porous materials. The coefficient of permeability is expanded to include the relevant
measurable properties of sedimentary rocks, and the upper and lower limits of
Darcy's law are examined.
In Chapter 4 we take a larger view, looking at the flow of water through aquifers
and comparing the influence of a producing water-well with the influence of a
stream or river. This is followed in Chapter 5 by some examples of the interaction
between aquifers, springs and rivers.
Chapter 6 examines the important topic of abnormally high pore pressures in
sediments, and their probable and possible causes. Interest in this topic should not
be confined to petroleum geologists: it is a common phenomenon of great con-
sequence in the early deformation of many sedimentary basins - which is the topic of
Chapter 7. This leads on, in the next chapter, to another important role of water in
geology: the sliding of large blocks of sedimentary rocks. Both lubricated and
un lubricated sliding are considered.
In Chapter 9 we conclude with two examples of the application of the principles
discussed in early chapters: a qualitative discussion of water flow along fault-planes,
and a semi-quantitative analysis of relative permeability when a porous material is
saturated with two immiscible liquids.
This work has, of course, involved the study of many works. In the Appendix I
have reviewed a number of papers that are commonly quoted in the geological
literature (and translated their equations into the notation of this book).
SI units have been used throughout. They are, I believe, a great help once one gets
used to them. All students are brought up with them in school and university: the
older geologists will have to conform. But I have not used all the prefixes. Con-
version tables giving the SI units in various other measures are included. A simple
pocket calculator will enable the reader to make cross-conversions between non-SI
units.
A Glossary is also included. I have had great difficulty with this because many
accepted definitions in ground-water hydrology are inconsistent with the physical
principles that are involved. For example, it makes no sense to define 'hydraulic
gradient' as the rate of change of pressure head because pressure head changes in an
inclined aquifer in which the water is static.
XIV

ACKNOWLEDGEMENTS

During the preparation of this book I have received help, advice, and constructive
criticism from many people whose expertise lies in the fields in which I was seeking
to repair my deficiencies. I am conscious of having discussed some aspect of this
work with almost every chemical engineer, civil engineer, applied mathematician,
physicist, geophysicist and geologist in this University. To them all, I am deeply
grateful. I have also benefitted from the advice of others around the world. In
particular, I express my thanks to C.J. Apelt, J .H. Doveton, J.M. Fitz-Gerald, R.G.
Font, V.G. Hart, D.A.L. Jenkins, W. Kass, R.A. Nelson, N. Street, and J.P. Webb.
This work would still be incomplete were it not for the tenacity, tolerance and
good humour of Carolyn Willadsen (in the early stages) then Margaret Eva in the
Geology Library of this University. They tracked down incorrect and unverifiable
references, and acquired numerous works from other libraries. I am most grateful to
them, and to their assistants.
I am also very grateful to Mrs Irene Lenneberg for drawing almost all the
illustrations.
For data, and permission to publish their previously-published figures, I am
indebted to R.A. Downing for figure 5-4, W. Kass for Figures 5-5 and 5-6, K.
Lemcke for Figure 5-7, and M.A. Habermehl for Figures 5-17 and 5-18.
I acknowledge with gratitude the following for permission to include data they
have provided and material of which they hold the copyright:
The American Association of Petroleum Geologists, Tulsa, for Figures 1-17,7-5
and 8-4;
The British Petroleum Company, London, for Figures 7-8, 7-10, and 8-9;
The Director of the Bureau of Mineral Resources, Canberra, for Figures 5-17 and
5-18;
Elsevier Scientific Publishing Company for several figures used in my book
Petroleum Geology.
The Institution of Water Engineers and Scientists, London, for Figure 5-4;
The Royal Dutch/Shell Group for the data of Figures 6-3 and 6-12;
The Texas Department of Water Resources for Figures 1-16 and 5-2.
If I have inadvertently omitted any name, I apologize. All weaknesses ofthe book
are due to me, and not to those who have done so much to ensure that it is free of
error.

Brisbane, 29 June 1980

The following work, received when this book was in press, will be of interest to readers of Russian:
Gurevich, A.E., 1980. Prakticheskoe rukovodstvo po izucheniyu dvizheniya podzemiykh vod pri poiskakh
poleziykh iskopaemYkh. Nedra, Leningrad, 216 pp.
 SYMBOLS

A Area [U]
C,c constants
Cr drag or resistance, coefficient, dimensionless
D diameter available to fluid flow [L]
d diameter of granular material [L]
d* hydraulic equivalent size of quartz sphere [L]
E energy [ML 2 T- 2 ]
e base of natural or Napierian logarithms
F force [MLT- 2 ]. Formation Resistivity Factor,
dimensionless
f fractional porosity, dimensionless
Fr Froude number, dimensionless
g acceleration due to gravity [LT- 2]
h head, thickness [L]
K coefficient of permeability, hydraulic conductivity [Lr 1]
k intrinsic permeability [L 2 ]
ko effective permeability to oil [L 2 ]
k ro = k o/ k relative permeability to oil, dimensionless
L dimension of length
length [L]
M dimension of mass
m 'cementation' factor, dimensionless
N newton, SI unit of force (kg m s - 2)
Pa pascal, SI unit of pressure (N m -2 = kg m -lS-2)
P pressure [ML -IT- 2 ]
Pe pressure of ambient fluid
Q volumetric flow rate [L 3 T- 1 ]
q = Q/A specific discharge, or discharge velocity [LT- 1 ]
R hydraulic radius [L]
r radius, correlation coefficient
Re Reynolds number, dimensionless
S specific surface [L - 1 ], hydraulic gradient (dimensionless),
total normal stress [ML -1 T- 2]
XVI

s saturation, proportion of pore space occupied by fluid,

dimensionless (esp., water saturation)
T dimension of time, tangential component of total stress
[ML -1T- 2 ], tortuosity, dimensionless
t thickness [L], time [T]
U=gz gravity potential [L 2T- 2]
V velocity [Lr 1 ]
v volume [L 3 ]
W,w weight, partial weight [MLr 2 ], width [L]
X exponent in porosity term in intrinsic permeability
(dimensionless), horizontal co-ordinate [L]
y horizontal co-ordinate [L]
z vertical co-ordinate [L]

a,/3 Angles, degrees (dimensionless)

'Y weight density, specific weight [ML - 2 r 2 ]
LI difference of, e.g., Llh, Lit
f5 dimensionless parameter that takes pore-fluid and ambient-fluid
pressures into account in various mechanical contexts;
ratio of effective normal stress to active normal stress, a/ae
void ratio = //(1 - j), dimensionless
'1 dynamic viscosity [ML -1 r 1]
() angle of slope, degrees, dimensionless
A=p/S proportion of overburden supported by pore fluid, dimensionless
Ae = PelS proportion of overburden supported by ambient fluid
f.l, micro (10- 6 ) f.l,s, micro-second
V='1/p kinematic viscosity [L 2 T- 1 ]
II dimensionless numbers
p mass density [ML -3]
a effective normal stress [ML - 1T- 2],
surface tension [MT- 2]
active normal stress, in absence of abnormal pore-fluid pressures is
equal to effective normal stress
T tangential component of total stress [ML -1 T- 2 ]
frictional shear stress parallel to sliding surface
[ML- 1 r 2]

cohesive strength, shear strength when a = 0 [ML -1 T- 2]

fluid potential (Hubbert's) [L 2 T- 2 ]
angle of friction, grade scale
function of ...
 1. INTRODUCTION: LIQUIDS AT REST

INTRODUCTION

Geology is the study of rocks, and we tend to think of rocks in terms of the solid
mineral constituents. This is natural because almost the rocks we examine, in
outcrop or in hand speciment, are dry. We are apt to forget that, at depths below
only a few metres from the surface, almost all rocks in nature are saturated with
water (exceptionally and locally, also oil or gas). Some of this water is important to
us as a source of fresh water for drinking, agriculture and industry: much of it is not,
being too salty. But whether we can use it or not, it is there as an integral part of the
rocks.
Of recent years, more and more emphasis has been placed on the water in the
rocks. The stimulus has come mainly from petroleum geologists and geophysicists.
This is also natural, not only because their prime interest is in the fluids but also
because the petroleum geologist these days rarely examines rocks either in hand
specimen or in outcrop because the rocks he studies are out of reach at considerable
depths below the surface, commonly under the sea. They must rely on indirect data
on the rocks in situ, mainly in the form of electrical and acoustical measurements.
These measurements depend largely on the nature of the fluids in the rocks.
Progress in the understanding of the role of water in geology has been hampered
by widespread misunderstanding of the physics involved. Our purpose is to seek a
basic understanding of the physics of pore water and its movement, at least to the
point where we can use that understanding intelligently in geology and the applica-
tions to ground water. In seeking this goal, we acquire the tools we need for an
understanding ofthe more complex role that pore water plays in structural geology
- from the deformation of sedimentary basins to the sliding oflarge rock masses. We
therefore start with the most elementary considerations.

Properties offluids
A fluid is a substance that yields at once to shear or tangential stress. It differs from a
solid in a number of obvious respects, but it is important for the geologist to be more
precise than the engineer in these concepts. Engineers sometimes define fluids as
material that yields in time to the slightest shear stress. This definition is perfectly
2

satisfactory provided the time-scale is short; but it obscures the distinction between
fluids and solids for geologists because many geological materials that the engineer
regards as solid will yield to slight stress over long periods of time. For example,
pitch can be broken with a hammer, but it flows (as in Pitch Lake, Trinidad) if
smaller stresses are applied for a longer time.
However, we cannot at the outset become distracted from our main purpose,
which is to examine the role of fluids, as distinct from solids, in geology. So we shall
regard as solid those materials that resist shear stress applied for a short time,
measured in hours or days: and we shall regard as fluid those that do not.
This fundamental property of fluids is a consequence of relatively wide molecular
spacing, compared to solids. Fluids may be liquids or gases, and these too differ in
the matter of molecular spacing, so that gases yield to shear stress more readily than
liquids. The molecular spacing also accounts for the fact that gases are more
compressible than liquids, and liquids more compressible than solids. By compress-
ible, we mean that a volume of the material can be reduced by the application of an
exterior force.
A fluid, as the name implies, can flow. Some fluids flow more readily than others.
Viscosity ('1, v) is the property of internal resistance to flow; the greater the resistance
to flow, the higher the viscosity. An ideal fluid is one in which there is no internal
resistance to flow: it has no viscosity. Real fluids have viscosity, and this property is
important in determining the rate at which fluid flows.
Consider two adjacent layers within a liquid, or a liquid between two closely
spaced plates (Fig. 1-1) of which the top plate moves at constant velocity, V, relative
to the lower plate. Fluid adheres to each plate, so that the fluid against the upper
plate moves at velocity V, while that against the lower plate does not move. Between

dz

Fig. 1-1. Newtonian concept of viscosity.

 3

the two, we can visualize progressive slippage of one layer over the adjacent layers.
This is Newton's conception of viscosity.
The coefficient of viscosity, 1], (also known as the absolute and dynamic viscosity)
is defined as the ratio of shear stress to the rate of shear strain:

(1.1)

In words, the coefficient of viscosity is the tangential force per unit area that
maintains unit relative velocity between two parallel planes unit distance apart. It
. . ML -IT- 2
has dImensIOns LT 1 L 1 = ML -1 T- 1
The unit of viscosity is newton second per square metre (N s m -2) or the poise, P,
which is dyne second per square centimetre in units of force, length and time; or
grams per centimetre second (g em - 1 S - 1) in units of mass, length and time. (The
poise is named after the French physician, 1.L.M. Poiseuille, 1799-1869, whose
interest in blood flow led him to conduct a beautiful series of experiments on flow
through small pipes.)
If the coefficient of viscosity remains constant through the fluid, the fluid is said to
be newtonian, and its viscosity newtonian. Non-newtonian fluids, of course, are those
in which the coefficient of viscosity does not remain constant. We can see the mutual
relationships in Fig. 1-2. The ideal fluid is represented by the horizontal line at r = 0
(no shear stress, finite rate of shear strain). The ideal elastic solid is represented
by the vertical line at dx/dz = 0 (finite shear stress, zero rate of shear strain).

Ideal elastic solid

Non-newtonian fluid

IJ)
IJ) Newtonian fluid
........
Q)

IJ)

...
Q)
Q)
.s::
(J)

Ideal fluid
~-----------------------.
Rate of shear strain (dx/dz)
Fig. 1-2. Schematic relationships between shear stress and rate of shear strain for various types of
materials.
4

Newtonian fluids are represented by straight lines passing through the origin, while
non-newtonian fluids are represented by curved lines. Viscosity, with time as one of
its dimensions, is a concept that links fluids and solids.
The ratio of dynamic viscosity to mass density, 11/ p, appears in many expressions
in fluid mechanics. It has dimensions L 2T- 1 , and is known as kinematic viscosity (v)
because it concerns motion without reference to force. The unit of kinematic
viscosity is m 2 s- 1 in SI units, or the stoke, S, which is cm 2 /s. (The stoke is named
after the British mathematician and physicist Sir George Stokes, 1819-1903, whose
work on the effect of the internal friction of fluids on the motion of pendulums led,
amongst other things, to what is now called Stokes' law for the terminal velocity of a
sphere falling through a fluid.)
Density (p) is defined as the mass of unit volume of the substance. Its dimensions
are ML -3. The weight of unit volume of the substance (pg, where g is the accelera-
tion due to gravity) is called the specific weight or weight density (1'). Its dimensions
are ML -2T- 2 . The unit of mass density (it is better to qualify it so, and avoid
ambiguity) is kilogramme per cubic metre (kg m - 3) in SI units, or gramme per cubic
centimetre (g cm - 3). The unit of specific weight or weight density (as distinct from
mass density) is newton per cubic metre (N m - 3). In many practical applications the
specific weight is converted to units of pressure per unit of vertical length, such as
kg-f/cm 2 per metre or psi/foot.
The reciprocal of mass density is specific volume.
Density also involves a matter of scale in the dimensions of the 'unit volume': it
must be large relative to the particle size in order to give a representative figure. For
example, consider the density of a heterogeneous material such as a porous sand.
The unit volume can be taken so small that it only represents the pore fluid, or the
grain material. There is some minimum size that statistically represents the material,
with its bulk proportions of solid and fluid.
Specific gravity, the ratio of the density ofa substance to that of pure water at a
standard temperature (4 DC for scientists, 60 OF for engineers) is a dimensionless
quantity that will not be used because it is both inconvenient and commonly
ImprecIse.
The pressure (p) at a point in a fluid is defined as the force per unit area on a very
small area surrounding the point. Its dimensions are ML - 1 T- 2. The SI unit of
pressure is the pascal, or newton per square metre (pa = N m - 2 = kg m -1 s - 2).
Other units in use are kg-f/cm 2 and psi (pounds per square inch).
Stress is also a force per unit area, but it is not strictly synonymous with pressure
because the force leading to a state of stress is a surface force that can be resolved
into a normal component (a) and a shear or tangential component (r). Stress is a
tensor quantity, and the definition of a state of stress at a point in a solid requires six
quantities to be known (see any text book on structural geology or rock mechanics).
 5

STATICS

The inability of fluids to resist shear stress leads to a number of fundamental

propositions.
1) The free surface of a liquid in static equilibrium is horizontal. This is a matter of
common observation and practical application. To demonstrate this, we postulate
an angle of slope eto the free surface (Fig. 1-3) (the condition of static equilibrium
requires the fluid above the free surface to be less dense that the liquid we are
considering). The weight W of a small element of the liquid at this surface can be
resolved into a normal component, W cos e, and a shear component, W sin e. Since
liquid cannot sustain a shear stress, emust take the value at which the shear stress is
zero, i.e., e = o.

Fig. 1-3.

2) Surfaces of equal pressure in static fluids are horizontal. Consider a volume of

liquid in a right-cylindrical container (Fig. 1-4a + 1-4b). The weight of this liquid is
pgAd and it is supported by the base of area A. The pressure on the base is therefore

pgAd
p=--=pgd (1.2)
A

The pressure is dependent only upon the density of the liquid and its depth (we
assume g to be constant for practical purposes). If p is not constant and not a
function of d, shear stresses will exist that will cause flow until they are eliminated
and p becomes a function of d only. The shape of the container does not alter this
(Fig.l-4b).
3) The pressure at a point in a liquid in static equilibrium is equal in all directions.
Consider a small prism of the liquid (so small that its weight is insignificant
compared to the pressure) of which side b has unit area (Fig. 1-5). The area of side a
is then sin a; and of c, cos a. The force (pressure x area) exerted by the liquid on side
a is Pa sin a, and on side c it is Pc cos a. The force Ph on unit side b can be resolved into
6

(a)

1=------------------- = =

Fig. 1-4. A

Fig. 1-5.

horizontal and vertical components, [lb sin a and Ph cos a respectively. Since the
prism is in static equilibrium, the sum of the forces acting on it must be zero. Hence,
Pa sin a ~ [lb sin a = 0; and Pa = [lb,
pc cos a - [lb cos a = 0; and Pc = [lb,

and the forces acting on the two parallel sides are clearly equal and opposite from
the second proposition. Since Pa = [lb = Pc is independent of the angle a, which may
 7

arbitrarily assigned a value and orientation, the pressure at a point in a static liquid is
equal in all directions. (Note carefully that the pressure on a submerged object is not
equal in all directions.)
This leads to a further proposition:
4) The horizontal force acting on a surface in a static fluid is the product of the
pressure and the vertical projection of the area. Consider the prism in Figure 1-5
again. The force acting on side b, of unit area, is [Jb, the horizontal component of
which is as before [Jb sin a. But sin a is the area of side a, which is the area of the
projection of b onto a vertical surface. This proposition can also be shown to be true
for curved surfaces by considering a very small surface area and its tangential slope.

BUOYANCY

It will be convenient at this point to consider buoyancy, lest the reader think that we
are concerned with trivialities. Buoyancy is a superficially simple process that acts
on all materials immersed in fluids, and intuitive reasoning can be misleading.
The matter of the pressure exerted by a static liquid on a point P within it is
usually demonstrated as follows (Fig. 1-6):
Consider a small vertical cylinder of the liquid of cross-sectional area A, extending
from the point P to the surface of the liquid d units above. The forces acting on the
vertical surface ofthe cylinder are all horizontal, and since no shear stress can exist,
there is no vertical component. The weight of the cylinder, pgAd, is therefore
supported by an equal force acting upwards on its base,

Fig. 1-6.
8

pA = pgAd.
Hence
p = pgd,
which is the result we had before. Superficially, this seems satisfactory enough. Now
extend the argument to the pressure exerted by the liquid at a point on the lower
boundary surface, in contact with the container. This pressure (Fig. 1-7) is clearly
pgh; but what do we mean by the force acting on the base of the cylinder when the
base of the cylinder is the container?
If a container similar to that in Figure 1-7, but with a very thin, flexible but
impermeable base, is inserted into a larger container with identical liquid until the
levels in both coincide (Fig. 1-8) the physical situation of Figure 1-6 is restored, and
we can visualize the effect in terms of equilibrium and vertically-directed forces.
There is really no doubt that the pressure in the liquid at the base of the container is
pgh.
Now, instead of a conceptual cylinder we insert a real solid cylindrical rod of
impervious material, with length h and mass density ps that is greater than the
density of the liquid, p. It is a simple matter to verify that as the rod is inserted, so its
weight (as measured by the tension on the top end) decreases. What are the forces
acting on the rod when it is in the position comparable to Figure 1-6?
The weight of the rod in air is PsgAh. Over the immersed portion, liquid pressure is
acting on the vertical surface without shear components, so this cannot reduce the
weight. But at the base ofthe rod there could be an upward vertical force pgAd. It is a
High School experiment to show that the weight on immersion to depth d in the
liquid is given by

Wd PsgAh - pgAd (1.3)

Fig. 1-7.
 9

or, in words, the weight of the rod is reduced by the force due to the liquid at depth d
acting upwards on the base of the rod. Alternatively, since pgAd is the weight of the
liquid displaced by the rod, the weight of the rod is reduced by the weight of the
liquid displaced (Archimedes' Principle). These are valid alternative numerical state-
ments, but are they valid alternative physical statements?
Consider the total immersion ofthe rod, with the liquid level adjusted to coincide
with its top, and the base of the rod now attached to the container in such a way that

Fig. 1-8.

Fig. 1-9. Solid cylinder inserted into liquid.

 10

the liquid is totally excluded, and the rod becomes an integral part of the container
(Fig. 1-10). There is no doubt that its weight is now

W h = PsgAh - pgAh = (Ps - p)gAh (1.4 )

but all forces are apparently normal to the vertical surface of the rod, and there is no
basal surface on which the liquid pressure can act: the weight of the rod has been
reduced as if the liquid acted on the basal surface. The physical validity of the
concept of an upward force on the base of the rod is compromised: the real force is
due to the density difference, and it acts over the whole submerged volume of the
rod. We must regard the hydrostatic pressure field as being continuous throughout
the space occupied by the rod, as well as the liquid. It is supporting the partial
density p of the rod: the remainder, (Ps - p), is as if it were in a vacuum. Con-
firmation ofthis can be found in the observable results of running drillpipe into deep
boreholes filled with mud.
In the mid-1950s, development drilling in several oil fields was hampered by what
was known as 'wall-sticking' or 'differential sticking'. The pressure of the mud
acting on the drillpipe opposite depleted reservoirs with pore pressures considerably
below normal hydrostatic, held the pipe firmly to the wall of the hole. To try to free
the pipe, it was pulled at the surface. The amount of pull was limited by the tensile
strength of the pipe. The question was: what is the weight of the drillpipe above the
stuck point? That is, does the force of buoyancy act on the drillpipe above the stuck
point? If so, the pipe could be pulled that much harder: if not, as was commonly
believed in the early 1950s, the pipe could only be pulled to its tensile strength (less
some safety factor).
Consider the forces acting on a open-ended pipe hanging freely in a borehole filled
with mud - this is the so-called 'drillpipe fallacy' (see Hubbert and Rubey, 1961,

Fig. 1-10. Solid cylinder integral part of container.

 II

p. 1593, for a more detailed discussion). For this, we shall consider 3,000 m of pipe
hanging open-ended in mud of mass density 1,500 kg m -3 (1.5 g/cm 3 ). The pipe
weighs 26.9 kg/m in air, displaces 3.3 x 10 - 3 m 3 /m, and has a cross-sectional area
of 3.3 x 10- 3 m 2 . The pressure exerted by the mud at the bottom of the pipe is
pgd = 1,500 x 9.8 x 3,000 = 4.41 X 10 7 Pa (450 kg-f/cm 2 ). This is providing an
upward force of 1.46 x 105 N (14,850 kg-f) on the bottom of the pipe, equivalent to
the weight of about 550 m of the pipe in air. Since we can see the bend in 30 m of pipe
standing in the rack on the rig floor, it is clear that 550 m of pipe standing on end in
air would buckle. But it does not buckle in the borehole; so either the physical
interpretation must be wrong or there are factors that have not been taken into
account.
Alternatively, the pipe displaces 9.9 m 3 of mud that weights 14,850 kg; so the
weight of the pipe in the borehole is 80,700 - 14,850 = 65,850 kg. The effective
weight per metre is 22.0 kg, and the tensile load on the pipe varies continuously from
its total effective weight at the top to zero at the bottom, and the whole length of the
pipe is in tension. Buoyancy, like weight, must be a body force.
Reverting to the original question, whether the force of buoyancy acts on pipe
above the stuck point, the answer is yes: the weight of the pipe above the stuck point
is reduced by the weight of the mud displaced.

LIQUID-FILLED POROUS SOLIDS IN ST A TIC EQUILIBRIUM

If we consider unit bulk volume of a porous solid (the volume being large relative to
the pore and grain sizes), it will be evident that the proportion of pore space
determines the amount of liquid that can be contained. Porosity (f) is the ratio of
pore volume to bulk or total volume: vp/(v p + vs ). In some geological contexts it may
be convenient to use the engineers' void ratio (e) which is the ratio of pore volume to
the volume of the solids: f/ (1 - f). This has the merit of being the ratio of a variable
to a constant in most geological contexts.
Porosity has a statistical component grains and pore spaces are never identical,
seldom similar. Any arbitrary plane through the material will intersect both grains
and pore spaces. In this plane, the area of pore space over the total area is a measure
of the porosity of the material. This will vary slightly in planes in different positions
and with different orientations.
Porosity and bulk density of a material are related:

(1.5)

where Pp is the pore-fluid density, and Ps the grain density. Since the mean grain
density is commonly about 2 650 kg m- 3 (2.65 g/cm 3 ) an estimate of porosity or
bulk density can easily be made.
The propositions treated at the beginning of the chapter have their counterparts in
12

porous solids provided the pore spaces are interconnected so that a continuous
liquid phase exists. The interconnected porosity is known as effective porosity.
Take a container and fill it with loose coarse sand *. Water can be poured into the
container to the extent of the pore space before the surface of the water rises above
the sand. When the water is at rest, its upper surface is horizontal. This can be
measured by inserting manometer tubes (oflarger internal diameter than the pore
spaces) into the sand at the edge of the container (Fig. 1-11). Since the water cannot
sustain a shear stress there will be a uniform pressure field extending throughout the
inter-connected pore space, the pressure increasing downwards as the depth. For
the same reason, the pressure will be normal to the surfaces of the sand grains.
Close examination of the surface of the grains at the interface between the water
and air will, however, reveal that the surface of the water between grains is not
horizontal, but concave upwards into contact with the solid particles. This is due to
surface tension, and is a departure from the fundamental property of liquids that we
started with - that the surfacc ofliquid at rest is horizontal. But is is a departure on a
small scale that only applies at contacts between solids, liquid and air, or with solids
and two different liquids (when it is known as interfacial tension). The reality of
surface tension can be seen by inserting a tube of very small internal diameter into
water, and the water level in this capillary tube rises abov~ the surface of the free
water (hence the qualification above, that the manometer tube should be of larger
diameter than the pore spaces).

· .~.: ............:. '0'::":' ::...... :. ~

.............................
..... ..... ...... ... .... ..... ..
.......
.... ..... ................
.... ............
·...
................
.... ...... ....... .... ..
...........................
.... ...... ........ .... ... ..
• • • • • • • • • ,0 • • • • • • • • • • 0 ' • • • • • •
..............
........
0
.........
.... .....
••••••••••••••••••••

... ....
....
......
.....................
....................
.... .
.......
'0' •••••

..... ...
. .... .
·.. ...........................
.......... ...... ...... ... .. .
...... ......... ... ...
... ... ........ '" ..... .
..... .. ..
·.......
...........................
........... ............... ....
.....
.... .... .......
................ .... ... ......
.........
...........
........ ... .... ..........
......... ....... ......
...............
....... .............. ...
...... ... ....
·.......
...........................
.........
..... ..... ....... ..... .......
........... ...... ........
............... ....... .... . .... .
..........
..... ..... ...........
....... ...........
... ......
.....
.. ' . ' .................... .
........ .... ...... ........
..... ....... .......
........... ................
........... .. . ... .
........
........ ......................
..... ... ..... ...... ..
... ........ ....
............. ..... ... ....... ..
..............
... ..... .............
....... ......... ... .... .... ..
.......
......
.............
........ ..... ........
.............. ....
.......... ... ........
...... ..... .... ........
...................... ........
.... .. . .... .
......

Fiq. 1-11. Water-level in porous sand.

* This discussion and these diagrams represent mental experiments, or experiments on a very large
scale. Isaac Roberts discovered nearly a century ago that pressure on the bottom of a bin did not increase
after the grain in it had reached a depth of more than twice the width of the bin. See Cowin, 1977, for the
modern theory.
 13

Surface tension distorts the free surface of water in fine-grained sediments: the
finer the grain, the smaller the pore spaces and the greater the elevation of the
apparent water level above the level it would have had without surface tension.
There are real difficulties in visualizing this effect because, if we regard water as
incompressible over the small pressure range involved, and as having negligible
tensile strength, there cannot be an increase in the volume of water. The apparent
increase must be due to the upper fluid's presence, leading eventually to a transition
from 100% water to 100% air over a vertical distance that is determined by the
capillary pressure of the water in the material.
Interfacial tension between water and oil or gas is of considerable importance in
petroleum reservoir engineering and the theoretical approach to the problems of
petroleum migration. However, we shall ignore the effects of surface tension and
capillary pressure on the interface between water and air for the time being.

Bulk density

Take a container of known volume. Fill it with a known weight of clean sand,
agitating it to achieve a stable packing. Pour in a known weight of water until the
water surface coincides with the top of the sand. Clearly the weight of the water-
saturated sand exerts a force on the bottom ofthe container, and that the total force
is the sum ofthe force exerted by the water and that exerted by the sand. (Sand could
sustain a slight shear stress, but we shall assume that it does not, and that the whole
weight is supported by the bottom of the container.)
The porosity of the sand in the container is the ratio of the pore-water volume to
the total volume, vw/v. The bulk weight is the sum of the weights of solid and water,
and can be expressed PbgV; and the bulk density is the bulk weight divided by the
bulk volume, Pbg (= Yb). Again, there is a scale effect: the bulk volume considered
must be large enough to be representative.
The question arises: what are the relative contributions to the vertical force, and
hence pressure, exerted by water-saturated sediment? (To avoid ambiguity, we shall
consider the vertical forces on a horizontal plane a few grains above the contact
between the base of the sand and the container.)
The total weight of water-saturated sediment in the container is made up of the
weight of solids

Ws = psgv (1 - f) (1.6a)
and the weight of water in the pores

Ww = Pwgvj (1.6b)

the sum of which is

PbgV = Ws + Ww = [Ps(1 - f) + Pwflgv. (1.6c)

14

This can be verified experimentally. What about buoyancy? The effective weight of
the solids is reduced by the weight of the water displaced. Should we not write for the
weight of solids
Ws = Psgv(1 - f) - Pwgv(1 - f)
= (Ps - Pw)gv(1 - f)?
Adding this term to Pwgvj gives a total that is less than the total weight of the
materials put into the container by the amount pwgv (1 - f), which is the weight of
water displaced by the solids. If we had filled the container to the brim with water
and then put in the sand, these last expressions would have correctly led to the total
weight (this is how the volume of solids is usually measured). By putting the sand
into the container first and then adding sufficient water to fill the pore space, the
volume of water displaced by the solids is not, in effect, added. Buoyancy does playa
part, as we shall see.

Pressure and stress

We have found that the total weight of a water-saturated porous solid is the sum of
the weight of the solids and the weight of the pore water, as one would suppose. The
question now arises: what are the pressure or stress relationships?
If we divide equation (1.6c) by A, the total area of the base, we get

(Jbgh = Psgh(1 - f) + Pwghf (1.7)

This suggests that the pressure or stress due to the solids is only acting on the
proportion of A that is occupied by solids, (1 - f), and that the water pressure is
only acting on the proportion of A occupied by pore space.
Terzaghi (1933) examined the effect of buoyancy on concrete dams. He postu-
lated that buoyancy only acts on solid surfaces exposed to water in the pore spaces,
and set up a buoyancy coefficient

Ac
m=1--
A'

where Ac is the area of solids in the undulating surface 8' that only cuts the contacts
between grains (Fig. 1-12), projected onto the horizontal plane surface 8 of area A.
Thus the term Ac/ A is the proportion of the area not exposed to buoyant forces, and
m is a measure of the area over which buoyant forces do act. In two series of
experiments he found that the value of m varied but little from unity. Even with
dense concrete of low porosity, the buoyant force was found to be
F B = yv(m - f) ~ yV(1 - f),
where y is the specific weight of the water.
 15

s'
s-----'----I----s
Fig. 1-12. Terzaghi's concept of areas on which pore pressure might be exerted. (After Terzaghi, 1933,
fig. 1).

This paradoxical result is of some interest because it means that the force of
buoyancy acts on the total area, (1 - f), of solids irrespective of the contact areas
between grains. Clearly the value of Ac is not zero because concrete's tensile strength
depends on bonding between the grains.
Later, he found the same results in plastic clays (Terzaghi, 1936), and came to the
conclusion "The strain in clay and in concrete exclusively depends on the differences
between the total stresses and the neutral stresses. In every point of the saturated
material the neutral stresses act in every direction with equal intensity and they are
equal to the pressure in the water at that point." (Terzaghi, 1936, p. 875). It is hard to
escape the conclusion that Terzaghi regarded buoyancy as a body force, and that the
field of fluid pressures (neutral stresses) extends throughout the porous solid,
through solids as well as the pore liquids*.
* Terzaghi seems to have retreated from this position in later work. Harza's paper on the significance
of pore pressure in hydraulic structures, in which he came to conclusions similar to those of Terzaghi
(1936), was first submitted for publication in 1937, and resubmitted with revisions 10 years later (Harza,
1949). The discussion takes up more than three times the space of the paper itself, and about half of this is
adverse to Harza's thesis. See also Hubbert and Rubey, 1959, pp. 129-142; Hubbert and Rubey, 1960, pp.
621-628; Hubbert and Rubey, 1961, for more detailed and extensive discussion. The view that buoyancy
is a surface force and that the fluid pressure acts only on the area fA still appears to be widely held.
 16

The partition of stresses and pressures in porous solids is now generally called
Terzaghi's relationship:
(J = S - p, (1.8)
where (J is the effective stress, S is the total vertical stress (Pbgh) and p is the fluid
pressure or neutral stress (Pwgh when the pressures are normal hydrostatic).
Reverting now to equation (1.7), we see that this is but one possible partition.
Substituting into equation (1.8) the expressions for Sand p, we obtain

(J = Pbgh - Pwgh = [Ps(1-.n + Pwf - Pw]gh

-, [Ps(1 - f) - Pw(1 - f)]gh.
The effective stress in porous solids is due to the partial density of solids less the
water they displace; and it is this partition of total stress that agrees with experi-
mental results (see also Hubbert and Rubey, 1959, pp. 139-142).
The effective stress, then, is the difference between the total stress and the pore-
liquid pressure; and it is important in geology because it is this stress that causes
mechanical compaction of sediments under the force of gravity. (The neutral stress
is so called because it does not cause deformation.) To explore this further, we shall
consider the effective stress in a water-saturated sediment when there is a depth of
water on top of the sediment (Fig. 1-13).

Fig. 1-13.
 17

If the water-saturated sand that we have been considering is overlain by a depth d

of water, the total weight W now consists of the weight of water
(1.9a)
and the weight of solids
Ws = psghA (1 - f) . (1.9b)
Adding these, and using equation (1.6c),
W = PsghA (1 - f) + PwghAf + PwgAd
= PbghA + PwgAd. (1.9c)
The total pressure or stress at the base ofthe container is Pbgh + Pwgd, and the water
pressure is, of course, Pwg(h + d); hence the effective stress is
(1.10)

which is the same as before. In words, the effective stress is independent ofthe depth
of water over the porous water-saturated sediment. The pore pressure is higher, the
total pressure is higher, but the effective stress remains the same.
This result is important in a number of geological contexts; most obviously, it
means that deep-water sediment that is porous and permeable compacts under the
force of gravity in the same way and to the same extent as an identical sediment in
shallow water.
Finally, consider now the consequences of loading a water-saturated porous
sediment, as in Figure 1-14. A rigid, impermeable piston is placed on top of the
sediment, and it is loaded. Assuming the container to be rigid, what are the conse7

Fig. 1-14.
18

quences of this loading on the fluids and the solids in the sediment. The answer
depends on the compressibility of the solid framework. If the framework is incom-
pressible, it will carry the full load and the pore-fluid pressure remains unaltered.
For any degree of compressibility of the solid framework, the consequence of the
load Fz will be a rise in the pore-fluid pressure above that given by Pwgh: the pore
fluid will bear part of the load. The least compressible component bears most ofthe
load.

Field confirmation of Terzaghi's relationship

Terzaghi's partition of total stress in a rock into effective stress (0") and neutral stress
or pore pressure (p) is not merely a theoretical conception: evidence in its support
comes not only from the experiments of Terzaghi and others, but also from field
observations in context of ground water and petroleum.
Before the turn of the century, it was discovered that the load of a train on the
railway line near a well could raise the level of the water in the well (King, 1892,
p. 67; Veatch, 1906). More recently, Jacob (1939) reported on this effect near the
Smithtown railway station, Long Island, New York. The well is located 16.5 m (54
ft) north of the main line, 110 m (360 ft) east of the station. It is 27.1 m (89 ft) deep,
with the water-level 10.7-12.2 m (35-40 ft) below the surface. The sequence of
sediments in a nearby well is sand (12.2 m) on clay (9.1 m) on gravel (at least 9 m),
which is the aquifer. The aquifer is seen to be confined.
Each day there were 11 eastbound and 11 westbound trains weighing anything
from 260,000 to 540,000 kg. The same train returns, so the average weight in each
direction was about equal. It was found that the westbound trains caused the water-
level to rise 9.1 mm, on average, while the eastbound trains caused it to rise 13.7 mm.
The difference between the two seems to be due to the different speeds - westbound
trains averaging about 12-15 m S-1 past the well; eastbound, about 6 m S-1.
Of particular interest are the records of the arrival and departure of a freight train
with 19 wagons and a caboose. The train stopped, the locomotive and four wagons
were detached, one wagon was shunted to a siding, the locomotive and four wagons
were reconnected, and the train left. The changes in water-level during these oper-
ations were recorded, and are shown in Figure 1-15.
It is evident that the weight of the train added to that ofthe overburden increased
the effective stress and so reduced the pore volume. The symmetry of the changes
associated with arrival and departure suggest elastic recovery of the aquifer.
On a larger scale, we find that the abstraction of ground water in the Houston
district of Texas (Gabrysch, 1967) between 1890 and 1961led to a fall in the static
waterlevels. Over the years 1943 to 1964, the land surface subsided up to 1.5 m (5 ft).
Figure 1-16 shows maps ofthe decline in static water level and of the subsidence, and
they leave little doubt that the decline of pore-fluid pressure led to an increase in
effective stress, further compaction of the aquifer, and surface subsidence.
 19

TRAIN TRAIN ft.

mm. STOPS STARTS
+9 +0.03

+6 +0.02

+3
\ ~ +0.0

-"-- - - -~ I- _ _
f-_
0

-0.0
-3 ----COCOMOTIVE
-6 OPPOSITE WELL ~
-0.02
1/
-9 Time in V Minutes -0.03
5 10 15 20 25 30

Fig. 1-15. Fluctuations in water-level in a well due to a train (after Jacob, 1939, fig. 5).

The Wilmington oil field in California, discovered in 1932, is one of the world's
giant oil fields. It lies below the southern coastal districts of Los Angeles and the
city of Long Beach. This is an industrial (and residential) area that includes Long
Beach harbour and naval shipyard. During the first 30 years to the end of 1967,
more than 1.84 X 10 8 m 3 (1.156 X 109 bbl) of oil and 23.8 X 10 9 m 3 (840 X 10 9 ft 3 )
of gas were produced (Mayuga, 1970). Its importance from our point of view is that
within a decade of major production (which began about 1937, when the signif-
icance of the discovery became apparent) subsidence of the land surface began
threatening coastal installations. Total subsidence up to 1967 (including a little due
to ground-water extraction) is shown in Figure 1-17, and the deepest part of the
depression so formed lies over the crest of the oil field. Maximum subsidence
amounted to 8.8 m (29 ft). This vertical movement also led to horizontal movements
up to 3 m, which also caused extensive damage on the surface and to oil wells at
depth. Measurements in boreholes (see Mayuga, 1970, for details) indicate that
almost all the compaction had taken place in the producing zones.
The correspondence between production rate and subsidence rate left no doubt as
to the cause of the subsidence, for the maximum subsidence rate of 0.71 m/year (2.3
ft/year) occurred in 1951 about 9 months after the production rate peak.
Pilot water-injection schemes, intended both to halt subsidence and to improve
oil production, began in 1953; and in 1958, when the major scheme started, produc-
tion could be increased while the rate of subsidence further decreased. In areas of
greatest water injection, the subsidence was even reversed, with raising ofthe surface
20

-tOO

SCALE
o 10 MILES
1=====' ': .
o 10 KILOMETRES

Fig. 1-16a. Decline in static water-level (in feet) in the Houston district, 1943-1964 (after Gabrysch, 1967,
fig. 4, by courtesy of the Texas Department of Water Resources).

I ----- --
--- ------- --- -- - ....... ............
..... '\
,,-/ ..... ~----~
\
,,-
/'
J

---
/
/ /
/ ~-----~ /
/
/ ~
/ \
/ "- \
I
I
I \
\
f
I I
\ I
\ I

,I
\
\
\
\ I
\ I
\
SCALE

10 MILI!S
\~
,
!="===="F'.10=="'==1.
KILOM!Tflts
\\
"
Fig. 1-16b. Subsidence of land surface (in feet) in same area as Figure 1-16a over same period (after
Gabrysch, 1967, fig. 19, by courtesy of the Texas Department of Water Resources).
 21

150
35
>-
c
"'C

125 "'.....E
""
c
30
...
0
,.,...
"'C
en
E a
25 100 ze::[
'0 en
en
a
z ~l-
e::[ 20
en
2'";-
75
~a:
w 15 z
0
~
a: 50 §
~
-.J
10 ~
0
a:
w
25 ~
~
5

Fig. 1-17. Wilmington oil field, California: oil production, water injection and land subsidence (after
Mayuga, 1970, fig. 16).

by up to 25 cm*. A full account of this field is due to Mayuga (1970).

These three examples show that loading a porous and permeable sedimentary
rock reduces the pore volume, that abstraction ofliquids leads to compaction, and
that injection of liquids can de-compact a sedimentary rock. There is little doubt
that Terzaghi's relationship is real (at least qualitatively) and that it is the effective
stress that compacts detrital sedimentary rocks mechanically.
Quantifying this aspect of geology is not easy. The deformation of solids can be
characterized by parameters known as the elastic moduli (see any modern text-book
on structural geology), which can be determined by laboratory measurement. But
the dimension oftime is critically important. Elastic behaviour on a short time-scale
may well be replaced by plastic or fluid behaviour on a longer time-scale. With
porous materials, this means that as the period of time under load increases, so the
compressibility of the solid framework increases (the compressibility of fluids does
not change with time).

* It is normal practice now to maintain oil reservoir pressures by water injection, both to prevent
subsidence and to maintain production. An earlier example of oilfield subsidence is mentioned by M. ap
Rhys Price in his discussion of a paper by Kugler (1933, p. 769). Subsidence over the Lagunillas field on
the shore of Lake Maracaibo, Venezuela, was found to be 'in direct proportion with the production taken
out'.
22

DIMENSIONAL ANAL YSIS

We introduce this powerful tool of fluid mechanics now because readers may be
unfamiliar with it, and an elementary understanding of it helps in the handling ofthe
various equations we shall be using.
If an equation is to be generally valid, we must be able to use any consistent set of
units, and any constants must be dimensionless. Such an equation is said to be
dimensionally homogeneous. All physical quantities, such as volume, density, ve-
locity, can be expressed in terms of the basic units of Mass, Length, and Time. Thus,
velocity is distance divided by time, or in the notation for dimensions, LT- 1 . Any
equation relating such quantities must be such that exponents or powers of M, L,
and T are equal on both sides of the equation. For example,
weight density mass density x acceleration of gravity
y p x g

or, in dimensions,
ML -2T- 2 ML- 3 x LT- 2.

To check this, we set up what are called indicial equations.

For M: 1 1
For L: -2 -3 +1
For T: -2 -2.
The equation thus checks, and it is therefore dimensionally homogeneous. Dimen-
sional analysis uses the requirement of homogeneity to predict the form of an
equation relating various physical quantities. An example will make this clear.
Imagine we did not know the relationship between pressure and depth in an
incompressible fluid. We might well assume that the pressure p is a function of mass
density p, g the acceleration due to gravity, and the volume, v (Fig. 1-18), and write

p=!/J(p,g,v).

This equation can be written in dimensional form:

ML -IT- 2 = (ML -3y
= M"-L -3a

from which we write the indicial equations

for M: 1= a
for L: -1 = - 3a + b + 3c
forT: -2=-2b.
 23

I
I
Id

Fig. I-IS.

The solution of these simultaneous equations is, a = 1, b = 1, and the substitution

of these in the indicial equation for L gives c = 1. Hence the equation in dimen-
sional form is

The (L) is clearly vertical depth, so the functional relationship between pressure,
density, acceleration due to gravity, and depth is

p=Bpgd,
where B is a dimensionless constant. A few experimental measurements would
establish that the dimensionless quantity

p/pgd = B = 1
for a liquid of constant density.
All valid equations can be written in dimensionless terms, with the great ad-
vantage in experimental work that only two variables in each have to be determined
for different values, and the simplest can be chosen.
While dimensional analysis does not necessarily give the full function, it indicates
the form it will take. It will have been noted that our original assumption that
volume would be a parameter above was rejected by the analysis.
A list of common physical quantities and their dimensions are given in Table 1-1.
The reader is referred to Buckingham (1914, 1921), Brinkworth (1968, pp. 79-89), or
Pankhurst (1964) for a more detailed account of dimensional analysis.
24

Table 1.1. Dimensions of some common physical quantities.

Acceleration L T- 2
Bulk modulus (& other moduli) M L- 1T- 2
Compressibility M-1L T2
Density, mass M L -3
Density, weight M L- 2T- 2
Energy M L2 T- 2
Force M L T- 2

}
Permeability, coefficient of
L T- 1

Hydraulic conductivity

Permeability, intrinsic L2
Pressure, stress M L -IT- 2
Specific discharge L T- 1
Surface tension M T- 2
Velocity L T- 1
Viscosity, absolute or dynamic M L-1T- 1
Viscosity, kinematic L2 T- 1
Weight M L T- 2
Work M L2 T- 2

Temperature L2 T- 2
Quantity of heat M L2 T- 2
Thermal conductivity (thermal units) M L-1T- 1

GENERAL REFERENCES
Chow, V.T.,1959. Open-channel hydraulics. McGraw-Hill, New York, 680 pp.
Daugherty, R.L., and Franzini, J.B., 1965. Fluid mechanics with engineering applications (6th edition).
McGraw-Hill, New York, 574 pp. (International Student Edition.)
Forchheimer, P., 1914. Hydraulik. B.G. Teubner, Leipzig and Berlin, 566 pp.
Franks, F., (Ed.), 1972. Water: a comprehensive treatise, Vol. 1. The physics and physical chemistry of
water. Plenum Press, New York and London.
Giles, R.V., 1962. Theory and problems of fluid mechanics and hydraulics (2nd edition). McGraw-Hill,
New York, 274 pp. (Schaum's Outline Series.)
Riddick, J.A., and Bunger, W.B., 1970. Organic solvents (3rd edition). In: A. Weissberger (Ed.), Tech-
niques of chemistry, Vol. 2. Wiley-Interscience, New York, London, etc.
Robinson, J.L., 1963. Basic fluid mechanics. McGraw-Hill, New York, Toronto, and London, 188 pp.
Sorby, H.C., 1908. On the application of quantitative methods to the study of the structure and history of
rocks. Quarterly Journal Geol. Soc. London, 64: 171-233.
Terzaghi, K., 1943. Theoretical soil mechanics. John Wiley & Sons, New York; Chapman & Hall,
London, 510 pp.
Terzaghi, K., and Peck, R.B., 1948. Soil mechanics in engineering practice. John Wiley & Sons, New
York; Chapman & Hall, London, 566 pp.
Tuma, J.J., 1976. Handbook of physical calculations. McGraw-Hill, New York, 370 pp.

SELECTED BIBLIOGRAPHY

Bridgman,P.W., 1926. The effect of pressure on the viscosity offorty-three pure liquids. Proc. American
Academy Arts Sciences, 61(3): 57-99.
 25

Brinkworth, B.J., 1968. An introduction to experimentation. English Universities Press, London, 182 pp.
Buckingham, E., 1914. On physically similar systems; illustrations of the use of dimensional equations.
Physical Review, 2nd series, 4: 345-376.
Buckingham, E., 1921. Notes on the method of dimensions. Lond. Edtnb. Dub!. Phil. Mag., 6th series, 42:
696-719.
Cowin, S.c., 1977. The theory of static loads in bins. J. Applied Mechanics, 44: 409-412.
Gabrysch, R.K., 1967. Development of ground water in the Houston District, Texas, 1961-65. Texas
Water Development Board, Report 63: 35 pp.
Harza, L.F., 1949. The significance of pore pressure in hydraulic structures. Trans. American Soc. Civ.
Engineers, 114: 193-214. (Discussion: 215-289.)
Hubbert, M.K., and Rubey, W.W., 1959. Role of fluid pressure in mechanics of overthrust faulting, I.
Mechanics of fluid-filled porous solids and its application to overthrust faulting. Bull. Geo!. Soc.
America, 70 (2): 115-166.
Hubbert, M.K., and Rubey, W.W., 1960. Role ofITliid pressure in mechanics of overthrust faulting: a
reply [to Laubscher, 1960]. Bull. Geo!. Soc. America, 71 (5): 617-628.
Hubbert, M.K., and Rubey, W.W., 1961. Role of fluid pressure in mechanics of overthrust faulting: a
reply to discussion by Walter L. Moore. Bull. Geo!. Soc. America, 72 (10): 1587-1594.
Jacob, C.E., 1939. Fluctuations in artesian pressure produced by passing railroad-trains as shown in a
well on Long Island, New York. Trans. American Geophysical Union, 20: 666-674.
King, F.H., 1892. Observation and experiments on the fluctuations in the level and rate of movement of
ground-water on the Wisconsin Agricultural Experiment Station Farm and at Whitewater,
Wisconsin. U.S. Dept. Agriculture, Weather Bureau, Bulletin no. 5 (75 pp.)
Kugler, H.G., 1933. Contribution to the knowledge of sedimentary volcanism in Trinidad. J. Instn
Petroleum Technologists, 19: 743-760. (Discussion: 760-772.)
Laubscher, H.P., 1960. Role of fluid pressure in mechanics of overthrust faulting: discussion. Bull. Geol.
Soc. America, 71 (5): 611-615.
Leliavsky, S., 1947. Experiments on effective uplift area in gravity dams. Trans. American Soc. Ciu.
Engineers, 112: 444-487.
Mayuga, M.N., 1970. Geology and development of California's giant - Wilmington oil field. In: M.T.
Halbouty (Ed.), Geology of giant petroleum fields. Memoir American Ass. Petroleum Geologists, 14:
158-184.
Moore, W.L., 1961. Role of fluid pressure in overthrust faulting: a discussion. Bull. Geol. Soc. America,
72 (10): 1581-1586.
Pankhurst, R.C., 1964. Dimensional analysis and scale Jactors. Chapman & Hall, London, 152 pp. (Inst.
Physics Monographs for Students.)
Pratt, W.E., 1927. Some questions on the cause ofthe subsidence of the surface in the Goose Creek field,
Texas. Bull. American Ass. Petroleum Geologists, 11 (8): 887-889.
Pratt, W.E., and Johnson, D.W., 1926. Local subsidence of the Goose Creek oil field. J. Geology, 34 (7):
577-590.
Skempton, A.W., 1970. The consolidation of clays by gravitational compaction. Quarterly Journal Geol.
Soc. London, 125 (for 1969): 373-411.
Terzaghi, K., 1933. Auftrieb und Kapillardruck an betonierten Talsperren. 1" Congres des Grands
Barrages (Stockholm, 1933),5: 5-15.
Terzaghi, K., 1936. Simple tests determine hydrostatic uplift. Engineering News Record, 116 (June 18):
872-875.
Veatch, A.C., 1906. Fluctuations of the water level in wells, with special reference to Long Island, New
York. U.S. Geol. Surv. Water Supply and Irrigation Paper 155 (83 pp.).
 2. LIQUIDS IN MOTION

Just as it is our experience that the upper surface of a liquid at rest is horizontal, so is
it our experience that the upper surface of a liquid in motion is inclined from the
horizontal, and inclined in the direction of flow. Hydrostatics, the science offluids at
rest, is a special case of hydrodynamics, the science of fluids in motion.
The flow of water in an open channel is governed by two principal forces: the
component of the force of gravity along the channel bed tending to accelerate the
water, and the frictional resistance between the water and the material of the
channel tending to slow it down. The water accelerates until these two forces are
equal, and it then flows at a constant rate (in terms of the quantity of water passing a
transverse plane in unit time).
All liquid flow falls into one of two categories: laminar flow, when the paths
followed by the particles are essentially parallel, and turbulent flow. In geology we
are more concerned with laminar flow, although turbulent flow is important in
sediment transport mechanisms. The criterion that determines whether flow is
laminar or turbulent is a function of the velocity* of the water and the dimensions
of the channel. For the present we shall only consider laminar flow, and leave
discussion of this criterion for pipe flow.
Water overflowing a dam into an open channel (Fig. 2-1) will accelerate until the
resisting forces become equal to the propelling forces, and some time and distance is
required for this to be achieved. Flow nearest the dam is nonuniform; the water is
accelerating and its depth in the channel is decreasing in the direction of flow. As
soon as the resisting forces become equal to the propelling forces,the flow becomes
uniform, with constant rate or velocity, and constant depth. Uniform flow persists
until the conditions change - such as a change in slope or of channel dimensions -
but will be re-established in these new conditions if they persist for long enough.
Flow may also be described as steady if, in any position, the velocity remains
constant with time.
The study of liquids in motion centres around three basic variables: a charac-
teristic dimension of the flow, the velocity or rate of flow (which also involve the
physical properties of the liquid, mainly viscosity), and the hydraulic gradient.

* A useful distinction between velocity and speed is not always easily maintained, as we shall see. Here,
the velocity is the volumetric flow rate divided by the transverse area.
 27

Fig. 2-1. Schematic section through water overflowing a dam into an open channel.

Analysis ofliquid flow in pipes and channels proceeds on the assumption that where
the liquid is in contact with a solid, the relative velocity of the liquid is zero (as we
saw in the discussion of viscosity on p. 2. Consider liquid flow down a straight
channel, as in Figure 2-2. The motive force is the component of weight of the water
along the channel, and it is resisted by the shear forces generated from the confining
solid surfaces. When these two forces are equal and opposite, the water ceases to
accelerate and the flow is said to be uniform. Then,
Iwh pg sin e- to l(w + 2h) = 0, (2.1)

where to is the boundary shear stress and wand h are the width and depth of water
(normal to the surface) in a reach of the channel of length I. The characteristic
dimension of the channel has been found to be the ratio of boundary shear stress to
the component of unit weight down the channel,
to/pg sin e= lwh/l(w + 2h)
= volume/wetted surface area
= area of flow/wetted perimeter. (2.2)
This is known as the hydraulic radius (sometimes, the hydraulic mean depth because
this quantity approaches the depth in a wide, shallow channel), and is given the
symbol R. It has the dimension of length. It is by means ofthe hydraulic radius that
flow of different cross-sectional shapes is compared.

Velocity or rate offlow: the speed of a particle of water or other liquid is a variable of
position and depends partly on the viscosity of the liquid. It can usually be seen that
the speed of the water at the surface near the sides of a channel is less than that near
the middle (Fig. 2-3a). This profile, usually called a velocity profile, results from the
28

Fig. 2-2. Forces acting on unit length of Jiquid in uniform laminar flow in an open channel.

frictional resistance along the sides of the channel, and its shape is a function of the
viscosity of the liquid, the width of the channel, and its slope. It will readily be
appreciated that it is also a function of the depth of the channel, and that there are
also analogous profiles in planes normal to the water surface, parallel to the channel
sides (Fig. 2-3b). These complexities dictate the practical expedient of defining mean
velocity as the volumetric flow rate divided by the transverse area of flow. It has
the dimensions of velocity (q = Q/A: LT- 1 = L 3 T- 1 /L 2 ). This is satisfactory for
steady flow and for all computations involving quantities proportional to velocity;
but for quantities such as momentum, which varies as the square of velocity, the
velocity profile should be taken into account.
The third basic variable, hydraulic gradient, is concerned with the energy gradient
down which the liquid flows. Energy exists in various forms, but it is measurable by
the amount of work (force x distance moved) that it can do, and it is relative. It has
dimensions ML 2 T- 2 . Water has three principal energies: kinetic, due to its move-
ment; potential, due to its elevation in the gravitational field; and thermal, due to its
heat. To these we add pressure because it can be considered as an energy (work done
in compression) and it has the dimensions of energy /volume. Water moves unless it
is in mechanica equilibrium with its physical environment and constraints. The
principle formulated by Lagrange, late in the 18th Century, is helpful: a mechanical
system is in stable equilibrium when the kinetic energy is zero and its potential
energy is at a minimum with respect to its physical environment and constraints.
For example, if you pour water into a basin, the kinetic energy will be converted into
heat by friction and will decrease; mechanical equilibrium will be achieved when the
water is at rest and occupying the space of minimum potential energy within the
basin. We must clarify the concept of potential energy.
When it has come to rest, the water that was poured into the basin occupies a
position that is determined by the shape (i.e., constraints) of the basin, and the
 29

Fig. 2-3a. Speed of water at surface is less at the sides (plan).

0.5

I
I
I I
:~

Fig. 2-3b. Speed of water at the bottom is less than that at the top (profile). The mean velocity (q) is
approximately the velocity at lie of the depth from the bottom.

density of the water relative to the other fluid (air). Although it has come to rest, the
water has an energy due to its position in the gravitational field: if the plug is
removed, it will 'seek a lower level' and in so doing, acquire kinetic energy. The
energy the water has in the basin is not used while the plug remains, but it is potential
energy, and it is potential energy due to position in the gravitational field.
Energy, as be have seen, is measurable by the amount of work it can do, and is
 30

relative. An identical basin of water at a higher elevation in the earth's gravitational

field has a higher potential energy because it can do relatively more work. If it were
connected to the lower basin, and the upper plug removed, the water from the upper
would displace that of the lower basin, and the combined volume of water would
eventually reach mechanical equilibrium (unless it overflowed, in which case some
of the potential energy would be realized as kinetic energy). We can therefore take
an arbitrary horizontal plane as datum, and the potential (mechanical) energy of
unit mass of water would be given by
Ep =gz.

This is equal to the work required to move unit mass of water a vertical distance z
from the reference or datum plane to the position it occupies, by a frictionless
process.
If there are changes of pressure and volume, then work is involved in these
changes. The movement itself involves kinetic energy; and the work done in over-
coming frictional resistance will generate heat (which is a dissipation of mechanical
energy). Hence, the water in the basin has potential energy due to its position, and if
it is free to flow, this energy will be converted to kinetic energy and heat. This heat
loss is irreversible, so that flowing water continuously loses energy.

Bernouilli's theorem
Ifwater is regarded for practical purposes as being incompressible over the pressure
changes that may take place in relatively short and slow movement, the principles of
conservation and equivalence of energy allow the formulation of an energy balance.
Consider an ideal liquid flowing from station 1 to station 2 along a pipe of
decreasing diameter, as in Figure 2-4. The work done on the liquid in the pipe is
equal to the change of energy of the liquid. At station 1, the force applied to the
liquid is equal to PIAl; and in moving the liquid a small distance J1 I in the small
interval of time Jt, the work done is equal to PIAIJh. The weight ofliquid passing

®
Fig. 2-4. Liquid flowing along converging pipe.
 31

station 1 during the interval of time 6t is pgA16l1; and the same weight of liquid
leaves station 2, that is, pgA 26l2.
The potential energy Ep per unit of mass ofliquid entering at station 1 is gZl, so the
potential energy of unit weight is Z1 and the potential energy of the liquid entering at
station 1 is pgA16l1Z1 and that leaving at station 2 is pgA26l2z2.
The kinetic energy Ek per unit of weight is proportional to q2 /2g, so the kinetic
energy of the liquid entering at station 1 is pA 161 1qi /2, and that leaving at station 2 is
pAz6Izq~/2.
Finally, there is thermal energy Et . The effect of friction is to dissipate mechanical
energy by raising the temperature of the liquid, with some loss of heat to the pipe-
but we are considering an ideal liquid of zero viscosity.
The principle of conservation and equivalence of energy requires that the sum of
the work done, the potential energy, and the kinetic energy, shall be constant, i.e.,

P1A1611 + pgA 1611z 1 + pgA1611q~/2g = p zA 2612 + pgA 2612z z +

pgAZ612q~/2g

(in which each term has the dimensions of energy, M L 2 T- Z).

Since pgA1611 ~o = pgAz61z,

-PI + Zl + -qi P2
=c - + Z2 + -q~
pg 2g pg 2g
or,
P q2
- + Z + 2: = constant. (2.3)
pg 9

This is known as Bernouilli's theorem, and it is valid for incompressible, frictionless

liquids. Each term has the dimension of length. The reader will be uncomfortable
over the term q2/2g, wondering what meaning is to be attached to the velocity in
view of the definition of mean velocity (p. 28). Strictly, there is a coefficient in this
term, to take the velocity profile into account; and the value of this coefficient in
laminar pipe flow is 2. But it is normally neglected, and the mean velocity q taken for
this term.
The first term in equation (2.3), p/ pg, is known as the pressure head and is the
height of a column ofliquid of density p that can be supported by a pressure p. The
second term, z, is merely the elevation of the element ofliquid above (or below) an
arbitrary horizontal datum plane: it is called the elevation head or, less desirably, the
potential head. The third term, qZ /2g, is known as the velocity head. The sum ofthese
is known as the total head; and with an ideal, incompressible liquid of zero viscosity,
the total head would remain constant. In ground-water geology, the total head is
commonly known as the hydraulic head.
Figure 2-5 shows these heads diagrammatically superimposed on Figure 2-4.
Note that as the velocity increases, so the pressure decreases - a conclusion that is
32

TOTAL HEAD

Fig. 2-5. Heads superimposed on Figure 2-4.

probably contrary to the intuitive, but experimentally verifiable and the basis of the
venturi flow meter.
Real liquids, of course, are not frictionless; and in motion they lose energy in the
form of heat generated by friction within the liquid. So with a real liquid, the total
head decreases in the direction of flow, and equation (2.3) becomes

PI
-pg + Zl
qi P2
+ -2g = -pg + Z2
q~
+ -2g + hL' (2.3a)

where hLis the head loss incurred between stations 1 and 2.

We revert now to the matter of flow down an open channel of constant slope eand
uniform cross-section. If the liquid in the channel were an ideal liquid, without
viscosity, without friction, Bernouilli's theorem tells us that the total head would
remain constant. The elevation head is decreasing as sin e. The component of the
force due to the weight of the liquid along the channel, pg sin e, induces as
acceleration g sin eper unit of mass of the liquid, and the velocity head increases as
the square of the velocity. The velocity of an ideal liquid in the channel increases
indefinitely (there will come a time when the rate ofincrease of velocity head exceeds
the rate of decrease of the elevation head) and the pressure head will corresponding-
ly decrease. In other words, the liquid will become shallower in the channel in the
direction of motion.
With water, a real liquid, in the channel, the water will accelerate until the
resisting frictional force equals the gravitational force inducing acceleration, and
 33

thereafter the water moves at a constant velocity (by which we mean that the volume
of water passing any transverse plane in unit time is constant and equal). From this
point (Fig. 2-6) the total head decreases as the elevation head because both pressure
and velocity heads remain constant. This loss of total head represents the energy
dissipated, and is irrecoverable. It is equal to the loss of elevation head and is
proportional to the frictional forces, or viscosity: it is a measure ofthe departure of
the real liquid from an ideal liquid in that channel. The shape of the channel is a
factor because the larger the ratio of water volume to wetted surface area, the more
closely the water approaches the behaviour of an ideal liquid.

Arbitrary Datum

Fig. 2-6. Heads of uniform steady flow in open channel.

Channel-flow equations in common engineering use are not very satisfying (they
are, of course, satisfactory or they would not be in common use). Probably the most
commonly used equation for turbulent flow in channels and rivers is the Manning
equation, which in its metric form is

1 2 1
q =-R 3 SI, (2.4)
n
where q is the mean velocity, R the hydraulic radius, and S = iJhll is the slope of the
water surface. n is a roughness coefficient known variously as Manning's or Kutter's
n (these values are not altered according to the system of measurement used, but the
numerical factor is, being 1 for metric, 1.49 for foot-pound-second units). This
formula works well, but it is unsatisfying having and empirical equation with a
1
coefficient having dimensions TIL?;, and no sign of g, p and 1], which are hidden in
there somewhere.
The role of these hidden components can be indicated by taking the earliest
34

empirical channel-flow formula due to Chezy (in an unpublished note of 1775):

q = cJRS. (2.5)

J
Now, c can be shown to be equal to (8g If), where fis known as the friction factor.
Also, for laminar flow, f ex I1IRqp (in other words, roughness is not a factor, in
laminar flow: see, for example, Ackers, 1958, p. 2). Substituting these terms into
Chezy's formula,

q= (CP~Rq)! (RS)! = (Cpg;ZqS)!

from which

(2.6)

Note carefully that equation (2.6) is valid for laminar flow only, and also that it is
dimensionally homegeneous. It could have been derived from dimensional analysis,
and the reader may care to try it.
We shall leave channel flow with the question, what would be the law oflaminar
flow through a channel filled with sand?

PIPE FLOW

Equations seeking to describe the flow of liquids in pipes are associated with the
names of Poiseuille, a French doctor interested in blood flow through veins and
arteries (1840, 1841), and Darcy, a French engineer employed in municipal water
supply (1858), and others. It will better serve our purpose here to consult the work of
Osborne Reynolds (1884, 1901), a British professor of Engineering, because his
experiments covered the same topic, and introduced important new ideas.
Reynolds was aware that there are two types of flow in pipes, laminar and
turbulent (or, as he put it, direct and sinuous), and he was well aware that the
transition from one type of flow to the other depends on the velocity of the water
in the pipe, the diameter of the pipe, and the viscosity of the liquid. He sought
the criterion that determines whether flow is turbulent or laminar, and reasoned
that since there is no measure of absolute time or absolute length, any length
must be relative to some other length, and speed, to some other speed. He noted
that the ratio of dynamic viscosity to mass density, 111 p, has the dimensions of a
length multiplied by a speed or velocity. He inferred that the criterion would be
some critical value of the ratio of the products of two lengths and velocities,
Dql(l1lp) = Dqpll1, where q is the mean velocity ofthe water (as before, volumetric
discharge rate divided by cross-sectional area) in a pipe of diameter D, and I1lp = v
is the kinematic viscosity of the liquid. The number is dimensionless, of course,
 35

so it does not matter which system of units is used provided it is consistent.

Figure 2-7a sQows the mean velocity plotted against the hydraulioj gradient
as obtained from Reynolds' experiments with a lead pipe of 0.615 em 'diameter
(Reynolds, 1884, Table IV). The striking feature of the plot is the linear relation-
ship between the hydraulic gradient (friction losses) and the mean velocity, with a
change of slope at a velocity of about 0.54 m s -1. Figure 2-7b shows the lower-
velocity data of his Tables IV and V plotted logarithmically. This technique
(introduced by Reynolds some years earlier) puts the data in the form log x =
a log y + log b, corresponding to x = b ya. The slope of the log-log plot in Figure
2-7b indicates that at low velocities the exponent is close to unity, while the slope of
the turbulent data is close to t.
In other words, the friction losses are roughly

•
0.15
••
~
.t:
•
-c::3
_- 0.1 •
c •
.CD
- •
"'C
...
to •.... Re=2080
C)

.-
U
:; 0.05
m...
"'C
~
J:

0.5 1.0
mean velocity, q m 5- 1
Fig. 2-7a. Relationship between mean velocity and hydraulic gradient found by Reynolds for a pipe of
0.615 cm diameter (Reynolds, 1884, Table IV).
36

•
••
•
-1
• •
•
Re = 2080 .... •

-2

-1 -0.5
log q
Fig. 2-7b. Reynolds' data of Figure 2-7a plotted as logarithms.

proportional to the mean velocity up to some critical mean velocity, and then, after
a transition zone, roughly, proportional to the square of the mean velocity*.
Reynolds determined from his own experiments and those of Poiseuille and
Darcy, that the velocity criterion is
q = Kry/Dp, (2.7)

where K is a constant with a value about 2000 for pipes with circular cross-section.
Turbulent flow occurs when Dpq/ry > ~ 2000 when q is the mean velocity and D is
the diameter of the pipe. (This is not the number in his 1884 paper: he used a relative
viscosity in that work and so found another number. His 1884 experimental data
indicate a 'real critical value', as he called it, rather higher than 2000. His famous
experiments with a coloured dye indicated a critical number averaging nearly 13,000
in 29 experiments!)

* The raw data in Reynolds' Tables III, IV, and V, is not linear in laminar flow, but this is because the
water was cooling during this part of the experiments. When the viscosity is corrected for temperature,
the linear relationship is found.
 37

The number found by evaluating Dpq/11 has come to be called a Reynolds number,
designated Re or N R . Note most carefully the indefinite article - a Reynolds number.
If D is taken to be the hydraulic radius of the pipe (nr 2 /2nr = r/2 = D/4) rather than
its diameter, the critical Reynolds number has another value (~500), but the
custom of writing 4R for D leads to the same number. A characteristic velocity and a
characteristic dimension of length must be identified or chosen, and recorded.
Dimensional analysis helps us to an expression for the friction losses incurred
during laminar flow ofliquids in pipes. Reynolds' experiments indicate that the head
loss is a function of kinematic viscosity (v = 11/ p), the mean velocity (q = Q/A), and
a linear dimension characterizing the pipe. Bernouilli's theorem also indicates the
presence of g, the acceleration due to gravity. We therefore postulate that the head
loss incurred during laminar flow of water through pipe length I may be expressed.
L1h/1 = CD' qb VC if,
where C is a dimensionless coefficient. Dimensionally,

According to Buckingham's II theorem (see any text book on fluid mechanics, and
Buckingham, 1914, 1921) these five physical quantities expressed in two funda-
mental dimensions can be arranged into 5 - 2 = 3 dimensionless groups, of which
LL -1 is one, from which the form of the equation can be derived. Each of these
dimensionless groups is called a II-term. We take D and q as the dominant quanti-
ties, as Reynolds found, and make the two remaining II-terms from (D, q, v) and
(D, q, g).
We know that a Reynolds number can be formed from the first:

The indicial equations are,

for L: 0 = a + b + 2c
for T: 0 = - b - c
from which a = - c, and b = - c, and the dimensionless number or group is v/Dq
= II 2 • (This is, in fact, the ratio Reynolds used in 1884, not the inverse.)
The third group is formed from D, q, and g:
LOTo = (L)' (LT-l)b (LT- 2)d.

The indicial equations are,

for L: 0 = a + b + d
for T: 0 = - b - 2d
from which a = d, b = - 2d, and the dimensionless group is Dg/q2 = II 3 . So,
v Dg
t/J (Ill, II 2 , II 3 ) = t/J (L1h/l, -D '-2) = 0
q q
38

and

v
Ahjl=-l/12
Dg
(D9)
-2 .
q
Reynolds found that the hydraulic gradient in laminar flow was proportional to the
velocity, hence

Ahjl = c(~)
Dq
(L) Dg
= C~.
D2g
(2.8)

Analysis and experiment indicate that C has the constant value of 32 for pipes with
circular section and diameter D, and that for laminar flow the nature of the pipe
does not affect the frictional losses. The constant found for Reynolds' experiments
in his Table IV is 32.8, a discrepancy well within the accuracy of the experiment
because it could arise from a temperature error of less than 1 C. 0

The equation
Ahjl = 32qvjgD 2 (2.8a)
is known as the Hagen-Poiseuille, or the Poiseuille law for laminar flow in pipes *.
It is the basis of much work on the flow of fluids through porous media. This
equation can, of course, be re-arranged
D2 P Ah
q=--g- (2.8b)
32 1'/ I
and generalized by substituting 4R = D,

R2 P Ah
q=TYig-l . (2.8c)

Note that this equation is in the same form as that for laminar flow in open channels
(Eq. 2.6), and they could even be identical.

* It is worth noting in passing (although we shall not use it) that the hydraulic gradient can also be
expressed as a function ofyelocity head, q2/2g. Multiplying top and bottom of equation (2.8a) by q/2, we
obtain

LJh/i = 64q 2v _ 641) ~ q2 _ 64 ~ q2

2gD2q - pDq D 2g - Re D 2g'

This is the Darcy- Weisbach equation for laminar flow, the more general form being
1 q2
LJh/i =f--
D 2g

where f is here the friction factor.

 39

DIMENSIONLESS NUMBERS

We have seen that a Reynolds number, Dpqjy/, is dimensionless (and intentionally

so); and in the dimensional analysis above, two other dimensionless numbers were
found, Ahjl and Dgjq2. The first ofthese is the hydraulic gradient. The inverse ofthe
last, q2jDg, is known as a Froude number (Fr) - also its square root. A Froude
number is important in naval architecture, but it is usually regarded as irrelevant in
pipe-flow, or any other type of flow in which a free upper surface to the liquid is not
evident. Nevertheless, the analysis revealed it, and it is not irrelevant.
Reynolds numbers are rationalized as the ratio of inertial to viscous forces:
Froude numbers, as the ratio of inertial to gravitational forces. In our problems, a
Froude number appeared because a free surface is implied when pressures are
measured as manometer heights (or, in channel flow, because there is a free surface).
In any case, all these flows take place under the force of gravity. If laminar flow is
limited by a real critical value of a Reynolds number, it follows that in pipe flow
there is also a real critical value of the corresponding Froude number, but this is not
significant.
There are two great practical advantages of writing equations in terms of dimen-
sionless numbers: any consistent system of measurement may be used, and, their use
simplifies the experimental verification of the relationship. If measurements of pipe
flow are made, and the results plotted as Reynolds number versus Froude number,
the constant can be evaluated. In such experiments, it is only necessary to alter
one of the variables in each number, and it does not matter which: so the simplest
can be varied (in this case, the velocity and the hydraulic gradient, keeping the
other quantities constant). This aspect can be pursued in any text-book on fluid
mechanics.

PRESSURE

The reader will have noticed that pressure does not occur explicitly in any of the
preceding flow equations (2.6,2.8). Pressure is a force divided by an area on which
the force acts, and it occurs only in the expression for pressure head (equation (2.3)).
Pressure has the dimensions of (mass x acceleration)jarea, that is, MLT- 2 L -2 =
M L - 1 T- 2. Pressure head, on the other hand, has the dimension of length.
The whole subject of fluid movement has been plagued by confusion between
these two quantities, and it is essential to keep the distinction between them clearly
in mind. It is true that Poiseuille, Reynolds, Darcy, and countless later workers used
the word pressure when they meant pressure head. The early workers' descriptions of
their experiments leave no doubt about what they were meaning: this is not always
true of later workers.
If, therefore, pressure is used instead of pressure head in any ofthe previous flow
40

equations, they become physically erroneous because they become dimensionally

unhomogeneous. For example, the Poiseuille law for laminar flow in pipes (equa-
tion (2.8b)) is commonly written
D2 g Lip
q=12rj-1 .
This is wrong, and the reader who is not satisfied that it is wrong should not proceed
until he is. It is wrong because it implies that when q = 0, Lip = O. This is only true
for a horizontal pipe. It is also dimensionally unhomogeneous.

As with channel flow, the question now arises, what is the flow equation for a pipe
filled with sand? There will be little doubt that the form of the equation will be the
same as those for channel and pipe flow, equations 2.6 and 2.8c. It is evident that the
equation must take into account the greatly reduced area of flow (normal to the flow
direction) and the greatly increased wetted perimeter (that is, the greatly reduced
hydraulic radius). In other words, flow will be dominated by frictional resistance.

STOKES'LAW

In his investigations of the motion of pendulums, Stokes (1851; 1901, p. 60) devel-
oped analytically an expression for the terminal settling velocity of a solid sphere in
a fluid when the terms in V2 can be neglected:

v = 2g (£.s..
9v pw
_ 1) r2 = gd 2 (£.s.. -
18v pw
1), (2.9)

where v is the kinematic viscosity, ps and pw are the densities of the sphere and the
fluid respectively, r is the radius and d the diameter of the sphere, and g the
acceleration due to gravity. This equation is valid for any consistent set of units, but
it is not valid when the Reynolds number, Vd/v, exceeds one (approximately).
Stokes' law, as it has come to be called, must be used with these limitations in
mind - a single solid particle settling at such a velocity that the Reynolds number
does not exceed one. But the principle that settling velocity is determined partly by
size and mass density is a useful one, and settling samples of sediment can lead to a
rapid method of mechanical analysis, if the settling tube can be calibrated. There
have been various attempts at doing this (for example, Rubey, 1933, Watson, 1969,
and Gibbs et aI., 1971) all meeting with some success. Dimensional analysis can help
here.
We have seen that the six quantities, V, d, g, ps, pw, and v, with the three
dimensions of mass, length, and time, can be arranged into a function of three
dimensionless groups or n-numbers that are derived from the quantities involved.
In this case, they can be seen to be
 41

JI 1 Vdlv, which is a Reynolds number;

.~
JI l =Vljgd, which is a Froude number; and
JI 3= (PsIPw) - 1. (This form arises from the use of kinematic viscosity, rather

than dynamic viscosity, and therefore a density term (Ps - Pw)lpw.) Some combi-
nation of these three dimensionless numbers will give the form of the function.
One possibility is
(2.10)
which leads to Stokes' law when a = 1 and b = 18. But a and b must be constants, at
least over a finite and useful range of JI 1, so that
log JI 1 = a log (JI l IJI 3 ) + log b (2.11)

is the equation of a straight line. If experimental data in this form plots as one or
more straight lines, the constants a and b can be evaluated by regression analysis on
the data of each straight line, and all the equations are of the form of equation (2.10),
which can be expanded and re-arranged into

V= {(:: _ 1}l g du u+l 1b v} 1/(211-1) (2.12a)

and

d = {b V2a - 1 ( : : _ 1) II g - I I v} 1!(a + 1). (2.12b)

When a = 1 and b = 18, Stokes' law for single solid spheres results.
Riley and Bryant (1979) calibrated their settling tube for analysis of multi-grain
samples using industrial quartz spheres with diameters ranging from 74 to 710 ,urn,
sieved to i-phi intervals. Several samples, each weighing 1 to 5 grammes, from each
i-phi interval were settled. The grains landed on a tray to which was attached a
strain gauge, by means of which a continuous recording of cumulative weight
against time was obtained. The median velocity of the sample was then computed,
and the geometric mean (the mid-point of the i-phi interval) was taken as the
characteristic dimension of the sample. The water was at 20 C, giving a kinematic 0

viscosity of 1.007 x 10- 6 m l S-l. The density of the spheres was assumed to be
2,650 kg m -3.
Computing the three JI-numbers from Riley and Bryant's data, we find that the
logarithm of JI 1 versus the logarithm of JIll JI 3 plot as two straight lines (Fig.
2-8) with a change of slope at approximately log JI 1 = 0.37 (Reynolds number,
Re = 2.4). Regression analysis on the data of these two sets gives:

2 < Re < 60:a = 1.47,b= 55 (r ~ 0.9987)

0.25 < Re < 2: a 1.11, b = 25 (r = 0.9956).
Table 2-1 compares the experimental data with the predictions of equations (2.12).
42

2~----------~-----------'

1~-----------+-----Zf---~

o~------~~~------------~

-2 o

Fig. 2-8. Dimensionless plot of Riley and Bryant's (1979) settling tube calibration data for multi-sphere
samples.

Table 2-1. Data of Riley and Bryant (1979) and predictions of equations (2.12).

Vms- 1 dJ1.m V d

0.100 600 0.102 591

0.091 550 0.091 549
0.075 460 0.073 472
0.060 390 0.059 396
0.045 325 0.047 316
0.037 275 0.038 271
0.030 230 0.030 230
0.0250 195 0.0243 199
Re=2
0.0113 115 0.0114 114
0.0081 98 0.0086 94
0.0066 81 0.0062 84
0.0040 64 0.0041 63
 43

As we would expect with such high values of the correlation coefficient (r), the
agreement is good. We conclude that equations (2.12) are valid calibration equa-
tions for the analysis of multi-grained samples in a settling tube, within the ranges of
Reynolds number stipulated, and using the same technique of settling as in the
calibration.
Gibbs et al. (1971) calibrated their settling tube for single spheres of various sizes
and, in the larger sizes, different mass densities. The water in the tube was at 20 0 C.
Computing the three IT -numbers from their data, and plotting them as before (Fig.
2-9), we find a continuous curve that indicates that a and b are not constant. It is
perhaps worth noting, though, that the curve can be closely approximated by four
straight lines with changes of slope at log IT 1 = 0, 1, and 2, corresponding with
Reynolds numbers 1, 10, and 100. Regression analysis on these four sets of data
gives us:

100 < Re < 2300: a = 3.20, b = 40 (r 0.9964)

=

10 < Re < 100: a ~ 1.65, b = 67 (r= 0.9991)

1 < Re < lO:a = 1.21,b = 38.6 (r= 0.9988)

Re < 1: a = 1, b = 18 (Stokes' law).

Table 2-2 compares the observations with the predictions of equations (2.12), and
again, we find good agreement, and can conclude that these equations are suf-
ficiently accurate for single spheres in settling tubes, for the ranges of Reynolds
number given above.
The values of a and b for multi-spheres are different from those for single solid
spheres. This is not surprising (apart from the approximation used for single
spheres). The significant difference between single-sphere settling and multi-sphere
settling is the interaction between one grain and another, and the generation of
sympathetic water movement in the tube. Clearly, the diameter of the tube in
relation to the diameter of the particle or particles is also a parameter, because the
larger the particle, the greater the intcraction with the wall of the tube*.

Hydraulic equivalence

The sedimentological problem of assigning a diameter to grains of irregular shape is

relatively simple if we take behaviour in water to be the guide. The hydraulic
diameter is then the diameter of a sphere of the same material that has the same
terminal settling velocity. The hydraulic mean diameter of a multi-grain sample is,
by analogy, the mean diameter of a sample of spheres of the same material that has
the same median settling velocity. (There are two troubles with this approach:

* There is a small device for measuring wind velocity that is based on this principle. A small, light,
sphere is inside a narrow tube, the end of which is exposed to the wind. The stronger the wind, the higher
the sphere is suspended. If the tube were of constant internal diameter, the device would not work: but it
expands upward, and the wind velocity can be read against the scale opposite the small sphere.
44

3----1-----------1-----------1---~r_

2----~----------~----------~r-----

,....
1=
0) 1
o

O----~------~~~~----------4--------

-1-r~~----------~~----------~-------

Fig. 2-9. Dimensionless plot of Gibbs et al.'s (1971) settling tube calibration data for single solid spheres.

grains of very irregular shape, such as needles, may have a preferred orientation
while settling. And, more fundamentally, settling from suspension is not the usual
process leading to accumulation of sediment in the stratigraphic record. But there
are similarities between a grain's performance while settling, and when in traction
along a horizontal surface, so the concepts are still useful.)
Equation (2.12b) gives the hydraulic diameter of a single grain, or the hydraulic
 45

Table 2-2. Data of Gibbs et al. (1971) and predictions of equations (2.12) (single spheres).

Vms-I dJlrn V- a
0.4607 5000 0.4644 4941
0.3904 4000 0.3904 3994
0.2507 1670 0.2432 1735
0.1400 950 0.1422 931
Re = 100
0.1148 783 0.1173 768
0.0790 550 0.0781 555
0.0711 500 0.0700 507
0.0572 417 0.0568 420
0.0421 323 0.0423 322
0.0359 283 0.0363 280
Re= 10
0.0271 233 0.0278 229
0.0196 183 0.0191 186
0.0139 150 0.0140 149
0.0118 133 0.0116 134
0.00940 118 0.00964 116
Re= 1
0.00521 84 0.00568 80
0.001977 50 0.002011 50

Note: the density of the first two spheres was 2,240 kg rn -3, the third, 2,755 kg rn -3, and the rest
2,488 kg rn -3.

mean diameter of a multi-grain sample, when its density (Ps) is inserted and the other
quantities are known or measured.
There is an analogous concept: the diameter of a quartz sphere that would settle at
the same terminal velocity as the grain of other material. In natural sediments,
grains of heavy materials tend to accumulate with larger grains of quartz: this was
called the 'equivalent hydraulic size' by Rittenhouse (1943). Equation (2.l2a) leads
to simple expressions for this relationship.
Equating the right-hand side of equation (2.12a) for the density of quartz with
that for the material of density P*, we obtain

(2.13a)

where d* is the hydraulic equivalent size, d is the hydraulic diameter of the material
of density P*, and the appropriate value of a is taken according to the Reynolds-
number range (the exponent is not very sensitive, and will lie usually between 0.5
and 0.6). The ratio d*/dis constant within each range of Reynolds numbers. If this is
to be expressed in phi-units, the difference is constant:

3.32a p*
¢* - ¢ =--1loglO {1.65/(--1)}. (2.13b)
a+ pw
46

The use of these equations is best illustrated by an example. A sample of rutile

(P*/ pw = 4.2) is found to have a median settling velocity of 0.052 m s -1 in water
at 20° C (v = 1.007 X 10- 6 m 2 S-1). Inserting these values into equation (2.12b)
gives an hydraulic mean diameter of 240 11m (rounded off; Reynolds number 12).
Equation (2.13a) gives the ratio of hydraulic equivalent mean size to hydraulic
mean diameter, d*/ d = (3.2/1.65)°·6 = 1.5. So the hydraulic equivalent mean size is
1.5 x 240 = 360 11m.
The use of dimensionless numbers in such work leads to simpler expressions than
those derived by other methods, and there is the added advantage that not only may
the settling tube be used at whatever temperature is convenient (provided it is
known), but one is not restricted to water as the settling medium. Obviously there
are limitations on the size of material to be analyzed, but the smaller sizes may be
more readily analysed by using a liquid of smaller kinematic viscosity than water,
thus speeding up the analysis without loss of accuracy. By the same token, material
that would fall outside the range of calibration in water may be brought within it by
changing the liquid.
Fluidization is a process in which solid material is supported in an upward flow of
fluid, and it may be important in some aspects of volcanism. Strictly, it is the
'fluidizing' of an otherwise coherent porous solid, and so is a variety of Darcy's law
rather than Stokes'. The principle is the same, although the Reynolds number may
well be outside the realm of both laws.

NOTE ON POISEUILLE'S EXPERIMENTS

Poiseuille, in his studies of water flow through very small capillary tubes of di-
ameters 0.013 to 0.65 millimetres, found an expression for what may be regarded as
relative kinematic viscosity. His pipe flow equations (in his notation) were
PD 4 4klf PD 2
Q = kif ----r- and V= ----;: ----r-'
where Q is the discharge in cubic millimetres per second and V is the velocity in
millimetres per second (poiseuille, 1840, pp. 1046-1047). P is the pression, a ma-
nometer reading in millimetres corresponding to L1h because the discharge end of
the capillaries was open to the atmosphere ('Nous avons emprunte la pression, non
ala charge du liquide qui s'ecoule, mais a un manometre a air libre, soit aeau, soit a
mercure ... ', op. cit., p. 962). Since P, D, and L all have dimensions of length
(poiseuille used mm), Q of volume/time, and V of velocity, kif has dimensions
[L - 1 T- 1 ], that is, P9/11 or g/v (he realised that kif included the density of the liquid).
In part IV of his work, Poiseuille (1841, pp. 112-115) found that the relationship
between kif and temperature Twas
 47

k" = kl (1 + 0.033679T + 0.0002209936T2 )

where kl is the value ofk" at O°C, which he found to be 135.28 (mm- 1 sec-i) for
water manometers.
Comparing Poiseuille's equation for velocity with equation (2-8b), we find

4k" PD 2 D2 P L1h
nL = 12ti g -t
from which

k" = ~---.fL
4 32v'

So we can write

V
n 32k"
=4 n 32k
g = 4 g (1 + 0.033679T + 0.0002209936T2)-1 (mm 2; sec -1)
1

or
v = 1.779 X 10- 6 (1 + 0.033679T + 0.0002209936T 2 )-1 (m 2 S-l).

This formula for the kinematic viscosity of water at temperatures within the range of
Poiseuille's experiments is highly satisfactory, as Table 2-3 shows.

Table 2-3. Comparison of kinematic viscosities (m 2 S-1) predicted by Poiseuille's formula (column 1)
with values commonly accepted today (column 2).

TOC (1) (2)

10 1.31 x 10- 6 1.31 X 10- 6

20 1.01 X 10- 6 1.01 x 10- 6
30 8.05 X 10- 7 8.04 X 10- 7
40 6.59 X 10- 7 6.61 X 10- 7
45 6.00 x 10- 7 6.05 X 10- 7

SELECTED BIBLIOGRAPHY

Ackers, P., 1958. Resistance of fluids flowing in channels and pipes. Dept. Scientific and Industrial
Research, London, Hydraulics Research Paper 1 (39 pp.).
Buckingham, E., 1914. On physically similar systems; illustrations of the use of dimensional equations.
Physical Review, 2nd series, 4: 345-376.
Buckingham, E., 1921. Notes on the method of dimensions. Lond. Edinb. Phil. Dubl. Mag., 6th series, 42:
696-719.
Darcy, R., 1858. Recherches experimentales relatives au mouvement de l'eau dans les tuyaux. Memoires
presentes par divers savants aI'Academie des Sciences de I'Institut de France 2" serie, 15: 141-403.
Emery, K.O., 1938. Rapid method of mechanical analysis of sands. J. Sedimentary Petrology, 8: 105-111.
Gibbs, R.J., Matthews, M.D., and Link, D.A., 1971. The relationship between sphere size and settling
velocity. J. Sedimentary Petrology, 41: 7-18.
48

Poiseuille, J .L.M., 1840. Recherches experimentales sur Ie mouvement des liquides dans les tubes de tres
petits diametres. Comptes rendus hebdomadaires des Seances de I' Academie des Sciences, Paris, 11:
961-967,1041-1048.
Poiseuille, J.L.M., 1841. Recherches experimentales sur Ie mouvement des liquides dans les tubes de tres
petits diametres. IV. Influence de la temperature sur la quantite de liquide qui traverse les tubes de tres
petits diametres. Comptes rendus hebdomadaires des Seances de l' Academie des Sciences, Paris, 12:
112-115.
Reynolds, 0., 1884. An experimental investigation of the circumstances which determine whether the
motion of water shall be direct or sinuous, and the law of resistance in parallel channels. Philosophical
Trans. Royal Society, 174 (for 1883): 935-982.
Reynolds, 0., 1901. Papers on mechanical and physical subjects, Volume II (1881-1900). Cambridge
University Press, Cambridge, 740 pp.
Riley, SJ., and Bryant, T., 1979. The relationship between settling velocity and grain size values. 1. Geol.
Soc. Australia, 26: 313-315.
Rittenhouse, G., 1943. Transportation and deposition of heavy minerals. Bull. Geot. Soc. America, 54
(12): 1725-1780.
Rubey, W.W., 1933. Settling velocities of gravel, sand, and silt particles. American J. Science, 25:
325-338.
Stokes, G.G., 1851. On the effect of the internal friction of fluids on the motion of pendulums. Trans.
Cambridge Philosophical Soc., 9: 8-106. (Not sighted.)
Stokes, G.G., 1901. On the effect of the internal friction of fluids on the motion of pendulums. In: G.G.
Stokes, Mathematical and physical papers, Volume 3. Cambridge University Press, Cambridge, pp.
1-141.
Taylor, E.H., 1939. Velocity-distribution in open channels. Trans. American Geophysical Union, 20:
641-643.
Vanoni, V.A., 1941. Velocity distribution in open channels. Civil Engineering, 11 (6): 356-357.
Watson, R.L., 1969. Modified Rubey's law accurately predicts sediment settling velocities. Water
Resources Research, 5: 1147-1150.
 3. LIQUID FLOW THROUGH POROUS SANDS

We have considered so far those aspects of fluid statics and fluid dynamics that are
essential for an understanding of the main theme of this book - pore water and
geology - and we have seen that the topics can barely be considered without some
mathematics, but that the mathematics could be done by a High School student.
Mathematics is a language that more geologists understand nowadays, but it is
worth remembering that much of applied mathematics in the natural sciences is
either trivial or intractable. It is not our purpose here to derive the laws of flow
through porous solids from the Navier-Stokes equation, but rather to express them
in terms of the measurable and useful parameters of sedimentary rocks. Our pur-
pose is to understand the processes rather than to develop predictive equations or
formulae. Those who seek a higher goal may find the note at the end of the chapter a
useful starting point.

HENR Y DARCY'S EXPERIMENTS

Henry Darcy was an engineer concerned with the public water supply to the town of
Dijon, France, and in 1856 he published as an appendix to a broader report the
results of experiments conducted 'to determine the laws of water flow through
sands' (Darcy, 1856, pp. 590-594; reprinted in Hubbert, 1969, pp. 303-311. See also
Hubbert, 1940, p. 787 et seq., 1956 and 1957, for analyses of Darcy's experiments).
The experimental apparatus is shown diagrammatically in Figure 3-1. It consists
of a cylinder in which there was a measured column of sand supported by a filter.
Water was passed through the cylinder, and the volumetric rate of flow measured at
the outlet. Manometers were placed near the top and the bottom ofthe sand (Darcy
used mercury manometers, and computed the readings that would have been
obtained with water manometers). At various rates of flow, the difference in ele-
vation between the mercury levels in the two manometers was measured, and
computed for water. The results of three series of experiments are plotted in Figure
3-2.
Darcy observed that 'for sands of the same nature' the flow of water was propor-
tional to the difference in elevation between the levels in the manometers (as
computed for water) and inversely proportional to the length of the sand in the
 r----
50

Ah
1______ _

~
. . . . . . . ..
::1:1:11:111:1:::111:11:111111111

.e ~{{{{{{{{
1 .................
................
.................
................
.................
................
.................
.................
.................
.................
.................
................
.................
----~:.:.:.:.:.:.:.:.:.:.:.:.:.:.~:.:.:.~.

Fig. 3-1. Diagram of Darcy's apparatus.

direction of water flow. From his observations he deduced (in the notation used in
this book)
A
Q = KT (hi + I ± h2 ), (3.1 )

where Q is the volume of flow in unit time, I is the length of sand traversed, A is the
gross cross-sectional area normal to the flow, hi is the computed or equivalent water
height in the inlet manometer above the top of the sand, h2 is the computed water
height in the outlet manometer above ( - ) or below ( + ) the base of the supporting
filter, and K is 'a coefficient depending on the permeability of the bed' of sand.
We must pause here to note with some surprise that not all the experiments Darcy
reported support the linear expression he deduced. If a straight line is drawn on
Figure 3-2, 1st Series, with the constraint that it passes through the origin, and the
slope adjusted so that as many points fall above the line as below it, the first five
points fall below the line and the second five above it. Figure 3-3 shows Darcy's data
 51

Ah
m

10

OL-.-------------' 4
o 5x10-

Fig. 3-2. Results of three of Darcy's experiments.

(first two series, and Ritter's results) plotted on logarithmic scales: only the series
carried out by M. Ritter shows a strictly linear relationship (and the 4th Series, with
three points, which is not shown), the rest having slopes less than unity (between
0.87 and 0.92). Darcy probably relied on Ritter's experiments, and the linear
relationship has since been amply demonstrated, so we must guess that the water
was cooling down during the first three series of experiments (as indeed it was for
Reynold's experiments, with the same result).
It is critically important for the understanding of Darcy's experiments to note that
'Toutes les pressions ant he rapportees au niveau de laface inferieure dufiltre' (op.
cit., p. 592, my italics) - 'All pressures have been referred to the lower surface ofthe
filter' - and that he was therefore measuring what we would now call the sum of the
elevation and pressure heads. This is quite explicit in his equation in the term hI + I
because I is the elevation head (in his experiments) above the supporting filter and hI
is the pressure head above the top of the sand. He was not measuring the pressure
difference across the sand.
In the conventional modern notation, Darcy's law is written
52
L1h
q = Q/A = - K-
l ' (3.2)

where q is the specific discharge (or discharge velocity) and K is Darcy's coefficient of
proportionality that depends on the permeability of the porous material. The minus
sign follows the convention that q is in the direction of decreasing total head (h); but
it is commonly omitted (as we shall do) when the directon of flow can be obtained by
inspection.

Ah
m

1L---~~--~~~------------~
5x10- S 10- 4 5x10 4
Q m 3 s- 1
Fig. 3-3. Data of Figure 3-2 plotted on logarithmic scales.

THE COEFFICIENT OF PERMEABILITY

The coefficient of permeability, K, clearly takes several influences into account.

Darcy used water for his experiments, but had he used oil or any other liquid that is
inert to the solids he would no doubt have obtained quantitatively different results.
He did obtain different results with different sands. The coefficient K therefore
consists of at least two parts: one related to the material alone, the other to the
liquid.
The dimensions of K are [L 3T- 1 ] [L - 2] [LL -1] = LT- 1 , which are the dimen-
sions of a velocity. Since L1h/1 is dimensionless, q also has the dimensions of a
velocity, but it is a notional velocity in the sand because only part of the cross-
 53

sectional area is available for liquid movement. K therefore includes this effect, but
we shall consider it later. The obvious physical properties of the liquid are (from the
considerations of Chapter 2) its mass density p, with dimensions ML - 3, and
dynamic viscosity 11, with dimensions ML -1 T- 1 • Gravity is clearly a driving force,
so we include g, with dimensions LT- 2 (Q could have been measured as a weight
flow rate). There remains the significant dimension of the permeable material.
Many workers (e.g., Lindquist, 1933; Hubbert, 1940, 1956, 1957; Schneebeli, 1955)
have taken the characteristic dimension to be the mean grain diameter; but this is
unlikely to be adequate because it is a very indirect measure of the geometry of the
pore space through which the liquid passes. Let us try to clear this problem up.
For liquid flow in channels and pipes, as we have seen, the characteristic dimen-
sion is found to be the hydraulic radius

R= area of transverse section of liquid

wetted perimeter

volume of liquid
(3.3)
wetted surface area .

This approach to flow through porous solids was taken by Blake (1923), Kozeny
(1927a, 1927b) and Fair and Hatch (1933). The obvious difficulty is that part of the
surface area of the grains (for example, close to grain contacts) may not exert any
significant influence on the fluid flow; but the material constant should take this
into account.
If we consider bulk volume v of the permeable material, the product of the
effective porosity and the bulk volume, fv, is the volume of movable liquid. The
wetted surface area of the grains in the bulk volume can be calculated for grains of
simple geometry, leaving the contact areas to be taken into the material constant. If
the grains are all spherical and of equal diameter, the surface area of each grain is
nd 2 . With n grains in bulk volume v, their total surface area is

(3.4 )
The total volume of these grains is

n n d3
v n =-6- (3.5)

and they occupy (1 .n of the bulk volume of the material. The bulk volume is
therefore

n n d3
v = 6(1 .~ f) (3.6)

and the ratio of grain surface area to bulk volume, known as the specific surface (S)
is
54

n n d2 6(1 - f)
(3.7)
S = Sjv = (nnd3)/6(1 _ f) = d

The smaller the grain diameter, the larger the specific surface. For non-spherical
grains, the factor will be larger than 6 because spheres have the smallest ratio of
surface area to volume.
Our purpose being to understand the roles of the measurable and observable
parameters of granular sedimentary rocks, not to develop predictive equations, we
take the characteristic dimension of the pore space to be

R =f/S =fd/(1- f) (3.8)

leaving the dimensionless coefficient to be taken into a general coefficient.

What meaning is to be attached to the mean grain diameter d when the grains are
of unequal size? The mean diameter appears in the hydraulic radius in the term for
the specific surface (equation (3-7)), so it is the contribution to the specific surface
that is significant, and the interpretation of mean diameter must be that that gives
the true specific surface.
Consider a sediment composed of two fractions only, WI/ Wby weight of grains of
diameter d1, and W2/ W by weight of grains of diameter d2. If the grains are geo-
metrically similar, and of the same material or density, the total weight is propor-
tional to the total grain volume (1 - f), and the weight of each fraction proportional
to its grain volume. Following the argument in the derivation ofthe hydraulic radius
(equations (3.4) to (3.7), the contribution to the specific surface from the d 1 fraction
1S

(3.9)

and likewise, from the d2 fraction

(3.9a)

There exists a mean diameter d such that

Cl (1 - f) Cl (1 - j)Wl Cl (1 - f)W2
(3.10)
d d1 W + d2 W

or

1 1 WI 1 W2
(j=d 1 W+d 2 W· (3.11 )

This is the harmonic mean diameter weighted by the weight fraction.Note that the
harmonic mean takes size-sorting into account (to some extent at least) because the
 55

greater the dispersion (of positive numbers) the smaller the harmonic mean is
relative to the arithmetic and geometric means.

Tortuosity

There is yet another property ofthe pore space that must be taken into account. It
was noted earlier (as most previous investigators have noted) that the specific
discharge q is a notional velocity in the porous material (it is the real velocity of the
water above and below the sand in Darcy's experiments), so K contains a dimen-
sionless factor that takes this into account. Only the effective pore space is available
for water flow, so the mean water velocity in the sand is apparently qlf. The real
velocity through the pore space is greater than that because the pore passages are
not straight, and a particle of water must travel a mean distance It during its
displacement over the linear length I of a sand (parallel to the macroscopic flow
direction). This dimensionless property ofa sedimentary rock is known generally as
tortuosity, which we shall define T = It/I; it is difficult to define quantitatively and
evaluate.
Consider a cube of impermeable material penetrated by a single non-linear
capillary tube of circular cross-section that passes from one face to a similar position
on the opposite face of the cube (Fig. 3-4). It is clear a) that the length ofthe capillary
is greater than the distance between the faces, b) the area of intersection of the
capillary with the bounding faces is greater than cross-sectional area of the capillary
normal to the flow, and, on surfaces cut through the cube parallel to the bounding
faces, the area of intersection with the capillary will only be equal to the cross-
sectional area where the axis is normal to the surface; and c) that energy losses

.0. -- --------
-- ,
........ ...........
\
I
I

--- ---------- --
I
-'

Fig. 3-4.
56

incurred in water flow through the capillary are greater than those incurred in flow
through a straight capillary of the same diameter.
Now consider a cube of impermeable material of side I penetrated by n non-linear,
non-intersecting, capillaries of diameter x, the mean length of which is It, the volume
of the capillaries amounting to, say, 20% of the total volume of the cube. The
capillaries pass from one face to the parallel face, the bounding faces being normal
to the mean flow direction. We infer that
1) the mean inclination between the axes of the capillaries and the normal to the
bounding faces, a, is given by cos a = lilt.
2) the circular sections of the capillaries appear, in general, as ellipses on the
bounding faces (and any surface parallel to them) with the minor axis equal to the
diameter ofthecapillaries (x), and the major axis equal on average to x/cos a = xlt/I.
3) the surface porosity, defined as the proportion of pore space in the total area of a
surface - in this case the bounding faces - is therefore
nnx 2 1t 1
is = ----;rr- x r' (3.12)

This is seen to be also the bulk porosity f due to capillaries of diameter x and mean
length It in a cube of volume 13 .
There are several obvious defects in this model. In real porous materials, the
capillaries are not of constant or equal diameter, they intersect one another, and
they pass in all directions through the cube (although there may be a preferred
orientation in anisotropic material). But however we refine the model, it seems that
the term fA exaggerates the area of pores available for water flow by a factor of It/I
because of the 'mean inclination' of the pore channels. So the expression we sug-
gested for the real velocity (qlf) should be qlt/fl. The real pore velocity, as we have
seen, is greater than this by a factor It/I, so the real pore velocity Vp is related to q by

Vp =lor· (3.13)

Since q is defined by QIA, the coefficient of permeability K contains a factor f(lllt)2.

We are still not finished with tortuosity, however, because the hydraulic gradient,
t1hll, is also affected by it: the gradient is exaggerated because I < It. So it also
requires a factor lilt, which is also contained in K.
We conclude, therefore, that the total effect of tortuosity is

(3.14)

and that K contains the factor f as well as T- 3.

Reverting now to the components of the coefficient of permeability, K, and the
expansion of Darcy's law, dimensional analysis of the variables with dimensions (p,
1'/, g, R) leads to an expression similar to Equation 2.6; and, withfand T,
 57

(3.15)

and Darcy's law can now be written

_ CfT-3 R2 P Llh (3.15a)

q- fig-I·

For the evaluation of tortuosity we turn to the electrical measurement of resist iv-
ity of porous solids saturated with an electrolyte, in particular to Archie's Forma-
tion Resistivity Factor. Archie (1942), following on earlier work, found that the
resistivity of a porous solid, Ro, is related to the resistivity of the electrolyte satu-
rating the pore space, R w , by
(3.16)
where F is a dimensionless material constant known as the Formation Resistivity
Factor, or just Formation Factor.
By an argument strictly analogous to the one just used for permeability, the
Formation F actor can be expressed in terms of porosi ty , capillary lengths and cross-
sectional areas:

(3.17)

Following Street (1958) we see that

(Fj)-1.5 = (l/ld 3 = r3. (3.17a)
But Archie suggested that the data supported the relationship F = f -m, where m is a
number between about 1.3 for loose, unconsolidated sands to about 2.5 for con-
solidated sands. So
f r 3 =f(f-mf)-1.5
= fl.5m - 0.5 • (3.18)
Substituting equation (3.18) into equation (3.15) we obtain an expansion of the
coefficient of permeability in terms of all the foregoing considerations,

K = CfI.511l-0.5 R2 ~ g. (3.19)
1'/
(Although R = fd/(1 - f), it will be more convenient for the time being to keep
tortuosity separate from the hydraulic radius, which is a measure of pore size.)
Verification of this relationship is sought by plotting experimental data in dimen-
sionless form, the dimensionless groups or numbers being derived from the vari-
ables in the manner used to obtain equation (2.8) in the previous chapter.
The variables are K,fT- 3 , R, p, 1'/, g. Six variables with the three dimensions of M,
L, Time, can be arranged into (6 - 3 =) three dimensionless groups of which fT- 3,
58

being dimensionless, is one (II d. Choosing K, p, 1], as the repeating variables

common to both the other groups, and induding R in one and g in the other, we find

and the indicial equations are,

for M: 0 = b+c
for L: 0 = a - 3b - c + d
fot Time: 0 = - a - c
from which, b = - c = a = d. Since a II term can be replaced by any power of that
term,

which will be recognized as a Reynolds number with K a velocity. For II 3 we use K,

p, nand g, obtaining

These three groups are consistent with equations (3.15) and (3.19) if we write

II3/ II l = e IIi.
II 3 is found to be the product of the Reynolds number II 2 and a Froude number (the
ratio of inertial to gravity forces)

So writing Fr for K 2 T 3/gRjsince II3/IIl is also a Froude number,

Re x Fr = e Re 2
from which
Fr= eRe
or

(~;) e(K~R)-
= (3.20)

The dimensionless groups are thus found to be those in brackets in equation (3.20).
Schriever (1930) carried out some careful experiments on the permeability of
packings of glass spheres to hot oil, within the realm of Darcy's law (for there is a
realm, as we shall see). Each of four series of experiments was made with spheres ofa
single diameter, but the porosity was varied by altering the packing by hammering.
He thus obtained measurements of the permeability of each of four sizes for four
different porosities. Schriever's paper is of no theoretical interest now: he was
 59

seeking to evaluate the constants of Slichter's (1899) equation, which has been
abandoned in favour of those by Kozeny (1927) or Fair and Hatch (1933) or
variants of these.
Taking the cross-sectional area used in his formula on p. 335 of his paper, and
evaluating fT- 3 by assigning m the value 1.3 for loose sands, his data have been
tabulated in Table 3-1 and plotted in Figure 3-5 in dimensionless form, log Re versus
log Fr, revealing the relationship
log Fr = a log Re + log C
corresponding to Fr = C Rea. The slope of the line indicates that a = 1; and the
value of C (computed from Schriever's data) is found to be 8.42 x 10- 3 . The
number C is dimensionless, but it is not a true constant because it contains at least
one coefficient relating to the shape of the grains (in the expression for hydraulic
radius ).

Table 3-1.Grain size, porosity, and permeability. (Data of Schriever, 1930, p. 335, Table 1)

KpR K2
dcm f Kcm/s R-fd/(I-j) Re ~--
Fr~ Rf"45 C~Fr/Re
Ij g .

0.1025 0.3870 1.4411 x 10- 1 6.4710 x 10- 2 1.5592 x 10- 1 1.2959 x 10 - 3 8.3 X 10- 3
0.3777 1.2657 6.2212 1.3166 1.0771 8.2
0.3653 1.0986 5.8994 1.0836 8.9817 x 10- 4 8.3
0.3533 9.5533 x 10- 2 5.5997 8.9445 x 10- 2 7.5104 8.4
0.0528 0.3889 3.8869 x 10- 2 3.3602 X 10- 2 2.1838 X 10- 2 1.8026 X 10- 4 8.3 X 10- 3
0.3779 3.5590 3.2074 1.9086 1.6506 8.6
0.3689 3.1120 3.0863 1.6059 1.3582 8.5
0.3603 2.8071 2.9739 1.3958 1.1867 8.5
0.0443 0.3958 3.0159 x 10- 2 2.9020 x 10- 2 1.4634 x 10 - 2 1.2250 X 10- 4 8.4 X 10- 3
0.3849 2.6525 2.7721 1.2294 1.0329 8.4
0.3715 2.3087 2.6185 1.0108 8.7211 x 10- 5 8.6
0.3552 1.8463 2.4403 7.5332 x 10- 3 6.3871 8.5
0.0252 0.3934 9.7454 x 10- 3 1.6343 x 10 - 2 2.6630 x 10- 3 2.2913 X 10- 5 8.6 X 10- 3
0.38055 8.0996 1.5481 2.0965 1.7533 8.4
0.3690 7.0929 1.4737 1.7477 1.4770 8.5
0.3597 6.2742 1.4157 1.4851 1.2484 8.4

Notes: 1) First two columns, Schriever's data. Third column computed from his data by dividing by the
area A and the ratio of mercury density to his oil density (14.77 x 13.55/0.836 ~ 239.39). The area was
obtained from his formula on p. 335 and the data of his table (fJ is correctly given in spite of the error of
sign).
2) p/Ij ~ 16.72 s/cm 2 . The value assigned to m is 1.3 for loose sand, hence the exponent 1.45.
3) Dimensionless numbers are the same in any consistent set of units : Schriever used c.g.s., and the same
result is obtained with SI units.
60

Fr 104~---------+.o---------+--

105~----------~--------~~-
10- 3 10- 2 10- 1
Re
Fig. 3-5. Dimensionless plot of Schriever·s (1930) data (Table 3-1 ).

We therefore write Darcy's law

LJh
q= Q/A =K-Z

= Cfr- 3 R2 ~g LJt . (3.21 )

where fT- 3 = f 15m-OS, where m is the same 'cementation factor' used in Archie's
Formation Factor. And reverting to Equation 3.19, we see that the combined
porosity term is

where x varies from 3.5 to 5.3 as m varies from 1.3 to 2.5.

The coefficient of proportionality K, the coefficient of permeability, is also com-
monly called the hydraulic conductivity. We noted early on that this coefficient
consists of at least two parts, one due to the material alone, another due to the liquid.
The term CfT-3 R2 = Cjd 2 /(1 - ff relates solely to the porous material; it has
dimensions L 2 , and this quantity is called the intrinsic permeability, with the
sym.bol k:
 61

The term p/y/, the inverse of the kinematic viscosity, relates to the liquid (not
necessarily to gases). The last two terms, g(L1hfl), relate to the energy loss in the
system. We have already seen (pp. 30 to 32) that the hydraulic gradient, L1h/l, can be
expanded by Bernouilli's theorem in terms of pressure head, elevation head and
velocity head. So the complete expansion of Darcy's law, for our purposes, is

q = C:rT - 3 R 2P
- g -I [PI-P2 + (ZI - Z2) ]
y/ I pg
rd 2 ;Jp g I1 [PI pg
= C (1 _ f)2
- P2 + (ZI - Z2)
]
, (3.22)

where the suffixes refer to two points along the macroscopic flow line separated by
distance I measured also along the macroscopic flow line. The velocity head has been
omitted because it is negligibly small, being the difference of two very small quanti-
ties compared to the pressure and elevation heads.

Let us now review the terms in this expansion of Darcy's law, because it is important
to understand each.
- The specific discharge q is a notional velocity across a plane surface normal to the
macroscopic flow direction. It has a real value, of course, and it is the basis for
calculating ground water flow rates and well yields. It has the dimensions of length
divided by time, but the length is more readily understood as a volume divided by an
area.
- The coefficient C is dimensionless, but it is not a true constant because it includes
a coefficient from the hydraulic radius, the value of which in Clies between 1/6 2 for
spheres and 1/8 2 approximately. There is also a coefficient from Archie's relation-
shipF = bj -m, but this coefficient b is close to unity, varying from about 0.6 to 1.3 in
sands (see Keller, 1966, p. 563, Table 26-5). We shall examine the coefficient C again
before the end of the chapter.
- The tortuosity T = ltll, takes into account the non-linear pore paths, the mean
inclination of which to the macroscopic flow direction is given by cos a = l/T. From
equation (3.18), T- 3 =j1.5m-1.5, so Tvaries asj°.5-0.5mwhere m varies from 1.3 to
about 2.5. In other words, tortuosity is a function of porosity, and for 20% porosity,
T varies from 1.3 to about 3.3 and a varies from 40° to 70° .
The use of Archie's Formation Factor for the evaluation oftortuosity depends on
the assumption of strict analogy between the flow of electicity and liquids. There are
cogent arguments in support of this, for Hubbert has shown that Darcy's law and
Ohm's law are strictly analogous both physically and mathematically (Hubbert,
1940, p. 819; see also Versluys, 1930, p. 217).
Many sedimentary rocks are anisotropic with respect to permeability (and, in-
deed, to electrical resistivity). Of particular significance are the vertical and hori-
zontal permeabilities, the latter normally being greater than the former. It is evident
that the expression for intrinsic permeability must include a vector quantity, and the
62

only vector quantity appears to be tortousity.

The exponent m is called the cementation factor, and the name has been criticized
as inappropriate. The main reason why ordinary cementation seems to have little
effect on permeability (Ftichtbauer, 1967, p. 359) seems to be that it first forms
pendular rings around the points of contact, which is the volume of pore space that
contributes least to flow because of the velocity gradient to the static boundary
surface (Fig. 3-6) (see also Rose, 1959). It seems, therefore, that m is a factor that
takes pore shape into account, for porosities less than about 26% are impossible with
packings of spheres of similar sizes. For straight capillaries parallel to the macro-
scopic flow direction, m = 1, and the tortuosity is equal to 1.
- The hydraulic radius R = fd/(1 - f) takes the wetted surface area into account,
and is a function of both porosity and the harmonic mean grain diameter. The
harmonic mean diameter can be computed from the same data from which the
geometric mean is calculated, and with the same limitations (such as using the
bounding sieve sizes to obtain a measure ofthe mean size ofthe grains retained on
the smaller sieve).
The role of porosity in permeability is seen to be a composite effect due to
tortuosity, hydraulic radius, and the first correction to the notional velocity q.
Permeability varies asf1.5m-0.5f2/(1 - f)2 = f1.5m+ 1.5/(1 - f)2. As m varies from 1.3
to 2.5, permeability varies asf3.S/(1 - f)2 to fS.3/(1 - f)2.
- The kinematic viscosity v = 1]/ P inversely affects the hydraulic conductivity K: the
more viscous and less dense the liquid, the small~r the specific discharge for a
constant hydraulic gradient through a given material. Kinematic viscosity has the
dimensions of a length times a velocity, hence the units are m 2 s -1 in the SI system,
where the mass density is in kg m - 3 and the dynamic viscosity is in kg m -1 s -1 (the
dimensions also reveal why these two viscosities are called kinematic and dynamic).
- The hydraulic gradient ,1h/l is commonly misunderstood and misrepresented,
although its significance has long been recognized (Norton, 1897, Plate VIII and p.
173, drew an 'isopiestic' map of Iowa, with sea-level as datum, and used the term
'hydraulic gradient'). The total head h is the algebraic sum of the pressure head p/ pg
and the elevation head z: it has the dimension oflength (as all heads do). Versluys
(1917) called this the 'hydrostatic potential' or simply 'potential', and mentioned
that we are usually concerned with potential differences. The hydraulic gradient is

Fig. 3-6. Pendular cement occupies space that would have contributed little to liquid flow.
 63

the gradient of total head along the macroscopic flow path of length I: it is dimen-
sionless. Hubbert (1969, p. 15) drew attention to some early misunderstandings, but
the position is now so confused that it would probably be better if a whole new
terminology could be devised. The 1972 AGI Glossary of Geology is wrong in
defining hydraulic gradient as 'the rate of change of pressure head per unit of
distance of flow' (their italics). It is wrong because static pore water in an inclined
aquifer has a pressure-head gradient but no flow. This error arises from their
definition of pressure head, in which they wrongly refer it 'to a specific level such as
land surface'. Pressure head only makes sense when it is measured from the aquifer:
the introduction of an arbitrary datum immediately changes this to total head,
which is correctly defined in the 1972 AGI Glossary of Geology. Figure 3-7 makes
these distinctions clear.
The total head is an energy per unit of weight : the product gh is an energy per unit
of mass, with dimensions L 2 r- 2 • Hubbert (1940, pp. 796-803), in an analysis that is
much more rigorous and general than that here, called gh the potential of the fluid,
with the symbol ,p. So the term g,1h/1 is the potential gradient.

.-h

p/pg
-z

•
p
Fig. 3-7. Relationship between elevation head (z), pressure head (p!pg), and total head (h).
64

THE LIMITS OF DARCY'S LAW

The existence of a lower limit to Darcy's law has been postulated on the grounds of
liquid adsorption to solid surfaces, molecular and pore-throat sizes. Whether this is
really so on a practical scale is hard to determine, and in any case depends on what
we mean by a statement that liquid flow 'follows Darcy's law'. If we mean that there
is a linear relationship between the energy losses and the specific discharge. then
Darcy'S law holds for permeable glass (Fig. 3-8) with an intrinsic permeability of the
order of 10- 16 cm 2 or 10- 8 darcies (see Glossary) for water, acetone, and n-decane
(the permeabilities being computed from data reported by Nordberg, 1944, and
Debye and Cleland, 1959). This is near the lower limit of intrinsic permeabilities
measured in shales as reported by Bredehoeft and Hanshaw (1968) and Magara
(1971).
The matter is very complex because meticulous experimental techniques are
required for the measurement of such small permeabilities. Olsen (1965, 1966)
concluded that Darcy's law is valid for most clays, and that the deviations from
Darcy'S law observed by Hansbo (1960). Lutz and Kemper (1963). and von Engel-
hardt and Tunn (1955) could be attributed to the experimental conditions, but that
some of the threshold gradients for fluid flow found by Miller and Low (1963) were
larger than the possible experimental error. (It should be noted that Miller and Low
found a linear relationship for clays with about 70% porosity, non-linear for those
with about 85% porosity - well beyond the limits of sedimentary rocks.)

1 2 3 4 5
x 10-7 cmjs q
Fig. 3-8. Data of Nordberg (1944) on left, and Debye and Cleland (1959) show that Darcy's law applies to
permeable glass.
 65

Nordberg's data (Fig. 3-8) possibly indicates a threshold hydraulic gradient of

about 100 for liquid flow, a small intercept at q = 0 being indicated. Debye and
Cleland took evaporation into account, and no significant intercept is found. So
evaporation could account for Nordberg's intercept (that is, apparent q < 0 at L1h
= 0 rather than L1h > 0 at q = 0).
It is interesting to compare the values of the constant C found from Schriever's
(1930) and Nordberg's (1944) experiments. Expressing the hydraulic radius more
explicitly by R* = fI S where S is the specific surface, the expression for intrinsic
permeability may be written

k = c*fr3 f2 (3.23)
S2'

Nordberg reports the following data for VYCOR, the brand of porous glass he
used:f = 0.28, S = 1.9 X 106 cm 2 jcm 3 • The tortuosity T = It/I of another sample of
VYCOR is about 2.5 (Barrer and Barrie, 1952). The mean value of k from Nord-
berg's data is found to be 2.65 x 10- 16 cm 2 ; hence C* = 0.7.
The value of C found from Schriever's experiments (8.42 x 10- 3 ) included a
shape factor from the hydraulic radius (k)2; so for comparative purposes,
Schriever's C* = 0.3. The intrinsic permeability of Schriever's material varied from
9 x 10- 6 cm 2 to 4 x 10- 7 cm 2 . The value of C* from the experiments of Schriever
and of Nordberg are remarkably close in view of the great disparity of intrinsic
permeabilities, Schriever's material being of the order of 109 to 10 10 more perme-
able than Nordberg's porous glass. Could the true value be 0.5, and Darcy's law for
laminar flow ofliquids through porous solids identical with that for laminar flow of
liquids through pipes (Equation 2.8c)?

Upper limit of Darcy's law

There is an upper limit to Darcy's law. For a given material, there is a value of q, the
specific discharge, above which the energy losses cease to be linear with q and
increase more rapidly. In other words, there is an upper limit to the range in which
the coefficient of permeability K is constant. It is customary to relate this limit to a
Reynolds number based on q and the mean grain diameter d (rarely defined, but
geometric mean sometimes implied) - and in these terms, the upper limit of Darcy's
law is found to fall at a Reynolds number between 1 and 10. The actual range found
may be greater than that*, but this need not concern us greatly because a Reynolds
number based on q and d in this way is hardly likely to be satisfactory. The notional

* Not as great as that reported by Scheidegger (1960, p. 159). The low value ofO.l (Nielsen, 1951) is
based on a different Reynolds number from the high value of75 (plain and Morrison, 1954) - both being
different from the Re = qpd/11 used by most workers. Nielsen's Reynolds number is not readily con-
verted; but Plain and Morrison's 75 corresponds with a Reynolds number qpd/11 ~ 15. However, their
data do not permit reliable conclusions to be drawn concerning the true upper limit of Darcy's law.
66

velocity q is related to the mean pore velocity by the factor jT- 2 and the mean grain
diameter is a characteristic length of that part of the material through which the
fluid does not pass (it has been compared to taking the pipe thickness as the
characteristic dimension of pipe flow - unfairly, because the grain diameter is a
measure of pore size, even if a poor one). A more realistic Reynolds number for fluid
flow through porous solids would be
qT2 p fd
T Ii a(1- f)'
where d is the harmonic mean grain diameter and a takes the value 6 for spherical
grains, to about 8 for irregular shapes. Using the argument leading to Equations
3.17 and 3.18, we see that jT- 2 = r,
so there is a porosity factor in this Reynolds
number, f'-mja(1 - f). Iff = 0.3, m = 1.3, and a = 7, the porosity factor is 0.3.
Reynolds number based on q and mean grain diameter are too large, and do not take
all the significant parameters of the porous solid into account.
Lindquist (1933) made an interesting study of water flow through porous soils
and came to the conclusion that Darcy's law failed at a Reynolds number about 4
(based on q and mean diameter), but that flow was still laminar to Re = 180; and
that the failure of Darcy's law is due to inertial forces in the flow through irregular
pore spaces (Lindquist, 1933, pp. 89-91). Bakhmeteff and Feodoroff (1937) also
found Darcy's law to fail at a Reynolds number about 4, and they clearly under-
stood the nature of this failure.
Hubbert (1940, pp. 819-822) has emphasized that the first failure of Darcy's law is
due not to turbulence but to the inertial forces' becoming significant. Brownell and
Katz (1947), when allowance is made for their porosity term, also found a critical
Reynolds number about 5 above which Darcy's law did not apply; they thought that
this failure was due to turbulence in some paths, not in others.
Schneebeli (1955) found that Darcy's law failed at a Reynolds number about 5 for
spheres,2 for granite chips, but that turbulence set in at a Reynolds number about
60 for both materials. These results suggest that shape may influence the point at
which Darcy's law fails, as well as tortuosity; but the difference could be due to
porosity. The effect of porosity is shown in Figure 3-9, in which Schriever's data is
plotted against Schneebeli's dimensionless numbers. *
* Statistics is no substitute for thought. Regression analysis of Schneebeli's groups with Schriever's
data suggests the relationship Cf = 1433 Re- O. 99 with r = -0.994 and a level of significance < 0.1%.
From this one might conclude that Cf x Re = constant, and so d2 /K = constant for constant porosity.
However, regression analysis of the data of each diameter leads consistently to a (log/log) slope of
-2.000 and r = -1. For constant diameter, C (cx Re- 2 from which we find gd 3(p/Yf)2 = constant. This
tells us nothing we did not already know, because all the other quantities were constant in Schriever's
experiments.
Two valid conclusions can be drawn: Schriever's experimental techniques were meticulous, and at
least one pertinent dimensionless variable is missing from Schneebeli's groups. It is because there are only
limited ranges of porosity to be obtained from porous solids that the suppression of porosity in the first
result has little effect on the statistical significance.
 67

106~~~------~----------~-----

104+-----------~----------~~---

103~--------~~--------~-----
10- 3

Fig. 3-9. Schriever's (1930) data plotted against Schneebeli's (1955) dimensionless groups.

Fortunately, most ground-water flow takes place within the realm of Darcy's law,
and the real critical value ofthe Reynolds number that takes porosity into account,
and used the harmonic mean grain diameter (always smaller than the geometric or
arithmetic mean) appears to be close to 1.

The generalization of Darcy's law for liquids is an oversimplification in the geolog-

ical context; but it helps us to understand the more complex reality, the meaning of
permeability, and the factors that affect it. Real sediments display variable grain
size, variable sorting, porosity and tortuosity, and these may change along the flow
path. The movement of subsurface water through porous sediments (as distinct
from fissured rocks) may involve changing geometry ofthe bed as a whole, changing
intrinsic permeability and, through changing temperatures, changing viscosity and
density of the liquid.
Consider water flow through two layers of contrasting sand (Fig. 3-10) in appa-
ratus similar to Darcy's. The conservation of matter requires that the volumetric
68

Fig. 3-10. Apparatus for investigation flow through two contrasting sands.

rate of flow per unit of cross-sectional area in the first layer shall be the same as that
across the second layer. Hence,

P LJh 1 P LJh2 - P LJh

Q/A = q = K 1 - g - = k 2 - g - = k-g-. (3.24)
11 II 11 h 11 I
If pg/11 can be regarded as constant, then

k /k = LJh2/12 (3.25)
1 2 LJhl/ll

and the ratio of the intrinsic permeabilities (and their hydraulic conductivities) of
the layers is equal to the inverse ratio of their hydraulic gradients. If, for example,
the hydraulic gradient across the first layer is half that across the second, the
permeability of the second layer is half that of the first. The two layers together
behave as if they were one layer with permeability k and hydraulic gradient Jh/l.
Darcy's law, in practical terms, is a statistical relationship in which inhomogeneities
 69

are averaged out between manometer readings. The greater the number of measure-
ments of total head along a flow path, the more detailed the information we can
obtain about permeability changes. Pressure and total head measurements or esti-
mates are more reliably obtained at present than in situ permeability measurements.
In the geological context, the geothermal gradient makes variables of the dynamic
and kinematic viscosities, 1'/ and v (= 1'/1 p). Strictly, pressure also affects these,
increasing the density and increasing the viscosity of water at subsurface tempera-
tures (see Bett and Cappi, 1965). Figure 3-11 shows the approximate change of v
under subsurface conditions of normal hydrostatic pressure and a geothermal
gradient of 36°Cfkm. While these changes are important because increasing the
temperature from 25 ° C to 100 ° C increases the hydraulic conductivity of a perme-
able rock by a factor of 3, such temperature changes will occur usually over a depth
range of about 2 km. Vertical water movement during compaction will commonly
be through less than 500 m, with a temperature difference of less than 20°C. At
temperatures above 70°C, corresponding to depths below about 1,500 m, the
change of viscosity is relatively slight (about 25% for a 20°C change) and no

Kinematic Viscosity m 2 S-1

10-7 0.5 1 )(10- 6
I t......--"""

-"
Y
V" E
~V
/
1000 .c...
Q.
/ CI)
C

I
J
2000 ...
CI)

to
I E
><
I
I
...
0
Q.
3000 Q.
I
<t
I
o
150 c
Fig. 3-11. Approximate kinematic viscosity of water under subsurface conditions with a geothermal
gradient of 36 DC/km and a normal hydrostatic pressure gradient (interpolated from data of Gray, 1957,
p. 2-138, table 2n-l, and p. 2-209, table 2v-6).
70

significant error is likely if the average is taken. (If these sound like modern argu-
ments, the reader is referred to King, 1899, pp. 81-82!)
These principles can be extended to multi-layer sequences, and the relative
permeabilities of the layers deduced. The absolute values ofthe permeabilities, and
ofthe specific discharge q, are much more difficult to obtain, and no attempt to do so
will be made here.
Lateral migration of water through a sand may involve both changes of perme-
ability and of the gross geometry of the sand. These will be considered in the next
chapter after the introduction of flow nets; but it is worth noting here that the
specific discharge will not be constant in a permeable bed of variable thickness
because the cross-sectional area A changes. We therefore write equation (3.24)

P Llh1 P Llh2
Q = k1 A1 - 9 - = k2 A2 - 9 - (3.26)
'1 11 '1 12
and A is commonly proportional to the thickness of the aquifer.

SELECTED BIBLIOGRAPHY

Archie, G.E., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics.
Trans. American Illst. Mining Metallurgical Engineers (petroleum Division), 146: 54-62.
Bakhmeteff, B.A., and Feodoroff, N.V., 1937. Flow through granular media. J. Applied Mechanics, 4:
A97-A104. (Also in Trans. American Soc. Mechanical Engineers, 59. See discussion by L.P. Hatch, J.
appl. Mech., 5: A86-A90, which is also in Trans. Am. Soc. mech. Engrs, 60).
Barrer, R.M., and Barrie, J.A., 1952. Sorption and surface diffusion in porous glass. Proc. Royal Society,
A213: 250-265.
Bear, J., 1972. Dynamics offluids in porous media. American Elsevier, New York, 764 pp.
Bett, K.E., and Cappi, J.B., 1965. Effect of pressure on the viscosity of water. Nature, 207 (4997): 620-
621.
Blake, F .C., 1923. The resistance of packing to fluid flow. Trans. American Inst. Chemical Engineers, 14
(for 1922): 415-421.
Bredehoeft, J.D., and Hanshaw, B.B., 1968. On the maintenance of anomalous fluid pressures: I. Thick
sedimentary sequences. Bull. Geol. Soc. America, 79: 1097-1106.
Bridgman, P.W., 1926. The effect of pressure on the viscosity offorty-three pure liquids. Proc. American
Academy Arts Sciences, 61 (3): 57-99.
Brownell, L.E., and Katz, D.L., 1947. Flow of fluids through porous media. I. Single homogeneous
fluids. Chemical Engineering Progress, 43 (10): 537-544. (Discussion: 544-548.)
Cornell, D., and Katz, D.L., 1953. Flow of gases through consolidated porous media. Industrial and
Engineering Chemistry, 45 (10): 2145-2152.
Darcy, H., 1856. Lesfolltaines publiques de la ville de Dijon. V. Dalmont, Paris, 674 pp.
Debye, P., and Cleland, R.L., 1959. Flow ofliquid hydrocarbons in porous Vycor. J. Applied Physics, 30
(6): 843-849.
Fair, G.M., and Hatch, L.P., 1933. Fundamental factors governing the stream-line flow of water through
sand. J. American Water Works Ass., 25 (11): 1551-1565.
Fiichtbauer, H., 1967. Influence of different types of diagenesis on sandstone porosity. Proc. 7th World
Petroleum Congress, 2: 353-370.
Gray, D.E., (Ed.), 1957. American Institute of Physics Handbook (1957). McGraw-Hill, New York.
Hansbo, S., 1960. Consolidation of clay with special reference to the influence of vertical sand drains.
Sweden. Staten.~ Geoteklliska Illstitut. Proc., 18: 1-160.
 71

Happel, J., and Brenner, H., 1965. Low Reynolds number hydrodynamics with special applications to
particulate media. Prentice-Hall, Englewood Cliffs, N.J., 553 pp.
Hubbert, M.K., 1940. Theory of ground-water motion. J. Geology, 48 (8): 785-944.
Hubbert, M.K., 1956. Darcy's law and the field equations of the flow of underground fluids. Trans.
American Inst. Mining Metallurgical Petroleum Engineers, 207: 222-239. (Also in J. Petroleum Tech-
nology,8.)
Hubbert, M.K., 1957. Darcy's law and the field equations of the flow of underground fluids. Bulletin de
/' Association Internationale d' Hydrologie Scientifique, no. 5: 24-59.
Hubbert, M.K., 1969. The theory of ground-lVater motion and related papers. Hafner, New York and
London, 310 pp.
Irmay, S., 1953. Saturated steady flow in non-homogeneous media and its applications to earth em-
bankments, wells, drains. Proc. 3rd International Conference Soil Mechanics (Zurich), 2: 259-263.
Keller, G.V., 1966. Electrical properties of rocks and minerals. In: S.P. Clark (Ed.), Handbook of
physical constants (revised edition). Memoir Geoi. Soc. America, 97: 553-577.
Kennedy, G.c., and Holser, W.T., 1966. Pressure-volume-temperature and phase relations of water and
carbon dioxide. In: S.P. Clark (Ed.), Handbook of physical constants (revised edition). Memoir Geol.
Soc. America, 97: 371-384.
King, F.H., 1899. Principles and conditions of the movements of ground water. Annual Report U.S. Geol.
Surv. (1897-98), 19 (2): 59-294.
Klinkenberg, LJ., 1942. The permeability of porous media to liquids and gases. American Petroleum
Inst. Drilling and Production Practice 1941: 200-213.
Kozeny, J., 1927a. Dber Grundwasserbewegung. Wasser kraft und WasserlVirtschaft, 22: 67-70, 86-88,
103-104,120-122,146-148.
Kozeny, J., 1927b. Dber kapillare Leitung des Wassers im Boden (Aufstieg, Versickerung und An-
wendung auf die Bewiisserung). Sitzungsberichte del' A kademie del' Wissenschaften in Wien, Abt. IIa,
136: 271-306.
Lindquist, E., 1933. On the flow of water through porous soil. 1" Congres des Grands Barrages
(Stockholm, 1933),5: 81-101.
Lutz, J.F., and Kemper, W.D., 1959. Intrinsic permeability of day as affected by clay-water interaction.
Soil Science, 88: 83-90.
Magara, K., 1971. Permeability considerations in generation of abnormal pressures. J. Soc. Petroleum
Engineers, II: 236-242.
Miller, RJ., and Low, P.F., 1963. Threshold gradient for water flow in clay systems. Proc. Soil Soc.
America, 27: 605-609.
Nielsen, R.F., 1951. Permeability constancy range of a porous medium. World Oil, 132 (6): 188-192.
Nordberg, M.E., 1944. Properties of some Vycor-brand glasses. J. American Ceramic Soc., 27: 299-305.
Norton, W.H., 1897. Artesian wells oflowa. Iowa Geol. Survey, 6: 113-428.
Olsen, H.W., 1965. Deviations from Darcy's law in saturated clays. Proc. Soil Science Soc. America, 29:
135-140.
Olsen, H.W., 1966. Darcy's law in saturated kaolinite. Water Resources Research, 2: 287-296.
Owen, J.E., 1952. The resistivity ofa fluid-filled porous body. Trans. American Inst. Mining Metallurg-
ical Engineers (petroleum Branch), 195: 169-174.
Plain, GJ., and Morrison, H.L., 1954. Critical Reynolds number and flow permeability. American J.
Physics, 22: 143-146.
Rose, W.D., 1959. Calculations based on the Kozeny-Carman theory. J. Geophysical Research, 64 (1):
103-110.
Scheidegger, A.E., 1960. The physics of flo\\' through porous media (revised edition). University of
Toronto Press, Toronto, 313 pp.
Schneebeli, G., 1955. Experiences sur la limite de validite de la loi de Darcy et I'apparition de la
turbulence dans un ecoulement de filtration. La Houille Blanche, 10 (2): 141-149.
Schriever, W., 1930. Law of flow for the passage of a gas-free liquid through a spherical-grain sand.
Trans. American Inst. Mining Metallurgical Engineers (petroleumDivision), 86: 329-336.
Slichter, C.S., 1899. Theoretical investigation of the motion of ground waters. Annual Report U.S. Geol.
Surv. (1897-98), 19 (2): 295-384.
Street, N., 1958. Tortuosity concepts. Australian J. Chemistry, II (4): 607-609.
72

Versluys, J., 1917. De beweging van het grondwater. Water, 1: 23-25,44-46,74-76,95.

Versluys, J., 1930. The origin of artesian pressure. Verhandelingen der Koninklijke Nederlandsche Aka-
demie van Wetenschappen, Afd. Natuurkunde, 33: 214-222.
Von Engelhardt, W., and Tunn, W., 1954. Dber das Stromen von Fliissigkeiten durch Sandsteine.
Heidelberger Beitriige zur Mineralogie und Petrographie, 4: 12-25.
Von Engelhardt, W., and Tunn, W.L.M., 1955. The flow of fluids through sandstone. Illinois State Geol.
Surv. Circular 194 (17 pp.).
Weyer, K.E., 1978. Hydraulic forces in permeable media. Memoire du Bureau de Recherches Geologiques
et Minieres, 91: 285-297.
Winsauer, W.o., Shearin, H.M., Masson, P.H., and Williams, M., 1952. Resistivity of brine-saturated
sands in relation to pore geometry. Bull. American Ass. Petroleum Geologists, 36 (2): 253-277.
Zoback, MD., and ByerJee, J.D., 1975. Permeability and effective stress. Bull. American Ass. Petroleum
Geologists, 59 (1): 154-158.
 4. THE AQUIFER AND FIELDS OF FLOW

It is one thing to make experiments to determine, as Darcy did, the amount of water
that can be passed through a sand filter: it is quite another to apply these results to
the geological materials of an aquifer. Consider an artesian aquifer, as in Figure 4-1,
with water entering it in the intake area on high ground and leaving it by leakage or
extraction where the ground is lower (to the left of the figure). Any well drilled into
this aquifer will encounter water in it that will flow at the surface unless it is
restrained by wellhead equipment. The pressure of the water and its density can be
measured at the surface, and so the pressure head computed (if the density varies
well to well, mean or unit density is taken). We can assume that the velocity head will
be negligible, and the total head measured above our sea-level datum is

(4.1)

and the fluid potential ip in the position of the well is

ip = gh = gz + f. (4.2)
p

P!pg
~-- ------...:--

z
! Sea Level
Fig. 4-1. Schematic section through leaking artesian basin, showing elevation head, pressure head, and
head loss.
74

For each well the total head or fluid potential is .-:asily calculated; and if there are
enough wells, it will be possible to contour a surface that is the conceptual surface of
total head of the aquifer and, lacking only the factor g, its fluid potential surface.
This surface is known as the potentiometric surface (also, but less desirably, piezo-
metric surface: it is not a surface that is a measure of pressure only). It is important
for the reader to be absolutely clear as to what this surface is, and what it is not.
The potentiometric surface is a conceptual surface defined by the data of equa-
tions (4.1) or (4.2) in the aquifer. It is not the level to which the water would rise if it
were free to do so: it is the level to which water would rise in a manometer inserted
into the aquifer at any point. The shape of the surface is determined by the energy
losses due to movement ofthe water in the aquifer: if the water is not moving, there is
ho energy loss and the potentiometric surface is horizontal. Contours drawn on the
potentiometric surface are, of course, lines of equal potential that are known as
equipotential lines. They are the intersection of an equipotential surface with the
potentiometric surface. The equipotential surface only coincides with the potentio-
metric surface when the water is at rest: under all other conditions, they are inclined
to each other.
We concluded from Darcy's experiments that water flows from positions of
higher energy or potential to positions of lower energy or potential. Hence, where
the potentiometric surface is not horizontal, the aquifer water from which it was
derived is in motion, and it is moving in the direction of maximum slope down the
potentiometric surface. Intuitively, we can accept that the direction of flow at a point
is normal to the equipotential surface at that point, and so normal to the equi-
potential line at that point (it can also be proved analytically), so the macroscopic
streamlines or flowlines can be drawn from the equipotential lines and the two form
aflow net (Fig. 4-2). In general, flowlines converge where the liquid is accelerating,
diverge where it is decelerating.
Consider two wells sited down a single flow line (but not producing), as in Figure
4-3. The water is flowing because the potentiometric surface is not horizontal: it is
flowing from B·to A because that is the direction of decreasing energy. The questions
then arise, how much water is flowing, and how fast?
If we knew the hydraulic conductivity or intrinsic permeability ofthe aquifer we
could at least answer the first question from Darcy's law:

_ Q_ hB - hA _ k~ LJcp
q-A- K I - 11 1 ' (4.3)

(Note carefully that I is the macroscopic length through the aquifer, not the hori-
zontal distance, so the slope of the potentiometric surface does not necessarily give
the hydraulic or potential gradient.)
The practical problem here is that a measurement of permeability from a borehole
core is, in the first place, a very small sample ofthe aquifer; and in the second place,
it is likely to be mechanically disturbed, having been cut around by the bit and then
 75

flow lines.
d by equipotential and
Fig. 4-2. Flo w net forme

Ah
- hL
-- -= = ----------
B
A Ground le ve l ,..;;---

--=-=--------- -
...
..
'ig.4-3.
76

removed from its in situ temperature and pressure to atmospheric. So while its
permeability can be measured with great precision, it is an illusory precision. (This is
a common feature of subsurface geology that may beguile us in the belief that 'at last
we have some hard data'.) Ideally, we wish to know the bulk permeability of the
aquifer in the direction of flow, but must often be satisfied with estimates for lesser
volumes - but volumes larger than the hand specimen. The best estimates available
with present technology are from data obtained from pumping wells. Before con-
sidering this, it is necessary to look at the field of flow within the aquifer more
closely.

Fields of flow

Any physical quantity that can be mapped in 3-dimensional space can be said to
have afield of that quantity. Thus a body of water can be said to have a temperature
field, a pressure field, and, if the water is moving, a velocity field. Since temperature
and pressure are scalar quantities (for any position they are sufficiently defined by a
single number and a scale of measurement) and velocity is a vector quantity (for any
position, number, scale, and direction are required for its definition) the body of
water is more precisely said to have scalar fields of temperature and pressure, and a
vector field of velocity. We talk too of gravity and stress fields (the latter being
important in matters we shall come to in later chapters). Gravity is a vector quantity
that, for our purposes, can be regarded as a field in which the magnitude and
direction of the force remains constant (leaving to geophysicists the small variations
that are significant for them). Stress is a tensor quantity, requiring six quantities for
its definition (tensors are to vectors what vectors are to scalars). For the moment, we
shall confine ourselves to scalar and vector fields in the context of water flow
through porous rocks.
For every scalar field there is an associated vector field of the rate of change of the
scalar quantity in 3-dimensional space. For example, in a body of water there are
surfaces in the pressure field that pass continuously through points of equal pres-
sure. These surfaces of equal pressure (isobaric surfaces) in static water are hori-
zontal: but there is a vector field of pressure gradient in which the field lines are
everywhere normal to the isobaric surfaces. Likewise, there is a vector field of
temperature gradient in which the field lines are normal to the isothermal surfaces.
Potentials are scalar quantities: the potential at a point has magnitude but no
direction. Hence the fluid potential field in a body of water is a scalar field, with
which is associated a vector field of potential gradient. This field is of fundamental
importance in the analysis of ground-water movement because the fluid potential
gradient is, as we have seen, the source of the 'driving force' of the water through
porous rock.
There are four fields in a ground-water aquifer that are of particular importance:
the pressure field, the velocity field, the potential field - and all geological processes
 77

take place in the gravity field. We shall first consider the pressure field and the
gravity field.
When a body of liquid is at rest, the surfaces of equal pressure are all horizontal
and the vector field of pressure gradient has perpendicular field lines that are
sometimes said to represent vectors equal in magnitude but opposite in direction to
the force of gravity. This is wrong because the dimensions of pressure gradient are
[ML- I T- 2 ] [L- I ] = [ML- 2 T- 2 ], which are the dimensions of weight density,
pg, as we would expect, while the dimensions offorce (of gravity) are [MLT- 2 ]. We
must compare only quantities that are dimensionally alike. The force of gravity
gives to unit mass an acceleration g. What, then, is the pressure as a force on unit
mass? Pressure is a force per unit of area on which it acts - in this case, a horizontal
area. The pressure gradient is the change of pressure along a line normal to the
isobaric surfaces - in this case, a vertical line. Hence the net force acting on a unit
cube of the liquid (Fig. 4-4) is the difference between the pressures on the two
horizontal faces of the unit cube, (P2 - pd. The mass of this unit volume of the
liquid is p. Hence the net force acting on unit mass of the liquid is

1 P2 - PI 1
Fp = - d d = - grad p, (4.4)
p 2- 1 P
where (P2 - pd/(d 2 - d I ) is the pressure gradient, which we can conveniently
abbreviate by using the vector notation, grad p. This force is dimensionally identical

Fig. 4-4.
78

to the force of gravity acting on unit mass (that is, an acceleration, LT- 2) and since
grad p is a vector of magnitude pg in a fluid at rest,

1
- grad p = g (4.5)
p

and we can say that the vector field of pressure gradient in terms of force per unit of
mass in a body of water at rest is equal in magnitude and opposite in direction to the
vector field of gravity. The resultant force on the body of water is zero, and it is
therefore at rest (Fig. 4-5).
By continuing this line of argument we can say that the isobaric surfaces in a body
of water in motion are not horizontal, and the vector field of grad p is not vertical
(Fig. 4-6). The resultant force acts on the water to give it an acceleration that is
reduced to a velocity by the frictional resistive forces. But velocity of liquids in
porous solids, as we have seen, is difficult to deal with. Let us therefore look at the
fluid potential field.
The fluid potential <P is the scalar quantity of energy possessed by unit mass of the
liquid in a particular position. The associated vector quantity, grad <P, is the fluid
potential gradient and is the quantity that determines the direction offlow and, with
the coefficient of permeability or hydraulic conductivity K, the specific discharge of
water (volume across unit area in unit time). Hubbert (1940, pp. 794-802) showed
that the fluid potential is the sum of the work done against gravity and pressure
when transposing unit mass by a frictionless process from an arbitrary state and
place of reference to the particular state and place of interest;

<P = gz +p- Po . (4.6)

P
We have just considered the second term on the right of equation (4.6); we now look
at the first term on the right. In Chapter 2 (p. 30) we took gz to be the potential
mechanical energy of unit mass of water, which we shall now call the gravity
potential
U=gz. (4.7)
This scalar field has its associated vector field of gradient of gravity potential with
dimensions [LT- 2]. Clearly, the surfaces of equal gravity potential are horizontal,
and the field lines of gravity potential gradient are vertical and equal to g. If the body
of water is static, the surfaces of equal pressure are horizontal, with vertical field
lines of pressure gradient in terms of force per unit mass; and in this special case, the
vector 1/ p grad p is equal in magnitude and opposite in direction to the gravity
potential gradient vector, grad U. The resultant force is zero. If the body of water is
moving, however, the vector quantity of force per unit of mass due to pressure is not
vertical, nor is it necessarily equal to grad U; and the resultant is the vector quantity
of fluid potential gradient (Fig. 4-7).
 79

~ grad p

g
Fig. 4-5.

~ grad p
I
I
I
I
I
I
I
I

/
/
/
/
/
/
/
/
g
Fig. 4-6.

We thus have three superimposed fields of which the field of fluid potential
gradient is the resultant of the other two. Surfaces of constant pressure, gravity
potential, and fluid potential are normal to their gradient vectors (Fig. 4-8); and, in
theory at least, they are mappable from data acquired from boreholes. Ground
water moves in the direction of the fluid potential gradient vector, from positions of
higher fluid potential to positions of lower fluid potential.
Reverting now to the artesian aquifer with which we started this chapter, we can
infer the fields of flow within the aquifer from the measurements of depths and
pressures in boreholes, and the computed potentiometric surface (Fig. 4-9). Because
the water is confined to its aquifer, its macroscopic flow lines are restricted by the
80

I ~ grad p
I
I
I
I
I
I
I
grad cP
/
/
/
/
/
/
/
/
grad U
Fig. 4-7.

"-
"- P2/P
P1/P
- U1
-U 2

Fig. 4-8. Surfaces of constant pressure, gravity potential, and fluid potential are normal to their gradient
vectors.

upper and lower boundaries of the aquifer, and the equipotential surfaces are
normal to this flow. The surfaces of equal pressure are not horizontal, but inclined in
the direction of motion by an amount that is determined, in effect, by the fluid
potential gradient.
It will be remembered that grad qJ is a vector, which has both direction and
magnitude. The direction conforms to the aquifer, as it must: the magnitude may be
 81

Fig. 4-9. Gradient vectors of pressure, gravity potential, and fluid potential in a confined aquifer
(exaggerated ).

large or small. In artesian basins, the fluid potential is usually relatively large, but
the fluid potential gradient is small unless there is significant leakage or production
from the aquifer.
Now, the volume of the aquifer enclosed by two flow lines projected through the
aquifer (Fig. 4-10) may be considered as a stream tube, or flow tube, enclosed by an
impermeable surface (because there can be no flow across tlow lines). It follows that
the mass or volume discharge in steady flow across any transverse area of the flow
tube per unit of time is constant. Thus, from Darcy's law,
Q = Kl Al ,dh1/l1
= K2 A2 ,dh 2 /1 2 . (4.8)

If we assume that the hydraulic gradient, ,dh/l, is approximated by the contour

spacing (as it is in all gently-dipping aquifers) we can see that
1) if A remains constant, K is greatest when the hydraulic gradient is least;
2) if A decreases downstream and K remains constant, the hydraulic gradient
increases; and
3) if the hydraulic gradient remains constant, K increases as A decreases.
Thus a map of the potentiometric surface of an aquifer allows us to infer the
directions of water movement within the aquifer, and to make certain general
deductions about the changes in aquifer properties. It does more: a closed area of
high potential can only exist where there is a source, and water is being added to the
82

Fig. 4-10. Stream tube in confined aquifer enclosed by two flow lines projected through the aquifer.

aquifer. And a closed area of low potential can only exist where there is a sink, and
water is being taken from the aquifer. The latter is of present interest to us because it
happens around a producing well, and the form of the sink (known as the cone of
depression) enables us to obtain a measure of permeability over a larger area than
the hand specimen.

PRODUCING WATER WELL

Consider a well that penetrates a horizontal isotropic confined aquifer completely,

the water in the aquifer being static (Fig. 4-11). Before the well produces, the level of
the water in it coincides with the potentiometric surface of the water in the aquifer.
Isobaric surfaces are horizontal; the water is everywhere within the aquifer at the
same potential, so there is one horizontal equipotential surface: the water does not
flow. When water is produced from the well, the well becomes a sink, and water
 83

f-::. -_--=-_--=-_--=-_-_-_--=-_---=-=--=
f-=---=- -=-_--=-_-_-_-_-_-_-_-_---=-=

Fig. 4-11.

flows radially through the aquifer to the well down a potential gradient that is
created by the reduced potential in the well. The volume of water, Q, produced in
unit time (when steady flow has been achieved, which may take some time) is
produced into the borehole across an area

(4.9)

where r w is the radius of the borehole, and t the thickness of the aquifer. The same
volume of water is produced in unit time across all concentric areas of greater
radius, r, so that

A=2nrt. (4.9a)

Letting qw represent the specific discharge, Q/ Aw, into the well; and q, the specific
discharge across the concentric area A at radius r, then
qw x 2nrwt = q x 2nrt

and

(4.10)

From this we see that the specific discharge, which is a notional velocity, varies
inversely with distance from the well. From Darcy's law

(4.11)

from which the hydraulic gradient is seen to be

84

Ah qw rw
Ar = K r· (4.11a)

Indicating the total head at the wall of the borehole by hw, the total head at radius r is
given by

hr = hw qwrw
+-K
- f rw
r 1
-dr
r

qwrw I r
hw + (4.12)
K- nrw-.

Since qw = Q/2nrwt, we may substitute Q/2nt for qwrw in equation (4.12):

Q r
(4.l2a)
hr=hw+--In-
2ntK rw
and the total head is seen to increase as the natural logarithm of the distance from
the borehole, approaching the total head of the static aquifer (prior to production
from the well).
Equation (4.12a) indicates that h increases indefinitely as r increases, that is,
steady flow is theoretically impossible in a finite aquifer.
More generally,

Q r2
h 2 -h 1 =2 K ln - , (4.13)
nt r1
where the suffix 2 represents a station further from the well than the station
represented by the suffix 1 (Fig. 4-12), which may be at the producing well. The
boundary conditions for these equations are that h2 = h at r = ri, where ri is the
'radius of influence'. Equation (4.13)is due to Thiem (1906) and bears his name. The
product tK, with dimensions [L 2 T- 1 ], is known as the transmissivity of the aquifer.
The hydraulic conductivity is therefore given by

K= Q ln~ (4.13a)
2nt(h2 - hI) r2 .
From this it can be seen that if we can determine the elevation of the potentiometric
surface in two positions when the well is producing steadily at rate Q, then the value
of the hydraulic conductivity can be determined. The elevation of the potentio-
metric surface can be measured in what are called 'witness wells' (boreholes of small
diameter drilled specially for this purpose).
Equation (4.13a) is strictly valid only for steady flow, and this is theoretically
impossible in a finite aquifer. But it shows that the larger the value of K, the
shallower the cone of depression and the larger the radius of influence of the
producing well. Witness wells should be drilled well within the radius of influence, in
the pronounced part ofthe cone. The producing well itself may be used to determine
 85

-
--- --- --- --- --- --- ---
- --- --- --- --- --- --- --

Fig. 4-12.

hI: but while this practice leads to a risk of error <;lue to possible non-Darcy
behaviour near the borehole, it should never be forgotten that it is better than no
data at all. The producing well need not be produced at the maximum rate for the
purposes of determining the hydraulic conductivity, but it must be at a rate high
enough to generate a measurable cone of depression or drawdown in the well. Steady
flow may be assumed when the water level (or pressure) in the witness wells stabilize.
If the water in the aquifer is not static, that is, the potentiometric surface is
inclined from the horizontal, then the symmetry of the cone of depression is lost
when the borehole is put on production. But potentials and heads are scalar
quantities, so the observed potentials or heads are the result of the superposition of
one field (due to pumping) on another:

<Ptotal: <Paquifer + <Pwell . }

(4.14)
htotal - haquifer + hwell .
The potential or head due to the well will be negative when water is being extracted
(positive, if being injected).
If this refinement is to be pursued, and it need only be pursued when the incli-
nation of the aquifer's potentiometric surface is considerable, at least two witness
wells will be required, and they must not be in a straight line with the well. If such a
configuration as in Figure 4-13 is used, we can aquire valuable information on the
aquifer. In the first place, the static water levels in the three wells (static, in this
context, refers to the levels when the well is not producing, or the aquifer has
86

-
350
- 30

- 10
90
Fig. 4-13.

---------------

--------------
-
----
----
----
----
----
--------- --- ---
-
----
----
----
----
----
------- --- ---- --
---
---
--
- --
- --
- --
- --
- --
- --
- --
- -
-

Fig. 4-14a. Potentiometric surface in profile across well producing from a confined aquifer in which the
regional flow is from right to left.
 87

-------------------

Fig. 4-14b. Flow net obtained by superimposing the effect of the well on the aquifer.

stabilized after production) define the slope of the potentiometric surface, and hence
the direction of water movement in the aquifer.
When steady production at rate Q has been attained, the cone of depression that
results is superimposed on the potentiometric surface of the aquifer. There are, of
course, sophisticated methods of solving these problems, but a simple approach
may still be useful. The drawdown (s) is the difference between the static and
pumping levels in the wells (Equation (4.13a) could have been written in terms of
drawdown rather than head). Hence, an approximation to the hydraulic conductiv-
ity can be obtained by substituting (Sl - S2) for (h2 - hd in equation (4.13a); and
from equation (4.13), the radii of equal intervals of drawdown can be estimated. The
equipotential lines ofthe combined field are obtained from the algebraic sum of the
two components, as in Figure 4-14. The flow net can be completed by sketching in
the flow lines normal to the equipotential lines.
88

There is a relatively narrow path to follow in these matters: on the one hand, there
is the danger of drawing inferences that are not justified by the data, and, on the
other hand, the danger of being so scared by the lack of precision that no assessment
is made. For example, if the hydraulic conductivity can be estimated from the
drawdown data, this also allows us to estimate the natural flow through the aquifer
in the vicinity of the well when it is not on production. This estimate may well be
inaccurate; but if reasonable care is taken, it is surely better than no estimate.

Unconfined aquifers

We have considered confined aquifers so far because they are more easily treated
quantitatively. Unconfined aquifers are more difficult because the water table, or
free water surface, is the potentiometric surface (Fig. 4-15). Not only is the flow then
three-dimensional near the well, but also the area of concentric cylinders across
which the water flows decreases towards the well (c.f., t, the aquifer thickness,
assumed constant in the confined aquifer).
The equation for unconfined flow, analogous to equation (4.13), is

h 22 _ h 21 _
--
Q 1nr2- (4.15)
nK rl

but this equation lacks the rigour of equation (4.13) because it depends on assump-
tions known as the Dupuit assumptions, which are: that the flow is uniform and
horizontal, and that the flow rate is proportional to the tangent of the free surface
slope, Jh/ Jr, not the sine. These assumptions are clearly reasonable for flow at some

Fig. 4-15. Well penetrating unconfined aquifer: production induces a cone of depression on the water-
table.
 89

distance from the well, where the slope of the free water surface is small, but they
become quite unacceptable near the well, where there is a considerable vertical
component of flow.
Equation (4.15) is derived as follows:
Assume a horizontal isotropic aquifer with an impermeable base and static water to
be penetrated completely by a well producing steadily at rate Q. With the assump-
tion of horizontal flow, this rate Q flows across any concentric cylindrical surface
below the free water level: the area of such a surface is is 2nrh where h is the elevation
of the free water surface above the base ofthe aquifer. Darcy's law, with the Dupuit
assumptions, is then written

Q=KA dh
dr
dh
= 2nKrh dr. (4.16)

Integration of this equation with respect to h and to r, for the limits h = hI when
r = rl and h = h2 when r = r2, gives us

Q = nK h~ - hi
In (r2/rd
which can be re-arranged in the form of equation (4.15).
Provided the slope dh/dr is not large, an estimate ofthe hydraulic conductivity can
be obtained from a producing well and two witness wells. If rl is taken as the
borehole radius, rw , or rl is small (witness well near the water well), such estimates
can be misleading because the Dupuit assumptions are no longer valid.
It will occur to the reader that when a well ceases production from either a
confined or an unconfined aquifer, the rate at which cone of depression is elim-
inated, and the rate of drawn-down water levels in the wells returning to normal, will
also be a measure of the hydraulic parameters of the aquifer (Horner, 1951).
Likewise, if water is injected into the aquifer, there will be a cone of impression above
the original free water surface; and its shape will depend on the hydraulic param-
eters of an unconfined aquifer (for a confined aquifer, the cone of impression is
superimposed on the potentiometric surface).
It is not our purpose here to pursue these matters to the point that the reader will
be equipped to perform the duties of a ground-water hydrologist; for that, he must
consult such standard texts as Todd (1959), De Wiest (1965), and Davis and
DeWiest (1966). Our aim is an understanding of the phenomena.

NATURAL SINKS

The phenomena of drawdown and discharge associated with water wells producing
 90

from unconfined aquifers have their natural counterpart in the flow of streams and
rivers that penetrate an unconfined aquifer (Fig. 4-16). However, it is in the nature
of an unconfined aquifer that it is subject to recharge from rainfall: so the conditions
change with time, or from season to season.
To understand these processes, we shall look at water flow into an open channel
that partly penetrates an unconfined aquifer. Consider the channel in Figure 4-16 to
be partly penetrating an isotropic uniform unconfined aquifer of great thickness
relative to the depth of the incision; and assume that the aquifer is uniformly
recharged by rainfall over the area so that the water-table is maintained at a
constant level in each position. The aquifer has a source, therefore, and the channel
is a sink. Water is flowing through the aquifer from the ground-water divide to the
channel, and our purpose is to sketch in the equipotential lines and the flow lines.
At equal elevations above the channel, on each side, an equipotential surface
intersects the water table (Fig. 4-16). Since the water in the channel is at the lowest
potential in the section, an equipotential surface close to the channel must pass
beneath it, and flow will be radial to the channel. At the divide, the direction offlow
changes from flow to the channel to flow to the adjacent channel: here the equipot-
ential surfaces are horizontal, with flow vertically downwards. In the region of the
divide, therefore, the equipotential surfaces are concave upwards.
We therefore infer that flow takes place in all directions from vertically downward
at the water-table divide to vertically upward under the channel, and, following
King (1899, p. 99, fig. 14) and Hubbert (1940, pp. 166-170) sketch the flow net as in
Figure 4-17. On account of the rainfall recharge, assumed uniform over the area, the
flow lines near the water table are steeper than the water-table slope.

Fig. 4-16. Schematic section showing a stream penetrating an unconfined aquifer.

 91

Fig. 4-17. Schematic flow-net in section for Figure 4-16.

A difficulty with this presentation is that one wonders how the equipotential
surface lines between <Pi and <P2 should be drawn, because these have curvature in
opposite senses, with <Pi never passing under the divide and <P2 never passing under
the channel in this line of section. This difficulty can be understood to some extent
by considering that the mean potential lies somewhere between <Pi and <P 2 , and the
divide disturbs this in one direction, the sink in the other.
In plan (Fig. 4-18) we can also sketch the flow-net and see that equipotential
surface pass under both divide and channel, and that in general, the flow lines follow
the topographical slopes. This was clearly understood by Latham (1878), who wrote
' ... water standing at such a declivity is clear evidence of movement' and 'The
greatest elevation of the subterranean water is usually found under the highest
lands, and the least elevation under the lands having the lowest level. The flow of
water laterally is from the hills to the valleys, and longitudinally down the valley-
lines; therefore, as a general rule, the flow of subterranean water conforms to the
surface-falls of the country' (Latham, 1878, p. 207).
In nature, of course, recharge is not uniform, nor are aquifers usually of very great
thickness, nor are they isotropic. Nevertheless, the corollary of our first funda-
mental proposition (p. 5) is reinforced: when the free surface of a liquid is not
horizontal, that liquid is in motion. And we infer motion throughout the aquifer, not
just near the water table.
92

Fig. 4-18. Schematic flow-net in plan for Figure 4-16.

SELECTED BIBLIOGRAPHY

Davis, S.N., and DeWiest, R.J.M., 1966. Hydrogeology. John Wiley & Sons, New York, London, and
Sydney, 463 pp.
De Wiest, R.J.M., 1965. Geohydrology. John Wiley & Sons, New York, London, and Sydney, 366 pp.
Heath, R.C., and Trainer, F.W., 1968. Introduction to ground-water hydrology. John Wiley & Sons, New
York, London, and Sydney, 284 pp.
Horner, D.R., 1951. Pressure build-up in wells. Proc. 3rd World Petroleum Congress (The Hague, 1951),
Section 2: 503-521.
Hubbert, M.K., 1940. The theory of ground-water motion. J. Geology, 48 (8): 785-944.
King, F .H., 1899. Principles and conditions of the movements of ground water. Annual Report U.S. Geol.
Surv. (1897-98), 19 (2): 59-294.
Latham, B., 1878. Indications of the movement of subterranean water in the Chalk Formation. Report
British Ass. Advancement of Science (47th Meeting, Plymouth, 1877): 207-209.
Meinzer, O.E., 1923. Outline of ground-water hydrology with definitions. U.S. Geol. Surv. Water-Supply
Paper 494 (71 pp.). (Read definitions critically.)
Norton, W.H., 1897. Artesian wells ofIowa. Iowa Geol. Survey, 6: 113-428.
Thiem, G., 1906. Hydrologische Methoden. (Inaugural dissertation, Stuttgart). Gebhardt, Leipzig, 56pp.
Todd, D.K., 1959. Ground water hydrology. John Wiley & Sons, New York and London, 336 pp.
5. AQUIFERS: SPRINGS, RIVERS, AND MAN-MADE
DRAINAGE

Let us remind ourselves at the outset that rivers are not merely channels that carry
rain-water runoff to the sea. Perennial rivers are fed by ground water, streams, and
occasionally by rain-water runoff during and after storms: perennial streams are fed
by ground water, smaller streams, and occasional runoff: the source is a spring or a
zone of seepage, and is the highest intersection of the ground surface with the water-
table. Common causes of springs are illustrated in Figure 5-1. When there is a
depression in the ground that penetrates the water-table, there is a lake: but a lake
may also be fed by a stream, and it may overflow as a stream. In these matters we are
nearly always concerned with unconfined aquifers, which are recharged by rainfall
that percolates downwards through the soil.
The flow of intermittent rivers may be due largely to runoff from storms (as desert
wadis) or to the seasonal raising of the water table due to the wet season (as in many

...... . ....... .
......
........... .
........... . .
•••••••• o •

.............
..............
................
.................
....................
....................
......................
:::::::::::::::::::::::::::::::::::::::::::: ~/~' .
•.•••.•...•••••.•...•••••••••..•.•••.....•.• . :;i ....•

{:.::::::{{{::::rrr{ ~q}:/::/::::::::::-:.
Fig. 5-1. Common causes of springs, in section.
94

Australian rivers). Also, reaches of a river may be intermittent where it passes over
the outcrop of a porous and permeable formation in which the water table may from
time to time lie well below the bed of the river (thick limestones are commonly the
cause of this).
So the activity of springs and the flow of rivers and streams is related to rainfall
and the resulting variations in the level of the water table.
These observations are not new, of course. The ancients thought that springs and
rivers were fed by subterranean reservoirs (it would be arrogant of us to suppose
that they all misunderstood the nature of these reservoirs), but the first scientific
observations were carried out by Pierre Perrault in the upper Seine river basin in
France between 1668 and 1670, and published in his De l'origine des fontaines in
1674. He measured the rainfall and estimated the river flow, and found that the
quantity of water falling as rain was about six times the quantity carried by the river.
These measurements dispelled any belief that rainfall was inadequate to account for
the flow of rivers (see Perrault, 1967, for reference to an English translation of
Perrault, 1674; but Perrault cannot be credited with understanding springs because
throughout his book he insists that rain-water cannot penetrate deeply into the soil:
he thought the waters in wells came from rivers - that if there were no rivers, there
would be no water for the wells).
A little later, Edme Mariotte verified Perrault's results in Paris (his work being
published posthumously in 1686, two years after he died). He estimated the annual
rainfall in the Seine catchment area above Paris to be about 7.14 x 1011 cubic pies (a
little larger than an English foot). The river below Pont-Royal was 400 pies wide,
and five deep, flowing at an average speed of about 150 pies/minute (250 when in
flood); 'but,' he wrote, 'because the water at the bottom does not move as fast as that
in the middle, nor the middle as fast as that at the surface ... one can take as the
mean speed 100 pies in a minute'. Rainfall exceeded the river flow by a factor of
nearly seven (Mariotte, 1717, pp. 338-339).
Mariotte also observed that the infiltration of water into a cellar at the Paris
Observatory varied with the rainfall. He wrote (op. cit., p. 336) 'The summer of 1681
was very dry in France, so much so that most wells and springs in many areas dried
up; and although the end of October and the beginning of November were rather
cold, the waters continued to diminish, which they would not have done if water had
been formed by vapours rising from below ground and condensing from the cold at
the surface of the Earth. There is a hollow in the cellars of the Observatory in which
there had been water continuously from 1668 to 1681: but the drought of 1681 dried
it up entirely, and there was no longer a single drop in it in February 1682 although
there had been much rain over several days at the beginning of this month; and in
spite of the following summer having been very rainy, the water did not return in the
month of September, nor even during the two following years'.
About the same time, the British Astronomer Edmond Halley determined the rate
of evaporation from a pan of water salted to the salinity of the sea - and extrapolated
 95

his results to the Mediterranean Sea. The Mediterranean is, as its name suggests, a
sea that is almost enclosed by land. Into it flow several large rivers: from the north,
the Ebro, Rhone, and Po flow into the Mediterranean itself, and the Danube,
Dniester, Dnieper, and Don rivers flow into the Black Sea, the only outlet from
which is through the Bosphorus into the Mediterranean. From the south flows the
Nile. Nor is the Mediterranean free from rain. The only connexion with an ocean is
to the Atlantic through the Straits of Gibraltar, where the surface current also flows
into the Mediterranean. Halley (1687) extrapolated his experimental results, made
some gross approximations to the volumes of water contributed by the rivers, and
concluded that evaporation was entirely adequate to remove the waters contri-
buted by rivers and rainfall. He estimated that the contribution of the rivers was
only one third of the amount removed by evaporation. This proportion is the same
as that arrived at in modern times (see G [orgy] and S [alah], 1974, p. 856, Table 1).
A few years later, Halley (1691) gave a brief but accurate 'account of the circulation
of the watry Vapours of the Sea, and of the cause of Springs'.
Thus the essential ingredients of what is now known as the hydrological cycle (Fig.
5-2) were established more than two centuries ago. We shall concentrate on that
part of the cycle that lies in and on the land, and the inter-relationships between
aquifers, rivers, and rainfall, and some cases of human interference in these that
serve to illustrate the inter-relationships.
It has no doubt been observed ever since wells were constructed that the water-
level in wells fluctuates; but precise observations had to await the development of
instruments capable of measuring these fluctuations with time. Such instruments
enabled King (1892) and Veatch (1906) to publish accounts of detailed observations
concerning the fluctuations of water-levels in wells - fluctuations mainly over
periods of a few days, measured with great precision. These are still of interest
(indeed, we mentioned King's observations on changes of water-level due to loading
by a railway train at the end of Chapter 1).
Veatch found that production from an artesian well between tide levels on the
coast at Huntingdon, Long Island, New York, fluctuated with the tides with a delay
of about 2 minutes at low water and 8 minutes at high water (Veatch, 1906, pp. 10-
13). The weight of the sea water clearly loaded the aquifer. This conclusion is not
contrary to the conclusion reached in Chapter 1 (p. 17), that effective stress is
independent of the depth of water over a sand. The significant parameters here are
permeability and time: the tide rises and falls too quickly for equilibrium to be
reached, and so some of the load is borne by the confined aquifer.
One of Veatch's more interesting observations (op. cit'. p. 24) was that the water-
level in a 4.3 m (14 ft) well rose rapidly when it rained, but that this was not due to
infiltration (it rose before the adjacent brook), nor to barometric or temperature
changes. A similar, but reduced effect was observed in a 21.9 m (72 ft) well. He
inferred that the rise was due to the weight of the rain-water on the ground. (If this
seems unlikely, note that 1 cm of water on a hectare is 100 m 3 , which weighs about
 \0
0\
)::---.'~
. ~... '-D .' . ." ._~/~/l?.r:~- ") !....,. / c' ,
/. -, :-:'---:-.l:~\\::--I/I/
,i. .' J ., ~?.../. y~,
Clouds a~~ water ;~~~ ~l~.:~:.. );:T '-, /~; -- .,..."..,~.'., ,. ~-.'\.~.,,:: .
k.: : ,. J~
.r ). - /. - '"
"';;:~::~ ....
.
t .:;,-./,...-::::-- ,. . ~,:;~*~~~~J~:\::~, ~.'-
.~)

Water table

Explanation
Precipitation
Permeable
Relatively impermeab
le
m;'im Impermeable
~ Springs or seeps
1~ DIrection of movement
of water or vapor
Fig, 5-2. (Courtesy of the
Texas De par tme nt of Wa ter Resour
ces.)
 97

100,000 kg - one inch on an acre weighs about 100 tons - and recall the effect of the
freight train on the water-level in a well).
As an example of the early human influence on the relationship between
an aquifer and a river, we may cite the well-known case of the river Thames
and London's ground-water supply (Whitaker and Tresh, 1916; Buchan, 1938).
London's main ground-water source has been the Chalk (Cretaceous) aquifer that
underlies it. This is a synclinal artesian aquifer fed from intake areas over the
outcrop north and south of the city, confined by relatively impermeable Tertiary
beds above, and the Gault Clay (Cretaceous) below (Fig. 5-3). As London's demand
for water grew with the city, so the pressure head in the wells declined (Buchan,
1938, pp. 28, 32) and the aquifer's potentiometric surface was lowered: the aquifer
was produced for years beyond its safe yield. In the Thames estuary on the down-
stream side of London, the Chalk crops out on the river bed over a distance of some
kilometres. In earlier times, this area acted as a leak from the aquifer, with fresh
water entering the river. Springs could be seen at low tide. With the decline in head,
the flow here was reversed, and the aquifer was contaminated with saline water (at
best!). This consequence was predicted by the Local Government Board reporting
to Parliament in 1899 (See Whitaker and Tresh, 1916, pp. 53-54)*.
The normal interrelationship between aquifer and river flow was nicely studied by
Ineson and Downing (1965) in two English rivers, the Itchen and the Stour. Figure
5-4 shows the Itchen hydrograph they obtained. Storm-water runoff accounts for
the peaks: but the general pattern of flow corresponds to the seasonal water-table

NW SE

LON DON
Chlltern River North
Hills Thame. Downs
London Clay

Gault Clay

Fig. 5-3. Simplified geological cross-section through the Chalk aquifer beneath London, England.

* In the Arabian Gulf (I was told some years ago) the pearl fishermen were able to stay at sea for
protracted periods because they collected fresh water from submarine springs. They would dive down
with a jar, or skin, invert it over the spring, then close it. The growth of Bahrain and the concomitant
increase in demand for water depleted these aquifers so much that the sea-water entered where fresh
water had emerged.
98

8 700

-
>-
(II

I
II)
7 600 c;;-"E
'"E
6 Q
500 Z
w c(
C) C/)
a: 5 ;:,
c(
l:
400 0
l:
(J I-
C/) 4
Q
50
3
,.
..................
.",
m
A.O.D.
",
/

2 .'"
./ 45
/
"

".
I .............. 40
40 ,/' ...... -._.-._._._._._.-
.-.-._._...",/
35L-~__~~__~~__~~__~~__~__~~
35
1959 0 N 0 1960 J F M A M J J A 5
Fig. 5-4. Hydrograph ofthe river Itchen, England, and (below) water-levels in a near-by well (after Ineson
and Downing, 1965, fig. 1).

fluctuations, as shown by the levels in a nearby water well. They estimated that
between 75% and 85% of the flow in these rivers was due to ground-water flow; and
we may take these figures as representative of rivers in temperature climates.
We shall now look at some further examples that illustrate the variations on the
theme of water flow in nature, with special reference to the aquifer.

WATER FROM THE DANUBE TO THE RHINE

There can be few more impressive interactions between rivers and ground water
than the transfer of water from the Danube to the Rhine through the MaIm (Upper
Jurassic) limestone aquifer (Fig. 5-5). The loss of Danube water into the 'briihl'
between Immendingen and M6hringen has presumably been known for some
centuries: in 1719 Breuninger published an account of the head-waters of the
Danube with the thesis that the water that entered the sink-holes in the Danube
between Immendingen and M6hringen emerged as the spring at Aach, the largest in
Germany, and so reached the Rhine.
The loss of water from the Danube was, of course, important to the towns and
 99

I main paths of tracers

------ dry volleys, streams
+ sinks, sink- holes ,,
.;
, .... ,----'

Fig. 5-5. Subsurface flow directions in area between the Danube and the spring at Aach (after Kiiss and
Hotzl, 1973).

villages below the sink, particularly when at unusually low flow rates in the head-
waters all the water disappeared underground (as happened in 1874 for the first
time, it is said). This led in 1877 to what was perhaps the first use of tracers in
ground-water studies*, when an oil tracer was introduced into the water at the sink
and recovered at the spring above Aach (see Kass, 1969, 1972; Kass and Hotzl,
1973). The oil was introduced in the afternoon of Saturday, 22 September 1877, and
* An earlier inadvertent use of tracers occurred in the Teutoburger Wald, central Germany: two ducks
vanished down a sink near Neuenbecken, to emerge after some days about 5 km away at the source of the
Lippe!
 100

from 6 am on the Monday morning, 24 September, the spring water at Aach had a
distinct smell of creosote, which lasted about 6 hours (Knop, 1878, reported in Kiiss,
1969 - the year of the experiment is wrongly recorded here as 1878). At 4 pm on
Tuesday, 9 October 1877, an organic dye sodium fluorescein (uranin) was intro-
duced into the water at the sink, and recovered at the spring above Aach at dawn on
12 October.
The number of days per year during which the Danube ran dry between Immen-
dingen and M6hringen, all the water being lost to the sinks, had been increasing
during the period 1885 to 1929 at an average rate of 2.55 days/year (Wolf, 1931,
reported in Kiiss, 1969). It seemed entirely reasonable to attribute this to the
enlargement of the joints and fissures by solution of the limestone. However, Kiiss
(1969, p. 243, Fig. 5-6) brought unpublished data of Pantle up-to-date, and the
seven-year average of days of total loss per year shows a peak about 1950 of a little
over 200 days/year, with a decline after that. This decline suggests a cause outside
the aquifer, perhaps in the long-term weather changes. It also suggests that 1874
is unlikely to have been the first year in which the Danube ran dry between
Immendingen and M6hringen.

300

0-
ra
cu
200
--=--
~

1J ~
M

=
ra
J
j~ 1
M
M
a
~
11 or

-
ra
a
to-
100
,
1[
- 1900 -----. - --
- - 1950 ----
Fig. 5-6. Number of days the Danube ran dry at Immendingen each year since 1884 (figures by courtesy of
DrW. Kiiss).
 101

In recent years, the flow of water to Aach has been amply demonstrated by a
dozen tracers, not only from the Immendingen-M6hringen area, but also from
further downstream at Tuttlingen (Duttlingen of Breuninger, 1719) and Fridingen.
We have no means of knowing the precise paths of the water; but the spring at Aach
acts as a major sink to the aquifer, with water flowing to it over an arc of about 90° .
The volumetric rate offlow to the Aach spring, estimated to average 6 m 3 s -1, varies
according to the conditions - as does the velocity. A velocity of 0.035 m S-l was
recorded in October 1971 (106 hours to traverse 13.3 km straight-line distance);
and in August 1969 the recorded velocity was 0.070 m s -1. The 1877 experi-
ments indicate about 0.09 m s -1 in September, and about 0.06 m s -1 in October.
Immendingen is close to 650 m above sea-level: the spring at Aach, 467 m above sea-
level. The hydraulic gradient is therefore about 0.014. It is hard to understand how
these ranges of velocities occur when the constraints seem to be so limited (Darcy's
law does not apply to flow through fissures, but the laws for pipe and channel flow
are, as we have seen, very similar with similar constraints).
The MaIm aquifer extends over a wide area of the eastern Molasse basin (at least)
and is of some interest. Along the southern outcrop ofthe MaIm (Fig. 5-7) the water-
table elevation is given approximately by the level of the Danube: at the south-
western end, the water-table is not higher than 650 m at Immendingen or lower than
467 m at the Aach spring. To the south-east of Aach, the water-table elevation
appears to be limited by the Bodensee, 395 m above sea-level, although the MaIm
does not crop out around it. Drilling for petroleum has provided other data points
by virtue of pressure and depth measurements. Figure 5-7 shows the potentiometric-
surface map of the MaIm aquifer in the Molasse basin drawn by Lemcke (1976, p.
11, fig. 1). The most striking feature of this is the apparent sink in the aquifer east of
Munich (Munchen). If the aquifer is in physical continuity over this area, we infer
that water is not only flowing from the Danube to the Rhine over a wider area than
from near Immendingen to Aach, but also flowing into the deeper part of the basin.
The point of interest here is that the elevations of the potentiometric surface
indicated by the borehole data are considerably less than the ground-surface eleva-
tions (around 600 to 700 metres above sea-level) and there is no obvious surface
outflow from this sink. Other aquifers in the Molasse basin also have pore-water
pressures well below the normal hydrostatic pressure.
If this were the only case of its sort, it might seem that there must be an error in the
data or its interpretation. Figure 5-8 shows the potentiometric-surface map of a
Mesaverde sandstone in central U.S.A., and Figure 5-9 that of the Viking Sandstone
in Canada (Hill et aI., 1961). Perhaps they are right in suggesting osmosis as the
cause.
If liquids of different salinities are separated by a semipermeable membrane, the
less saline solution moves through the membrane into the more saline solution; and
if equalization of the salinity cannot be attained, a pressure difference results. This
process is called osmosis. If the sub-normal pressures are one side of this coin, the
other has yet to be found.
 .....
o
tv
seA L E

o 10 20 30KM.
. . .
,~'., South East edlle of the Maim outcrop

Doto point

II

\)
o
5
-

) r-~-"
) \..~? f' s
J-....r~-.j -
R~V~1
to. ·"JL./ \.../"l.t''s" (,

Fig. 5-7. Map of potentiometric surface of the Maim aquifer in the Molasse basin of southern Germany (after Lemcke, 1976, fig. 1).
 103

I~
o
::J:I~ 732!1

~Ia::0
I-
:::> ...J
o
U

I
o
o
~
z x
o UJ
N:!:
a::
~ ~
I~
I

"",~o
684%~&~ . _• v .'~_
I
I

I
seA L [

10 20 30 40 !l0 MILES
I I II I I
o ~O KILOMETRES

Fig. 5-8. Map of potentiometric surface ofa Mesaverde sandstone, central U.S,A. (after Hill et aI.,
1961, fig, 3),
104

!l00

150~

o I!lP Iyo MILES

I
o 50 KILOMETRES

Fig. 5-9. Map of potentiometric surface ofthe Viking Sandstone in Canada (after Hill et a!., 1961, fig. 2).
 105

OCEANIC ISLANDS AND COASTAL AQUIFERS

The behaviour of coastal aquifers is important to many countries because of

population concentrations and their demands for water for industry, agriculture,
and domestic use. Perhaps no country is so conscious of the sea as The Netherlands,
nor as conscious of the proper use of fresh water. Levels are important to the Dutch
because 'downhill' does not lead very far. It is hardly surprising, therefore, that it
was in Holland that it was first noticed that water wells encountered the water-table
not at sea-level but some distance above it. In a long note, Drabbe and Badon
Ghijben (1889)* published details of water-levels near Amsterdam, noted that the
water-table was commonly above sea-level, and proposed the explanation that the
denser sea water supports a longer column of the less dense fresh water. They took
the specific gravity of Zuiderzee water (then open to the sea) to be 1.0238 and
concluded that if a is the elevation of the water-table above sea-level, the depth of
'balance' between the two waters is given by a/0.0238 = 42a (Drabbe and Badon
Ghijben, 1889, p. 21).
Some years later, Herzberg (1901), working on the North Sea coast of Germany,
found the same features and drew the same conclusions.
Briefly stated, their conclusions were that the ratio of the elevation of the water-
table above sea-level to the depth of the fresh/salt water interface below sea-level is
equal to the ratio of the difference between the densities of the sea water and fresh
water to the density of the fresh water, i.e.,

h/z = (Psw - Pfw)/Pfw ~ 1/40. (5.1)

This has become known as the Ghijben-Herzberg relationship or 'law' (one wonders
why Drabbe was left out, and Herzberg put in).
The Ghijben-Herzberg relationship implies static equilibrium between the lens
of fresh water and the sea water: that the mass or weight of fresh water in a column
h + z is equal to the mass or weight of sea water displaced, or

(p fw - Pair)gh + (p fw - Psw) gz = O. (5.2)

Drabbe and Badon Ghijben probably understood that this is unstable, but Versluys
(1919) pointed it out clearly. He went further and argued that the consequences of

* This reference is almost invariably given as Badon Ghijben, 1889, or Ghijben, 1889. W. Badon
Ghijben was in fact the junior author (Badon is part of the surname). There are several matters of interest
in this note: for example, they clearly understood the significance of the sloping water-table because on
p. 16 they wrote 'Neemt men in aanmerking, dat de grondwaterstand in de duinen in het algemeen de
golvende zandoppervlakte op eene geringe diepte voIgt en hier dus zeker verscheidene meters boven den
waterspiegel der Zuiderzee is verheven, dan schijnt het zeker, dat er in den ondergrond een voortdurende
stroom moet zijn van de duinen naar de Zuiderzee' - 'When one considers that the ground-water level in
the dunes follows in general the wavy surface of the sand at shallow depth and thus here is certainly
elevated several metres above the level of the Zuiderzee, then it appears certain that there must be a
continuous underground flow from the dunes to the Zuiderzee' (their italics).
106

this instability were that the relationship was much more complicated, and that the
fresh water flowed into the sea over a zone, as in Figure 5-10. He clearly understood
that the fresh water is not at uniform potential * in the lens, so it is in a state of
constant motion; and if no rain rell, the lens would dissipate by flow to the sea near
sea-level.

.... - .. _-- .... .., .. -----

Fig. 5-10. Versluys' diagram showing flow to the sea in a fresh-water lens under coastal dunes (Versluys,
1919, fig. 4).

The Ghijben-Herzberg lens is usually shown diagrammatically as in Figure 5-11,

but there is great distortion in such diagrams due to the grossly exaggerated vertical
scale. This falsely suggests that the water flow is mainly vertical. Figure 5-12 shows a
half-lens on a more natural scale.
The shape of the water table indicates that the water in the lens is flowing away
from the centre line, in general (as Drabbe and Badon Ghijben appreciated), and
considerations of symmetry require the flow at the centre line to be vertically
downwards initially (as Versluys appreciated). The matter of equipotential surfaces
in 2-fluid systems was treated by Hubbert (1940, pp. 868-870,924-926), the essence
of which is this: if the sea water is static and the fresh water flowing, the sea water is
at constant potential so the fresh-water/salt-water interface is an equipotential
surface of the sea water. On the fresh water side of the interface, the equipotential
surfaces of the fresh water are normal to the interface because it is a boundary to the
flow. If L1h is the vertical interval of equipotential contours on the water-table,
L1hp fw/(Psw - P fw) is the corresponding vertical interval of equipotential contours
on the interface. The equipotential surfaces between the corresponding contours are
curved, meeting the interface at right angles; so the interface is displaced outwards
from the centre line from the position predicted by the Ghijben-Herzberg 'law', and
is therefore deeper than that prediction.
* Versluys (1917, p. 24, eq. 6) had already explicitly defined 'hydrostatic potential', or 'potential' for
short, as the sum of the elevation and pressure heads. This differs from Hubbert's fluid potential only in
the factor g (Hubbert, 1940, p. 802).
 107

I
,,,--_ .... /
I
I
I
I
I
I
I
I

,,
I
I

I
I

Fig. 5-11. Schematic section through Ghijben-Herzberg lens in an island, vertical scale grossly ex-
aggerated. Sketched in is the influence of a producing water-well.

!COAST LINE

Fig. 5-12. Half-lens at more natural scales, with equipotential lines.

108

If the slope of the water-table between two equipotential contours is a below the
horizontal in the direction of flow, and the material is isotropically permeable, we
have from Darcy's law

. L1h q
sIn a c= - =-
L1[ K'

where L1[ is the corresponding length along the watertable. At the interface, the flow
is inclined upwards: the flow is converging. The same quantity of water crosses each
equipotential surface in the fresh water, so q (= Q/A) increases towards the coast,
and the slope of the watertable increases towards the coast. By a similar argument,
the slope of the interface above the horizontal also increases towards the coast. The
surfaces bounding the fresh water are curved, concave to sea-level (as Versluys
appreciated) .
In practice, it is so undesirable to drill water wells to produce from near the edge
of the lens that its precise shape is not very important, and the Ghijben-Herzberg
'law' can be taken as a limit. In any case, there are many disturbing influences -
irregular rainfall, tides, anisotropic porous solids, and perhaps diffusion at the
interface with the salt water*. The effect of producing water from a well in a
Ghijben-Herzberg lens is to superimpose another field of flow. The field imposed by
a producing well far from the coast is much stronger than the natural field, and it will
be appreciated that a small apparent depletion that lowers the water-table slightly
has a far more serious hidden effect at the bottom of the lens (sketched on Fig. 5-11).
The Pacific islands are particularly dependent on their Ghijben-Herzberg lenses
because they, like all oceanic islands, are entirely dependent on rainfall for water.
The lens is their natural storage reservoir for times of drought. The Second World
War imposed enormous strains on many of these resources.
The island of Oahu in Hawaii, on which Honolulu and Pearl Harbour are
situated, is an interesting example of a Ghijben-Herzberg lens because erosion has
led to the accumulation of a relatively impermeable layer of sediment over the zone
of natural seepage at and below sea-level (Fig. 5-13). This caprock, by distorting the
Ghijben-Herzberg lens, gives to it a local artesian quality. According to Ohrt (1947)
and to Wentworth (1952) the (total) head above mean sea-level in a well drilled
through the caprock in the central Honolulu area was about 12.8 m (42 ft) in 1880,
implying by the Ghijben-Herzberg relationship a depth below sea-level to the
interface of 486 m (1600 ft). Wells drilled here to 366 m (1200 ft) below sea-level
produced fresh water. The uncontrolled drilling of artesian wells had led to a serious
decline in the total head by 1910; and by 1926 it was down to 7.6 m (25 ft) when a
conservation programme, aided by rainfall, slightly reversed the decline. By 1947,

* Versluys (1919) argued persuasively that the brackish zone near the interface is the result not so
much of diffusion but rather of incomplete washing by fresh water, because the salinity does not increase
with depth throughout the lens, but only in a transition zone near the interface.
 109

POTENTIOMETRIC SURFACE

--L - -- --
!
Spring
-- Sea Level

Fig. 5-13. Schematic section showing artesian zone on a low-lying coastal area covered by relatively
impermeable sediments.

nearly 500 artesian wells had been. drilled through the caprock and, in spite of the
strains of war, the total head was only slightly below the 1926 level.
The loss of fresh water in storage in the Ghijben-Herzberg lens was therefore
enormous, estimated at 40% of the natural quantity, and it is generally considered
that such losses are irrecoverable in the time-scale of our lives. The expansion of the
lens probably leaves a thicker transition zone of brackish water. In the central
Honolulu area, salt water has encroached to 305 m (1000 ft), with local indications
of encroachment to 245 m (800 ft).
Ghijben-Herzberg lenses are not the only source offresh water in volcanic oceanic
islands. Perched aquifers may be contained between dykes, and water may flow
down old lava channels, contained by relatively impermeable tuffs. On the Atlantic
island of Tenerife (Canary Islands, Spain) the main water production comes from
galerias, tunnels about It x 1 m dug into the hillside with a slight upward slope.
These penetrate progressively older strata, and so may encounter percolating water
or a perched aquifer. These galerias are purely speculative ventures because water is
not known to exist along their paths beforehand. Some are about 4 km long, and
represent years of work and investment. Naturally, the failures tend to be longer
than the productive ones. Around 1960,98% ofTenerife's water supplies came from
galerias: 162 million cubic metres a year from 386 producing galerias with a total
length of 763 km (Canarias. anexo al Plan de Desarrollo Economico y Social
1964-1967: Presidencia del Gobierno, Edici6n del Boletin Oficial del Estado, 1964,
pp. 65, 68, 72). These figures suggest an average length of about 2 km.
110

ARID REGIONS: THE QANAT OF IRAN

In arid regions, as Sandford (1935) pointed out, the rainfall may be slight but it is not
everywhere insignificant. Working in the desert to the west of the Nile, around the
common borders of Libya, Egypt, Sudan and Chad, he measured the elevation of
water wells with an aneroid barometer and the depth to the water table, and so
produced a contour map of the water table (Fig. 5-14). The shape of the water table
indicates a southerly source for the ground water, with flow to the west south of
Tibesti, and to the north-east to the Nile. It is the development of aquifers in arid
regions that is of interest rather than the aquifer itself, and the most important
method of development is due to the Persians.
Throughout the area of the former Persian and Arab Empires, ground water is
developed for agricultural and domestic use by a system of wells joined together by a
tunnel or gallery that brings water from the foothills to the plains (Fig. 5-15). The
system is of great antiquity, for it had been developed by the year 800 BC, and was
introduced into Egypt in 512 BC (while Egypt was a Persian province, during the
reign of Darius ) in the oasis of Kharga - to such effect that the event was recorded by
an inscription (Butler, 1933).
Essentially the same system is found from Morocco to Afghanistan, and in
southern Arabia around the Oman mountains and in the Yemen. Different names
are used in different parts: shat-at-ir in Morocco, foggariur in Algeria, qandt or
kandt in Iran and Baluchistan, kariz in Iraq and Afghanistan, falaj (plural, ajlaj)
around the Oman mountains, and shariz in the Yemen. They are of interest because
they illustrate a practical understanding of ground-water hydrology amongst the
ancients that is still of inestimable value. We shall call them qanat, generally, as they
are called in their country of origin. They do in arid climates by the labour and
ingenuity/of Man what rivers do in more humid climates.
The first stage in the construction of a qanat (Noel, 1944; Beckett, 1953) is the
sinking of exploratory and appraisal wells in the area thought to contain a suitable
aquifer that could supply water to a settled area (or land that could be settled). These
are hand-dug, about 0.75 m in diameter, and may be about 100 m deep. When
satisfied that there is sufficient water, a survey is made to determine the level at
which the qanat and the water will emerge. The traditional method of survey is to
lower a line down the 'mother well' (the one furthest away from the area of intended
use) and to tie knots at the ground surface and the water table. A line is then strung
from the ground surface at the well to the top of a pole sited in the direction in which
the qanat will run. This line is levelled by wetting it and adjusting the level at the pole
until the water drops collect at the middle of the line. The difference in elevation is
then subtracted from the cord that records the depth to the water table, and marked
with another knot or pin. This process is continued until the loss of elevation equals
the original length between the knots (illiteracy is not to be confused with igno-
rance!). The gradient of the qanat is then decided (usually about 1:1500).
 24 24
+
·154
00.....

242 .240"",, .174

2~' "'-,166
o
215' ~ ~
<"
~
,466

TIBESTI

20· zo·
+
.448 o
59~ ~
497
284~ 140' 509'
2TS. 510' '"0 ~
I;)
0
504 ...
m )365 521
• 205 C
~~'6' ~ m/
185') 250) 380'

410

16·
+ + +
eTSO'

20· 24· 28" ......

......
Fig. 5-14. Map of water-table in Nubian Sandstone in desert west of the Nile, elevations in metres (after Sandford, 1935).
112

Fig. 5-15. Schematic section through a qanat.

Shafts are dug to the required level along the planned route of the qanat, every 300
m or so, and the construction of the qanat itself is begun in the dry section, working
from the lower end. Shafts are dug at intervals of about 25 m where the depth is
slight, for ventilation and the extraction of spoil. The interval is increased as the
depth becomes greater. The qanat itself is dug from shaft to shaft: it is about 1.25 m
high and 0.75 m wide, and Noel (1944) reports that these dimensions are considered
to be the largest consistent with safety. Two lamps are used to keep the qanat on
course (as they are in the galerias of Tenerife ). Soft sections are lined with oval rings
of baked mud, straw, and dung. Once the dry section has been completed, construc-
tion into the aquifer can be started.
Production usually declines rapidly at first, but if the aquifer is good, soon
stabilizes. Qanat may be deepened to offset decline in production. Beckett (1953)
and Smith (1953, p. 133) reported that a qanat near Kirman in SE Iran had its yield
stimulated by an earthquake, the source being changed from a seepage to a spring.
Ideally the water flows at t to t m s -1 . At greater rates of flow, erosion of the qanat
may occur, leading ultimately to its collapse.
Small fish swim in all but the saltiest qanat, and are said to have come from the
rocks. They are also found in the aflaj of Buraimi in southern Arabia, and the same
explanation is given there. Smith (1953, p. 81) ascribes their presence to their being
washed in from rivers in flood; but fish have been reported from water wells in many
countries. Lyell (1867, p. 393) reports live fish from artesian water wells from 47.5 m
in Germany and 53.3 m in the Sahara: Norton (1897, pp. 167-168) reports live
crustaceans from an artesian well in Texas, and live fish from wells near Aberdeen,
Dakota, U.S.A. Smith (1953, p. 135) collected one insect larva from the source of a
qanat.
Butler (1933) records that the water supply for the entire city of Tehran came from
 113

36 qanat, ranging in length from 13 to 26 km and passing in places more than 150 m
below the surface. These supplied 0.17 m 3 s -1 in autumn. A qanat near the foot of
Kuh-i-Kurgis was said to be 90 km long, while lengths of 40 to 45 km are not
uncommon. The yield from a qanat may be uniform throughout the year, or show
seasonal variations. Thesiger (1959, p. 290) found during his 1948-49 journey
through southern Arabia that some villages in the southern Oman mountains at the
northern end of the Wahiba sands had been abandoned because their aflaj (as the
qanat are here called) had run dry. His description indicates that these aflaj did not
reach the surface: they were probably like those in Iran illustrated by Graves (1975,
p. 43). Further to the north-west from the villages visited by Thesiger, on the west
flank of the Oman mountains, the villages of the Buraimi oasis derive most of their
water from aflaj that pass into the foothills of the Oman mountains. These were
collectively producing about 0.4 m 3 s -1 in the mid-1960s.
The question arises, how does the water reach the land surface in a qanat? The
short answer is that the loss of head incurred in channel flow is very much less than
that in the aquifer, so that if the aquifer somewhere has its water table above the level
of the land to be irrigated, it can be exploited. That part of the completed qanat that
is in the aquifer is a sink to that aquifer, and water flows into the qanat down the
potential gradient that is set up by its construction (Fig. 5-16). The large surface area
exposed to the aquifer accounts for its productivity, even when the permeability is
low. The reader will recognise that in a transverse section through the qanat in its
producing region, there will be a drawdown analogous to that of a producing water
well; but the qanat is a sub-horizontal line sink. The relatively high initial produc-

Fig. 5-16. Schematic cross-section through a qanat where it penetrates the aquifer, showing equipotential
lines from which radial flow is inferred.
114

tion, and rapid decline, is due to the production of the volume of water during
drawdown before steady flow has been achieved.
Clearly the flow is not like a fully-penetrating well, but like the flow to a perennial
river (discussed on p. 90). The potential in the qanat at any point is the lowest in the
cross-section through that point, so there is radial flow within the aquifer to the
qanat, normal to the equipotential surfaces that pass under the qanat. The rate of
flow in the qanat increases downstream until the qanat leaves the aquifer.
If the construction of qanats could be automated, they could still serve the arid
regions of the Earth.

THE GREAT ARTESIAN BASIN

The Great Artesian Basin (Figs. 5-17, 5-18, 5-19) lies under nearly two million
square kilometres of eastern Australia, under parts of the States of Queensland,
New South Wales, South Australia and the Northern Territory. It has many aquifer
in five main groups, ranging in age from Triassic to Cretaceous, the sediments older
than Cretaceous being dominantly non-marine. It owes its artesian quality to
Tertiary and Cretaceous marine shale aquicludes, intake areas in the east elevated
about 500 m above sea-level, substantial rainfall on the intake areas, and a top-
ographic slope to the west of the intake areas that is steeper than the hydraulic
gradient. Its importance for the rural development of interior eastern Australia is
inestimable.
The first artesian well was drilled in 1887 near Cunnamulla in Queensland, after
a geological report indicated the possibility of artesian supplies: it found a
flow of 4.2 x 10- 3 m 3 s- 1 at 393 m (Queensland, Artesian Water Investigation
Committee, 1955). By 1914, 1229 wells had been drilled, producing a total of
18.5 m 3 s- 1 . Thereafter the total flow began to decline: by 1950, it was 12 m 3 s- 1 .
The decline of head amounted to over 100 m in places, but the residual head
averaged 15 to 30 m above ground surface. Between 1890 and 1950, the total volume
of water withdrawn was about 3 x 10 10 m 3 (30 km 3 ) from the Queensland portion,
and the investigating committee (op. cit.) estimated that the final 'steady state' flow
would be reached with a production of 11 m 3 s -1 from 520 permanent artesian
bores. Other wells would become subartesian and would have to be pumped.
The average depth to the aquifers in Queensland (if we take the average depth of
bore holes drilled to 1953) was 439 m, with the shallowest being only 2 m and the
deepest, 2136 m deep. Over 1000 km of artesian hole have been drilled; registered
boreholes, artesian and subartesian, total over 2000 km.
As we saw in previous chapters, the total head of an artesian aquifer that is
perfectly sealed is horizontal, at the elevation of the intake area: but any leakage
leads to flow towards the leak, and a corresponding slope to the potentiometric
surface. The Great Artesian Basin had many leaks before humans started making
 115

them. There were springs in the intake area (Whitehouse, 1955) as a result of erosion
reaching the water table, as a result of bedrock reaching the surface, and less-readily
definable springs known as mound springs. The mounds of mound springs are built
from debris and sediment carried to the surface by the water, with a result similar to
mud volcanoes. These mound springs also occur outside the intake area. They are of
particular interest, and we shall return to them shortly. The natural leakage from the
Great Artesian Basin was considerable, and so its potentiometric surface was not
horizontal. Its form has been deduced from a simulation model by Habermehl
(1980) and is reproduced here as Figure 5-17.
At Cunnamulla (elevated about 189 m above sea-level) the original natural loss of
total head was about 137 m in 400 km (a gradient of3.4 x 10- 4 ) towards the south-
west from the intake area, leaving a total head of about 320 m above sea-level,
locally about 130 m above ground level. We do know that water production has
considerably altered the shape of the potentiometric surface. Near Cunnamulla,
where there was early development of this ground water, the 1,050-foot (320-m)
equipotential line retreated about 165 km northwards between 1899 and 1950,
with a reduction of head of122 m (Ogilvie, 1955, p. 46 and fig. 94). And Spring Creek
(90 km south-east of Boulia) originally flowed for more than 130 km to
the Diamantina River, the water emerging from springs at a temperature of
18°C to 38°C with 0.61 kg m- 3 (about 600 ppm) dissolved salts. By 1896 it had
been reduced to about 30 km of flow, and by 1950 it had ceased flowing.
The general form of the potentiometric surface in 1970 is shown in Figure 5-17
(general, because different aquifers have different heads or potentials), and this
indicates that the regional flow patterns have not been changed, the main flow
being westwards, then south-westwards to Lake Eyre (which is 12 m below sea-
level) and beyond. In the north (off the map) flow is to the north under the Gulf of
Carpentaria.
That there has long been natural leakage in the south-western part of the basin is
shown by the presence of numerous mound springs. These mound springs build up a
cone of sediment, debris, and salt until they reach the potentiometric surface, at
which level all kinetic energy is, of course, lost. Whitehouse (1955) observed that
there are extinct mounds in the far south-west, the tops of which are about 30 m
above the level of the active mounds (one must assume that he was of the opinion
that they could not build up a further 30 m).
The temperature of the artesian water varies over the basin, and the geothermal
gradient varies from about 30°C/km to 80°C/km, the highest being apparently
166°C/km near Julia Creek. West of Winton, Queensland, 2.6 x 10- 2 m 3 s- 1
of boiling water was produced from 1372 m, indicating a geothermal gradient of
55-58°C/km if the mean surface temperature is taken as 25-20°C. However, ac-
curate temperature measurements of the aquifers are lacking. The gradient in-
creases from east to west in Queensland.
The quality of the water is also variable (Fig. 5-18). Airey and others (1979), in a
116

o ~
g
CD
CD '"
.....
o

o
co
!

3:
en
z

o
'"~

o
'"'" I-
Z
o
'"'"

Fig. 5-17. Potentiometric maps of the main Lower Cretaceous-Jurassic aquifers for 1880 and 1970,
modified from computer-simulated model results. Datum is mean sea-level, and potentials relate to pure
water at 15°C. (Habermehl, 1980, fig. 15, reproduced by courtesy of the Bureau of Mineral Resources,
Canberra.)
 1360
1970
o o 200km
<I
PotentiometriC contour
- / 2 0 - (m) above MSL

22°

NT
~

SA
30° ~
\\.,
Lake Torrens\ \
\ \
<:"' \
\' ........... "I )1
NSW
'? I,
~ 1 I"
~J • Broken Hill
Lake GOirdner') ~
!.{v

NEWCASTLE AUS 1/905 --..)

 136 0 ~
148 0
pH
Townsville
" -_ _,-_...:2:..:;90 km
:. < (1
() 0 ."
No:: :"'04
b :;;~.
K~C 00
u . . . ·t../1 .: CD 3
,,~ ,
'"1 ::: ...... Mg Mt Iso
,.., (1)
po ,.., 00 _
t'" Co ~<C03 .
o
~~9 % fotoleqUIvalents T%l sum of Ions 79
--.J" per million (epm) In mg/Z (ppm)
- a
~(i • Well locatIOn
po :::.
vo 22°
"
,. , 0a
~'O
~ ~. NT
:0 g.
00::>
Oen
" 0
~.......,
~~
- ,..,
--.J:::
o
" ::>
,.., 0..
.g :;;
(3 ~
0.."
::: ,..,
~ ~
0..0
va
#
'-< en
o !:!..
o "
'" 0
(1) 0..
_,.., -"
~ :::>
0
o-,:;;
;. 5'
,,~

t:Ci';
g,..,'" "~. -
'" ::>
o :;;
-'" 30°
3;::::::
S'
(1)
:po
""'0
Po '0
;;; 5-
(l)~
en po
0,.0
16!
'f6,..," --'~" NSW
:;n en

NEWCASTLE AUS 1/907

 119

GU LF OF

CAR PEN TAR IA

MT .ISA • JUlia Ck .
•

• Win ton

• ROCK HAM PTO N

I
I

I CUN NAM ULL A.

BRISBANE •
V
I
, __ _ ... (
j- -- -- -- -- - -- -- -- -- --
I \_r "'

n.
d portion of the Grea t Artesian Basi
Fig. 5-19. Location map, Queenslan
120

study of a large area of the eastern part of the Great Artesian Basin about Cunna-
mulla, concluded that the variations in the principal Jurassic aquifer reflect varia-
tions in the rate of infiltration of recycled salt throughout the late Quaternary.
We choose two aspects of the Great Artesian Basin to illustrate general principles:
the age of the water, and the initial decline in production.
Habermehl (1980) quotes water velocities of 1 to 5 m per year (q/f), so the age of
the water may be up to 100,000 years for every 100 km from the intake area. Water
can be dated up to 50 years using tritium eH), an isotope of hydrogen with two
neutrons in addition to the proton in its nucleus; and from 600 to about 30,000 years
using carbon 14. Thus the age, or residence time, of the water must be estimated
from hydraulic data for the time being (there is hope that chlorine 36, with a half-life
of 308,000 years, will become a reliable tool for dating old water).
It is sometimes argued (and has been argued informally for the Great Artesian
Basin) that old water indicates that it is a non-renewable resource that should be
treated the same as mining any other non-renewable resource. There is not the
slightest doubt that the water in the Great Artesian Basin is being renewed from the
intake area - the potentiometric maps show that - and it could even be argued that
production will improve the quality of the water with time.
The matter of high initial production and subsequent decline to a stable rate can
be considered to have three components. The potentiometric surface on completion
of the first well is at its natural level, and during the early period of production, this is
drawn down to a level that depends on the hydraulic properties of the aquifer and its
depth below the surface. The aquifer's energy is therefore initially high at the well-
site, but declines as it adjusts to steady flow and the development of a stable
hydraulic gradient to the well.
The other two components of this decline come under the heading of specific
storage. Specific storage is the sum ofthe water stored in the aquifer by virtue ofthe
compressibility of the solid framework of the aquifer, and water stored by virtue
of its own compressibility. When the aquifer pressure is relieved by production,
Terzaghi's relationship comes into play, and the decline in water pressure is com-
pensated by an increase in the effective stress - which in turn compacts the rock a
little, expelling a further quantity of water (S = (J + P ~ constant). The water also
expands a little as a result of this relief of pressure. Production stabilizes when the
aquifer stabilizes under the new conditions.

SELECTED BIBLIOGRAPHY

Airy,P.L., Calf, G.E., Campbell, B.L., Hartley.P.E., Roman, D., and Habermehl, M.A., 1979. Aspects
of the isotope hydrology of the Great Artesian Basin, Australia. Isotope Hydrology 1978, I: 205-219.
(proc. Internat. Symp. on Isotope Hydrology, International Atomic Energy Agency and UNESCO,
Neuherberg, Federal Republic of Germany.I.A.E.A., Vienna.)
Badon Ghijben, W., 1889. See Drabbe, 1., and Badon Ghijben, W., 1889.
 121

Beckett, P., 1953. Qanats around Kirman. J. Royal Central Asian Soc., 40: 47-58.
Breuninger, F.W., 1719. Fons Danubii primus et naturalis odeI' die Urquelle des weltberiihmten Donau-
stroms. Tiibingen, 388 pp.
Buchan, S., 1938. The water supply of the County of London from underground sources. Memoir Geol.
Surv. Great Britain, 260 pp.
Butler, M.A., 1933. Irrigation in Persia by kanats. Civil Engineering, 3 (2): 69-73.
Day, J.W.B., and Rodda, J.C., 1978. The effects of the 1975-76 drought on groundwater and aquifers.
Proc. Royal Society London, A363: 55-68.
Dickey,P .A., and Cox, W.e., 1977. Oil and gas reservoirs with subnormal pressures. Bull. A merican Ass.
Petroleum Geologists, 61 (12): 2134-2142.
Drabbe, J., and Badon Ghijben, W., 1889. Nota in verband met de voorgenomen put boring nabij
Amsterdam. Tijdschrift van het Koninklijke Instituut van Ingenieurs, 1888-1889: 8-22.
Freeze, R.A., and Witherspoon, P.A., 1966. Theoretical analysis of regional groundwater flow.
1. Analytical and numerical solutions to the mathematical model. Water Resources Research, 2 (4):
641-656.
Freeze, R.A., and Witherspoon, P.A., 1967. Theoretical analysis of regional groundwater flow.
2. Effect of water-table configuration and subsurface permeability variation. Water Resources
Research, 3 (2): 623-634.
Freeze, R.A., and Witherspoon, P.A., 1968. Theoretical analysis of regional groundwater flow.
3. Quantitative interpretations. Water Resources Research, 4 (3): 581-590.
Ghijben, W.B., 1889. See Drabbe, J., and Badon Ghijben, W., 1889.
Glover, R.E., 1959. The pattern of fresh-water flow in a coastal aquifer. 1. Geophysical Research, 64 (4):
457-459.
G lorgy], S., and S [alah], M.M., 1974. Mediterranean Sea. In: The New Encyclopaedia Britannica (15th
ed.), Macropaedia, Vol. 11, pp. 854-856. Encyclopaedia Britannica Inc., Chicago, London, etc.
Graves, W., 1975. Iran, desert miracle. National Geographic, 147 (1): 2-47.
Habermehl, M.A., 1980. The Great Artesian Basin, Australia. Australia, Bureau of Mineral Resources J.
Australian Geology and Geophysics, 5: 9-38.
Halley, E., 1687. An estimate of the quantity of Vapour raised out of the Sea by the warmth of the Sun;
derived from an Experiment shown before the Royal Society, at one of their late Meetings. Philos-
ophical Trans. Royal Society London, 16 (189): 366-370.
Halley, E., 1691. An account of the circulation of the watry Vapours of the Sea, and of the cause of
Springs. Philosophical Trans. Royal Society London, 17 (192): 468-473.
Herzberg, B., 1901. Die Wasserversorgung einiger Nordseebader. Journal jur Gasbeleuchtung und ver-
walldte Beleuchtungsarten sowie fur Wasserversorgung, 44: 815-819, 841-844.
Hill, G .A., Colburn, W .A., and Knight, J.W., 1961. Reducing oil-finding costs by use of hydrodynamic
evaluations. In: Economics oJpetroleWl1 explomtion, development, and property emluation. Prentice-
Hall, Englewood Cliffs, N.J., pp. 38-69.
Hitchon, B., 1969a. Fluid flow in the western Canada sedimentary basin.!. Effect of topography. Water
Resources Research, 5 (1): 186-195.
Hitchon, B., 1969b. Fluid flow in the western Canada sedimentary basin. 2. Effect of geology. Water
Resources Research, 5 (2): 460-469.
Hitchon, B., and Hays, J., 1971. Hydrodynamics and hydrocarbon occurrences Surat Basin, Queens-
land, Australia. Water Resources Research, 7 (3): 658-676.
Hubbert, M.K., 1940. The theory of ground-water motion. J. Geology, 48 (8): 785-944.
Ineson, J., and Downing, R.A., 1964. The ground-water component ofriver discharge and its relation-
ship to hydrology. 1. Instil Water Engineers, 18 (7): 519-541.
lneson, J., and Downing, R.A., 1965. Some hydrogeological factors in permeable catchment studies. 1.
Instil Water Engineers, 19 (1): 59-80.
Kass, W., 1969. Schrifttum zur Versickerung der oberen Donau zwischen Immendingen und Fridingen
(Siidwestdeutschlandl. Steirische Beitriige zur Hydrogeologie, 21: 215-246.
Kass, W., 1972. Die Versickerung der Oberen Donau, ihre Erforschung und die Versuche 1969.
Geologisches Jahrbuch, C2: 13-18.
Kass, W., and Hiitzl, H., 1973. Weitere Untersuchungen im Raum Donauversickerung-Aachquelle
(Badcn-Wiirttemberg). Steirische Beitriige zur Hydrogeologie, 25: 103-116.
122

King, F.H., 1892. Observations and experiments on the fluctuations in the level and rate of movement of
ground-water on the Wisconsin Agricultural Experiment Station Farm and at Whitewater,
Wisconsin. U.S. Dept. Agriculture, Weather Bureau, Bulletin no. 5 (7< pp.).
Kraus, L., 1969. Erdol- und Erdgaslagerstiitten im ostbayerischen Molassebecken. Erdoel-Erdgas-
Zeitschrift, 85: 442-454.
Lemcke, K., 1976. Dbertiefe Grundwiisser im siiddeutschen Alpenvorland. Bull. Vereinigung
Schweizerischer Petroleum-Geologen und -Ingenieure, 42 (103): 9-18.
Lemcke, K., and Tunn, W., 1956. Tiefenwasser in der siiddeutschen Molasse und in ihrer verkarsteten
Malmunterlage. Bull Vereinigung Schweizerischer Petroleum-Geologen und -Ingenieure, 23 (64): 35-56.
Lyell, C., 1867. Principles of Geology or the modern changes of the Earth and its inhabitants (10th ed.),
vol.!. John Murray, London, 671 pp.
Mariotte, E., 1717. Traite du mouvement des eaux et des autres corps fluides. In: Oeuvres de Mr.
Mariotte, Pierre van der Aa, Leiden, pp. 321-476.
Noel, E., 1944. Qanats. 1. Royal Central Asian Soc., 31: 191-202.
Norton, W.H., 1897. Artesian wells ofIowa. Iowa Geol. Survey, 6: 113-428.
Ogilvie, C., 1955. The hydrology of the Queensland portion of the Great Australian Basin. In: Queens-
land, Artesian Water Investigation Committee, Artesian water supplies in Queensland. Government
Printer, Brisbane, Appendix H.
Ohrt, F., 1947. Water development and salt water intrusion on Pacific islands. J. A merican Water Works
Ass., 39 (10): 979-988.
Perrault, P., 1674. De l'origine des fontaines. Pierre Ie Petit, Paris, 353 pp. (Not sighted.)
Perrault, P., 1967. On the origin of springs. (English translation by A. LaRocque.) Hafner, New York and
London, 209 pp.
Queensland, Artesian Water Investigation Committee, 1955. Artesian water supplies in Queensland.
Government Printer, Brisbane, 79 pp + 2 appendices.
Sandford, K.S., 1935. Sources of water in the north-western Sudan. Geographical Journal, 85 (5):
412-431.
Schrockenfuchs, G., 1975. Hydrogeologie, Geochemie und Hydrodynamik der Formationswiisser des
Raumes Matzen-Schonkirchen Tief. Erdoel-Erdgas-Zeitschrift, 91: 299-321.
Smith, A., 1953. Blind white fish in Persia. Allen & Unwin, London, 231 pp.
Spain, Presidencia del Gobierno, 1964. Canarias: anexo al Plan de Desarrollo Econ6mico y Social
1964-1967. Edici6n del Boletin Oficial del Estado, Madrid, 518 pp.
Thesiger, W., 1959. Arabian sands. Longmans, London, 326 pp.
T6th, J., 1962. A theory of groundwater motion in small drainage basins in central Alberta, Canada. J.
Geophysical Research, 67 (11): 4375-4387.
T6th, J., 1963. A theoretical analysis of ground-water flow in small drainage basins. 1. Geophysical
Research, 68 (16): 4795-4812.
T6th, J., 1979. Patterns of dynamic pressure increment of formation-fluid flow in large drainage basins,
exemplified by the Red Earth region, Alberta, Canada. Bull. Canadian Petroleum Geology, 27 (1):
63-86.
Van Everdingen, RD., 1968. Studies of formation waters in western Canada: geochemistry and
hydrodynamics. Canadian J. Earth Science, 5: 523-543.
Veatch, A.C., 1906. Fluctuation of the water level in wells, with special reference to Long Island, New
York. U.S. Geol. Surv. Water-Supply and Irrigation Paper 155 (83 pp.).
Versluys, J., 1917. De beweging van het grondwater. Water, 1: 23-25, 44-46, 74-76, 95.
Versluys, J., 1919. De duinwater-theorie. Water, 3 (5): 47-51.
Versluys, J., 1920. The theory of dune water. Water Services, 22: 182-184. (Incomplete translation into
English of Versluys, 1919, without figures.)
Walpole, G.F., 1932. An ancient subterranean aqueduct west of Matruh. Survey of Egypt Paper (Egypt,
Maslahat al-Misahah), 42 (40 pp.).
Wentworth, C.K., 1952. The process and progress of salt water encroachment. Union Geodisique et
Geophysique Internationale, Association Internationale d' Hydrologie Scientifique, Assemblee Generale
de Bruxelles 1951, 2: 238-248.
Whitaker, W., and Tresh, J.C., 1916. The water supply of Essex from underground sources. Memoir
Geol. Surv. Great Britain, 510 pp.
 123

Whitehouse, F.W., 1955. The geology of the Queensland portion of the Great Artesian Basin. In:
Queensland, Artesian Water Investigation Committee, Artesian water supplies in Queensland.
Government Printer, Brisbane, Appendix G.
Wulff, H.E., 1966. The traditional crafts ofPersia: their development, technology, and influence on eastern
and western civilizations. M.LT. Press, Cambridge, Massachusetts, 404 pp.
Wulff, H.E., 1968. The qanats ofIran. Scientific American, 218 (4): 94-105.
 6. MOVEMENTOFPOREWATER,
AND ABNORMALL Y HIGH PORE PRESSURES

The distinction between ground water and pore water may not be very logical, but it
has the merit of distinguishing the readily-exploitable fresher pore water near the
surface from the brackish to salty water in the pore spaces of most sedimentary
rocks at greater depth. As always, such distinctions recognize tendencies only, for
there are areas (such as the Niger delta; see Dailly, 1976, p. 96, fig. 3) where fresh
water is found in sands to depths of two or three kilometres. The distinction is also
seen as a distinction between meteoric water and what is called connate, the former
being derived 'recently' from rainfall, the latter being defined as water that was
trapped in the pore spaces when the sediment accumulated.
The difficulty with the definition of connate water is that we cannot accept that
significant amounts of water were trapped in the pores when the sediment accumu-
lated. If one considers the bulk water in the pores of a rock unit, then it is an
acceptable interpretation of the pore water of muds and mudstones (although the
chemistry of the water may well have changed) but it is not acceptable for the pore
water in porous and permeable sediments such as sands and some limestones. The
doctrine of uniformitarianism suggests that some sediment masses with 'connate'
water passed through a ground-water stage earlier in their history. There is little
doubt that pore water in sedimentary rocks is normally in motion, and that it cannot
therefore be strictly 'connate' - and this applies also to the pore water of most
mudstones. An important mechanism for this motion is the compaction of the
sediments.

COMPACTION OF SEDIMENT

In its simplest terms, compaction is a diagenetic process by which sediment under

the weight of its own overburden is compressed so that its bulk volume is reduced
and its bulk density increased. Compaction can only be achieved by the reduction of
pore volume, so compaction leads to the migration of the pore fluids towards
positions oflower energy. The loss of porosity leads also to loss of permeability (as
we have seen in Chapter 3); and the concomitant increase in bulk density leads to a
more competent rock, but a thinner one.
While we shall concentrate on the mechanical aspects of compaction, it must be
 125

remembered that many other influences, related and unrelated, may be present.
Clearly, increasing the temperature of the rock during burial reduces the effective
viscosity of the solid constituents as well as the liquid, and time increases the strain
in the rock. Chemical changes alter both solids and liquids, leading to what some
call consolidation rather than compaction (but the nomenclature is confused).

Compaction of sands

Sand on the sea floor is moved and agitated by the waves and swell of the sea, by
currents, and by seismic shocks. This leads to the sorting ofthe sediments, and to the
arrangement of the grains in a more or less stable packing. These influences extend
to the edges of the continental shelves (the influence of swell, not waves, can be felt at
such depths). But we must remember that the sediment on the sea-floor is not
normally accumulating into the stratigraphic record as such: it is only that part of
the sediment that reaches a position in which the energy of the environment is
insufficient to move it further - that part that comes to be permanently below
baselevel - that accumulates into the stratigraphic record. This is the process of
winnowing, and the process behind facies changes. While it is a process that cannot
be observed because the human life-span is too short, we may assume that a
sediment passes into the stratigraphic record no worse sorted than it was before.
(See Chapman, 1973 or 1976, pp. 1-18, for a more detailed discussion of this difficult
point.)
Studies of sand bodies have revealed that variation of porosity due to facies is far
greater than variation due to compaction with depth; but that sands appear to have
a linear compaction trend with depth (Maxwell, 1964). As the load increases with
depth, so the effective stress increases and porosity is reduced. The grains deform
elastically at first, but, due to the time dimension, recovery is not complete if the load
is removed long afterwards. (See Stephenson, 1977, for an interesting discussion of
the maximum effect of anyone component, in particular, temperature.)
The loss of porosity is relatively small in sand compaction, and the loss of
permeability (provided there is no chemical precipitate or cement) is also relatively
slight. Taking representative compaction figures from Maxwell (1964) and eval-
uating the combined porosit term from p. 60, P'5 /(1 - ff, we estimate that
reduction of porosity from 35% to 25% results in a 75% loss of permeability to 25%
of its former value. While this is a large proportional reduction, the permeability at
25% porosity may still be large.
Under severe loading, the grains may fracture and change shape: quartz over-
growths and cements also change pore geometry. Fracture of the grains reduces
permeability to a much greater extent through the increase in specific surface as well
as the greater loss of porosity due to changed geometry.
126

Compaction of muds, clays, and 'shales'

The compaction of muds is a complex process that has both physical and chemical
components. We shall follow previous practice in regarding the mechanical pro-
cesses as dominant, and then consider the modifications due to other causes. In
many applied geological contexts, the term shale is used to embrace all fine-grained
compactible sediments other than carbonates.
Muds and mudstones contain a significant, but not necessarily dominant propor-
tion of minerals that deform plastically under load, that is, material that does not
recover its former dimensions when the load is removed. Many of them are platy in
form, so that contacts between grains tend to be surfaces rather than points of
contact. The most significant consequence of these properties from our point of
view is that compaction of mudstones not only reduces porosity permanently, but
also seriously and permanently reduces permeability on account ofthe restriction of
the passages from one pore to another. This effect has important consequences,
because the loss of permeability also inhibits the expulsion of the fraction of the pore
water that must escape if further compaction is to take place. We shall pursue this
matter later.
There are two practical approaches to the determination of the relationship
between porosity - and bulk density - and depth: direct measurement, and indirect
measurement by geophysical means. Direct measurement, which requires very
careful work, was carried out by Hedberg (1926, 1936) on Tertiary shales in bore-
holes in Venezuela; and by Athy (1930) on upper Palaeozoic shales in Oklahoma
(Fig. 6-1). Hedberg was careful to point out that the direct relationship is between
pressure and porosity, not depth and porosity; but he recognized the value and need
for depth-porosity relationships.
Hedberg (1936) came to a number of important conclusions:
a) there are four broad stages of compaction: first, a mechanical re-arrangement of
the particles; then a de-watering stage with expulsion of some of the pore water and
adsorbed water; then mechanical deformation of the grains; and finally, a recrys-
tallization stage. These stages overlap.
b) porosity is a better indicator of compaction than bulk density; and that reduc-
tion of porosity is related in the early stages to the overburden pressure. With
overburden pressures from 0 to 800 psi (to 5.5 x 106 N m- 2 ) he found an expo-
nential relationship
P = 67.214 G-O.I047,

where P is the porosity (%) and G is the overburden pressure in psi (estimated from
his measurements). From 800 psi to 6000 psi (to 4.1 x 10 7 N m- 2 ) a linear rela-
tionship
P = 34.86 - 0.00421 G.
 127

POROSITY f
oo ~----~------~----~------~--~~
0.5
km.

::I:
I-
0.. 1
W
C

2 x1000ft.
Fig. 6-1. Shale compaction curves of Athy (1930) and Hedberg (1936), the latter generalized.

From 6000 psi to 10000 psi (to 6.9 x 107 N m - 2), he found a different linear
relationship
P = 13.93 - 0.0006935 G.

c) he accepted that there is a need for a single function relating porosity to pressure,
and found that for depths greater than those at which the overburden pressure is
greater than 200-300 psi,
P = 40.22 (0.9998)G. (6.1 )
d) he found that depth in feet is approximately equal numerically to the overburden
128

pressure in psi (Hedberg, 1936, pp. 259-260).

Athy (1930) studied upper Palaeozoic (permian to base Pennsylvanian) shales in
Oklahoma, and proposed the relationship
f = fo e- cz , (6.2)
where f is the porosity at depth z,fo the porosity depth z = 0, and c, a coefficient
(with a value 1.42 x 10 - 3 m - 1 for his data). He had to extrapolate the top 1400 ft
(430 m) of his curve, and some doubt has been cast on his results due to possible
tectonic forces having contributed to the compaction.
Athy's curve has been widely used by subsequent workers (e.g., Rubey and
Hubbert, 1959; Chapman, 1973; Magara, 1978), and attempts at reconciling Athy's
curve with Hedberg's (e.g., Weller, 1959) have not been convincing. It has generally
been considered that Hedberg's curve for Tertiary shales would approach Athy's
curve for Palaeozoic shales with time.
It is important to realize that neither curve implics the history of porosity reduc-
tion with time and burial, but only the relationship between porosity and depth or
pressure now. Therein lies the obvious weakness of such empirical formulae, for
temperature and time do not appear explicitly in either. And one must always
remember that compaction is not always only a vertical process: there may be lateral
spreading also in some situations (such as deltas).
Recently, Chapman and Fitz-Gerald (in prep.) took a different approach. If rocks
in the context of geological time behave not as solids but as liquids, then compaction
can be modelled by regarding rocks as compressible liquids. They propose the
dimensionless expression (still without time or temperature)

.f*-l+(Y~I)eYZ" (6.3)

where f* =f1fo,Y = Ps/(Ps - Pw)fo and z* = z/a, where a is a scale length. In

theory, the scale length can be measured from experimental data: it is the length that
makes dp /d(J unity for small values of (J. Figure 6-2 shows f* plotted against z* for y-
values of 3,3.5, and 4, within which range most shale sections will fall.
While the dimensional plots of Hedberg's (1936) and Athy's (1930) data seemed
conflicting, the dimensionless plots agree well with the field data, the values of y
being 3.9 and 3.4 respectively, when a scale length of 2,000 m is assigned to Athy's
data, and 5,000 m to Hedberg's. The scale length is related to the mechanical
properties ofthe sediment by definition, but it may also take other influences, such a
temperature and time, into account.
The same model applies to the mechanical compaction of sand (or any other
geological material) as well as shale, but the scale length for sand is larger, and so the
dimensional plot is virtually a straight line.
These results have not yet been widely tested, and so should not be accepted
uncritically. Much work remains to be done on sediment compaction trends with
 129

f*
o 0.5 1
o.-~--~--~--~--.-~--~--~--~~

z* 1

2
Fig. 6-2 Dimensionless compaction curves of Chapman and Fitz-Gerald (in preparation). 1* ~ fifo,
y = p,1 (p, -- Pw)fo and z* = z Ia where a is a scale length.

depth, and the role of chemical diagenesis in these.

From a practical point of view, however, we can rarely obtain the data we need by
direct measurement, and must resort to indirect, geophysical, methods in boreholes
if we are to obtain the local compaction trend of shales with depth. For this, the most
reliable method is to use the sonic log, and to plot the logarithm of the sonic transit
time in shale (L'1t sh ) against depth. A great mass of data from various parts of the
world indicates that over the normally-compacted part of the sedimentary se-
quence, these points plot on a straight line (with some scatter). Figure 6-3 shows
 130

SHALE RESISTIVITY - Rsh SHALE TRAVEL TIME -Il tsh

ohm-metres /Ls/ft
1 2 3 4 5 80 100 120 140 160180
, i i i ' iii

-
....u
o

~
cu
Q,

5
-..
....u
....
CD
2 c:
o
....u
o
o
f
&0
o 0
o
o
--
~
cu
CD

3 10
m. ft. 6=1.0 0.8 0.6 0.4 0.2
)(10 3
Fig. 6.3. Shale resistivities and shale travel times in a well in Borneo (courtesy of the Royal Dutch/Shell
Group).

shale resistivity and shale transit time plotted on a logarithmic scale against depth
for a well in Tertiary sediments in Borneo. Such plots of shale transit time can be
expressed by an empirical formula of the same form as Athy's for compaction,
(6.4)

where Ato is the extrapolated value of Atsh at depth z = 0, and b [L -1] is the slope
(log At)/z. It is found that Ato is commonly about 540 J.lS m -1 (165 J.ls/ft) correspond-
ing to a sonic velocity in shales of about 1,850 m s -1 (6,000 ft/sec) and fo of about
0.5; but the slope varies from area to area.
It must be noted that the common assumption that porosity is proportional to
shale transit time is only true if the slope b is the same as the slope c in Athy's
equation. Very little data has been published, but it seems to the author that this is
 131

only true for very thick shale sections with little sand, and that the slope of the shale
compaction line, c, can be larger than b by a factor of at least 3 when there is a
significant number of sands in the section. (The slope c can be estimated from the
Formation Density Log if one assumes that the bulk density is inversely propor-
tional to porosity.)
Confidence in the use of the sonic log comes from the fact that when shales with
abnormally high pore pressures are encountered, the shale transit time increases
with depth (velocity decreases) in what is known as the transition zone (Fig. 6-3);
and this departure from the normal trend usually agrees with drilling data very
closely (more closely than that of the Formation Density Log, for reasons that are
not understood). Shales that show an increase in transit time with depth into a zone
of abnormally-high pores pressures are inferred to be undercompacted for their
depth - they have not compacted fully because the commensature part of the pore
water has not been expelled. We shall look, therefore, at pore pressures in strat-
igraphic sequences.

PORE PRESSURES

A very large number of water wells, and boreholes drilled for petroleum, provide
evidence that the pore pressures in the upper few kilometres of sedimentary basins
are normal hydrostatic: they will support a column of the water to near the land
surface, or to sea-level in low-lying areas (Fig. 6-4). These observations indicate that
the water in the pores at depth is in physical continuity to the surface, and we may
regard the sediment as having accumulated in water with as much logic as the more
usual statement that the rocks contain water. If we wish to call this water 'connate',
then we must redefine the term because this water is not trapped in the sediment.
Pore pressures may be approximated by the relationship found in equation (1.2):
p=pgz, (6.5)

where p is the mean mass density of the water, and z the depth at which the pressure
is required to be known. This is an approximation only: the density may vary with
depth due to varying salinity, its compressibility, and its tendency to expand with
depth due to increasing temperatures. Furthermore, the relationship implies that
the free water surface coincides with that at which z = O. This is usually sufficiently
nearly the case, but there are areas, as we have seen in Chapter 5, in which the free
water surface is not near the level where z = O. On the one hand, there are artesian
aquifers: on the other, aquifers with abnormally low pressures (such as those in the
Molasse basin of southern Germany).
There are also data obtained from many parts of the world from petroleum
exploration drilling that indicate abnormally high pore pressures, in some cases
approaching the pressure exerted by the total overburden:
132

PRESSURE
MPa 50

psi 10000

o~
1 $,..
~~
~~
$~
5
::I:
~
Q.
W 2
C

3 10
m. ft.
Fig. 6-4. Typical pressure-depth plot of sedimentary sequence.

(6.6)

where Pbw is the mean bulk wet density of the overburden.

This relationship is also an approximation because Pbw may very - and usually
does - not only with depth but also with lithology. The value of Pbw usually falls
between 2,000 kg m - 3 for unconsolidated sediments of high porosity and 2,400 kg
m -3 (2.0 g/cm 3 to 2.4 g/cm 3 ). Hence the overburden pressure gradient is commonly
about 22,000 N m - 2 per metre (0.22 kg/cm 2 per metre). These pressures are known
as geostatic pressures, and their gradients as geostatic gradients. Lithostatic is a
common but undesirable synonym: geostatic has priority (Dickinson, 1951; 1953,
p. 425). We shall, however, use the word overburden because it has the merit not only
of priority but also of implying that the tota110ad of solids and fluids is involved.
Thus, over the normally-pressured part of the sequence, we may regard the
 133

overburden pressure as the total normal stress on a horizontal plane at depth z. We

have seen in Chapter 1 that Terzaghi (1936), in the context of soil mechanics,
postulated that the total normal stress on such a surface consists of two partial
pressures: the pressure or stress transmitted through the solid framework, from
grain to grain, and the pressure exerted by the pore water,
S = (J + p. (Terzaghi's relationship) (6.7)

The stress transmitted through the solid framework, (J, which Terzaghi called the
effective stress, is the stress that compacts a sedimentary rock. Terzaghi called the
pore pressure the neutral stress because it does not directly deform the grains.
It will be recalled that the field of fluid pressures is inferred to exist throughout the
space occupied by the fluid and the contained solids, and that this force supports
part of the weight of the solids. Hence the effective stress may be seen as the
difference between the overburden pressure and the pore pressure:
(J = S - p. (6.7a)
This is shown graphically in Figure 6-5.
The question now arises: what is the nature of abnormally-high pore pressures,
such as are found in so many parts of the world?
Dickinson (1951,1953) showed in his classic paper that stratigraphy, not depth,
determined the presence of abnormally-pressured reservoirs in the Louisiana Gulf
Coast. The top of abnormal pressures occurs beneath the dominantly sandy se-

Pressure -

,
...
.c
Q.
Q)
Q

Fig. 6-5. Effective stress is the difference between overburden pressure and pore-water pressure (here,
normal hydrostatic).
134

quence, in sediments of progressively younger age from north to south. The strati-
graphic context of shales with abnormally-high pressures is overwhelmingly that of
a dominantly regressive sequence, in which paralic and neritic sands overlie neritic
shales (Chapman, 1973, 1976, 1977). The sand content of the sequence (in terms of
beds of permeable sand) is significant because abnormal pressures are not normally
found in regressive sequences where the percentage of sand is large. Harkins and
Baugher (1969) found 5% to 10% sand to be a regional indicator of impending high
pressures for drilling in the U.S. Gulf Coast region.
The pressure-depth plot of a typical well in a typical thick Tertiary regressive
sequence is shown in Figure 6-6a. Below the top of what is known as the transition
zone, pore pressures increase above hydrostatic on a gradient (LIp/LIz) typically
much greater than the overburden pressure gradient; but before reaching the over-
burden pressure, the gradient decreases and the maximum value of pore pressure is
commonly found to be about 90% of the overburden pressure. The ratio of pore
pressure to overburden pressure at a given depth, piS, is given the symbol A
(Hubbert and Rubey, 1959, p. 142): it is regarded as the proportion of the total
overburden supported by the fluid pressure. So, when p = S, A = 1, and the total
overburden is supported by the pore fluid.
Reverting to Terzaghi's relationship, Equation 6.7, and substituting p = AS, we
can write
(J = S(1 - A). (6.7b)
As the pore-fluid pressure, increases towards the overburden pressure, A ~ 1 and the
effective stress that compacts the rock approaches zero. Boreholes that pene-
trate abnormally-pressured shales below a normally-pressured sand/shale sequence
usually reveal that the shales are undercompacted for their depth, having greater
porosity that would be normal for their depth.
Undercompaction is indicated by the reduction of electrical resistivity and reduc-
tion of sonic velocity, as measured in the wall of the borehole by borehole logging
devices. Figure 6-3 shows the resistivities and sonic travel times measured in shales
in a borehole in Borneo. It will be seen that both depart from the normal trend at
about the same depth, and the departure is consistent with an increase in porosity
with increasing depth in the transition zone.
There is also some direct evidence of these features acquired during the drilling of
such sequences: the increase in drilling rate when abnormal pressures are reached
(known as a 'drilling break'), and the occurrence of 'heaving shales'. The rate of
drilling depends on a number offactors under the control ofthe driller (mud weight,
weight on bit, rotation speed, hydraulic energy of the mud circulation, type of bit),
and some not under his control (the type of rock and its drillability, the pore fluid
pressure). If those factors under the control of the driller are kept constant, then the
penetration rate is found to be a fair guide to the lithologies being drilled. Sand, for
example, drills more easily than shale. But it was found that as the bit entered the
 PRESSURE (p) Ze
50 MPa 1 2 3 103 m.
10000 psi 5 10 10 3 ft.

® [:::11 \ \ \ '\" @
~1+5 ~
::x::
ffi 2 t r:;/
c
~\\"\~~
3"1..10

m. ft. A=Ae 0.5 0.6 0.7 0.8 0.9 1.0 1.0 0.9 0.8 0.7 0.6 0.5 Ae

Fig. 6-6. Pressure-depth plot of typical young regressive sequence (A), and plot of equilibrium depth against true depth (B).
......
w
U>
136

transition zone to abnormally-high pressures, the penetration rate increased re-

markably (in many parts of the world this became the standard method of detecting
the top of abnormal pressures while drilling). The cause of this increase is seen as the
reduction or reversal of the fluid potential gradient at the bottom of the hole, from
being downwards to being upwards (or reduced downwards); and to some extent,
the increase in porosity, and decrease in cohesive strength of the shale.
It is evident that the pore water in the transition zone is at a much higher potential,
as well as pressure, than that in the overlying normally-pressured sequence, and that
consequently there is a tendency for the pore water to flow down this potential
gradient (which we may take to be vertical if the cause of the abnormality is
gravitational). As the pore water flows down this gradient, so the effective stress
increases in the direction of flow (upwards), and the shale is more compact in this
direction. This loss of porosity in the direction of flow also involves a loss of
permeability. Hence the gradient of pore pressure in the transition zone is seen as a
potential gradient due to the flow of pore water. It is therefore independent of the
overburden pressure gradient, and assumes a gradient that is a function of perme-
ability - the lower the permeability of the shale, as a whole, in the transition zone,
the steeper the gradient L1p/L1z.
The corollary is that much of the pore water in abnormally-pressured shales is
original pore water (for which we can reasonably use the term connate), and that
undercompaction is, as we have inferred, largely due to the failure of the pore water
to move in sufficient quantities in response to the loading. This inference is support-
ed by petroleum exploration experience: commercial accumulations of petroleum
are rarely found where the pore pressures are abnormally high (see Fertl, 1976, pp.
300-311), but are common above the transition zone.
The mechanics of sediment compaction leads us to further inferences concerning
the origin of abnormally high pore pressures. There is little doubt that on the
geological time scale, the constituents of shales behave plastically, and that com-
paction of shales is irreversible beyond the extent that can be attributed to thermal
expansion ofthe pore water and the solid framework, or to possible expansion when
interlayer water is released to the pore spaces (these possible effects will be discussed
later). Hence, we cannot accept that an undercompacted shale has ever been signif-
icantly more compacted - that the porosity of an undercompacted shale has never
been much less than it is now.
The interdependence of pore pressures and effective stress is developed further by
arguing that the degree of compaction of a shale that is abnormally pressured
corresponds to the normal compaction of an identical sediment at a much shallower
depth (Figs. 6-7 and 6-6b). Rubey and Hubbert (1959, p. 174) called this shallower
depth the equilibrium depth, or effective compaction depth. Using the same symbols
as before, but adding the suffix 'e' to denote their value at the equilibrium depth,
Terzaghi's relationship can be written
 137

(J = Se(1 - Ae) . (y.7c)

Equating equations (6.7b) and (6.7c), and substituting S = {Jbwg z, and Se = Pbwg Ze,
we obtain

and so to the approximation

(1 - A)
Ze = (1- Ae)Z,

which we write as an equality because the mean bulk wet densities above the two
depths will not differ by much. The effective compaction depth is related to the actual
depth by a factor that takes the pore-fluid pressures into account. Denoting this factor
by J (Chapman, 1972a),

J=~- (I-A) (6.8)

Z - (1- Ae)'

Strictly, Ae is the proportion of the overburden supported by the ambient fluid (see
Chapman, 1979), sea water if submarine, air if subaerial; but almost all known
sediments with such abnormally-high pore pressures are submarine, so we may take
normal hydrostatic water pressures (Pe = pw g z) to define Ae (= Pel S) for the time
being.

Pressure-

ZI+---

Fig. 6-7. Equilibrium depth (ze) is the depth at which a normally-compacted mudstone would have the
same value of effective stress as the overpressured mudstone at depth z.
138

The parameter 0 is non-dimensional, and usually takes values between 0 (when A

= 1) and 1 (when A = Ae). In the present context of submarine abnormal pressures,
o may be taken as a non-linear measure of the degree to which the pore fluids have
achieved mechanical equilibrium with the ambient fluid, or as a measure of the
expulsion of pore fluids since the sediment accumulated. Equation (6.8) indicates that
if the effective compaction depth is known or can be estimated, then the pore
pressure can also be estimated.
It is also important to note that the effective compaction depth is an estimate of
the maximum overburden at the time compaction equilibrium was lost, that is, the
depth at which abnormal pressures began to develop.
As a practical example of the use of these concepts we can take Figure 6-3 (Sonic
data). We assume that the straight-line part of the sequence is normally pressured,
and that the value of the parameter 0 is 1. Using the ratio Ze/ Z, we can construct lines
of equal value of 0, the normal trend being the line 0 = 1, and the vertical from L1to
being 0 = O. The smallest value of 0, the greatest abnormality, is about 0.2 (the
smaller points are probably in borehole mud only, where the shale has washed out).
This suggests that the equilibrium depth of the shale at the bottom of the transition
zone is 0.2 x 1,600 = 320 metres, and that the abnormality began to develop soon
after the accumulation of the more permeable sediments began. The pressure range
estimated from mud-weights at a depth of 2,475 m was between 4.28 x 10 7 N m - 2
and 4.66 x 10 7 N m - 2 (436 and 475 kg/cm 2 ). At this depth, the value of J is about
0.36 (885 divided by 2475). Since Ae is approximately 0.46, and the overburden
pressure gradient is about 2.26 x 10 7 N m - 2 per metre, the pressure estimated from
the sonic log would be about 4.52 x 10 7 N m -2 (460 kg/cm 2 ).
Now, the normal slope can also be taken, for practical purposes, as the slope
corresponding to zero fluid potential gradient - the pore fluids being close to
hydrostatic equilibrium. It follows that in sections in which the gradient of change of
L1 t/ L1z is smaller than the normal (more nearly vertical on the plot) the pressures are
increasing with depth more rapidly that the hydrostatic, and the fluid potential
gradient is upwards. Conversely, when the gradient of change is less than the
normal, the fluid potential gradient is downwards. In Figure 6-3, there is evidence of
both upwards and downwards migration of pore water within the main shale
section penetrated, with a zone of separation a little deeper than 2,000 m. The
environmental interpretation indicates a shallower environment for the sediments
below this depth: this is a transgressive/regressive cycle, and there are probably
more permeable rock units at greater depth, with pore fluids at lower potential.
These trends are mirrored in the shale resistivity plot.
On another scale, we may apply the principles to the migration of pore water
induced by compaction of any alternating sequence of campactible, relatively im-
permeable, beds and permeable, relatively uncompactible, beds - such as the alter-
nating shales and sands in ,the neritic and paralic part of young regressive sequences.
In terms of the pore water, then, compaction of a water-saturated sediment
 139

creates differences of fluid potential that tend to drive the water in the direction of
lower energy. The rate at which this movement can take place depends on the
permeability of the sedimentary rock that is compacting and the resistance to all the
water movement that results from the displacement of water from the compacting
rock.
Thus a very permeable sand lens that is entirely surrounded by a relatively
impermeable shale will only be able to compact to the extent that the pore water can
escape from the total system.
In an alternating sequence of shales and sands compacting under gravity, the
more compactible shales acquire a higher pore-fluid potential than the sands, and
the pore fluid tends to migrate both upwards and downwards into normally-pres-
sured adjacent sands (Fig. 6-8), and this movement will continue until compaction
equilibrium has been reached - which, if the sequence is subsiding, may not be for
many millions of years.
The commonly-held view that shale pore water migrates along the bedding
'because that is the direction of best permeability' is not in accord with what we
know of the mechanics of pore water movement. Shale may well be anisotropic with
regard to permeability; but water movement depends not only on permeability but
also on the existence of a fluid potential gradient. When compaction is due to
gravity, the equipotential surfaces in the shale pore water are, ideally, horizontal.
The geometry of the shale may disturb this horizontality to some extent, but
significant lateral migration requires that the ratio of lateral length of migration
path to vertical length be of the same order as the ratio of vertical permeability to
lateral permeability. At least for the thinner shales, we may take the vertical path to
be half the thickness. So shales ofthe order of a few hundred metres thick and lateral
extent of the order of several kilometres would require a vertical/horizontal perme-
ability ratio of the order of a few thousand, if lateral migration is to be important.
While there is great difficulty in estimating the permeability of shales in situ, there is

Fig. 6-8. Schematic pressure-depth plot through sand/mudstone/sand sequence.

140

no evidence to suggest that the transverse to lateral permeability ratio exceeds 10

with any significant frequency - two orders of magnitude too small.
The quantity of water expelled depends also on the area normal to the flow. It
seems self evident that most shales have a very much larger area normal to vertical
migration paths than to horizontal migration paths.
The evidence for vertical migration of shale pore water lies, as we have seen, in the
observed and inferred potential gradients that are generated by gravitational com-
paction. The pore water moves to a space of lower energy. A sand intercalated in a
shale may therefore act as a drain, receiving water through its lower surface from
below, through its upper surface from above. Within the sand, the water movement
must be lateral, probably towards its surface outcrop, or other leakage to the
surface. In a typical regressive sequence, in which the sands tend to wedge out in the
direction of the regression, the water in the sands will tend to move towards the
land-mass from which the regression came. This water can hardly be regarded as
connate.

OTHER POSSIBLE CAUSES OF OR CONTRIBUTORS TO

ABNORMALLY-HIGH PORE PRESSURES

Many processes have been postulated for the generation of abnormally-high pore
pressures - geometry of gas reservoirs, uplift, clay mineral diagenesis, thermal
expansion of water, osmosis, and tectonic activity - in addition to the mechanical
that we have adopted. There is some merit in each, of course. Let us review them
briefly.

Reservoir geometry. There is no doubt whatever that a thick gas column can
generate abnormally high pore pressures within the reservoir. This is due to the
difference between the normal hydrostatic water pressure gradient outside the
reservoir, and the gas pressure gradient within it (Fig. 6-9). This has been recognized
from the beginnings of serious study of subsurface pressures. Several reservoirs in
Iran have severely abnormal pressures on this account, and a relationship has been
observed between reservoir pressures and seeps (see discussion of Dickinson, 1951,
pp. 16-17). The effect of thick oil reservoirs is less pronounced, but not always
insignificant.
Such abnormally high pressures are particularly difficult to drill successfully
because, as can be seen in Figure 6-9, the greatest abnormality is at the top of the
reservoir, and the top ofthe structure, decreasing with depth and with distance from
the crest, and vanishing at the gas/water contact.

Uplift. Various authors at various times have suggested that abnormally-pressured

shales could result from normally-pressured shales being uplifted to a shallower
 141

Pressure -

,
-
.c
0.
CI)
Q

Fig. 6-9. Abnormal pressures developed in gas reservoir by virtue of the density difference between gas
and water.

depth without loss of pressure. The main objection to this hypothesis is that there is
unambiguous evidence of subsidence during the accumulation of regressive se-
quences in which abnormally-high pore pressures occur*, so this hypothesis must be
discarded.

Thermal expansion of water. It has, of course, long been recognized that thermal
expansion of water takes place - with some compression - when it is buried (' ... the
expansion ofthe water at high temperatures exceeds the decrease of volume caused
by compression .. .', Versluys, 1932, pp. 924-925), but it could not form part of any
hypothesis until the nature of abnormal pressures had been appreciated. This
appreciation can be said to have begun in the early 1950's. Thermal expansion of
water was suggested as a possible cause of abnormally high pressures by Smith and
Thomas (1971, p. 17), argued in more detail by Barker (1972), and then by Margara
(1978).11 appears to have received wide acceptance, but there are several difficulties.
If thermal expansion is the main cause of abnormally high pressures, it seems that
three features are required: the formation of a near-perfect seal to the shale, the
maintenance of constant pore volume, and the absence of mechanical compaction.
Objections to this hypothesis arise from these apparent requirements.
The most rapid heating of the pore water by burial in the Gulf Coast ofthe U.S.A.
would be about 1°C per 14,000 years (geothermal gradient of 36°Cjkm, burial at
* The accumulation of paralic and neritic sediments in any sequence more than a few hundred metres
thick must involve subsidence. The evidence of growth structures will be outlined in the next chapter.
142

500 years/metre). Thus the maximum rate of thermal expansion of water will be by a
factor of 1.0006 in 14,000 years. The intrinsic permeability required to allow the
pore water in a column of mudstone 1 m 2 in horizontal cross-section. 500 m deep or
thick, and 20% porosity to escape as it expands is ofthe order of 5 x 10- 21 m 2 , or
5 x 10- 17 cm 2 , or about 5 x 10- 6 millidarcies (Chapman, 1980). This is near the
lower limit of permeabilities measured in Tertiary mudstones in Japan (reported in
Magara, 1971, p. 241, fig. 9). Even under the extreme conditions postulated, such
permeabilities would be sufficient to dissipate the volume of water generated by
thermal expansion. So a near-perfect seal is required.
The maintenace of constant pore volume is intuitively unlikely. As the water is
heated by burial, so is the solid material. If this expands isotropically, with mainte-
nance of geometric similarity, the porosity remains unchanged but the pore volume
increases. So the pore volume will only remain constant if mechanical compaction
reduces the pore volume by the same amount.
That mechanical compaction does take place has already been discussed at some
length in the context of Terzaghi's relationship (Chapter 1).
There are other objections. If a perfect seal exists, the pore water pressure gradient
will be hydrostatic,but displaced to a higher pressure (Fig. 6-10). As with the thick
gas reservoirs, the greatest abnormality would be immediately beneath the seal.
Drilling and borehole logs indicate the reverse trend: greatest abnormality at the
bottom of the transition zone (Fig. 6-11). Also, abnormal pressures occur in places
at very shallow depth, above 1,500 m, where the temperature is not great and the
prospects of a perfect seal negligible.

Pressure - p

,
...
.c
Q.
GI
C

Z
Fig. 6-10. Pressure-depth plot when abnormal pore-water pressures are retained by perfect seal.
 143

Pressure - p

z
Fig. 6-11. Pressure-depth plot when abnormally-pressured pore water is flowing to positions of lower
potential through sediments of low permeability.

Clay mineral diagenesis. This hypothesis stems from a suggestion by Powers (1967)
that fluid release from clay minerals could act as a flushing mechanism for petro-
leum. For this mechanism to generate abnormal pressures, there must be a volume
increase when the interlayer water is released to the pore space. Magara (1978, pp.
100-109) has shown that the observed degree of under compaction could not have
been generated by this process, because the expansion, if it occurs, is insufficient to
account for the bulk density decreases observed.
Furthermore, the diagenesis of montmorillonite (more properly called smectite)
to illite, which could release water to the pores, is not regarded by Powers as
occurring above a depth of about 1,800 m (6,000 ft), nor by Burst (1969) between
800 and 2,500 m (2,600 to 8,500 ft). Comparable depths were found by Perry and
Hower (1970, p. 171) and rather greater depths by Weaver and Beck (1971. p. 18).
Abnormal pressures are common at depths shallower than these: for example,
Trinidad (Fig. 6-12, which also illustrates normal gradients in permeable beds),
New Zealand (Katz, 1974, p. 469), West Irian (Visser and Hermes, 1962, p. 230),
Papua New Guinea (Spinks, 1970), Kalimantan, and the Pleistocene of the U.S.
Gulf Coast.
While such fluid release may well be a reality at greater depths, the contribution
from this cause will only be assessed when we know with confidence whether
interlayer water has a higher density than pore water.
144

PRESSURE
MPa 20 SOOm.

:I:
I-
0..
W
C

m ft.)( 10 3 A= Ae O.S 0.6 0.7 0.8 1.0 0.9 0.8 0.7

Fig. 6-12. Pressure-depth plot of shallow oil reservoirs in Trinidad, with their corresponding equilibrium
depths. (Courtesy of the Royal Dutch/Shell Group.)

Osmosis. We have seen in Chapter 5 that this may be a cause of abnormally low
pressures in pore water, and that it could therefore lead to abnormally high pore
pressures in the more saline formation to which the water is flowing. It is a matter of
contrasting salinities. As a process for the generation of abnormally-high pore
pressures in shales, it is unconvincing. Indeed, Margara (1978, pp. 283-284) has
argued that the osmotic gradient in a shale opposes the generation of abnormally
high pressures by assisting the expulsion of the pore water.

Tectonic. There is ambiguity in those areas of obvious present-day tectonic activity

(California, New Zealand) because gravitational influences cannot be eliminated.
Nevertheless, when tectonic forces introduce strain in relatively impermeable sedi-
ments, this strain is likely to lead to abnormally-high pore pressures rather than a
reduction of pore volume, until water can escape. Such a mechanism has been
proposed for overthrust faulting (Roberts, 1972) and for the mineralization along
faults (Sibson et aI., 1975). Tectonic force is probably oflittle direct significance in
many areas of abnormal pressures (such as the U.S. Gulf Coast, the Niger delta, and
many areas of S.E. Asia) although, as we have noted, any tendency to slide may
induce strain that elevates the pore pressures.
In summary, the mechanical hypothesis may not account for all the abnormality
we observe, but it does account qualitatively for the observations. Sediments do
tend to compact under load.
 145

There is little doubt that thermal expansion of pore water contributes to ab-
normal pressures by increasing the volume of water that has to be expelled; but there
is considerable doubt that it is significant, let alone dominant.
Clay mineral diagenesis may change the quality of the pore water, but it is
doubtful whether it changes the quantity.
There is no doubt at all that thick gas columns can cause seriously abnormal
pressures, but this is a different process.

PORE-WATER MIGRATION IN SEDIMENTARY BASINS

We may now combine these various aspects of pore water migration to deduce the
general directions of pore water migration in sedimentary basins prior to orogeny.
We may take a simple transgressive-regressive cyle as our model (Figure 6-13, which
has a grossly exaggerated vertical scale) because the arguments can easily be extend-
ed to more complicated sedimentary basins.

Fig. 6-13. Schematic block diagram of transgressive sequence (upper) and transgressive-regressive cycle
(lower).
146

During the accumulation of the transgressive sequence, the basal permeable unit
(commonly a limestone or a sandstone) is in physical continuity with the sea close to
the shoreline, and so provides a conduit for the shale pore water that is expelled
downwards into it. Much of the compaction of this shale at this stage, though, is
seen as an extension of the settling process, with the bulk of the pore water moving
upwards relative to the grains. Whether there is absolute upward motion depends
on the rate of subsidence relative to the rate of pore-water movement. During the
transgressive phase, the shale (or more accurately, mudstone) is compacting from
the bottom towards the top.
With the onset of regression, significant changes take place in the pattern of
mudstone compaction. Loading the mudstone with a permeable rock unit - in the
regressive part of the sequence, almost invariably sand - has the effect of a filter
press, and the mudstone beneath it compacts from the top towards the bottom on
account of the potential and pressure gradients induced in the upper part of the
mudstone. The loss of permeability at the bottom and at the top of the mudstone
due to these opposing compaction patterns tends to seal the pore water in the
mudstone. Since this is a period of continued subsidence, and therefore of loading
by new sediment, the main period of generation of undercompacted, abnormally-
pressured 'shales' begins with the onset of regression. This conclusion is supported
by the geological evidence of deformation in the regressive sequence during
sediment accumulation, due to the mechanical instability of the sequence (this will
be discussed in some detail in the following chapter).
The dominant transgressive and regressive sequences characteristically have sec-
ond-order fluctuations, giving rise to alternating carbonates and shales in the
transgressive sequence (typically), or sandstones and shales, and almost exclusively
sandstones and shales in the regressive part of the sequence. This alternation is
particularly evident in the regressive part ofthe sequence until paralic and terrestrial
sands begin to accumulate. These second-order fluctuations give rise to wedges and
tongues of permeable sediment that are discontinuous seaward. The regressive
sequence also characteristically exhibits an increasing sand-shale ratio upwards
until paralic and terrestrial sands and conglomerates predominate. Harkins and
Baugher (1969) noted that in the U.S. Gulf Coast abnormal pressures are typically
below the 5% to 10% sand level. These sands, and those above, typically have pore
water at normal hydrostatic pressures, from which we deduce that there is hydraulic
continuity of pore space to the surface. We infer that pore water is expelled into
these sands through both top and bottom interfaces, and that the lateral migration
within the sands is, from their geometry, towards the land.
These patterns of pore water migration are of particular importance for the
accumulation of petroleum - and, one suspects, of base metals in the transgressive
carbonate sequence. Pore-water migration continues throughout the phase of ac-
cumulation of sediment in the sedimentary basin, and probably continues well into
the orogenic phase. If abnormal pressures exist when the orogenic phase begins, the
 147

pore water plays an important role in the nature of the deformation of the sedi-
mentary basin.

SELECTED BIBLIOGRAPHY

Athy, L.F., 1930. Density, porosity, and compaction of sedimentary rocks. Bull. American Ass. Peroleum
Geologists, 14: 1-24.
Barker, C., 1972. Aquathermal pressuring - role of temperature in development of abnormal-pressure
zones. Bull American Ass. Petroleum Geologists, 56 (10): 2068-2071.
Burst, J.F., 1969. Diagenesis of Gulf Coast clayey sediments and its possible relation to petroleum
migration. Bull. American Ass. Petroleum Geologists, 53: 73-93.
Chapman, R.E., 1972. Clays with abnormal interstitial fluid pressures. Bull American Ass. Petroleum
Geologists, 56 (4): 790-795.
Chapman, R.E., 1973. Petroleum geology: a concise study. Elsevier Scientific Pub!. Co., Amsterdam,
London, and New York, 304 pp.
Chapman, R.E., 1976. Petroleum geology: a concise study (paperback edition). Elsevier Scientific Pub!.
Co., Amsterdam, Oxford, and New York, 302 pp.
Chapman, R.E., 1979. Mechanics of un lubricated sliding. Bull. Geol. Soc. America, 90: 19-28.
Chapman, R.E., 1980. Mechanical versus thermal causes of abnormally high pore pressure in shales.
Bull. American Ass. Petroleum Geologists, 64: 2179-2183.
Dailly, G.C., 1976. A possible mechanism relating progradation, growth faulting, clay diapirism and
overthrusting in a regressive sequence of sediments. Bull. Canadian Petroleum Geology, 24 (1): 92-116.
Dickinson, G., 1951. Geological aspects of abnormal reservoir pressures in the Gulf Coast region of
Louisiana, U.S.A. Proc. 3rd World Petroleum Congress (The Hague, 1951), sect. 1: 1-16. (Discussion:
16-17.)
Dickinson, G., 1953. Geological aspects of abnormal reservoir pressures in Gulf Coast Louisiana. Bull.
American Ass. Petroleum Geologists, 37 (2): 410-432.
Fertl, W.H., 1976. Abnormal formation pressures. Elsevier Scientific Pub!. Co., Amsterdam and New
York, 382 pp. (Developments in Petroleum Science, 2).
Fiichtbauer, H., and Reineck, H.-E., 1963. Porositiit und Verdichtung rezenter, mariner Sedimente.
Sedimentology, 2: 294-306.
Harkins, K.L., and Baugher, J.W., 1969. Geological significance of abnormal formation pressures. J.
Petroleum Technology, 21: 961-966.
Hedberg, H.D., 1926. The effect of gravitational compaction on the structure of sedimentary rocks. Bull.
American Ass. Petroleum Geologists, 10 (11): 1035-1072.
Hedberg, H.D., 1936. Gravitational compaction of clays and shales. American J. Science, 5th series, 31:
241-287.
Hottman, C.E., and Johnson, R.K., 1965. Estimation of formation pressures from log-derived shale
properties. J. Petroleum Technology, 17: 717-722.
Hubbert, M.K., and Rubey, W.W., 1959. Role of fluid pressure in mechanics of overthrust faulting, 1.
Mechanics of fluid-filled porous solids and its application to overthrust faulting. Bull. Geol. Soc.
America, 70 (2): 115-166.
Katz, H.R., 1974. Recent exploration for oil and gas. In: G.J. Williams (Ed.), Economic geology of New
Zealand. Monograph Series Australasian Inst. Mining Metallurgy, 4: 463-480.
Magara, K., 1971. Permeability considerations in generation of abnormal pressures. J. Soc. Petroleum
Engineers, 11: 236-242.
Magara, K., 1978. Compaction and fluid migration: practical petroleum geology. Elsevier Scientific Pub!.
Co., Amsterdam, 319 pp. (Developments in Petroleum Science, 9).
Maxwell, J.C., 1964. Influence of depth, temperature, and geologic age on porosity of quartzose
sandstone. Bull American Ass. Petroleum Geologists, 48 (5): 697-709.
Perry, E., and Hower, J., 1970. Burial diagenesis in Gulf Coast pelitic sediments. Clays and Clay
Minerals, 18: 165-177.
148

Plumley, W.J., 1980. Abnormally high fluid pressure: survey of some basic principles. Bull American Ass.
Petroleum Geologists, 64 (3): 414-422.
Powers, M.C., 1967. Fluid-release mechanisms in compacting marine mudrocks and their importance in
oil exploration. Bull. American Ass. Petroleum Geologists, 51 (7): 1240-1254.
Roberts, J.L., 1972. The mechanics of overthrust faulting: a critical review. Proc. 24th International
Geological Congress (Montreal, 1972), Sect. 3: 593-598.
Rubey, W.W., and Hubbert, M.K., 1959. Role of fluid pressure in mechanics of overthrust faulting, II.
Overthrust belt in geosynclinal area of western Wyoming in light of fluid-pressure hypothesis. Bull.
Geol. Soc. America, 70 (2): 167-206.
Sibson, R.H., Moore, J.McM., and Rankin, A.H., 1975. Seismic pumping - a hydrothermal fluid
transport mechanism. J. Geol. Soc. London, 131: 653-659.
Smith, J.E., 1971. The dynamics of shale compaction and evolution of pore-fluid pressures. Mathemat-
ical Geology, 3: 239-263.
Smith, J.E., 1973. Shale compaction, J. Soc. Petroleum Engineers, 13: 12-22.
Smith, N.E., and Thomas, H.G., 1971. Origins of abnormal pressures. In: Houston Geol. Soc., Abnormal
subsurface pressure: a Study Group report 1969-1971. Houston Geol. Soc., Houston, pp. 4-19.
Spinks, R.B., 1970. Offshore drilling operations in the Gulf of Papua. J. Australian Petroleum Explo-
ration Ass., 10: 108-114.
Stephenson, L.P., 1977. Porosity dependence on temperature: limits on maximum possible effect. Bull
American Ass. Petroleum Geologists, 61 (3): 407-415.
Terzaghi, K., 1936. Simple tests determine hydrostatic uplift. Engineering News Record, 116 (June 18):
872-875.
Versluys,J., 1932. Factors involved in segregation of oil and gas from subterranean water. Bull American
Ass. Petroleum Geologists, 16 (9): 924-942.
Visser, W.A., and Hermes, J.J., 1962. Geological results of the exploration for oil in Netherlands New
Guinea. Verhandelingen van het Konnklijk Nederlandsch Geologisch-Mijnbouwkundig Genootschap
voor Nederland en Kolonieen, Geologische Serie, 20: 1-265.
Weaver, C.E., and Beck, K.C., 1971. Clay water diagenesis during burial: how mud becomes gneiss.
Special Paper Geol. Soc. America 134 (96 pp.).
Weller, J.M., 1959. Compaction of sediments. Bull. American Ass. Petroleum Geologists, 43: 273-310.
 7. ROLE OF PORE WATER IN DEFORMATION
OF SEDIMENTAR Y BASINS

We must make the distinction between pre-orogenic deformation and orogenic

deformation of sedimentary basins because there is a great deal of evidence that
important deformation occurs in sedimentary basins while they are accumulating
sediment on a subsiding sedimentary column - particularly in the terminal regressive
sequence. The regressive sequence itself is due to a neighbouring orogeny outside
the sedimentary basin: the creation of mountains creates sediment in increasing
amounts until the supply of sediment to a sedimentary basin exceeds the volume
created by subsidence of the sedimentary basin. The sea then tends to become
shallower, and the sediments prograde away from the orogeny.
Deformation in this part ofthe sedimentary basin that takes place while sediment
is accumulating is here referred to as pre-orogenic because it takes place before
orogeny of the sedimentary basin itself. (The term is not exclusive to this situation:
other parts, particularly the initial transgressive sequence also suffer pre-orogenic
deformation - but we are concerned with the role of pore water.) We shall later
discuss orogenic deformation, which takes place during orogeny ofthe sedimentary
basin itself. The relationship between the two is shown schematically in Figure 7-1.
As a modern example, we may take the U.S. Gulf Coast basin. Throughout the
Cenozoic, it has been accumulating sediment derived from the relatively distant
Rocky Mountains and Appalachians. It has not yet suffered orogeny, but it has
suffered extensive faulting and gentle folding apart from that due to salt diapirism.
We shall first consider the geology of pre-orogenic deformation, then its probable
causes.

PRE-OROG ENIC DEFORMATION

Growth structures

Structures that formed while the sediment in them accumulated are known collec-
tively as growth structures. Growth faults are faults (usually normal faults in the
context of regressive sequences, but not uniquely so) across which the thicknesses of
correlative rock units change abruptly (Fig. 7-2), the thicker being on the down-
thrown side. Likewise, growth anticlines are anticlines in which the rock units are
150

-SEDIMENTARY BASIN OROGENY

TRANSGRESSION

~~
ed\Olent

2.-------------~~~~=-----~~--
»»»»> I I

REGRESSION

u Deformation
due to
instability
in sequence

Fig. 7-1. Schematic relationship between orogenic and pre-orogenic deformation.

thinner over the crest than on the flanks (Fig. 7-3). In general, growth structures are
structures in which the thicknesses of rock units reflect the structure itself. Sedi-
mentary basins are growth structures on a large scale.
The first recognition of growth faults appears to be due to Tiddeman (1890), who
ascribed the lithological changes across the Craven fault in England to movement of
the fault concurrently with sediment accumulation. Subsequently, growth structures
were found almost exclusively in the domain of applied geology, first in coal mines,
then in oil and gas fields. By their nature, they are more easily recognized in
subsurface workings, with detailed data in three dimensions, than in outcrop. Dron
(1910) found such faults in the Scottish coal fields, and the existence and inter-
pretation of complicated growth structures became well accepted in the coal fields of
western Germany in the 1920s (Bottcher, 1925, 1927), later to be extended to
European oil fields (Stutzer, 1930).
They were rediscovered a decade or so later by the u.S. petroleum industry, and
given various names: Gulf Coast Type fault (!), synsedimentary, progressive, deposit-
 151

- - - - - - - - - - - - -... -: :_\. .;~ : ~

•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••

Fig. 7-2. Cross-section through typical growth fault.

Fig. 7-3. Cross-section through typical growth anticline.

152

ional, contemporaneous and growth. The terms depositional and growth were in
common use in the petroleum industry in the 1950s, but it was growth that was
recommended by Dennis (1967) in the International Tectonic Dictionary. It would
be hard to make a case for priority for growth, but it has its merits and we shall use it
throughout.
The interpretation of growth structures as structures that grew while the sediment
accumulated was clearly reached quite independently by the German coal geologists
and the U.S. petroleum geologists (there is no evidence that the U.S. writers had
read the German works - or, indeed, that either had read Tiddeman's paper).
Growth faults occur in two main stratigraphic associations: the initial transgres-
si ve sequence 0 f sedimentary basins formed by rifting and basement faulting (detect-
ed in recent years by marine seismic surveys), and the terminal regressive sequence
of sedimentary basins. In the latter, we have the coal-field occurrences already noted
(the major coal-bearing sequences of the world are dominantly regressive), and the
oil and gas field occurrences, which are also associated with shales with abnormally
high pore-water pressures. We are concerned mainly with the latter.
Growth faults in regressive sequences are typically slightly curved in plan (con-
cave to the direction of progradation of the regressive sequence) and curved in
section (concave up). The strike of the fault is roughly parallel to the depositional
strike, with the downthrown side on the side that is in the direction of prograda-
tion *. The dip of the fault typically flattens with depth from about 60° or 70° to 40° ,
or even less.
It seems to the writer that considerable misunderstanding has developed over this
flattening. Because abnormal pressures are found in the U.S. Gulf Coast in the
flatter part of a growth fault, and these pressures increase with depth as the fault
flattens, it has been implied (e.g., Bruce, 1973,p. 884) that the two observations have
a causal relationship. Apart from the theoretical and practical evidence that pore
pressures do not affect the angle of fracture (see Hubbert and Rubey, 1959, pp. 141-
142), it must be remembered that the angle of fault dip was determined at the time
the fault was initiated, which was soon after the beginning of accumulation of the
permeable sands of the regressive sequence - that is, in unconsolidated sediments at
shallow depth. This angle is typically about 60° to the bedding, but compaction of
the sediments (by definition, after the fault has been created) tends to flatten the dip.
Since the proportion of shale increases downwards in a regressive sequence, so the
relative compaction increases and flattens the dip of the fault. If the original angle
was 60° and the original porosity 50%,40% compaction of a faulted shale to 16%
porosity will reduce the dip ofthe fault to 45°. If the original fault dip was 50°,40%
compaction will reduce it to 35° .
So most of the flattening can be attributed to compaction. It is most unlikely that
* This is the characteristic that leads to the objectionable name 'down-to-the-basin fault'. It is
objectionable because it implies a position on the edge of a 'basin' and reveals a misunderstanding of the
nature of sedimentary basins.
 153

the fracture can be sustained in the undercompacted shale; but flow of shale away
from the loaded area may account for some extreme flattening reported. The rate of
movement of a growth fault is ofthe order of 100 m per million years or so -10- 4 m
per year - which is an infinitessimal movement for plastic shale.
One must also remember that any tilting or folding of the beds tilts growth faults
equally (see Murray, 1961, p. 183, fig. 4.27, for an example).
The movement of growth faults can be dated relative to the sediments they cut,
because thickness contrasts are only generated while the fault is moving. Plotting
the thickness ratio of correlative rock units across the fault against the stratigraphic
level dates the movement of the fault and its relative activity. No worldwide study of
the time of faulting relative to the transition zone to abnormal pressures has been
published, but Thorsen (1963) made an interesting study of growth faults in western
Louisiana.
Comparison of Thorsen's map (1963. p. 107, fig. 4) with Dickinson's (1953, pp.
416-417, fig. 3) shows that the periods of maximum growth-fault movement oc-
curred in Louisiana within a couple of biostratigraphic zones above the youngest
abnormally-pressured shales. Thorsen also noted that the time of maximum
growth-anticline movement generally was the same as for the faulting. This indi-
cates that growth-structure movement began soon after the accumulation of the
more permeable part of the sequence on top of the shales, and suggests that
abnormal pressures also developed early in these shales when the overburden was
relatively slight (cf. equilibrium depth in previous chapter). This is also supported
by Thorsen's observation that sand percentage is most closely related to contem-
poraneous structural growth near 'the basinward limit of sand deposition, that is, in
those areas of ten per cent or less sand', and Harkins and Baugher's (1969) obser-
vation that the top of abnormally-pressured shale normally occurs regionally below
the 5-10% sand level.
In the Midland field, studied by Fowler et al. (1971, p. 65), faulting began soon
after the permeable part of the sequence began to accumulate. The writer has seen
evidence of fault movement with less than 500 m of overburden on what are now
abnormally-pressured shales.
There is thus strong field evidence to support the theoretical arguments for early
generation of abnormally-pressured shales in regressive sequences, and the early
generation of all those parameters related to such abnormal pressures.
When more than one growth fault is found in an area, they tend to be younger in
the direction of progradation, so that growth faults on the down thrown side of a
growth fault tend to be younger that those on the upthrown side. They tend to begin
their movement and to stop their movement later in the direction of progradation,
so that the stratigraphic relationships of a sequence of growth faults may be as in
Figure 7-4. For examples of this, see Murray (1961, p. 190, fig. 4.33) and compare
with Dickinson (1953, pp. 416-417, fig. 3, and p. 419, fig. 5). Figure 7-5 illustrates
this relationship between growth faults and other features.
154

'.':'

Fig. 7-4. Cross-section through typical sequence of growth faults.

Growth anticlines, in which the thicknesses of rock units decrease towards the
crest, are mapped by the construction of isopach and cumulative isopach maps (see
Chapman, 1973 or 1976, pp. 221-223). The cumulative isopach contours for deeper
horizons (larger intervals) have a configuration resembling that of the structural
contours of that horizon. As a consequence of these thickness changes, growth
anticlines tend to become steeper-flanked with depth. This feature appears to have
been noted first by Stutzer (1930). Relatively little attention has been given to them
in the literature and, apart from Thorsen (1936) no explicit data has been published
on the timing of initial growth in relation to the overburden on the main shale part of
the sequence. The writer's experience agrees with Thorsen's, that the main period of
anticlinal growth embraces that of growth-fault movement, but that anticlinal
growth may continue after growth-fault movement has ceased. In other words, an
individual growth fault may have a shorter history of movement than its associated
anticline.

Causes of growth structures

The search for the causes of growth structures must begin with a mechanical
analysis. The mechanics of faulting is well understood, and there is no ambiguity
about the evidence that a normal fault develops in a stress field in which the least and
intermediate compressive stresses are in the horizontal plane, and the greatest
compressive stress is vertical (we take these principal stresses to be horizontal and
vertical because it is a close approximation under the continental shelves and low
lying coastal areas, where these phenomena are commonly found). A normal fault
makes an angle of about 30° with the axis of greatest principal stress, that is, its dip is
about 60° ; and the strike of the fault is normal to the axis of least principal stress.
 155

We speak of 'least' and 'greatest' compressive stresses because in unconsolidated

sediment with no tensile strength, all sediment below a few metres from the surface is
in compression (Hubbert, 1951, p. 367). But there is a component of tension (the
deviatoric stress of structural geologists) normal to the strike of a normal fault, in
the direction of lateral displacement: this component of tension is usually aligned
roughly normal to the depositional strike.
Thus growth faulting is seen, at least in part, as resulting from a tendency for the
materials to slide seawards in such open-ended situations as the U.S. Gulf Coast and
the Niger delta. We shall see later that bulk movement of the shale outward from
under the load may also contribute to growth faulting.
When growth faults are associated with growth anticlines, we may extend the
argument significantly. If growth anticlines were growing while the growth faults in
them were moving, then the growth anticline formed in the same stress field as the
growth faults, that is, with the greatest principal stress vertical. This conclusion is at
variance with orthodoxy, which would require the axis of greatest compressive
stress to be horizontal and normal to the axis ofthe fold. So it must not be accepted
without further scrutiny*. We therefore take a field example of which the data are
not in dispute - Seria oil field in Brunei.
The Seria field is an asymmetric anticline (Fig. 7-5) in Tertiary sediments. It is
about 20 km long and less than 5 km wide, with the axis parallel and close to the
present-day coastline (Schaub and Jackson, 1958; Liechti et aI., 1960). The strati-
graphic sequence is regressive, passing upwards from the neritic Setap Shale Forma-
tion (Oligo-Miocene), through the neritic shale/sand Miri Formation and the neritic
sand/shale Seria Formation, to the paralic Liang Formation. The regression is
apparently still going on, to the north or north-west.
The structure has no surface expression: it was revealed by shallow core-drilling
under a flat, low-lying, coastal swamp. The drilling of several hundred wells has
revealed beyond reasonable doubt that it is a growth anticline cut by growth faults,
some of which cut the youngest sediments. So the evidence is that the anticline was
formed in a stress field with a horizontal component of tension, the greatest princ-
ipal stress being vertical.
Schaub and Jackson (1958) attributed this folding ofthe Northwest Borneo basin
to orogeny towards the end of the Pliocene. They noted that while the folding is
more intense in the interior of Borneo, it decreases 'basinward' where broad gentle
synclines are separated by narrow, steep anticlines (of which Seria is one) that are
complexly faulted by normal faults in the main. These faults developed contem-
poraneously with the accumulation of the younger Neogene.
It will be noted that if the folding that is ascribed to a Pliocene orogeny implies
horizontally-directed compressional deformation, then the field data do not sup-
port such an hypothesis because normal faulting was taking place in the same
* They are not caused by the folding because they are not uniquely associated with the folding; but
folding may, of course, contribute to the faulting.
156

Paralic ~ Liono Formation

~nenr~~iC I}>}~<~ Ser i a Form at ion

c=J
seA L E
o I MILE
Neritic Miri Formation ====IF====I1
o
t=I

I KILOMETRE

Neritic g--~ Setap Shale Formation

Fig. 7-5. Cross-section through Seria field, State of Brunei (after Schaub and Jackson, 1958, fig. 3).

sediments in the same place at the same time. Lest it be thought that the reverse fault
in Miri field (Schaub and Jackson, 1958, p. 1332, fig. 2) supports the hypothesis of
horizontal compressive deformation, it must be pointed out that a reverse fault
dipping at about 60° is consistent with vertical deforming forces because fractures
make an acute angle with the axis of greatest principal compressive stress.
The following observations appear to be significant: a) The youngest formation,
the Liang, is folded (it was discerned from the original shallow core drilling). b) This
formation is also faulted by normal faults, and c) even over the crest, there are more
than 2,000 m of neritic and paralic sediments.
These observations surely lead us to the conclusion that the deformation of Seria
field took place during the Miocene to Holocene (at least) during subsidence in a
stress field with a component of horizontal tension, and the orthodox view cannot be
sustained.
Let us assemble all the pertinent data, to help us to a better understanding ofSeria
(and similar structures):
- The stratigraphic sequence is clearly dominantly regressive.
- The folding and faulting took place while the sediment accumulated and sub-
sided.
- The anticline is very gentle at shallow depth, but increases in steepness with
depth. Hence deformation continued to the very recent past, at least.
 157

- The Setap Shale is abnormally pressured in the Seria and Miri areas, and this is
probably a regional feature because it forms mudvolcanoes in the Jerudong-
Limbang area of Brunei and at Bulak Setap (from which it takes its name)
(Liechti et aI., 1960, pp. 326-329).
- The structural style is one of broad, gentle, synclines separated by narrow, steep,
anticlines.
These features - regressive sequence, growth structures, abnormally-pressured
shale - are found in many parts of the world in conjunction with this structural style.
It has been argued extensively that this association has important relations with
mechanical instability in young regressive sequences (Chapman, 1973, 1976, 1977a,
1977b). It is a diapiric* deformation.
The theory of deformation due to mechanical instability in simple sequences is
quite well known for simple geometry (Biot and Ode, 1965; Ramberg, 1972, for
example). Scale models agree well with theory (parker and McDowell, 1955; Ram-
berg, 1967; Tanner and Williams, 1968); and both the theoretical and physical
models resemble geological observations closely. A two-layered sequence in which
the upper layer is denser than the lower is mechanically unstable in the field of
gravity, and the interface will deform if the viscosities are sufficiently low. In
particular, Tanner and Williams (1968) found that anticlines tended to form with
their axes parallel to the intermediate principal stress.
When we speak of viscosity in rocks, we should qualify it with an adjective such as
equivalent or apparent to indicate that we are considering a property analogous to
viscosity in liquids. We must assume that in geology the strict analogy is with several
immiscible liquids of different viscosities - and therein lies the difficulty of quanti-
fying this property of rocks. In the present context the analogy is broadly justified on
the grounds that properly scaled physical models use materials that resemble liquids
more closely than the materials they represent, because of the dimension of time.
Some workers have used liquids and obtained results like those of nature (e.g.,
Ramberg, 1967, p. 53ff, using oils). Qualitatively, we must accept the field evidence
that materials that are brittle on the short time-scale have been folded without
fracture on the long time-scale (we see this property also in pitch, as in the Trinidad
Pitch Lakes, and sealing wax).
If we regard a young regressive sequence as a two-layer system of permeable sands
with normally-compacted shales overlying undercompacted over-pressured shales,
we are concerned with the relative viscosities of these two layers. The experience of
drilling for petroleum, as we saw in the previous chapter, indicates that the physical
properties of these two layers are very different, and that the pore-water pressure is
the significant parameter leading to these differences. When drilling through the
upper, normally-pressured layer, sands typically drill faster than shales, but the hole
* Purists are quite right to insist that a diapir is penetrative: but it is surely permissible to use 'diapiric'
for analogous processes. Salt diapirs are preceded by a non-penetrative deformation into salt waves. This
is the analogy.
158

remains stable through both. As soon as the transition zone is reached, the driller
observes what is known as a 'drilling break' - the rate of penetration increases
rapidly and soon exceeds that experienced in sands although now drilling in shales.
The two main reasons for this, both related to pore pressure, are the reversal or
reduction of the fluid potential gradient across the bottom of the hole, and the
reduction of the competence or coherence of the shale due to the reduction of
effective stress as the pore pressures increase above normal hydrostatic.
When undercompacted shales are encountered at relatively shallow depth (above
2,000 m approximately) they commonly tend to flow into the borehole ('heaving
shales', or 'gumbo' in the U.S. Gulf Coast). This effect is detected by increased
torque while drilling, or failure to reach bottom when going into the hole after
changing the bit. Electrical logs with a caliper for measuring the diameter of the hole
(such as the Microlog) may reveal shales sections with a hole diameter less than the
diameter of the bit that drilled the hole.
There is thus little doubt that abnormally-pressured shales that are undercom-
pacted also have a lower viscosity than normal for their depth. This conclusion is
supported by the evidence of mudvolcanoes, which occur in the same association
with young regressive sequences, and are true diapiric phenomena.
The matter of bulk density of these sediments is elusive. As we have seen, the bulk
wet density of a sedimentary rock can be expressed in terms of the mass densities
of the constituents and the porosity (all of which can be measured or closely
estimated),
Pbw = fpw + (1 - f)ps ,
= ps - f(ps - Pw), (7.1)
where fis the porosity, pw the mass density ofthe pore water, ps that ofthe solids. It
follows, therefore, that where the porosity is increasing with depth in the transition
zone to an undercompacted shale, so the bulk wet density ofthe shale decreases. If a
functional relationship can be found between depth, porosity, and pore-water
pressure, then the expression above can be modified to take these parameters into
account. Athy's formula (equation (6.2)) can be generalized by the introduction of
equilibrium depth (the parameter £5), and using this as the most unfavourable curve
for comparative purposes, we may substitute
f = fo e-c/)z (7.2)
into equation (7.1), obtaining
Pbw = ps - fo e-c/)z (Ps - Pw). (7.3)

This relationship is plotted in Figure 7-6 for ps = 2,650 kg m - 3, fo = 0.48,

c = 1.42 x 1O-3 m- l , Pw = 1,050 kg m- 3 ; and from it we see that if the mean
overburden bulk wet density is 2,300 kg m - 3, density inversions can exist in
abnormally-pressured shales down to depths of at least 2,000 m.
 159

BULK WET DENSITY, P b

2000 kg m- 3 2500

--
C ")
Q

- x

::I:
1 1

I-
0..
W 5
C

2 2
ft m
8 0 0.2 0.4 0.6 0.8 1.0
A,1 0.9 0.8 0.7 0.6 Ae
Fig. 7-6. Idealized relationship between depth, pore-fluid pressure (through the parameters c> and A) and
bulk wet density of a sediment.

From a practical point of view, there are several borehole logs that yield a
measure of bulk density ofthe rocks in the wall of the borehole (one problem with
these is that the correction factor to be applied is often a significant proportion of
the number obtained). Nevertheless, density inversions are revealed by such logs
when run in boreholes with abnormal pore pressures in the shales, and they can be
used for estimating the values of fo and c in equation (7.3). (There is a linear
relationship between bulk density and porosity; plot the natural logarithm of
porosity against depth and the slope gives the value of c, and the value of fo is
obtained by extrapolating the straight-line part to z = 0.)
The rate of deformation of an unstable sequence, and the geometry of the
deformation, depend on the density contrast between the two layers, their viscosity
contrast, and the thicknesses. It is found that the earliest deformation of the
interface between the two layers is of short wavelength and low amplitude, but, as
160

the deformation proceeds, so one wavelength is amplified at the expense of the

others. This is known as the dominant wavelength (see Biot and Ode, 1965). It is a
function of the viscosity ratio and the thickness ratio of the upper and lower layers.
The higher the viscosity of the overburden relative to that of the lower layer, the
longer the dominant wavelength and the slower the rate of deformation. Biot and
Ode found that with overburden/mother-layer thickness ratios between 1 and 5,
there was a significant rate of growth with a dominant wavelength about 10 to 20
times the thickness of the mother layer.
The main physical ingredients of gravitational instability - density inversion and
relatively low viscosity - are present in young regressive sequences with abnormally-
pressured shales. Also, the structural style so commonly observed - broad gentle
synclines separated by narrow steep anticlines - is in harmony with the development
of a dominant wavelength, as revealed by mathematical and physical models with
unstable sequences. Thus the argument for vertical rather than horizontal deform-
ing forces is supported by a rigorous qualitative argument around the field data of
growth structures; and the key to this argument is the role of pore water in
influencing the physical properties of the sediments. The problems of quantifying
the argument have not yet been solved. These difficulties relate largely to the
dimension of time, and the manner in which the physical dimensions and properties
of a three-dimensional geological space change with time.
With the passage of time, subsidence and the accumulation of sediment change
the geometry of the rock units seen as physical units; and the expulsion of pore water
changes both the densities and the viscosities of the rocks with time. These variables
are not amenable as yet to analysis. But it is worthwhile noting that these processes
have a general tendency to reduce the degree of mechanical instability with time,
and sooner or later the sequence will become stable. When or ifthe main shale wedge
has reached compaction equilibrium, it will be more dense than the overburden.
These tendencies may account for the observation that salt domes are much more
common than true shale diapirs, because the dimension oftime has little or no effect
on the physical properties of salt.
Finally, it will occur to the reader that certain 'open-ended' stratigraphic se-
quences will suffer a deformation due to the load on a relatively plastic mass of shale.
Such a situation, depicted in Figure 7-7, is relevant to the U.S. Gulf Coast and many
of the major deltas. The load will tend to squeeze the shale out laterally, generating a
bulk flow.
Dailly (1976) has presented a cogent argument for this occurrence in the Niger
delta; and it is clear that such outward flow may well contribute to the development
of normal growth faults, and that sliding in not the only possibility. On a small scale,
processes such as these have been studied in the Mississippi delta, where the
generation of ,mud lumps' has been attributed to an overburden load ofless than 100
m thickness. Folding and overthrusting were found in core-holes (Morgan et aI.,
1968; but see also Lyell, 1867, pp. 447-454 for what is probably the earliest inter-
pretation of mudlumps).
 161

Fig. 7-7. Schematic bulk movement of abnormally-pressured shale as a result of overburden load in a
delta.

OROGENIC DEFORMATION*

If orogeny of the sedimentary basin itself begins before compaction equilibrium has
been achieved, further deformation may take place in which the pore water plays an
important role. It is evident that this will be superimposed on any pre-orogenic
deformation, and that this superimposition may be a decisive factor in the style of
orogenic deformation. Most obviously, the creation of slopes may lead to sliding
(and equally obviously, there may be no hard and fast line to be drawn in time
between slopes created by basin subsidence and those created by the onset of
orogeny).
The previous argument included the matter of relatively low viscosity in the
abnormally-pressured shales: such shales will act as a lubricant for the sliding of a
regressive sequence, or any sequence that has the same physical characteristics.

Lubricated sliding

When a block is placed on a layer of material of uniform lower viscosity, and the
whole tilted, deformation of the less viscous underlying layer results in movement of
the block down the slope. For the same block on the same slope, the lower the
viscosity of the lubricant, the more rapid the movement. For the same block on the
same lubricant on the same slope, the thicker the lubricant, the faster the movement.
The steeper the slope, of course, other things being the same, the more rapid the
movement.
Kehle (1970) examine lubricated geological sliding in some detail in an interesting
paper, and concluded that such a mechanism could easily account for most cases of
* We use the term 'orogeny' in its original sense of 'mountain-building', not the modern variant
connected with the emplacement of granitic plutons.
162

tectonic translation. In spite of (or perhaps, because of) the simplifying assump-
tions, the computed velocities of sliding were far in excess of those required to
account for observed translations. We shall look first at some field examples of
deformation that have been attributed to sliding.
In western Papua New Guinea (Fig. 7-8), the Mesozoic and Tertiary sequence of
the Southern Highlands basin, south of the Kubor Range, appears to have slid
southwards as the Kubor Range was elevated, with deformation taking the form of
thrusting and overthrusting, and steep, asymmetric folding (Jenkins, 1974; Findlay,
1974; Ridd, 1976). This remote area has been mapped by the Australian Bureau of
Mineral Resources and BP geologists and geophysicists. There is also data from a
number of exploratory boreholes drilled for petroleum.
The stratigraphic sequence is summarized in Figure 7-9, the emphasis being on
the lithologies because these are the materials that are involved in the sliding. The
Southern Highlands basin sequence begins with a transgressive Jurassic sequence of
sands and mudstones (the Maril Shale) on granitic basement, followed by lower
Cretaceous mudstones, siltstones (Kerabi Formation) and upper Cretaceous mud-
stones (Chim Formation). Unconformably above these are Paleocene mudstones
and sandstones, Eocene limestone (deep-water micrites), and a lower Miocene
shallow-water shelf limestone in the southern part, deep-water greywackes and
mudstones in the northern part. In the thickest part of the basin, the Eocene lime-
stone, which is about 300 m thick, overlies a sedimentary sequence at least 7,000 m
thick that is composed largely of mudstone - in particular, the Chim Formation
with about 3,000 m of mudstone, and the Maril Shale with about 1,500 m.
Mapping has revealed a deformation that can be conveniently divided into four
zones, each with a characteristic style (Fig. 7-8). The northern zone (Zone IV of
Jenkins) consists of a broad syncline on the southern slopes of the Kubor Range,
with little obvious deformation. The next zone to the south, Zone III, is charac-
terized by broad gentle synclines separated by narrow steep anticlines with sinuous
trends. Some anticlines are overturned and overthrust to the south or south-west,
and Jenkins notes a diapiric character in places. The next is a narrow zone (17 km) of
imbricate thrusting in the Miocene limestones. The southernmost zone consists of
large folds overthrust to the south or south-west over the uplifted Muller Range.
While this structural configuration may suggest sliding to the surface geologist,
more positive evidence has been obtained from boreholes drilled for petroleum.
Several cross-section through structures that have been drilled can be found in
Jenkins (1974) and Ridd (1976). We take as an example the Puri anticline (Austra-
lasian Petroleum Co., 1961, p. 121, fig. 20), in the southernmost zone (I).
The well penetrated 616 m (2,022 ft) of Eocene and middle Miocene limestones
(Fig. 7-10), then 1,646 m (5,403 ft) of Lower Cretaceous mudstone and siltstone
before passing into Eocene and middle Miocene limestones again. The well termi-
nated at 3,078 m (10,100 ft) in Lower Cretaceous mudstones. The overthrust was
found by two side-tracked holes to be of very low angle of dip. The Mananda
 144° E 145°

PAPUA NEW GUINEA

B
-t-V<9, 60 S
0-9 194.

7°
seA L E
100 KILOMETRES
~~~~~~~~~~,
~
LOCALITY MAP

GULF OF PAPUA

Fig, 7-8a. Structural map of western Papua New Guinea. (Courtesy of the British Petroleum Co., Ltd,) Structural Zone I,
0'\
diagonal ornament: Zone II, plain (north ofI): Zone III, vertical: Zone IV, horizontal ornament. l;.)
 0'1
~
-

r-- Structural Zone I--+-Zonen~·~I·~---- Zone m -I- Zone N - - - - - - - -

METRES
S,OOO METRES

2,saO 5,000
o - 2,500
'" \. \. '\" "\ -~ ''\: -
2 sao -:;:::::::::--- SEA LEVEL
, .!I v--::::-"'. ~ 2 500
5,000 "':"
" " " , , " ',','" ",-,- -=----<:'
- -""""':' '
7,SOO ' ",' ' , ' , ' ,'," ~;",
' ,,', " ',',' 5,000
~,, " " , " " 7,500

Fig. 7-8b. Section B-B of Figure 7-8a.

 165

Shelf Lst. in S
Mi Gwke. and Mdst. in N

E Deep-water micrites

Pc Mdst./Sst.

CHIM Fm.
KU Mdst., Sltst.

KERABI Fm.
Mdst., Sltst.

J MARIL SHALE

BASEMENT

Fig. 7-9. Generalized stratigraphic sequence in area of sliding.

Cecilia, and Libano anticlines have very similar structural styles, also in Zone I but
about 250 km to the west-north-west.
Uplift perhaps of the order of 10,000 m is indicated for the Kubor Range during
the Plio-Pleistocene. The range stands now over 4,000 m above sea-level, with a
Miocene turbidite sequence in outcrop both to the south and the north. A smaller
uplift of 4,000 to 5,000 m is indicated for the Muller Range to the west, the axis of
which lies about 130 km to the south-west of the Kubor Range. Between these two
uplifts, basement reaches a depth of about 6,000 m, and this basement to the basin
changes elevation by approximately 11 km over a distance of nearly 50 km - a slope
of 12° to 13° - up to the Kubor Range on the north side of the basin. It is therefore
natural that this large area of structural complexity should be interpreted as due to
sliding towards the south down the slope as it was created. Jenkins (1974) suggested
two surfaces of detachment (sliding), the deeper one in the upper Jurassic sequence
leading to the thrusts of the southernmost structural zone, the other within the
Cretaceous sequence leading to the thrusts of the other three northern zones.
Findlay (1974), who studied an area adjacent to the east of Jenkins', postulated a
single detachment within the Jurassic sequence.
There is little data concerning the pore-fluid pressures in the sequence; and it is
 N 0-,
0-,
Purl No. I IA a IB

lit>
t>.,.
t,ooo
.., ; 0

4,000 c • ". /-4,000

tl,000 ',000

1,000 "", a,ooo

10,000
" 10,000

12,000
" "
,,~_:l!tS~ MESOZOIC
" ,
14,000 14,000

11,000 11,000

11,000
,"",~",I""\'-8ASE:MENT ~..!;~\I
_\ . . -, . . L~,.:-\- _~1:::- ":...'I-"j' . . I- ,- -: L'IPOO
__ ,,_\_"'\. . , ',. . _.!." . . I/~ .... \ - I \ - . , . . ....-l_\I\
,,',,1,"\-/,_ . . ·.. I_'_'-~\-'I'-
_;; ,\;'-:";/~:\" ICAlE I~
20,000 ,,_ ,_ ~_\ . . _\,... ¥ I
\ r" I , ,,'
J .... \ / _ , , _" *" .:,
'I .................. \ I f MILES
1\' -
2 KilOMETRE

Fig. 7-10. Cross-section through the Puri anticline (depth in feet). (Courtesy of the British Petroleum Co., Ltd.)
 167

unwise to infer pressures from the drilling difficulties that were sometimes encoun-
tered, because some of these were related to mudlosses in the limestone sections.
Nevertheless, Cretaceous and Tertiary mudstones in the Aure through, close to the
east, are abnormally pressured (Lepine and White, 1973), and it seems reasonable to
suppose that the thicker mudstones in the Southern Highlands basin are, or were,
abnormally pressured to some extent. And the deformation itself may have led to
higher pore pressures.
An understanding of this sliding must rest on an understanding of the physical
properties ofthe sedimentary sequence, and the slopes generated by subsidence and
uplift. The key properties are those of the dense limestones lying on Mesozoic
mudstones that we regard as lubricants for the sliding. While it may be misleading to
think of detachment occurring on thin surfaces like a fault, there is no difficulty in
regarding the Chim Formation as the upper zone of detachment. As regards the
limestone overburden, there is probably a quantitative distinction to be made
between the northern area with a thin Eocene limestone overlain by a Miocene
turbidite sequence, and the southern area of thicker Miocene limestones. We note
that this facies boundary is also close to the boundary between the diapiric-type
deformation to the north and the imbricate overthrusts to the south. As we have
seen, the Plio-Pleistocene uplift of the Kubor Range produced a slope of 12° or 13°
to the south; but southerly sliding was opposed by the smaller uplift along the
Muller Range axis. The zone of imbricate thrusting is on the northern side of this
axis, and thus appears to be related to fracture-failure of the more competent
limestones between these opposing slopes and their sliding forces. The southern-
most structural zone contains relatively undeformed limestones around the Muller
Range axis (Zone I of Jenkins).
If a block that is relatively rigid rests on a layer oflubricant, and there is no barrier
to sliding, the rate of sliding is a function of the weight of the block, the angle of
slope, the equivalent viscosity of the lubricant and its thickness. If all the dimensions
and parameters remain constant, there will be no deforming stresses within the
block, and the block will slide without folding or faulting. If, however, there is an
obstacle to sliding, such as a reversal of slope, then the forces opposing sliding create
a stress field within the block in which the component of lateral compression along
the block increases in the direction of dip. When this stress exceeds the strength of
the material, a thrust will develop, and the block will slide.
Such intuitive approaches to the problem of sliding lead to no great difficulties
when there is an obvious barrier to sliding: but in the case under discussion, there is a
more subtle influence. The weight of the block is its weight in the ambient fluid -air
if it is subaerial, water if submarine. Plio-Pleistocene uplift has elevated most of the
Eocene and Miocene marine limestones from below sea-level to above sea-level in
the area of structural complexity. The bulk weight of water-saturated limestone
above sea-level is about 25,000 N m- 3 , but only about 15,000 N m- 3 below sea-
level (2,500 and 1,500 kg weight per cuhic metre). So a uniform block oflimestone
168

resting on a uniform layer oflubricant, inclined so that the block is above sea-level at
one end and below it at the other, does not have an uniform sliding potential.
Unobstructed, a subaerial block will slide faster than an identical submarine block
on the same slope and lubricant, because the subaerial driving force is greater than
the submarine.
There is little difficulty in understanding the 'diapiric' structural zone quali-
tatively, with broad synclines separated by narrow anticlines or narrow belts of
disturbance. This has the same character as the pre-orogenic deformation discussed
earlier in this chapter, and a density inversion almost certainly exists between the
deep-water micrites and underlying Chim Formation. The writer had this type of
deformation in mind when he postulated that the dominant wavelength of diapirism
could determine the length of thrust sheets (in the direction of movement) in
unstable sedimentary sequences, the trend of such diapiric anticlines being parallel
to the depositional strike, normal to the slope (Chapman, 1974). Each line of
diapiric anticline forms a line of weakness, and any down-slope resistance to sliding
will tend to deform the diapiric anticline into asymmetry and overthrusting.
The zone of imbricate thrusting south of the 'diapiric' zone may also have had a
diapiric influence; but its position near the foot of the subaerial slope and the
opposing slope of the Muller Range axis suggests that this deformation is due to
fracture-failure in an area of high lateral compressive stress.
The sliding of the Tertiary, and perhaps Mesozoic, section in the Southern
Highlands basin is considered to be an example of orogenic deformation that is
gravity-induced, and, at least in part, superimposed on pre-orogenic deformation
due to mechanical instability in the sequence. In these processes, the pore water
almost certainly had the effect of reducing the equivalent viscosity of the thicker
mudstones. The mudstones were evidently of sufficiently low viscosity for diapiric
anticlines to form; and also the strain induced in the mudstones would have the
effect of reducing bulk volume, or increasing the pore pressure if the pore water
could not escape fast enough.
The role of sea-level in the style of deformation is a teasing problem of some
interest. In the East Borneo basin of Kalimantan, in the region of the Mahakam delta
(Fig. 7-11), the structural style onshore differs significantly from that found off-
shore. Onshore, the style is very similar to that of western Papua New Guinea, with
long, sinuous, steep anticlines separated by broad, gentle synclines. The anticlinal
trends tend to be asymmetrical, near the coast towards the coast, some with thrusts
in the core. These anticlines have long been regarded as due to lateral compression:
they formed an important part of van Bemmelen's gravitational tectogenesis (van
Bemmelen, 1949, pp. 352, 732). However, drilling in the Mahakam delta and off-
shore from it has revealed growth anticlines with normal growth faults in sediments
of similar (but rather younger, perhaps) age as those deformed onshore, from which
we infer a stress field with a component of horizontal tension (Chapman, 1977a).
The stratigraphic sequence is regressive, with a thick shale that is abnormally
 169

OATTAKA

l BEKAPAI
0
,
0 km. ,
25

Fig. 7-11. Sketch map of Mahakam delta, Kalimantan, Indonesia (after Chapman, 1977a, fig. 5).

pressured (at very shallow depth in the cores of the anticlines, generally). When a
map has been published of the contours on the top of abnormal pressures, we shall
perhaps be able to understand this paradox. It seems likely that the onshore
anticlines are diapiric (see Weeda, 1958, for example), and have been squeezed by
the sliding of the sandy overburden on abnormally-pressured shales against the
resistance offered by the submarine part of the sequence down-slope. Conversely, in
'the submarine part, if there is an increasing sliding potential seawards, the over-
burden may have a component oflateral tension.
In New Zealand, in the East Coast basin of North Island, Upper Cretaceous,
Eocene, and Miocene sedimentary rocks are found with great structural complexity
of the same general type as that found in Kalimantan and western Papua New
Guinea - broad gentle synclines separated by steep narrow anticlines, with thrusts,
overfolds, transcurrent faults, and mudvolcanoes (Ridd, 1970; Laing, 1972). Ridd
attributed this structural style to sliding on the abnormally-pressured shales of early
170

Tertiary age that have been encountered in boreholes drilled for petroleum. Ab-
normally-pressured shales occur as shallow as 356 m (1,168 ft) in Rotokautuku 1
and the well had to be abandoned at 627 m (2,057 ft) on account of the difficulties
(Katz, 1974, p. 469). It will be interesting to see if the offshore structural style is
similar.
Finally, it must be noted that sliding is not necessarily towards the sea: its
direction depends on the slopes generated. In the North Coast basin of West Irian,
under the Mamberamo delta, drilling has not only revealed very shallow abnor-
mally-pressured shales in a regressive sequence, but also that the basement is very
shallow near the present-day coastline (Visser and Hermes, 1962, Enclosure I-III).
Here it appears that the tendency is to slide towards the land; but there is too little
data to pursure this further.
Mechanical problems of such complexity cannot reliably be understood by in-
tuitive argument, so we shall seek a better insight into the role of pore water in such
deformation by more rigorous mechanical analysis (although necessarily sim-
plified ).

SELECTED BIBLIOGRAPHY

Australasian Petroleum Co. Pty. Ltd., 1961. Geological results of petroleum exploration in western
Papua 1937-1961. J. Geol. Soc. Australia, 8 (1): 1-133.
Barton, D .c., 1933. Mechanics of formation of salt domes with special reference to Gulf Coast salt domes
of Texas and Louisiana. Bull. American Ass. Petroleum Geologists, 17 (9): 1025-1083.
Biot, M.A., and Ode, H., 1965. Theory of gravity instability with variable overburden and compaction.
Geophysics, 30: 213-227.
Bishop, R.S., 1978. Mechanism for emplacement of piercement diapirs. Bull. American Ass. Petroleum
Geologists, 62 (9): 1561- 1583.
Bottcher, H., 1925. Die Tektonik der Bochumer Mulde zwischen Dortmund und Bochum und das
Problem der westfiilischen Karbonfaltung. Gluckauj- Berg- und Huttenmannische Zeitschrift, 61:
1145- 1153, 1189-1194.
Bottcher, H., 1927. Faltungsformen und primare Diskordanzen im niederrheinisch-westfiilischen Stein-
kohlengebirge. GlUckauj- Berg- und Huttenmannische ZeitschriJt, 63: 113-121.
Bruce, C.H., 1973. Pressured shale and related sediment deformation: mechanism for development of
regional contemporaneous faults. Bull. American Ass. Petroleum Geologists, 57 (5): 878-886.
Chapman, R.E., 1973. Petroleum geology: a concise study. Elsevier Scientific Pub!. Co., Amsterdam,
London, and New York, 304 pp.
Chapman, R.E., 1974. Clay diapirism and overthrust faulting. Bull. Geol. Soc. America, 85 (10): 1597-
1602.
Chapman, R.E., 1976. Petroleum geology: a concise study (paperback edition). Elsevier Scientific Pub!.
Co., Amsterdam, Oxford, and New York, 302 pp.
Chapman, R.E., 1977a. Subsidence and deformation of terminal regressive sequences in the Indonesian
region. Proc. Indonesian Petroleum Ass., 5 (1) (for 1976): 151-158.
Chapman, R.E., 1977b. Petroleum geology of young regressive sequences. Proc. S.E. Asia Petroleum
Exploration Soc., 3 (for 1976): 8-38.
Crans, W., Mandl, G., and Haremboure, J., 1980. On the theory of growth faulting: a geomechanical
delta model based on gravity sliding. J. Petroleum Geology, 2 (3): 265-307.
Dailly, G.C., 1976. A possible mechanism relating progradation, growth faulting, clay diapirism and
overthrusting in a regressive sequence of sediments. Bull. Canadian Petroleum Geology, 24 (1): 92-116.
 171

DaneS, Z.F., 1964. Mathematical formulation of salt-dome dynamics. Geophysics, 29 (3): 414-424.
Dennis, J.G., 1967. International tectonic dictionary English terminology. Memoir American Ass.
Petroleum Geologists, 7 (196 pp.).
Dickinson, G., 1953. Geological aspects of abnormal reservoir pressures in Gulf Coast Louisiana. Bull.
American Ass. Petroleum Geologists, 37 (2): 410-432.
Dron, R.W., 1900. The probable duration of the Scottish coalfields. Trans. Instn Mining Engineers, 18:
194-211. (Discussion: 211-212.)
Findlay, A.L., 1974. The structure of foothills south of the Kubor Range, Papua New Guinea. J.
Australian Petroleum Exploration Ass., 14 (1): 14-20.
Fowler, W.A., Boyd, W.A., Marshall, S.W., and Myers, R.L., 1971. Abnormal pressures in Midland
Field, Louisiana. In: Houston Geol. Soc., Abnormal subsurface pressure: a Study Group report
1969-1971. Houston Geol. Soc., Houston, pp. 48-77.
Gansser, A., 1960. Ober Schlammvulkane und Salzdome. Vierteljahrsschrift der Naturforschenden
Gesellschaft in Zurich, 105: 1-46.
Harkins, K.L., and Baugher, J.W., 1969. Geological significance of abnormal formation pressures. J.
Petroleum Technology, 21: 961-966.
Houston Geological Society, 1971. Abnormal subsurface pressure: a Study Group report 1969-1971.
Houston Geol. Soc., Houston, 92 pp.
Hubbert, M.K., 1951. Mechanical basis for certain familiar geologic structures. Bull. Geol. Soc. America,
62: 355-372.
Hubbert, M.K., 1972. Structural geology. Hafner, New York, 329 pp. (Contains Hubbert's papers on this
theme.)
Hubbert, M.K., and Rubey, 1959. Role of fluid pressure in mechanics of overthrust faulting, I. Mech-
anics of fluid-filled porous solids and its application to overthrust faulting. Bull. Geol. Soc. America, 70
(2): 115-166.
Hubbert, M.K., and Willis, D.G., 1957. Mechanics of hydraulic fracturing. Trans. American Inst. Mining
Metallurgical Petroleum Engineers, 210: 153-166. (Discussion: 167-168.)
Jenkins, D.A.L., 1974. Detachment tectonics in western Papua New Guinea. Bull. Geol. Soc. America, 85
(4): 533-548.
Katz, H.R., 1974. Recent exploration for oil and gas. In: GJ. Williams (Ed.), Economic geology of New
Zealand. Monograph Series Australasian Inst. Mining Metallurgy, 4: 463-480.
Kehle, R.O., 1970. Analysis of gravity sliding and orogenic translation. Bull. Geol. Soc. America, 81 (6):
1641-1664.
Laing, A.C.M., 1972. Geology and petroleum prospects of Ruatoria area, east coast, North Island, New
Zealand. J. Australian Petroleum Exploration Ass., 12 (1): 45-52.
Lepine, F.H., and White, J.A.W., 1973. Drilling in overpressured formations in Australia and Papua
New Guinea. J. Australian Petroleum Exploration Ass., 13 (1): 157-161.
Liechti, P., Roe, F.W., and Haile, N.S., 1960. The geology ofSarawak, Brunei and the western part of
North Borneo. Bull. Geol. Surv. Dept. British Territories Borneo, 3.
Lyell, C., 1867. Principles of Geology or the modern changes of the Earth and its inhabitants (10th edition)
volume 1. John Murray, London, 671 pp.
Morgan, J.P., Coleman, J.M., and Cagliano, S.M., 1968. Mudlumps: diapiric structures in Mississippi
delta sediments. In: J. Braunstein and G.D. O'Brien (Eds), Diapirs and diapirism. Memoir American
Ass. Petroleum Geologists, 8: 145-161.
Murray, G.E., 1961. Geology of the Atlantic and Gulf Coastal province of North America. Harper &
Brothers, New York, 692 pp.
Parker, TJ., and McDowell, A.N., 1955. Model studies of salt-dome tectonics. Bull. American Ass.
Petroleum Geologists, 39 (12): 2384-2470.
Ramberg, H., 1967. Gravity, deformation and the Earth's crust as studied by centrifuged models. Academic
Press, London and New York, 214 pp.
Ramberg, H., 1972. Theoretical models of density stratification and diapirism in the Earth. J. Geophys-
ical Research, 77 (5): 877-889.
Ridd,M.F., 1970. Mud volcanoes in New Zealand. Bull. American Ass. Petroleum Geologists, 54 (4): 601-
616.
Ridd, M.F., 1976. Papuan basin - on-shore. In: R.B. Leslie, HJ. Evans, and C.L. Knight (Eds.),
172

Economic geology of Australia and Papua New Guinea - 3. Petroleum. Monograph Series Austra-
lasian Inst. Mining Metallurgy, 7: 478-494.
Schaub, H.P., and Jackson, A., 1958. The northwestern oil basin of Borneo. In: L.G. Weeks (Ed.),
Habitat of oil. American Ass. Petroleum Geologists, Tulsa, pp. 1330-1336.
Selig, F., 1965. A theoretical prediction of salt dome patterns. Geophysics, 30: 633-643.
Stutzer,O., 1930. Absinken, Sedimentation und Faltung - gleichzeitige vorgange in manchen Erdol-
gebieten (Abstract). Geologische Rundschau, 21: 141.
Tanner, W.F., and Williams, G.K., 1968. Model diapirs, plasticity, and tension. In: J. Braunstein and
G.D. O'Brien (Eds), Diapirism and diapirs. Memoir American Ass. Petroleum Geologists, 8: 10-15.
Thorsen, C.E., 1963. Age of growth faulting in southeast Louisiana. Trans. Gulf-Coast Ass. Geol. Socs,
13: 103-110.
Tiddeman, R.H., 1890. On concurrent faulting and deposit in Carboniferous times in Craven, Yorkshire,
with a note on Carboniferous reefs. Report British Ass. Advancement Science (Newcastle-upon-Tyne,
1889), pp. 600-603.
Van Bemmelen, R.W., 1949. The geology ofIndonesia. Vol. 1A General geology ofIndonesia and adjacent
archipelagoes. Government Printing Office, The Hague, 732 pp.
Visser, W.A., and Hermes, J.J., 1962. Geological results of the exploration for oil in Netherlands New
Guinea. Verhandelingen van het Koninklifk Nederlandsch Genootschap voor Nederland en Kolonien,
Geologische Serie, 20: 1-265.
Weeda,J., 1958.0il basin of East Borneo. In: L.G. Weeks (Ed.), Habitat ofoil. American Ass. Petroleum
Geologists, Tulsa, pp. 1337-1346.
 8. PORE WATER AND SLIDING

Large-scale sliding of geological sequences, with little internal deformation, is not a

new idea. During the second half of the 19th Century, as the geology of the Alps,
north-west Scotland, and Scandinavia was being unravelled, evidence emerged of
lateral displacements of blocks many tens of kilometres long* in the direction of
movement. For example, T6rnebohm (1896, p.194) postulated movement of blocks
at least 130 km long on Caledonian thrusts. The main difficulty in these ideas was in
understanding the mechanics. The paradox was this: the strength of the rock limits
the length of the block that can be pushed along a horizontal surface because, if the
force applied to the end exceeds the strength of the material, the block will fail by
internal shear at the end being pushed. The strength of rocks is quite inadequate to
support the push required to move blocks longer than a few kilometres. On the other
hand, if the block slides down a slope under the force of gravity, the previous
difficulty is replaced by two others: the coefficient of friction of rock on rock
suggests that an angle of about 30° would be required for gravitational sliding - and
that also implies a vertical relief of about half the length of the block. The restric-
tions on relieflimit the length of blocks that slide under gravity to a few kilometres.
Smoluchowski (1909) calculated that the strength of granite was far less than that
needed for a 160-km (lOO-mile) block to be pushed, but continued
'Suppose a layer of plastic material, say pitch, interposed between the block and the underlying bed; or
suppose the bed to be composed of such material: then the law of viscous liquid friction will come into
play, instead of the friction of solids; therefore any force, however small, will succeed in moving the
block. Its velocity may be small ifthe plasticity is small, but in geology we have plenty of time ; there is no
hurry.'.

More and more evidence of large-scale translations was collected from different
parts of the world during the first half of the 20th Century. It is probably true to say
that the consensus of opinion was for pushing these large blocks or sheets along
overthrusts rather than sliding them down slopes, although Reyer (1888) had
suggested sliding as a cause of folding (Fig. 8-1). Indeed, evidence of uphill move-

* The conventional terminology of thrust sheets, long established, is length along the mapped thrust,
breadth or width at right angles to the mapped thrust. When discussing the movement of such sheets, it is a
needless distraction to have to remember that breadth or width is the length in the direction of movement
(we do not talk of a motor car's being 2 m long and 4 m wide). So throughout this chapter, length means
length in the direction of movement.
174

652

654

Fig. 8-1. Reyer's concept of sliding and folding (Reyer, 1888, figs. 652, 653, and 654).

ment in some areas, such as the Jura, apparently precluded gravitational sliding.
The problems of gravitational sliding lay not so much in the evidence as the angle
throught to be required.
The paradox appeared to be resolved by the fluid-pressure hypothesis of Hubbert
and Rubey (1959), which was an extension of Terzaghi's hypothesis for landslides
(Terzaghi, 1943, p. 235; 1950). In a theoretical analysis supported by experiment,
they argued that pore pressure can relieve the effective normal stress to such an
extent that longer blocks could be pushed without internal failure, and that blocks
could slide down slopes much gentler than the angle indicated by the coefficient of
friction.
This hypothesis has been widely accepted. However, it has usually been used out
of context. The main part of Hubbert and Rubey's paper was concerned with
subaerial sliding of water-saturated blocks; and in their application of the hypoth-
esis to the overthrusts of western Wyoming (Rubey and Hubbert, 1959) they found
it inadequate to account quantitatively for the inferred sliding, and so they con-
cluded that the process of effective-stress relief assisted the sliding.
The difficulty in their hypothesis lies in the fact that their process is much more
efficient subaerially than under water (by a factor of about two) and many slides
appear to have been submarine. Indeed, it is hard to believe that blocks about 6 km
thick in western Wyoming slid wholly subaerially, especially if they were sliding
 175

down the flank of a geosyncline (Rubey and Hubbert, 1959, p. 194). The less
favourable mechanics of submarine sliding reduces the contribution of effective-
stress relief through pore pressure. Moreover, almost all the sediments known to
have pore pressures high enough for low-angle sliding are below sea-level. The
paradox seemded to be re-appearing. Chapman (1979) discussed these difficulties at
some length, and concluded that catastrophic slides, such as the Grand Banks slide
reported by Heezen and Drake (1964), could have occurred by the Terzaghi/
Hubbert and Rubey process.
Greater promise of resolving the paradox lies in the point made by Smoluchowski
(1909): lubrication. The most widespread class of abnormally-pressured sediments
is the thick shale section of young regressive sequences, commonly at depths of 1 to 3
km below sea-level. These shales have the properties of a lubricant in geological time
-low equivalent viscosity and adequate thickness.
In this chapter we shall take a semi-quantitative approach to sliding, using
idealized models, in order to seek an insight into the processes, and the role of pore
water. We shall apply these to the area of Papua New Guinea described at the end of
the previous chapter.

GRAVITATIONAL SLIDING

Unlubricated

When a rectangular block ofthickness h is placed on a surface so inclined that it does

not slide (Fig. 8-2), its weight in the ambient fluid (water or air), per unit area of the
base,

Fig. 8-2.
 176

(8.1 )
can be resolved into a component normal to the surface

O"n = (Pb - Pa) gh cos B (8.1a)

and a shear component

T = (Pb - Pa) gh sin B. (8.1 b)

These may be regarded as active stresses; and they give rise to reactive stresses - a
normal reaction, and frictional resistance (T) that prevents sliding. As the angle of
slope Bis increased, so the shear component of stress increases; and when this stress
comes to exceed the frictional resistance, the block will slide. We shall assume that in
the geological context, sliding will begin slowly when these two stresses are equal.
The Mohr-Coulomb criterion for simple unlubricated sliding (see Hubbert, 1951,
p. 363; or any text-book on mechanics) is

T = To + O"n tan C/>, (8.2)

where To is the cohesive strength, or initial shear strength, of the materials at the
surface that will become the sliding surface, when the normal stress O"n is zero; tan c/>
is the coefficient of sliding friction, c/> commonly being about 30° for sedimentary
rocks. Hence, sliding will take place when

T=T,
(Pb - Pa)ghsin B = TO + (Pb - Pa)ghcos Btanc/>
(from equations (8.1a), (8.1b), (8.2)), or

TO
tan B = ( ) h B + tan. c/>. (8.3)
Pb - p. 9 cos
This equation indicates that sliding will usually take place at an angle B rather
greater than c/>.It was Terzaghi (1943,1950) who showed that for porous materials,
the total normal stress is the sum ofthe effective stress and the pore-fluid pressure, S
= 0" + p, and that if (S - p) is substituted for O"n in equation (8.2),

T = TO + (S - p)tanc/>, (8.4)

pore-fluid pressure in a subaerial water-saturated block is seen to reduce the effect-

ive normal stress so that sliding may take place on a slope B less than c/> if To is
sufficiently small.
Hubbert and Rubey (1959) argued that To is negligible in the geological context,
and, defining a parameter A = piS (the l;'roportion of the total load supported by
pore-fluid pressure) equated T = Stan B with T = (1 - A) Stan C/>, and so came to
the expression

tan B = (1 - A) tan c/> (8.5)

 177

for subaerial water-saturated blocks that may slide under gravity. From equation
(8.5) it can be seen that as p ---+S, so the angle at which sliding can occur approaches
zero. Numerous measured values of A = 0.9 indicate that sliding could take place on
slopes as gentle as 3!O .
The difference between subaerial and submarine sliding relates to the effect of the
ambient fluid: the subaerial block in toto receives negligible support due to buoy-
ancy in air, but the submarine block receives considerable support due to buoyancy
in water. Chapman (1979) derived the more general expression

+ t5tan4J,
II '0
tanu = ( ) h f) (8.6)
Ph - p. 9 cos

where t5 = (1 - A)/(1 - Ae), wherd eis the proportion ofthetotalload supported by

the ambient fluid pressure. Ae is sensibly zero when the ambient fluid is air, and
equation (8.6) reduces to equation (8.5) when '0 is negligibly small. In the submarine
environment, Ae has the approximate value of 0.5 ; so the critical angle for submarine
sliding is about twice that for subaerial sliding of an otherwise identical block. A
water-satmated subaerial block sliding slowly down its critical angle of slope cannot
slide far into the sea: the resistance to sliding remains the same, but the driving force
is reduced from Ph gh sin f) to (Ph - P.) gh sin f) (Fig. 8-3). If the initial shear strength
cannot be ignored, the critical angle for sliding is increased, as Hsii (1969) pointed
out. Table 8-1 shows the subaerial and submarine slopes required for blocks of
various thicknesses when the initial shear strength is 3 MPa, and the slopes for zero
initial shear strength, which are independent of thickness.
The fluid-pressure hypothesis seems inadequate as a general solution to the
paradox of overthrust faulting and sliding, but it is entirely adequate as an expla-
SEA LEVEL

S.L.

Fig. 8-3. When a sliding block enters the sea, the driving force is reduced but the frictional resistance
remains the same.
178

Table 8-1. Critical angle of unlubricated gravitational sliding (degrees) for identical water-saturated
subaerial and submarine blocks (italics). ro = 3 x 106 Pa (30 bar).

Subaerial ;. 0.5 0.6 0.7 0.8 0.9 1.0

Submarine b 1.0 0.8 0.6 0.4 0.2 0.0

hkm
1 23 20 17 14 II 8
42 37 32 26 20 14

2 20 17 14 10 7 4
36 31 26 20 13 7
3 19 15 12 9 6 3
34 29 23 17 11 5
4 18 IS 12 8 5 2
33 28 22 16 10 3
5 18 14 II 8 5 2
32 27 22 16 9 3
ro = 0 16 13 10 7 3 0
30 25 19 13 7 0

nation of some slides that have been discovered on the continental margins by
marine seismic surveys. For example, the Grand Banks slide was attributed by
Heezen and Drake (1964) to the 1929 earthquake, which led to a sequence of events
well known to students of geology. The Grand Banks slide (Fig. 8-4) was about 50
km long (in the direction of movement), 400 m thick and lay on a slope of about 10.
We assume Pb = 2,000 kg m -3 and ¢ = 30°. Re-arranging equation (8.6),
To = (tan e - 6 tan ¢ ) (Pb - Pw) g h cos e,
and inserting the above values, we find that TO could not have been much greater
than 7 x 1Q 4 Pa (0.7 bar), nor 6 much greater than 0.03. In other words, the material
at the sliding surface had very little cohesive strength, and the pore water was
bearing almost the total overburden at the time of sliding.
These findings support the conclusion of Heezen and Drake. If the sliding surface
were at the top of a poorly sorted, but porous, sand or silt, the seismic shock could
have re-arranged the grains into a more stable packing. Such a change requires
reduction of porosity - but porosity can only be reduced if pore water can be
expelled. Until that happened, the pore water bore the overburden.
It is important to appreciate that the sliding surface is more likely to have been a
sand or silt because good permeability is required for a short-distance slide: the
excess pore pressures must dissipate quickly. Had the sliding surface been a clay, the
slide would probably have travelled further because the excess pore-water pressure
in clays cannot be dissipated quickly.
Chapman (1979) also developed expressions for the critical length of slide-blocks
that are stopped by resistance at the down-slope end (a variation on the theme of
 O~NOAR~T~HttW~CS~TnA~~~~EL------------------------------------------------·~~~~--
~AURENTIAN CHANNEL SOUTHEAST 0

1-
-500
VERTICAL EXAGGERATION IN WATER 22:'

2-

-1000
if)
(f)3- :!!-
~ - 0-
Q - I-
U - ~-1500
L..J -
(f)4- LL. -

-
5- -2000
VERTICAL EXAGGERATION
IN SEDIMENTS ABOUT 15: I
-
6-
-2500

7-, , , 1 1 1 1 1 1

-20 -10 0 10 20 30 40 50 60 70 80 90
NAUTICAL MILES
Telegraph cables broken during earthquake x
Telegraph cable broken 59 minutes later ®

-...l
Fig. 8-4. Tracing of seismic reflection profile made on continental slope southeast of Laurentian Channel (Heezen and Drake, 1964, fig. 1). \0
180

pushing), and concluded that long blocks not only can but must slide on very small
angles of slope. Steeper slopes have much shorter critical lengths and are likely to
lead to chaotic slides and, in the extreme, turbidites. The length and slope of the
Grand Banks slide are in good agreement with those predicted by theory.
This is possibly the mechanism by which turbidity currents and turbidites are
caused. A sheet of sediment resting on a layer of water created by reduction of
porosity in the underlying sand is momentarily lying on a very efficient lubricant.
The expulsion of this lubricant implies a fluid-potential gradient directed towards
the leading and trailing edges (at least; perhaps also laterally). Thus the excess
pressure dissipates first at the margins of the slide. At the up-slope end, sediment will
be stripped from the block. At the down-slope end, if the length is greater than the
critical length (approximated in these conditions by Ie = 3h/2 sin 8), the braking
leading edge will not be able to sustain the shear component of weight of the block,
and the slide will over-ride the leading edge. With material oflittle cohesive strength
(a likely condition following a seismic shock), the block may then disintegrate, and
the slide will continue as a turbidity current.

Lubricated sliding
If the block in Figure 8-2 is placed not on a rigid planar base but on a thickness of
material with finite viscosity 'then the law of viscous liquid friction will come into
play ... therefore any force, however small, will succeed in moving the block'
(Smoluchowski, 1909). Let us examine this.
We must consider extensive sheets because a lubricant under a small block will be
displaced with radial components, and the block will sink into the lubricant *. We
follow Kehle (1970) in treating this as laminar flow down a slope.
An extensive thin sheet of a single liquid in uniform laminar flow down a gentle
slope on a rigid planar surface (Fig. 8-5) can be regarded as obeying Newton's law of
viscosity
T = I]dV/dh, (8.7)

where I] is the dynamic, or absolute, viscosity. The units and dimensions of this
relationship are
T: Pa=Nm- 2 =kgm- 1 s- 2 M L -1 T- 2
dV: m S-l L T- 1
dh: m L
1]: Pa s = N s m- 2 = kg m -1 S-l M L -1 T- 1

* This is indeed a real geological situation, but not the one we are considering. It has been examined
experimentally by Ramberg (1967, p. 133). The sandy sequence of a delta may sink into its underlying
clays or shales, and tend to extrude them. This seems to be happening in the Niger delta (Dailly, 1976).
 181

5I 1~ m /y e a r
!
!
!

T
h

Ny. \\-S. Id a" ,"

,,1 =' " pm fil ' <h
mo,h , "b ", ", 1
equivalent viscosit ,bo ct or m o' ,,"
y 10' 6 Pa s, 2 km " of m = deoe'"
thi ck , on slope of 2,300 kg m -, ,
N ot e th at dy na m 5° .
ic vi ,o o, ity ha s
,,, es , X time. So th e di m en ,ion , of
a \a,gOl "fOSS ov ,, ,, ,, X Je "i lth le
sm al le l stress oV el a ,h or te r tim e ", od ty , 0'
e! a lo ng e! tim e - will ha ve th e ,a m
a plOpcfty u, .,d in e elfec1 a' a
th e co n, tm ct io n
scal of pr op cd y
eduim
Eq vaod
leel
nts.viscosity i, a m
e" ''" '' of th e in te rn
s\ ra in <a t" . In th al f,i et io n 0 f th e lO
e geological co nt cks at very small
ex t, w ith very la e
lg vi sc o, iti " an
velocities of slidi d v" y ,m al l
ng, we assume th
at
(8.8)
whefO h, is th e to
ta l thickness of th
co nt fib ut in g to th e m at er ia l flowing
e ,h ea ' ,t le " at th , an d h. - h is th
e level 0 f in te ", t (F e th ie kn es s
s" os ' i, at a m ax im ig. 8- 5). N ot e th at
um on th e rigid ba th e sh ea '
sa l su fla c e , w be re
at th e to p of th e th e velocity is ze ro
fluid where th e ve , an d Z0 10
locity is greatest.
Eq ua tin g eq ua tio
ns (8.7) an d (8.8),

11~~ = (Pb - pa)g (h*

- h) si nB ,
an d in te gr at in g th
~. 0 when h - 00) we obtain
is w ith «' pe et to
h (o ot in g th at V

V=(pb - po)
~g (h * h - 2
2 (8.9)
h )sm
. B
.
Fi gu re 8-5 ,b oW
' th e th eo re tic al ve
de n, ity 2,300 kg locity pl O fik th ro
m -, , as su m ed eq ug h a ,u ba el ia l ,h
ui va le nt viscosity al e of mass
of 10 " Pa " t hi ck
n' " 2 km ,
182
on a ,lo pe 0 f 5' . K
oble fo un d th at th
th e flow la w u, ,:d o qu an tit at iv e r" ul
, '0 we ,h al l confin t< a<e rd at iv el y i"
m ", t be a, "u m ed o ou =l vO ' to th e ,o n, iti ve to
'" a m at to r of oo ur 'im pk eq ua tio n (8
way a" w m ng : " th at th e de ta il, of .9). Bu t it
tho vd oc ity pm bk pw fi l" oo n, tm etod
oq ui va kn t ,",,"o,it in a "a l ,h al o wi in th i'
y m ay well va<Y ll be di ff "o nt be
,ta nd th e na tu 'e of th N U gh th e "o tio <; au " the
lu bricated ,li di ng n. O ur pu rp o, " i,
vd oc ity ac ro " a , an d th e or de r of m to unde<-
lu br ic at in g be d. ag ni tu de of th e di
Eq ui va le nt ,",,"o ff er en '" of
newtonian viscos ,ity ca n bo re g. ,d
Co n, id er noW aity that would give th
rigid un ifo rm be d e sa m e effect as reality
ed '" tho
th e "m e shale w ith on e ki lo m et re thiok r" .
th e "m o eq ui va le ti ng on on e kilom
nt vi,,"o,ity of 10 etCf of
(Fig. 8-6). W0 'h al " P a' on tho "m e ,lo pe
l a<sumo a co ns ta of 5'
velocity i, relativelY nt ma SS dc u, ity of
in "" "i tiv o to "n ai 2,300 kg m " beca
only pa ra m et or th l ch an g" 0 f mass de ", ity uS C the
at ca n chango th (tho vi,eosity i, th
m ag ni tu de ). Th e ,h e vd oc ity by m uc h e
ea r ,treSS is tho "m m o" th an an or
2-km shale w n, id e at tho to p of th e d" of
e" d earlier, an d th ,h al o a.s in th e mid
th e ,am" th e up p" bed will o ve lo ci
slide over th or igid
ty profile in ou<ideal
dle of tho
ized model will be
h = 1,000 m, that ba " at th e vd oe it
is, at about 3 X 10 - 7 Y for h, --- 2,000 m,
Th o sequ m s , or about 9 m
enco an d slope of FigUfo 8-6 are-1sim lyear.
,Iiding of f the K ub ilar to tbo,", foun
or Range in Pa pu d in th e ar ea of
m at ed th e di ,ta ne a Now G ui ne a (J
o slid to be ab on t enkin" 1974). Je nk
x 10 ' years). U si !3 km ,inco th ee "r ly in s "t i-
ng th i' velocity, w plioceno (a pp ro 'im
eq ua tio n (8.9) im hieh i, 7 x 1 0 '" at el y 6
pl i" an oquivalent m so , 0, 2 x 10 " M
This i, within th e vi ,o o, ity of th e oe m ly e ,
range of vi so o, iti do c of 10 " pO ', 0' 10
an d th e po in t ,h ou " inferrod for th e ma " po i" .
ld no t be m i" ed th ntic (sco Cr itt en de
at ,ig ni fie an t tra n, n, 1963) .
la tio n i, po "i bl o
large viscosities. w ith ,u ch
 183

Ifwe accept Kehle's figure of 10 16 Pa s for the viscosity of shale, the velocity at the
interface will be nearly 10 metres/year if the whole system is subaerial. This is,
geologically, a catastrophic rate of sliding that would have completed the 13 km
movement in about 1,500years. Reducing the slope to 1° only reduces the velocity to
2 metres/year (7,000 years to complete the slide): reducing the thickness of the shale
lubricant to 100 m on a 1 only reduces the velocity to 1/4 metre/year (54,000 years).
0

If the whole system were submarine, these velocities would be about halved.
The idealized conditions assumed above only approach reality is so far as we can
say that there exists a newtonian viscosity that would lead to the observed results.
Qualitatively, the equivalent viscosity of a sedimentary rock depends on the effect-
ive stress, porosity, temperature, grain size and shape, and the material; it may also
depend on the strain rate. It may therefore be very variable. When salt is the
lubricant, some of these variables will be less important; but the viscosity of salt in
the subsurface is unlikely to be much greater than 10 16 Pa s, and may be as low as
1013 Pa s (see Ode, 1968, p. 74, Table IV).
The problem is seen to lie not so much in the answer to the question "How can
blocks slide?" but rather "What slows them down?". Short answers to the second
question are: a) surfaces are not planar, and non-planar surfaces consume energy,
b) part of the sliding block, through fracture or otherwise, may come to slide over an
unlubricated surface, and c) down-slope leads to the sea, where buoyancy roughly
halves the driving force and halves the velocity.
The first two are rather outside the scope of this book, but let us look briefly at the
question of un lubricated surfaces, because unlubricated sliding has been found to be
inadequate for geological translation on a large scale. A combination oflubricated
and unlubricated sliding may enlighten us.
Consider a block that slides slowly down a lubricated slope and meets an obstruc-
tion. If the block is short, it will merely stop at the obstruction; but there will be
some critical length of the block which, if exceeded, will lead to failure at the
obstruction in the form of a thrust fault (Fig. 8-7). The more the length exceeds the
critical length, the greater the movement that will occur on the thrust fault. Such
movement tends to lift the block off the thrust plane, but on geological scales the
effect is to deflect the sliding surface upwards along the thrust plane, leading to an
overthrust (Fig. 8-7b). As this movement continues, the frictional resisting forces on
the unlubricated surface increase, while the driving force decreases. The slide will
come to rest when these two forces are equal.
The driving force of an inclined block on a lubricated surface is simply the shear
component of weight along the base of the block, that is
(8.10)
where 11 is the length of the block above the base of the thrust. The frictional
resistance to sliding is given by the modified Mohr-Coulumb criterion,
F2 = 12w {r' + £5 (Pb - pa)ghcos8tan ¢}, (8.11)
184

Fig. 8-7.

where 12 is the length of the block below the base of the thrust, " is the shear stress
that corresponds, during sliding, to the initial shear strength '0 before sliding (,' is
usually rather smaller than '0), and J (Pb - Pa) gh cos e is the effective normal stress
(equations (8.4) and (8.6)) with the factor J taking the pore-fluid and ambient-fluid
pressures at the sliding surface over 12 into account.
Sliding is also opposed by the shear component of weight down the thrust plane.
The angle of the thrust will be about (45 0 - ¢/2) to the longitudinal greatest
principal stress (which we take to be parallel to the base of the block), that is, about
30° . So we can take the length of the thrust plane to be about 2h. This force is
F3 = 2 h W (Pb - Pa) gh sin (30 - e). (8.12)
The slide will therefore come to a halt when
Fl = F2 + F3 ,
11w(Pb - Pa)ghsine = 12 w{,' + J(Pb - Pa)ghcosetan¢}
+ 2hw(Pb - Pa) gh sin (30 - e)
from which

I -I {
1 - 2
"
(Pb - Pa)ghsine +
Jcos etan
sine
¢} + 2hsin (30 - e)
sine . (8.13 )

The minimum length of 11, the maximum length of over-thrust, occurs when,' is
negligibly small compared to (Pb - Pa) gh, so, for fully submarine or fully sub-aerial
slides, we can write for the maximum length of overthrust and thrust

11sine
I2:(~-~--c- 2hsin(30-e)
(8.14)
J cos etan ¢ J cos etan ¢ .
 185

This is a linear relationship in II and 12 for constant slope and conditions, and gives
the maximum likely length of an overthrust when the block length is I = II + 12 and
the overthrust is unlubricated. Figure 8-8 shows the maximum length of the over-
thrust plotted as a percentage of the total length for blocks of various thicknesses on
slopes of 10° .
If the weight of the block is to cause the thrust and overthrust, then the total
length of the block above the obstruction must exceed its critical length (see
Chapman, 1979, p. 26, eq. 30). But this is not the only possibility. The dominant
wavelength of diapiric anticlines may provide natural lines of weakness and divide
the block into lengths shorter than its critical length (Chapman, 1974).
Figure 8-9 shows the section between the Lavani anticline in the Muller Range
and the Cecilia anticline in the area of Papua New Guinea studied by Jenkins (1974).
The total length of this limestone block is about 30 km, 23 of which are on a slope of
10° above the base of the thrust Ud. Roughly 1/3 of the block is subaerial. The
limestone thickness averages ~ 1,250 m. Figure 8-10 is a plot of percentage over-
thrust against total length of block (11 + 12 ) computed from equation (8.14). The
maximum length ofthrust and overthrust expected fom this theory is therefore 9 or
10 km (32%) for a fully subaerial slide, or 6 km (20%) for a fully submarine slide.
This is in satisfactory agreement with the field observations of 7 km - and the slide
may still be going on. In the upper part, there are about 7 km ofthe block above sea-

____--------------=::::1 500m.
_ -------------- 1000 m.
Subaerial

30%

20%
____----------======:==:==~ 500m.
_---------------- - 1000 m.
__ - - - Submarine
""",-",."

"
10%

o 10 20 30 40 50
Total length, .t km

Fig. 8-8. Maximum length of thrust and overthrust expresses as a percentage of the total length of the
block on slope of 10° .
 ......
00
0\

Fig. 8-9. Cross-section A-A of Figure 7-8a.

 -...
187

tn

-
:;,
.c
..... 30
0
Q)
tn
co
.c
E 20

-
........
0

e:;,n

-...
.c
...
10
~
0
?fl.
E
:;,
0
.-E
>< 0 10 20 30
co
:IE Total length, .e km
Fig. 8-10. Maximum length of thrust and overthrust for block 1250 m thick on slope of 10°, such as the
limestone block in Figure 8-9.

level, above a small thrust fault. Here theory predicts an overthrust ofless than 10%.
Ifwe take Jenkins' (1974, p. 544) generalized dip of 4° off the Kubor Range at the
base of the Tertiary, the maximum length ofthrust and overthrust in a block 125 km
long would be between 11 km (submarine) and 22 km according to theory if the
average thickness of the block is between 1 and It km. This is also in satisfactory
agreement with the field observation of 13 km.
It seems therefore reasonable to conclude that the reason why sliding has not been
more extensive is that the obstructions to sliding have caused thrusting and over-
thrusting, part of which is unlubricated. We can therefore agree with all Jenkins'
conclusions except that sliding is probably still going on.

EFFECT OF SEA-LEVEL

The effect of sea-level on sliding is of interest. Not only may blocks slide into the sea,
but they may also be raised above sea-level during orogeny (as in the example from
Papua New Guinea).
188

Consider a block sliding slowly down a constant slope on a lubricant of constant

properties (Fig. 8-11). While it is entirely subaerial, the velocity of the block
(equation (8.9» is given by

V. = Pbgh sin (} (h h _ h2 )
sa 11 * 2'
and when it is entirely submarine, its sliding velocity is given by

Sea-level is a natural impediment to sliding because buoyancy reduces the shear

component of weight, reducing the velocity of a lubricated slide by about half - and

SEA LEVEL

Fig. 8-11. Effect of sea-level on block sliding slowly down constant slope into sea. The lengths of the
arrows are proportional to the component of weight down the slope.
 189

normally stopping an unlubricated slide. As a lubricated block slides into the sea, a
component oflongitudinal compression is induced in the block. This increases to a
maximum near sea-level when half the block is submerged. This force is the differ-
ence between the shear components of subaerial and submarine weights. It seems
that this force alone will not cause significant deformation unless the sliding block is
thin and weak.
Elevation of a block from below sea-level to above is another matter because, if
equation (8.9) has any validity, blocks will slide off even gentle slopes much faster
than geological rates of elevation. A sliding velocity of 10 m/year on a slope of SO
implies an elevation loss of nearly 1 m/year, which is several orders of magnitude
greater than normal elevation rates (the Kubor Range rate of uplift has been of the
order of 10- 3 m/year). So, obstruction to sliding must exist from early on if a
submarine block is to be elevated above sea-level. In Papua New Guinea, it seems
essential for the preservation of the Tertiary sequence that the Muller Tange uplift
was contemporaneous with the Kubor Range uplift.
If two contemporaneous uplifts retain a sequence that includes a lubricant, then
the obstruction to sliding is the opposing slope, and the obstacle that we considered
earlier is the part of the block on the opposing slope. In this case, there may be some
up-slope sliding if the slopes are not in mechanical equilibrium. The component of
longitudinal compression induced by these slopes (above or below sea-level) is
greatest at the inflexion at the bottom of the slopes, and first failure is likely to occur
there once some critical relief has been achieved (the critical relief being (Ie sin 8).
Once the overthrust requires for its maintenance of force greater than the strength
of the block, another thrust will develop on top of the first. This may be repeated
until the critical relief or length is no longer exceeded. It is tempting to speculate that
this is the process by which overthrusts may be superimposed and imbricate struc-
tures developed.

SELECTED BIBLIOGRAPHY

Birch, F., 1961. Role of fluid pressure in mechanics of overthrust faulting: discussion. Bull. Geol. Soc.
America, 72 (9): 1441-1444.
Chapman, R.E., 1974. Clay diapirism and overthrust faulting. Bull. Geol. Soc. America, 85 (10): 1597-
1602.
Chapman, R.E., 1979. Mechanics of unlubricated sliding. Bull. Geol. Soc. America, 90: 19-28.
Chapple, W.M., 1978. Mechanics ofthin-skinned fold-and-thrust belts. Bull. Geol. Soc. America, 89 (8):
1189-1198.
Crittenden, M.D., 1963. Effective viscosity of the Earth derived from isostatic loading of Pleistocene
Lake Bonneville. J. Geophysical Research, 68 (19): 5517-5530.
Dailly, G.C., 1976. A possible mechanism relating progradation, growth faulting, clay diapirism and
overthrusting in a regressive sequence of sediments. Bull. Canadian Petroleum Geology, 24 (1): 92-116.
De Jong, K.A., and Scholten, R., (Eds), 1973. Gravity and tectonics. John Wiley & Sons, New York, 502
pp.
Guterman, V.G., 1980. Model studies of gravitational gliding tectonics. Tectonophysics, 65: 111-126.
190

Heezen, B.C., and Drake, C.L., 1964. Grand Banks slump. Bull. American Ass. Petroleum Geologists, 48
(2): 221-225.
Hsii, KJ., 1969a. Role of cohesive strength in the mechanics of overthrust faulting and oflandsliding.
Bull. Geol. Soc. America, 80 (6): 927-952.
Hsii, KJ., 1969b. Role of cohesive strength in the mechanics of overthrust faulting and of landsliding:
reply (to discussion by M.K. Hubbert and W.W. Rubey]. Bull. Geol. Soc. America, 80 (2): 955-960.
Hubbert, M.K., 1951. Mechanical basis for certain familiar geologic structures. Bull. Geol. Soc. America,
62: 355-372.
Hubbert, M.K., 1972. Structural geology. Hafner, New York, 329 pp.
Hubbert, M.K., and Rubey, W.W., 1959. Role of fluid pressure in mechanics of overthrust faulting, I.
Mechanics of fluid-filled porous solids and its application to overthrust faulting. Bull. Geol. Soc.
America, 70 (2): 115-166.
Hubbert, M.K., and Rubey, W.W., 1960. Role of fluid pressure in mechanics of overthrust faulting: a
reply (to H.P. Laubscher]. Bull. Geol. Soc. America, 71 (5): 617-628.
Hubbert, M.K., and Rubey, W.W., 1961a. Role of fluid pressure in mechanics of overthrust faulting, I.
Mechanics of fluid-filled porous solids and its application to overthrust faulting: reply to discussion by
Francis Birch. Bull. Geol. Soc. America, 72 (9): 1445-1451.
Hubbert, M.K., and Rubey, W.W., 1961b. Role of fluid pressure in mechanics of overthrust faulting: a
reply to discussion by Walter L. Moore. Bull. Geol. Soc. America, 72 (10): 1587-1594.
Hubbert, M.K., and Rubey, W.W., 1969. Role of cohesive strength in the mechanics of overthrust
faulting and of lands Iiding: discussion. Bull. Geol. Soc. America, 80 (6): 953-954.
Jenkins, D.A.L., 1974. Detachment tectonics in western Papua New Guinea. Bull. Geol. Soc. America, 85
(4): 533-548.
Kehle, R.O., 1970. Analysis of gravity sliding and orogenic translation. Bull. Geol. Soc. America, 81 (6):
1641-1664.
Laubscher, H.P., 1960. Role of fluid pressure in mechanics of overthrust faulting: discussion. Bull. Geol.
Soc. America, 71 (5): 611-616.
Moore, W.L., 1961. Role of fluid pressure in overthrust faulting: a discussion. Bull. Geol. Soc. America,
72(10): 1581-1586.
Normark, W.R., 1974. Ranger submarine slide, northern Sebastian Vizcaino bay, Baja California,
Mexico. Bull. Geol. Soc. America, 85: 781-784.
Ode, H., 1968. Review of mechanical properties of salt relating to salt-dome genesis. In: J. Braunstein and
G.D. O'Brien (Eds), Diapirism and diapirs: a symposium. Memoir American Ass. Petroleum Geo-
logists, 8: 53-78.
Price, N J., 1977. Aspects of gravity tectonics and the development oflistric faults. J. Geol. Soc. London,
133: 311-327.
Raleigh, C.B., and Griggs, D.T., 1963. Effect of the toe in the mechanics of overthrust faulting. Bull. Geol.
Soc. America, 74: 819-830.
Ramberg, H., 1967. Gravity, deformation and the Earth's crust as studied by centrifuged models. Academic
Press, London and New York, 214 pp.
Reyer, E., 1888. Theoretische Geologie. E. Schweizerbart'sche Verlaghandlung, Stuttgart, 867 pp.
Rubey, W.W., and Hubbert, M.K., 1959. Role of fluid pressure in mechanics of overthrust faulting, II.
Overthrust belt in geosynclinal area of western Wyoming in light of fluid-pressure hypothesis. Bull.
Geol. Soc. America, 70 (2): 167-206.
Smoluchowski, M.S., 1909. Some remarks on the mechanics of overthrusts. Geological Magazine, new
series, decade V, v. 6: 204-205.
Terzaghi, K., 1943. Theoretical soil mechanics. Chapman & Hall, London; John Wiley & Sons, New
York, 510 pp.
Terzaghi, K., 1950. Mechanism oflandslides. In: S. Paige (Chairman), Application ofgeology to engineer-
ing practice (Berkey Volume) Geol. Soc. America, Boulder, pp. 83-124.
Tiirnebohm, A.E., 1896. Grunddragen af det centrala Skandinaviens bergbyggnad. Kongliga Svenska
Vetenskaps-Akademiens Handlingar, 28 (5): 1-178. (Resume in German: 179-197.)
 9. CONCLUSION

We revert in conclusion to the central theme of this book - the movement of water in
the subsurface. The principles we have developed and discussed can be applied to a
range of geological problems either as qualitative or as semi-qualitative arguments.
We take one of each for illustrative purposes.

FAULTS AND WATER MOVEMENT

An important contemporary debate amongst petroleum geologists is the question of

faults' acting as conduits for subsurface fluids, particularly with regard to the
migration of oil and gas, and abnormal pressures. It may seem at first sight that it is
eminently plausible that fluids could be conducted upwards along faults (or down-
wards): but plausible arguments must be supported by physical principles if they are
to be elevated even to the rank of pure speculation.
If a fault is to be a conduit for fluidjlow (as distinct from some other mechanism),
then it must have permeability, and there must be a potential gradient in the fault
plane. If it is to be more important as a conduit than the adjacent sediments, the
permeability must be better, or the potential gradient greater, in the plane of the
fault than in the sediments. This implies a potential gradient between the fault plane
and the adjacent sediments, and so requires a permeability barrier between the fault
plane and the adjacent sediments.
It is evident that there are difficulties to be overcome if we are to accept faults as
conduits for subsurface fluids in many geological contexts.
One possibility is that the fault plane develops void-channels, or channels of
enhanced permeability, by separation of the walls. In such a case, the resistance to
flow in these channels will be so much less than that through the sediments that a
significant contribution could come from this cause. Normal faults are the most
likely candidates for such a process in most areas, with transcurrent faults more
likely in some. The stress field of a normal fault is such that the least principal stress
is approximately horizontal, normal to the trace of the fault. But the least principal
stress is compressive at all but relatively shallow depths (some hundreds of metres,
not thousands, for sediments: Hubbert, 1951, p. 367; Hubbert and Willis, 1957),
and soon exceeds the cohesive strength of the sediments. The tendency will be to
192

close any gap quickly, if one can form at all.

A variation of this possibility is the development of a fault breccia that is more
permeable than the adjacent sediments, and sealed from them to some extent by
fault gouge. While the formation of breccia implies greater cohesive strength, it is
unlikely to provide a conduit over a great depth-range.
More fundamentally, we have seen that the quantity of water that flows through
permeable material is proportional to the area of cross-section normal to the flow.
The cross-sectional area of a fault is infinitessimal compared to the area of adjacent
sediment, and so the fault would have to have a vastly greater permeability or
hydraulic gradient or both to move significant quantities of water upwards by this
conduit.
It is very difficult to understand, with present knowledge, how a fault could
possibly be an important conduit for subsurface fluids from depths greater than
about one kilometre (to take a conservative round figure). The writer knows oftwo
···........................................
.......................................
........................................ ..
.
·..................................
.......................................
. .
·····................................
.................................
.................................
................................ ..
....................................
··..............................
··.............................
...............................
"
•••••••••••••••••••••• ••••••••••••••••••••••••••••••••••••••••••J-..........~~......a...

····.........................
............................
........................
......................... .
····.........................
..........................
.............................
..........................

...... . . ...... ....... ......

····..........................
........................
· ........................
.........................
........................ ...
. . .. .. .... ...... ..... .......
. . . . .
.·.......................... . . . . . . . . . . . . . . . . . . .
. .... .. .. .. .... ...... .. .. .. ....
...................................................................................................................................
....,......-rM.;f.............., •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••

···...·..........................................
............ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....
....... ......................................
.................................
................................. .
.........................................................................
Fig. 9-1. Schematic section through fault cutting abnormally-pressured shale, showing pore-water flow
directions.
 193

oil-well blowouts that caused fluids to follow a fault to the surface from shallow
depths, one of which repressured petroleum reservoirs at greater depths. Indeed,
faulted oil fields would be rare, one imagines, if faults acted as conduits.
There are many examples of faults in petroleum fields acting as seals to the lateral
flow of oil and gas.
Consider a fault penetrating a sequence in which there is a thick shale with pore
pressures above normal hydrostatic (Fig. 9-1), the formations above and below
having normal hydrostatic pressures. The potential gradients can be readily in-
ferred, and the fault seems to playa part only in the lateral migration of fluids.
However, faults do not always dispace the abnormal pressures by the same
amount as the sediments (see, for example, Fowler et aI., 1971, p. 56, fig. 3). A
situation such as that depicted in Figure 9-2 leads to a slightly different pattern of
potential gradients, but there is still no obvious gradient up the faults.
It is important to understand that speculation devoid of physical support, or in
seeming conflict with physical principles, may still turn out to be correct. But if we
ignore our understanding of physical principles when speculating, we are more
likely to be led into error. Temperature or geochemical anomalies near a fault must
have explanations that are consistent with physical principles .

• ' • • • • • • • • • • • • • • s • • • •• ••••••••••• ••••••••••••

. ::::.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.":':':':':':':':':':':': ,I:,':,':,':,':,',
··.·.·:·:·:·:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.··:·:.:·:i
I
.'..'........................
.. .............
.:.:.:.:. ..........
••• ' . ' •••• ••••••••••••••••••••••••••••••
A' 'p":':':':'
' ...............
............ • • •••• • • •• • • •
I, •••• II II II I, •••••••• I ,
•••••••••••••

:.:. :.:.:.:.:.:.:.:.:.:
'. '. ...... ........................
·'.:.:.:. ...........................
.............. ..
..:.:.:.:.:.:.:.:.:.:.:.:.:
. . . . . . . ... ... ..........
.·..................
:.: . . ...:.:....
... ...................
. ...............
.............................
.. .........
T.A.P.:·:·:·:·:·:·:·:·:·:·:·:·. ..................................
.................
................................................
.

Fig. 9-2. Schematic section through faults cutting abnormally-pressured shales in which the degree and
stratigraphic extent of the abnormality changes across the faults. Arrows show the direction of pore-
water flow.
194

FLOW OF TWO IMMISCIBLE LIQUIDS IN POROUS

SEDIMENTS

If analysis of the flow of a single liquid through porous materials has difficulties,
that of flow of more than one liquid presents formidable problems. It is, of course, a
matter of central interest to petroleum reservoir engineers, and much experimental
and theoretical work has been done on it. But interest should not be confined to
petroleum reservoir engineers because two-phase flow can occur in a variety of
geological situations. We take a semi-quantitative approach because we are more
interested in acquiring a feel for the parameters that affect the phenomenon than in
developing predictive equations required for proper petroleum reservoir manage-
ment. We shall take a simplistic approach along the lines of Chapter 3, and we shall
regard both water and oil as incompressible liquids, evenly distributed in the sense
that the proportion of either liquid in the total pore volume is considered to be
present in each pore.
We cannot see two-phase flow in the subsurface, so its character must be inferred.
When an oil well is put on production, oil typically flows with virtually no water,
and it will produce a significant proportion of its total ultimate yield with very little
water. Towards the end of the life of the well, the water content in the production
increases - slowly at first, but accelerating - until the well is no longer economical;
and when the well is abandoned, there is residual oil left in the reservoir amounting
typically to about 65% ofthe pore space (Bridgeman, 1969). This suggests that the
wetting liquid, water, is essentially static around the grains when the oil saturation
of the pore spaces is high. As the water saturation increases, the reservoir rock
becomes more permeable to water, while still retaining some permeability to oil.
There is clearly an effective permeability to oil in the wetted reservoir rock that is
analogous to intrinsic permeability.
Consider a porous, permeable, isotropic and homogeneous reservoir rock that is
entirely saturated with water. We can determine from the flow parameters the
intrinsic permeability of the material and its hydraulic conductivity. When oil
occupies part of each pore (the material being water-wet), each globule of oil
occupies the space of minimum potential in each pore - that is, the central part. As
the globule grows, three distinct stages are recognizable:
1) the globules are smaller than the pore-throats connecting one pore with another,
and there will be no great impediment to the flow of this oil with the water;
2) the globules are larger than the pore-throats, and work will have to be done to
force the globule into the next pore against the capillary pressure;
3) the globules coalesce to form a continuous oil phase, and the flow of oil will be
comparable with (but not identical to) that of a single liquid in a material of smaller
porosity.
Production of an oil well, we infer, takes place mainly in the third stage, if not
entirely.
 195

The question is, how does the effective permeability to oil change as the water
saturation changes? We are interested in relative changes, so we express the effective
permeability relative to the intrinsic permeability. The relative permeability to oil is
therefore defined
(9.1)
where k is the intrinsic permeability and ko is the effective permeability to oil in the
wetted material. ko = k when the material is 100% saturated with water or oil.
It might seem that we should make the distinction between intrinsic permeability
and the coefficient of permeability (or hydraulic conductivity), but experiments
suggest that relative permeability is not sensitive to changes of viscosity or density
(see Leverett, 1939). The reason for this apparent paradox is that the intrinsic
permeability k is a property of the material only, independent of the fluid, and so can
be determined at 100% saturation of either liquid.
From Darcy's law,
(9.2)
when the material is entirely saturated with either liquid. The analogous expression
for effective permeability to oil when both immiscible liquids are present is
(9.3)
Hence,
(9.4)

We would therefore expect relative permeability to be independent of viscosities and

densities *.

* It has also been reported that relative permeability is independent of the specific discharge, q = Q IA.
This is as one would expect, because these are in the nature of intrinsic permeabilities, which can be
measured over a range of q provided the flow is within the realm of Darcy's law.
Reports that relative permeability varies with the pressure gradient arise, it seems, from the use of an
incomplete form of Darcy's law (see Appendix, p. 207) and a failure to appreciate that k can be
determined with any liquid.
They write

and

so that

kw = kolk = q~/qwl'/w'
But q~ is proportional to L1ho and the larger the quantity L1ho, the larger the quantity q~. But L1ho does not
appear in their formulation because ifwas cancelled out on the assumption that L1p was the same for both
liquids. This is not necessarily so in an experiment using various pressure gradients on the mixtures.
196

Relative permeability is determined experimentally, and it is found that the

relative permeability to oil (or any other immiscible, non-wetting liquid) reduces
sensibly to zero before the oil saturation is reduced to zero - typically before it is
reduced to 15%. There have been various explanations for this (see Bear, 1972, pp.
459-466, for a recent discussion of the whole subject), but clearly they are related to
the second stage outlined earlier - the work required to force discrete globules of oil
from one pore to another.
A point of fundamental importance must now be raised. We found in Chapter 3
that liquid flow through porous glass with very small passages obeyed Darcy's law,
and therefore conclude that the static wetting layer may be very thin indeed, to be
measured in molecules. This is in conflict with our inference that the wetting liquid,
water, is static when the oil saturation is high. It seems likely, therefore, that the real
situation is that most of the flow through porous material is in the central regions of
the pores (as Lindquist, 1933, pp. 87-88, suggested) with trivial contribution from
the peripheral pore space*. As the oil saturation decreases, so more and more water
moves into the space of important flow. Once the oil loses physical continuity, more
energy is required for its movement - and while oil globules remain, they impede
water flow much as if there is a reduction of porosity (there is a reduction of effective
porosity). There is a concomitant increase in tortuosity, because oil occupies the
central space of the pores; and when there are discrete globules, they act as if they
were small solid grains, so reducing the parameter d (the harmonic mean diameter of
the matrix through which the liquid flows). The presence of two immiscible liquids, or
a liquid and a gas, reduces the effective permeability to both.
We therefore approach relative permeability in a semi-quantitative manner
through the porosity component of intrinsic permeability, assuming that the hy-
draulic radius term, {fd/(1 - f)Y, (as we have just inferred) and the shape factor C
are independent of saturation while oil is flowing.
Let s be the water saturation, or proportion ofthe pore space occupied by water as
the wetting liquid, then sf is the loss of effective porosity for oil due to the water
saturation. From equation (3.18),
(9.5)
where m is the cementation factor that normally ranges from about 1.3 for uncon-
solidated sands to about 2 or a little more for consolidated sands.
For the analogous expression for effective permeability to oil, we note that the oil-
water interface is not stationary while the oil is flowing, and so write
(9.6)
and the relative permeability to oil while the oil is flowing is

* cf. Versluys' (1919) concept of washing the salt water out in the transition zone of a Ghijben-
Herzberg lens -a case of flow of two miscible liquids.
 197

(f - sf) 1.5m. - 0.5

k ro = ko / k = f1.5m _ 0.5 ' (9.7)

where mo is analogous to m, but is strictly a variable depending on the saturation s.

For water flow, the oil is an effective reduction of porosity that causes an increase in
tortuosity by denying to water the central spaces of the pores. For water saturations
that allow the oil to flow, effective permeability to water will be approximated by
(9.8)
where mw is larger than mo for water saturations less than one (and greater than zero)
on account of the greater tortuosity. Hence, the relative permeability to water is
given by
(sf) l.5m. - 0.5
k rw = kw/k = fl.5m 0.5 (9.9a)

When the water saturation is such that the oil ceases to be a continuous phase
throughout the pore spaces, the oil not only reduces the effective porosity for water
but also acts as a static boundary and so leads to an effective reduction in hydraulic
radius. The harmonic mean diameter of the static components is also reduced.
Ignoring changes in the harmonic mean diameter, we therefore write
(9.9b)
We must now examine the quantities m, mo and mw. While m is a material constant
determined from the Formation Resistivity Factor, mo and mw are variables that
depend on the saturation. Each approaches m as its saturation approaches unity.
Since oil flows in the central spaces of the pores, we can probably take rna to be
approximately equal to m without serious error for our purposes. By virtue of the
water's exclusion from the central spaces of the pores, mw will be larger than m -
considerably larger when the water saturation is low. We can get an idea ofthe range
from Leverett's (1939, p. 153, fig. 3) calibration curve for relative electrical con-
ductivity versus saturation, on the assumption that relative electrical conductivity
(Cr ) is analogous to relative permeability to water. His data are closely approx-
imated by Cr = 1.1 S2.2, but the exponent varies from about 1.5 as s ---+ 1 to about
for s = 0.2. This suggests (by equating the exponent to 1.5mw - 0.5) that mw in his
unconsolidated sands ranged from about 1.3 for s = 1 (as it should be) to about 2 for
s = 0.2 We therefore arbitrarily take mw = 2m - ms, so that mw = m when s = 1
and mw ---+2m as s ---+0.
Introducing these substitutions into equations (9.7) and (9.9), we now write as our
approximations to relative permeability:
kro = (1 - s)1.5m-O.5; (9.10)
(sf) 1.5(2m - ms) - 0.5
k rw = f1.5m-0.5 ; (9.l1a)
198

(9 .11 b)

Figure 9-3 is a plot of these hypothetical curves for m = 1.6 and f = 0.3, such as we
might expect for a clean sandstone. The lower curve on the right is for k7w when the
'irreducible' oil saturation has been reached. The dashed curve above the others is
the sum ofthe two relative permeabilities while oil is flowing. Figure 9-4 is a plot of
the family of curves for m = 1.3 and m = 2.0, the normal range, and porosities of
35% and 20% respectively.
The porosity hypothesis for relative permeability leads to curves very similar in
shape and position to those found by experiment. The asymmetry is in the same
sense as that found by experiment, with the curves crossing when the water satura-
tion is more than 50% (see Leverett, 1939; Levorsen, 1967, p. 111, fig. 4-5; Bear,
1972, p. 460). There are several points of interest.
It has long been noted that the sum of the two relative permeabilities, kro + krw, is
less than unity; and this has been attributed variously to differences of viscosity and
to the entrapment of oil globules in 'dead-end' pores. It is sometimes referred to as

Oil saturation
1 0.5 o
1

-"
>-

.c ::::s
\
\
\ , kro k rw J
J
I

(
"JII
('0
CT
CI)
IJ
E
~
CI)
C)
c 0.5 \
a.
"- \... IJ
-- --
~

A'1.
CI) 0
>

,"" 'l
~
('0 0 1- .......
CI)
~~if

-
CC
,/
~
f ' ..........
o
o 0.5 1
Water satu ration
Fig. 9-3. Hypothetical relative per me abilities of two immiscible liquids plotted against saturation.
Dashed line is the sum of k m and k cw . Porosity 30%, m ~ 1.6. (Note: These curves should apply to any two
immiscible liquids because the relationship is independent of liquid properties: for discussion, we have
taken water to be the wetting liquid, oil to be the non-wetting liquid.)
 199

Oil saturation
0.5 o

-"
~

.a :::s
ca
~
'\ ~
~\ VI
J
CD C"
E
\ '\ I I
~ C')
\ \'1.3 1.3/ II
CD c 0.5
Q. .-
\ ~
I{ r

-- --
~
CD 0
> 2 I, "r\.. / I2
ca 0
~ "rv " 1/
~V ~
CD
a:
-
io""""
~ ~ ~ '-
o
o --- ~
0.5 1
Water satu ration
Fig. 9-4. Hypothetical relative permeability curves for m = 1.3,1 = 0.35, and m = 2.0 and! = 0.2.

the Jamin effect. Figure 9-3 shows the sum of the hypothetical relative permeabili-
ties plotted against saturation: this curve is also very similar in shape and position to
those obtained experimentally. It is therefore interesting to note that our simplistic
porosity hypothesis predicts for two-liquid flow in homogeneous, isotropic materials
that the sum of the relative permeabilities should be less than unity - indeed, very
much less than unity over most of the range of saturations - without recourse to any
other effects.
The curves (both experimental and these) suggest that capillary action on the
discontinuous oil phase is not necessarily important except for the highest water
saturations, at least 75%. But we see why the 65% oil saturation level is the
approximate economical and practical limit of oil-well production. The water-cut
(proportion of water in the total being produced) is given by the ratio k rw/
(krw + kro ), and the water/oil ratio by krw/ kro. These are plotted as percentages in
Figure 9-5 for the data of Figure 9-3.
Early production contains very little water, but soon after the water saturation
reaches 35% the water-cut increases rapidly, and this salt water must be separated
from the oil and disposed of. This effect is superimposed on the decline in total
production (rapid decline in oil production) due to the reduction in the sum of both
200

Oil saturation
1 0.5 o
100
/
~
%
I
I
(
Water/oil ratio .... ............

50 ~
'[water cut

J
I
o .. /
o 0.5 1
Water sa tu ration
Fig. 9-5. Hypothetical proportions of water and oil produced at different saturations (m = 1.6,1 = 0.3).

relative permeabilities. The 100% water production level is reached effectively

before 100% water saturation in the pores is reached; but the model predicts that
there will be a loss of relative permeability to water when the 'irreducible' oil
saturation is reached. As this oil saturation is approached in real sediments, there
will probably be a gradual shift to a point on the lower curve.
If a rock is oil-wet, then the pattern of relative per me abilities will not be the same,
but rather like the mirror image. However, in this case the water is for practical
purposes an unlimited resource, and only rapid production without pressure main-
tenance is likely to recover a significant proportion of the oil in place.
It will occur to the reader, as it has occurred to the oil companies, that signif-
icantly greater reserves of recoverable oil would be made available if a method could
be found for recovering the residual oil. The term 'secondary recovery' applies to
methods of reducing the amount of residual oil in final stages of normal production:
'tertiary recovery' for the recovery of the rest. Secondary recovery methods are
mainly pressure maintenance by water injection into the reservoir outside the oil
limits (and/or gas into the cap of the reservoir) with the object of sweeping the oil
upwards into the structure as depletion proceeds and the oil is replaced by water.
Tertiary recovery, still in the pilot-project stage with most companies, usually
involves the injection of a solvent into the water, with the intention to taking oil into
 201

solution, and recovering it at the surface (and re-using the solvent). Our model does
not suggest any practical alternative, unfortunately.
Interest in relative permeability should not be confined to petroleum reservoir
engineers. If pore water migrating from abnormally-pressured shales contains gas in
solution, the upward component of migration may take it to pressures at which the
gas comes out of solution (Chapman, 1972; Hedberg, 1974). This will immediately
reduce the effective permeability to water in a material of already low permeability
(at first, down the lower line on the right of Figure 9-3, for example), and further
expulsion of water will be impeded. Its pressure will therefore rise, and some or all of
the gas may be taken back into solution - allowing the pore water to move again
more freely, until the gas again comes out of solution. (If the gas coalesces to form a
space of continuous gas phase, it is not clear what will happen to this gas unless its
perimeter reaches a permeable bed, when it will be relatively easily expelled.)
Whatever the final outcome of such a process may be, it is clear that upward
migration of water will be seriously impeded, so enhancing the mechanical instabil-
ities discussed in Chapter 7. Note, however, that the same tendency would appear to
be absent from the downward-migrating pore water at the bottom of the shale
unless the absolute pressure is decreasing. Perhaps this is why so many oil reservoirs
seem to have their petroleum source rock stratigraphically higher in the sequence.
Infiltration 0 f rain-water in to soil and the under! ying rocks is also a case 0 f flow 0 f
two immiscible fluids, the second being air. Although this is a most complex
phenomenon, with the air moving in the opposite direction to the water, the
principles are the same as those we have discussed. It is a subject that has been
intensively studied by many people for many years (see Bear, 1972, p. 439ff, for a
recent detailed review and discussion).
Consider rain falling on a thin permeable soil that rests on a homogeneous,
isotropic sand that is water wet. Between the surface and the water-table the pore
spaces are, in general, partly filled with water (called vadose water), and partly with
air. Once the water saturation exceeds some minimum level (the specific retention),
water will flow downwards under gravity, displacing the air upwards. The relative
permeability of the sand to water would follow a curve similar to those in Figure 9-4
on the right, but modified somewhat by the opposing air flow.
As a first approximation, we would suppose that soon after rain starts falling, the
relative permeability of the sand to water will be greater near the surface than at
some greater depth in the vadose zone. This will cause a 'front' of water to move
downwards to the water-table. Once this front reaches the water-table, flow will
tend to stabilize over the vadose zone, with a fixed proportion of the rain-water
infiltrating into the sand. The water saturation in the vadose zone typically increases
downwards to the water-table and with it, we suppose, the relative permeability to
water until the point is reached that the air becomes a discontinuous phase, and the
capillary pressure must be overcome to expel the remaining air. We are therefore
forced, at last, to consider capillary pressure and surface tension.
 202

Capillary pressure and surface tension

It is well known that if a glass tube of small internal diameter is inserted into a bowl
of water (Fig. 9-6) water rises in the tube above the level in the bowl, and the shape of
the air/water interface (the meniscus) is curved, as in Figure 9-7.
As regards the elevation ofthe water in the tube, it is evident that the air pressure is
essentially the same in the capillary tube as on the surface ofthe water in the bowl;
and since the pressure in the water at the level ofthe water surface in the bowl is zero
(taking atmospheric pressure as datum, that is, gauge pressure), and the pressure in
the water column in the tube decreases upwards, the pressure in the elevated water
column must be negative, and there is a pressure differential at the meniscus. This
interface therefore acts as if it were an elastic membrane in a state oftension, with the
greater pressure on the concave side. This is the surface tension.
The amount of elevation, he, leads to the concept of capillary pressure;
Pc = -pghe = -pw (9.12)

The maximum amount of capillary rise is that at which the upward component of
capillary force and the weight of the elevated water are equal, that is;
pghenr2 = (J2nr cos e
and
2(J 2(J
' he
Pc =-cose· = -cose, (9.13 )
r ry
where (J is the surface tension, y the specific weight of the water, r the radius of the
capillary tube, and eis the contact angle (Fig. 9-7) (cos e ~ 1 for water on quartz).
For pendular water around the point of contact of two spheres, it can also be
shown that the capillary pressure is related to the two radii of curvature on the
surface of the water;

Fig. 9-6.
 203

Fig. 9-7.

Pc = (J(~ + r~') = 2(J/r* (9.14)

where r* is the harmonic mean of the two radii, taken to be negative since Pc is
negative). (J has dimensions MT- 2: it is defined in terms of work per unit of area.
Surface tension is known as interfacial tension between two liquids.
Sediments are not measurable in these terms. The contact angle () is only constant
when the fluids are static: it is greater for an advancing water surface than for a
retreating surface (examine a dew drop on a petal). It varies with water quality, as
does surface tension, and the nature of the solid. But the phenomenon of capillary
rise of water in sediment has long been known (perrault, 1674, reported experi-
ments), and the capillary rise and capillary pressure are inversely proportional to a
dimension r* that can be regarded as a measure of pore size. We have seen that
intrinsic permeability (dimensions L2) contains a measure of pore size (the hy-
draulic radius) and so would expect that
he ex (1 - f)/fd, (9.15)
where d is, as before, the harmonic mean grain diameter.
The amount of capillary rise (known qualitatively as the capillary fringe) there-
fore varies according to the nature as well as the texture ofthe sediment. It amounts
to less than one metre for sands, to a few metres for sediments of small hydraulic
radius. (A similar effect is present at the oil/water contact in an oil field that bounds
an oil accumulation: the transition is not at a surface, but over a zone - thin in \oil
reservoirs - in which the saturations change.)
We therefore see that the water-table should be defined not as the surface of 100%
water saturation, but the surface in the water at which the pressure is atmospheric.
The pores are 100% saturated for a short distance above this level, and above that,
the water saturation decreases.
It therefore seems unlikely that capillary forces are important in water infiltration
through a water-wet sand much above the level of 100% saturation, unless the
204

water-table is near the land surface. The infiltration rate will be determined by the
minimum relative permeability of the sand to water in the profile near the surface,
and any rainfall in excess of that amount will become surface run-off. This minimum
relative permeability in the profile will remain less than unity while air remains to
be expelled.
In reality, of course, there are many complicating variables involved, such as
irregular rainfall, dissolution of air in water, evaporation and condensation of water
in the vadose zone - but these are variations on the theme.

It has not been my purpose in this chapter to present a detailed analysis of the
phenomena discussed, but rather to illustrate through these phenomena two im-
portant applications of some of the concepts developed in earlier chapters. Further
study of two-phase flow can begin with the references listed, and the general
references at the end of Chapter 1. Such applications, though, are not unrewarding.

SELECTED BIBLIOGRAPHY

Bear, J., 1972. Dynamics offluids in porous media. American Elsevier, New York, London, and Amster-
dam, 764 pp.
Bear, J., Zaslavsky, D., and Irmay, S., 1968. Physical principles of water percolation and seepage.
UNESCO, Paris, 465 pp.
Bridgeman, M., 1969. Recent advances and new thresholds in petroleum technology. (8th Cadman
Memorial Lecture.) J. Inst. Petroleum, 55: 131-140.
Chapman, R.E., 1972. Primary migration of petroleum from clay source rocks. Bull. American Ass.
Petroleum Geologists, 56 (11): 2185-2191.
Fowler, W.A., Boyd, W.A., Marshall, S.W., and Myers, R.L., 1971. Abnormal pressures in Midland
Field, Louisiana. In: Houston Geol. Soc., Abnormal subsurface pressure: a Study Group report 1960-
1971. Houston Geol. Soc., Houston, pp. 48-77.
Hed berg, H.D., 1974. Relation of methane generation to undercompacted shales, shale diapirs, and mud
volcanoes. Bull. American Ass. Petroleum Geologists, 58 (4): 661-673.
Hubbert, M.K., 1951. Mechanical basis for certain familiar geologic structures. Bull. Geol. Soc. America,
62: 355-372.
Hubbert, M.K., and Willis, D.G., 1957. Mechanics ofhydraulicfracturing. Trans. American Inst. Mining
Metallurgical Petroleum Engineers, 210: 153-166. (Discussion: 167-168.)
Leverett, M.C., 1939. Flow of oil-water mixtures through unconsolidated sands. Trans. American Inst.
Mining Metallurigcal Engineers (Petroleum Division), 132: 149-169. (Discussion: 169-171.)
Levorsen, A.I., 1967. Geology of petroleum (2nd edition). W.H. Freeman & Co., San Francisco, 724 pp.
Lindquist, E., 1933'. On the flow of water through porous soil. 1" Congres des Grands Barrages
(Stockholm, 1933), 5: 81-101.
Perrault, P., 1674. De /'origine des fontaines. Pierre Ie Petit, Paris, 353 pp.
Rose, W., 1957a. Fluid flow in petroleum reservoirs.!' - The Kozeny paradox. Illinois State Geol. Surv.
Circular 236 (8 pp.).
Rose, W., 1957b. Fluid flow in petroleum reservoirs. II. - Predicted effects of sand consolidation. Illinois
State Geol. Surv. Circular 242 (14 pp.).
Versluys, J., 1916. De capillaire werking en in den bodem. (Thesis, Technische Hoogeschool, Delft, The
Netherlands.) W. Versluys, Amsterdam, 136 pp.
Versluys, J., 1919. De duinwater-theorie. Water, 3 (5): 47-51.
Versluys, J., 1931. Can absence of edge-water encroachment in certain oil fields be ascribed to capillarity?
Bull. American Ass. Petroleum Geologists, 15 (2): 189-200.
 APPENDIX. REVIEW OF COMMONLY-QUOTED
WORKS ON DARCY'S LAW

Our purpose here is to review, concisely, the main results of commonly-quoted

authors of the last 60 years or so, confining ourselves to those who sought to develop
the theme, rather than use the work of others*. Their expressions have been trans-
lated into the notation of this book.

Blake (1923), working in the context of chemical engineering, was apparently the
first to use dimensionless groups for the plotting of experimental permeability data.
Hi's groups lead to the following form of Darcy's law:

(A.l)

Although he was concerned with both air and water, his groups are strictly valid
only for gas or horizontal flow ofliquid, and there is no useful way in which his work
can be modified for the earth sciences. He did, however, recognize that q is a
notional velocity and substituted for it the term q!!; and he introduced the concept
of hydraulic radius, though not by name, by substituting!! S for the grain diameter.

Kozeny (1927b) worked in the context of soils and irrigation. His work is of central
importance, not only because it was analytical and thorough, but also because his
results have been incorporated into both chemical ingineering and earth sciences.
Much of the paper is devoted to capillary movements, and is peripheral to our
theme.
He first obtained an analytical expression of Darcy's law (without, however,
quoting him directly: he does not seem to have been aware of Blake's work). He
begins the development with the equation

where K 'die DurchHissigkeit und L1h die auf dem Stromlinienstiick I verbrauchte
Druckhohe darstellt' ~ 'where Kis the permeability and L1h the loss of pressure head
* It is regretted that works published in languages other than English, German, and French, have not
been consulted because translations cannot be relied upon if the translator does not have an intimate
knowledge of the subject.
206

over the stream length l' (quoted with change of notation). It is unfortunate that he
used the word Drukhohe because he clearly understood the nature of h (Kozeny,
1927a, p. 67). We shall take Ito be the hydraulic gradient.
Kozeny (1927b) recognized that q is a notional velocity, and inserted correcting
factors to write his equation (20),
pg I f3
q = YiICy; S2 , (A.2a)

where C is a shape factor of restricted range of values, S is the specific surface, and I
is the gradient of total head. He continued his equation (20) (our (A.2a)) for the case
of spherical grains,
pg I f3 2
q= ... = Yi ICy; 36 (1 _ f)2 d . (A.2b)

Accepting I as the hydraulic gradient, Kozeny's equation (20) (our (A.2a) and
(A.2b)) is essentially correct, although the role of tortuosity has not been taken fully
into account.
Kozeny also examined the meaning to be attached to the mean diameter and that
of the fraction between two sieves, and suggested the harmonic mean and two
variants (op. cit., pp. 305-306).
He perceived the effect of porosity and wrote (op. cit., pp. 277-278)
f3
K = Kl (1 _ f)2 (A.3)

having shown that experimental data supported the porosity factor f3 f(1 - f)2
better than others that had been proposed even for grains of irregular shape.

Fair and Hatch (1933), working in the context of hydraulic engineering and sand
filters, used dimensional analysis of pipe flow as the basis, and so also the concept of
hydraulic radius. They considered the meaning of d and concluded that there is
some shape factor that, when multiplied by lid, will give the correct specific surface.
They used the geometric mean of adjacent sieve sizes, d n = jd'd", for the mean of
the fraction on the assumption of a 'geometric size distribution'. They derived the
expression (op. cit., p. 1558, equation (3))
n

L1h = Cl!1 (1- ff q [~ \:WiJ2 (AAa)

I g p p W Ld
=
i 1
i

which can be re-arranged

(AAb)
 207

where C combines 11cl (they found Cl to have a value of about 5) and the geometric
shape factor 1/c~ (where C2 has a limited range of values from 6 for spheres to about
7.7): and d is the weighted harmonic mean grain diameter.
This expression differs from that of our equation (3.15a) only in neglecting the
tortuosity factor, and the relatively trivial difference of mean for the sieve fraction.

Wyckoff, Botset, Muskat, and Reed (1933, 1934), working in the context of petro-
leum reservoir engineering, proposed to define a unit of permeability and call it a
'darcy' (although the basis of the unit had been suggested by Nutting, 1930); and
they described the apparatus to measure permeability. (The papers cover the same
ground: the 1934 paper is the main one, to which we shall refer.)
They defined the permeability of a porous medium as 'the volume of a fluid of unit
viscosity passing through a unit cross section of the material under a unit pressure
gradient in unit time', and gave an incomplete form of Darcy's law

q= 11k -1-
,1p
(dimensionally correct) (A.5a)

which they develop into

(A.5b)

They proposed a unit of permeability to be called' "darcy" after D' Arcy' to be equal
to
1] x cm 3 x cm 1] x cm 2
sec x cm 2 x atmospheres = sec x atmos·
poises x cm 4
(A.6)
sec x dynes

with dimensions [L2], to be measured only within the realm of Darcy's law (they
regarded turbulence as setting in as Darcy's law fails at the upper limit).
Although they wrote 'The rate of flow of a fluid through any system is dependent
upon two fundamental quantities: the pressure gradients applied, and the resistance
to flow developed along the channel' (Wyckoff et aI., 1934. p. 161), and wrote their
expression for Darcy's law accordingly, it seems that they were actually measuring
the difference of total head in their apparatus. In the description of the apparatus,
they emphasized (op, cit., pp. 171-172) the use of manometers rather than gauges
'with the zero pressure points properly referred to the zero-flow levels in the
columns'. The sample holder is vertical in their apparatus.

Carman (1937, 1938, 1939). Carman's papers are in the context of chemical engi-
neering (1937, 1938), the results of which were offered to agricultural science (1939).
20S

He started with an incomplete statement of 'd'Arcy's law', modified Kozeny's

results, and wrote (1937, p. 152, equation (S))

q =f R2 g LIp
C1l1 Z
(i)It 2
(A.7a)

and, making the substitution R = f/ S, wrote what he called Kozeny's equation

q =
f3 LIp
Cll1S2 g -Z-
(1)2
It . (A.7b)

This is not what Kozeny wrote, nor is it dimensionally correct.

He then reviewed the experimental work and concluded that the value of Cl is
constant at about 5 (as Fair and Hatch, 1933, had found), so the surface area of
powders could thus be estimated since all the other quantities in the expression are
measurable.
Were it not for the influence Carman exterted on subsequent work, we would not
pursue the matter further: Kozeny's original expression is better. As Carman stated
in his introduction (1937), his purpose was to emphasize 'the importance of the
method of plotting by dimensionless groups introduced by Blake', and to use
permeability methods to determine 'the surface of powders': he introduced nothing
new.
In his 1935 paper, Carman may have stated Darcy's law correctly when he
referred to 'head loss', but before the end of the page, pressure gradient has re-
appeared. He modified 'Kozeny's equation' to the 'more convenient form'
f3 Llh
q=-S2Pg-[ . (A.S)
clI1
He then claimed that the dimensionless hydraulic gradient, Llh/l, is equal to LIp/pI.
This implies that pressure has the dimensions M L - 2.
These difficulties notwithstanding, he obtained good agreement with experi-
mental results.

Hubbert (1940, 1956/1957) examined Darcy's law from an earth-science point of

view, specifically ground-water hydrology, as part of a broader study. The relevant
results may be summarized:
a) liquids move down an energy gradient, the energy per unit of mass at a point in
the liquid being defined as its potential

1>lKJ.=gh=g(:g +z). (A.9)

(His discussion and derivation were more general, applying to any fluid: Hubbert,
1940, p. S02.)
b) Darcy's law can be expanded and generalized
 209

(A.IO)

where N is a dimensionless shape factor, and Nd 2 = k, the intrinsic permeability

(op. cit., p. 816).

c) He wrote 'the' Reynolds number applicable to flow through porous media qpd/n
(op. cit., p. 811), and
d) he pointed out that the unit of permeability, the darcy, was defined on the basis
of an erroneous expression of Darcy's law (equation (A.5) above) (op. cit., pp. 921-
922).
Hubbert's 1957 paper, written by invitation for the celebration of Darcy's centen-
ary, examined in detail Darcy's experiments, their physical meaning, and the con-
stitution of K. The expressions he derived (in the present context) are essentially
those of his 1940 paper, but he elaborated on the shape factor N (Hubbert, 1957, p.
42-46) introducing a porosity factorJ3/(1 - J? (op. cit., p. 44, equation (63); and p.
45, equation (73)). (See Note in Selected Bibliography.)

Krumbein and Monk (1942) looked at intrinsic permeability as a function of the size
parameters of unconsolidated sand from the sedimentologists' point of view. Using
Hubbert's (1940) work as a basis they derived the expression 'for size distributions
that satisfy the normal logarithmic probability law' (Krumbein and Monk, 1942,
pp.9-1O)
(A.11)
where k is the intrinsic permeability, Nis a shape factor, GM~ is the geometric mean
diameter ofthe particles, (J</> is the log standard deviation. The constants (k/GM~2)o
and a are determined experimentally.
It should be noted that porosity was kept constant close to 40%, and is not
included explicitly in their equation. It should also be noted that the range of
permeabilities studied, 57 to 1,555 darcies, exceeds the usual range of sediment
permeabilities by about three orders of magnitude.

Rose (1945a, 1945b) took a dimensional approach in the engineering context, but
does not seem to have reached any new conclusions of interest to geologists. He did
consider the upper limit of Darcy's law, plotting his own and other workers' data as
a Reynolds number against a coefficient of resistance, and found a departure from
linearity at a Reynolds number of about 10. He regarded this (1945a) as being the
onset of turbulence.

Schneebeli (1955) reported experiments conducted in the hydraulic-engineering

context near the upper limit of Darcy's law, using the dimensionless groups
210

(g~~t)and ( q~p ). (A.12a)

The first was called the coefficient of resistance (frottement): like Rose's, it is the
reciprocal of a Froude number. The second is, of course, a Reynolds number. From
these groups, the following expression can be derived for the realm of Darcy's law,

(A.12b)

In Figure 3-9 we plotted Schriever's (1930) data against these groups. It is clear that
the porosity term is required.
The main importance of Schneebeli's work is his confirmation of the fact that
Darcy's law begins to fail well before the onset of turbulence. Darcy's law failed with
spherical material at a Reynolds number (qdp/ry) of about 5, while turbulence began
at Re = 60. With angular fragments, Darcy's law failed at Re = 2, with turbulence
at about 60, as with spherical material. The limiting Reynolds number for Darcy's
law found by Schneebeli is substantially the same as that found by Lindquist (1933),
who also concluded that this was well before the onset of turbulence (op. cit., p. 91).

DISCUSSION

This brief review of selected, but commonly-quoted literature dealing with the flow
of fluids through porous media indicates a substantial consensus in the approach
and the results. Recognition that q is a notional velocity dates back further than the
period reviewed, at least to Slichter (1899). It was early recognized that there is a
component of permeability due to the material alone, and that the surface area of
solids bounding the flow is important among the quantities affecting resistance to
flow. Unfortunately, the seeds of an error were also sown early.
Darcy himself used the word pression, but he qualified it quite unambiguously.
Slichter (1899, p. 329) wrote Darcy's law as a function of pressure difference,
although he recognized the role of gravity (op. cit., p. 331, equation (7)) and
introduced a potential function. But the confusion persisted because on the next
page he wrote of 'equipotential or equipressural surface' as though the two were
synonymous. Not all the subsequent papers using pressure alone were in error in the
context in which they were written: the error is introducted when expressions
intended for gas, or the horizontal flow of liquids, are applied to non-horizontal
flow ofliquids. In many, but not all, earth science applications, no significant error
is introduced by assuming horizontal flow.
It is a tragedy of science that Darcy's name should be linked to the unit
of permeability with such an imperfect definition - the more so since the pro-
posers were probably measuring what Darcy himself measured. If Wyckoff et al.
(1933,1934) had gone from equation (A.5a) to q= (k/ry)pg(Jh/l) and so to
 211

k = Qy/l/Apg (hI - h2 ), they would have defined a unit of permeability of greater

value.
Had earth scientists been content with the papers of Darcy and of Fair and Hatch
(both written in an hydraulic engineering context, as was Darcy's, but one suffi-
ciently close by analogy to ground-water and petroleum geology) or those of
Kozeny, the topic might have been better understood. Much ofthe difficulty stems
from Carman's papers - but Carman is not entirely to blame for this. He was writing
for chemical engineers; and if they find his expression useful and wish to call it the
Carman-Kozeny equation, that is their affair. Nevertheless, one would expect
engineers to use expressions that are dimensionally homogeneous.
The choice of 'mean grain diameter' as the characteristic dimension by all but
Kozeny and Fair and Hatch has obvious disadvantages in that it is only indirectly
related to the 'mean pore diameter' through which the liquid flows. IfKrumbein and
Monk (1942) had taken d to be the weighted harmonic mean grain diameter, they
would have found that the greater the variance, the smaller the harmonic mean is
relative to the geometric mean. In their samples 10 to 20 (op. cit., p. 5, table 1), all
with a geometric mean diameter of 1 mm, they would have found a steadily
decreasing harmonic mean as the variance increased, and it is possible that the
observed decrease in permeability with increasing variance would have been ac-
counted for. Their equation seems to involve an empirical correction to the choice of
central tendency for statistical description of a sedimentary rock.
It seems that most of those who have used Krumbein and Monk's equation have
failed to notice that porosity is not included, and that the porosity for which the
equation was derived was about 40%.
There is no doubt that Hubbert's treatment of Darcy's law (see Hubbert, 1969) is
rigorous and useful, nor is there any doubt that lumping the various dimensionless
quantities into a single shape factor N has practical advantages. But the porosity
factors determined earlier by Kozeny (1927) and Fair and Hatch (1933) have their
advantages also.
Kozeny's and Fair and Hatch's recognition of the nature of the grain diameter
was also a significant step that has largely been neglected. It may be argued that
sedimentologists have become so settled on the median and the geometric mean as
their measures of central tendency that the harmonic mean is a trivial refinement of
no practical use. This is hardly so.
There is, unfortunately, no means of converting one measure of central tendency
to another: but the harmonic mean grain size of a sediment is as easily computed
from the mechanical analysis as the geometric mean - it is merely a different
operation on the same numbers.
212

SELECTED BIBLIOGRAPHY

Beard, D.C., and Weyl, P .K., 1973. Influence of texture on porosity and permeability of unconsolidated
sand. Bull. American Ass. Petroleum Geologists, 57 (2): 349-369.
Blake, F .C., 1923. The resistance of packing to fluid flow. Trans. American Inst. Chemical Engineers, 14
(for 1922): 415-42l.
Burke, S.P., and Plummer, W.B., 1928. Gas flow through packed columns. Industrial and Engineering
Chemistry, 20 (11): 1196-1200.
Carman, P.C., 1937. Fluid flow through granular beds. Trans. Instn Chemical Engineers, 15: 150-166.
Carman,P.C., 1938. The determination ofthe specific surface of powders. J. Soc. Chemical Industry, 17:
225-234.
Carman, P .C., 1939. Permeability of saturated sands, soils and clays. J. Agricultural Science, 29 (2): 262-
273.
Corrsin, S., 1955. A measure of the area of a homogeneous random surface in space. Quarterly ojApplied
Mathematics, 12 (4): 404-408.
De Ridder, N.A., and Wit, K.E., 1965. A comparative study on the hydraulic conductivity of uncon sol-
idated sediments. J. Hydrology, 3: 180-206.
Fair, G.M., and Hatch, L.P., 1933. Fundamental factors governing the stream-line flow of water through
sand. J. American Water Works Ass., 25 (11): 1551-1565.
Fraser, H.J., 1935. Experimental study ofthe porosity and permeability of clastic sediments. J. Geology,
43 (8, part 1): 910-1010.
Graton, L.C., and Fraser, H.J., 1935. Systematic packing of spheres -with particular relation to porosity
and permeability. J. Geology, 43 (8, part 1): 785-909.
Green, W.H., and Ampt, G.A., 1911. Studies on soil physics. Part I. - The flow of air and water through
soils. J. Agricultural Science, 4 (1): 1-24.
Green, H., and Ampt, G.A., 1912. Studies on soil physics. Part II. - The permeability of an ideal soil to air
and water. J. Agricultural Science, 5 (1): 1-26.
Hubbert, M.K., 1940. The theory of ground-water motion. J. Geology, 48 (8): 785-944.
Hubbert, M.K., 1956. Darcy's law and the field equations of the flow of underground fluids. Trans.
American Inst. Mining Metallurgical Petroleum Engineers, 207:222-239. (Also in: J. Petroleum Tech-
nology, v. 8, October 1956.)
Hubbert, M.K., 1957. Darcy's law and the field equations of the flow of underground fluids. Bulletin de
/' Association Internationale d' Hydrologie Scientifique, no. 5: 24-59. (Note: in this Appendix, reference
is made principally to Hubbert, 1957, rather than 1956. The papers are the same, but 1957 will be found
in Hubbert, 1969, together with Hubbert, 1940, and Darcy, 1856.)
Hubbert, M.K., 1963. Are we retrogressing in science? Bull. Geol. Soc. America, 74: 365-378.
Hubbert, M.K., 1969. The theory oJground-water motion and related papers. Hafner Publ. Co., New York
and London, 310 pp.
Kozeny, J., 1927a. Uber Grundwasserbewegung. WasserkraJt und WasserwirtschaJt, 22: 67-70, 86-88,
103-104, 120-122, 146-148.
Kozeny, J., 1927b. Uber kapillare Leitung des Wassers im Boden (Aufstieg, Versickerung und Anwen-
dung auf die Bewasserung). Sitzungsberichte der Akademie der WissenschaJten in Wien, Abt. IIa, 136:
271-306.
Krumbein, W.C., and Monk, GoO., 1942. Permeability as a function of the size parameters of un con sol-
idated sand. Technical Pubis American Inst. Mining Metallurgical Engineers, 1492 (11 pp.)
Lindquist, E., 1933. On the flow of water through porous soil. Premier Congres des Grands Barrages
(Stockholm, 1933), 5: 81-101.
Morcom, A.R., 1946. Flow through granular materials. Trans. Instn Chemical Engineers, 24: 30-36.
(Discussion: 36-43.)
Nutting, P.G., 1930. Physical analysis of oil sands. Bull. American Ass. Petroleum Geologists, 14: 1337-
1349.
Pryor, W.A., 1973. Permeability-porosity patterns and variations in some Holocene sand bodies. Bull.
American Ass. Petroleum Geologists, 57 (1): 162-189.
Rose, H.E., 1945a. An investigation into the laws of flow of fluids through beds of granular materials.
Proc.lnstn Mechanical Engineers, 153: 141-148.
 213

Rose, H.E., 1945b. On the resistance coefficient-Reynolds number relationship for fluid flow through a
bed of granular material. Proc. Instn Mechanical Engineers, 153: 154-168.
Rose, H.E., and Rizk, A.M.A., 1949. Further researches in fluid flow through beds of granular material.
Proc. Instn Mechanical Engineers, 160: 493-51l.
Rose, W.D., 1959. Calculations based on the Kozeny-Carman theory. J. Geophysical Research, 64 (1):
103-110.
Schneebeli, G., 1955. Experiences sur la limite de validite de la loi de Darcy et l'apparition de la
turbulence dans un ecoulement de filtration. La Houille Blanche, 10 (2): 141-149.
Schriever, W., 1930. Law of flow for the passage of a gas-free liquid through a spherical-grain sand.
Trans. American Inst. Mining Metallurgical Engineers (petroleum Division), 86: 329-336.
Slichter, C.S., 1899. Theoretical investigation of the motion of ground waters. Annual Report U.S. Geol.
Survey (1897-98), 19 (2): 295-384.
Wyckoff, R.D., Botset, H.G., and Muskat, M., 1932. Flow ofiiquids through porous media under the
action of gravity. Physics, 3: 90-113.
Wyckoff, RD., Botset, H.G., Muskat, M., and Reed, D.W., 1933. The measurement ofthe permeability
of porous media for homogeneous fluids. Review Scientific Instruments, new series, 4: 394-405.
Wyckoff, R.D., Botset, H.G., Muskat, M., and Reed, D.W., 1934. Measurement of permeability of
porous media. Bull. American Ass. Petroleum Geologists, 18 (2): 161-190.
 CONVERSIONS

PREFIXES FOR SI UNITS

10- 18 atto a
10- 15 femto f
10- 12 pico P
10- 9 nano n
10- 6 mIcro f.1
10- 3 milli m

}
10- 2 centi c
10- 1 deci d Further use
10 deka da discouraged
10 2 hecto h
103 kilo k
106 mega M
109 giga G
10 12 tera T

AREA

1.55 X 10 3 square inches

10.764 square feet
1.196 square yards
2.4711 X 10- 4 acres

DENSITY

10- 3 g/cm 3
6.2428 x 10- 2 pounds/cubic foot
1.9420 x 10- 3 slugs/cubic foot
 215

FLOW RATE

35.31445 cubic feet/second (cusec)

1.3198 x 104 gallons/minute (GPM), British or Imperial
1.5850 x 104 U.S. GPM
7.9189 x 105 Imp. gallons/hour (GPH)
9.5102 x 105 U.S. GPH
1.9005 x 10 7 Imp. gallons/day (GPD)
2.2825 x 10 7 U.S. GPD

FORCE

1 newton (N) 1 kg m 8- 2
105 dyne
0.1020 kgf
0.22481bf
7.233 poundals

GRADIENT

6.8519 X 10- 2 lbs/ft

9.80665 N m- 1
0.6720Ib/ft
1Nm- 2 m- 1
0.1020 kgfm- 2 m- 1
1.0197 x 10- 5 kgfcm- 2 m- 1
4.4208 x 10- 5 psi/ft
0.55 of/WOO ft

LENGTH

1m 39.37 inches
3.2808 feet
1.0936 yards
0.5468 fathoms
6.2137 x 10- 4 miles
106 microns (/l)
1010 angstrom (A)
216

PERMEABILITY (intrinsic)

1 darcy 10- 8 cm 2 (for practical purposes)

1 md 10- 11 cm 2

PRESSURE

1 pascal (pa) 1 Nm- 2

1 x 10- 5 bar
1.0197 x 10- 5 kg cm 2
9.8687 x 10- 6 atmospheres
10 dynes/cm 2
1.4504 x 10- 4 pounds/sq. inch (psi)
2.0886 x 10- 2 pounds/sq. foot

TIME

1 second 1.1574 X 10- 5 days

1.6534 x 10- 6 weeks
3.8023 x 10- 7 months (30.44 days)
3.1688 x 10- 8 years (365.25 days)

VELOCITY

3.6 km/hour
3.2808 feet/second
2.2369 miles per hour
1.9426 knots

VISCOSITY

1 Pa s 1 N m - 2 S = 10 poise = 103 cP
0.6720 pounds/foot· second (lb mass)
2.0886 x 1O- 2 1bf· sec/sq. ft
104 stokes = 106 cSt
1.5500 X 10 3 square inches/second
10.7639 square feet/second
 217

VOLUME

1 m3 219.969 Imperial gallons

264.173 U.S. gallons
8.107 x 10- 4 acre-feet
6.2898 barrels (bbl) of oil

WEIGHT (see also FORCE)

1 kg weight = 2.2046 pounds weight

 GLOSSARY

AQUICLUDE. A relatively impermeable rock unit that confines an aquifer (above,

below, or both).
AQUIFER. A porous and permeable, or fissured, rock unit that contains exploitable
water. Originally used for any water-bearing stratum, but probably meant what it
now means. Norton (1897, p. 130) revived 'a term of Arago's': but Arago (1835, p.
229) uses the normal French adjective 'aquifere' for water-bearing.
Arago, D.F.L. 1835. Sur les puits fores, connus sous Ie nom de puits artesiens, de fontaines artesiennes,
ou de fontainesjaillissantes. Annuaire, Bureau des Longitudes, 1835: 181-258.
Norton, W.H., 1897. Artesian wells oflowa. Iowa Geological survey, 6: 113-428.
AQUITARD. A rock unit that is not as permeable as an aquifer nor as impermeable as
an aquiclude.
ARTESIAN. Wells or aquifers in which the energy of the water is sufficient to raise it
above the land surface without the use of pumps are said to be artesian. An
aquifer is artesian if its total head is greater than the elevation of the land above
the same datum.
Some would define it in terms oflocal water-table rather than land surface, but
compare subartesian.
Name derived from Artesium, latin name for Artois, historical province of
northern France, where early artesian wells were constructed.
CAPILLARY FRINGE, RISE. The uppermost part of the zone of saturation above the
level in the water that is at atmospheric pressure; or the zone in which the water is
at less than atmospheric pressure; or the zone of 100% water saturation above the
water-table (sensu stricto). The c. rise is the thickness of the c. fringe.
COEFFICIENT OF PERMEABILITY. The coefficient K in Darcy's law when written
q = K L1h/1.1t is related to intrinsic permeability by K = kpg/'YJ. Dimensions LT- 1 .
Synonym: hydraulic conductivity.
DARCY. Unit of permeability based on incomplete expression of Darcy's law (see
Appendix, p. 207). Common unit, millidarcy (md). Dimensions L2.
DYNE. Unit of force in c.g.s. system. It is that force that gives to a mass of 1 gramme
an acceleration of 1 cm per second per second. Dimensions MLT- 2 .
ELASTIC STORAGE. See Storage coefficient.
EQUIPOTENTIAL SURFACE, LINE. A surface or line on which points of equal fluid
potential lie. Not synonymous with potentiometric surface. Synonym: isopoten-
 219

tial, which is not to be preferred, but is too commonly used to be dropped now.
FIELD CAPACITY. The water remaining in soil above the capillary fringe after extend-
ed period of drainage, expressed as percentage by weight of dry weight of soil. Cf.
Specific retention, Pellicular water. Dimensionless.
FROUDE NUMBER. Ratio of the products of two lengths and accelerations, Ll Ll T12 I
L2L2T:;2. In general, it is the number V'lldg, where Vis a velocity, d a character-
istic length, and g the acceleration due to gravity. It is interpreted as the ratio of
inertial to gravitational forces. In engineering practice, the Froude number is
usually taken as the square root of the number given above, i.e., Vljdg.
GEOSTATIC. Pressures and pressure gradients due to the gravitational load of the
total overburden are called geostatic. Undesirable synonym: lithostatic.
GROUND WATER. In practice, term restricted to usable subsurface water, fresh or
brackish.
HEAD. Refers to the vertical length of a column of fluid, usually liquid: ambiguous
when not qualified. It is an energy per unit of weight, with dimensions of Length,
not pressure.
Elevation head (less desirably, potential head): the potential energy per unit weight
due to elevation, i.e., the elevation of the point at which pressure p is measured,
above (+ ) or below (- ) an arbitrary horizontal datum. he = z.
Pressure head: the vertical length ofa column ofliquid supported, or capable of
being supported, by pressure p at a point in that liquid. h p = pi pg.
Velocity head: head due to the kinetic energy of the liquid. Negligibly small in
most geological contexts. h v = V2 12g.
Total head: the algebraic sum of the pressure and elevation heads (neglecting
velocity head).
Static head: pressure head in a borehole that is not producing, cf. static level.
HYDRAULIC CONDUCTIVITY. The coefficient K in Darcy's law when written
q = kAhl1. It is related to intrinsic permeability by K = kpglY/. Dimensions LT- 1 .
Synonym: Coefficient ofpermeability.
HYDRAULIC RADIUS. A measure of the size of the passage for fluids when the shape is
irregular. It is the area of transverse section of liquid divided by the wetted
perimeter, or equally the volume of liquid divided by the wetted surface area.
Dimension L.
HYDRAULIC GRADIENT. The difference of total head (see Head) divided by the
macroscopic length of porous material between the points where the total head is
measured - properly, the steepest gradient at a point in the fluid. Note that the
gradient of a potentiometric surface is not strictly the hydraulic gradient unless the
aquifer is horizontal (but it may be a sufficiently close approximation). Hydraulic
gradient is not the same thing as pressure gradient. Dimensionless.
INTERFACIAL TENSION. When two immiscible fluids are in contact in a capillary tube
or fine-grained porous material, the interface between the two liquids acts as if it
220

were an elastic membrane in a state of tension. Cf. surface tension. Dimensions

MT- 2 •
INTRINSIC PERMEABILITY. The component of permeability that is ascribable to the
porous material alone, independent of the physical properties of the fluid passing
through it. Dimensions L 2.
ISOPIESTIC LINE. Line of equal potential or total head. Synonym of equipotential,
which is to be preferred for the same reasons that potentiometric is to be preferred
over piezometric.
ISOPOTENTIAL SURFACE, LINE. A surface or line on which points of equal fluid
potential lie. Not synonymous with potentiometric surface. Synonym: equipoten-
tial, which will be preferred by those who do not mix greek and latin roots.
ISOPYCNIC SURFACE, LINE. A surface or line of equal density.
LITHOSTATIC. Undesirable synonym of geostatic and overburden (adj.) in context of
pressures and pressure gradients.
PELLICULAR WATER. Water in the vadose zone that does not move because it adheres
to the solid surfaces. Cf. field capacity.
PHI (¢) UNITS. Grade scale in sedimentology. ¢ = -log2 (diameter in mm).
PHREATIC WATER. Water in the zone of saturation, i.e., it is now synonymous with
ground water.
PIED. Old French measure of length. 1 toise = 6 pieds = 1.949 m, so 1 pied =
0.325 m. Pies were presumably about the same.
See letter to The Times, May 2, 1903, in: Gregory, K., (Ed.), 1978. The first cuckoo: letters to The
Times since 1900. Unwin Paperbacks, London, p. 45.
PIEZOMETRIC SURFACE. See potentiometric surface.
POISE. Unit of dynamic or absolute viscosity in c.g.s. system. A force of one gram will
maintain unit rate of shear between two surfaces of unit area unit distance apart
when the viscosity of the liquid between the surfaces is one poise (P). Units are:
dyne-second per square centimetre == gram per centimetre second == poise.
Dimensions ML -1 T- 1 •
POROSITY. A rock is said to have porosity if there is space between the solid grains. If
space is due to fractures, it is called fracture porosity. Quantitatively, it is the
volume of void space divided by the total volume of solids and void space, f
= vp/(vp + vs ). Usually spoken of as a percentage. Dimensionless. Related to void
ratio by f = 8/(8 + 1).
POTENTIOMETRIC SURFACE. The surface obtained by mapping the total head of an
aquifer. The potentiometric surface can be contoured (equipotential lines) and
fluid flow is in the direction normal to these contours, in the direction of decreas-
ing total head or potential. The synonym piezometric surface is not to be preferred
because it is a matter of potential, not pressure.
PRESSURE. Force per unit of area. The pressure at a point in a fluid is the force per
unit area acting on an element of volume that is small in relation to the bulk of the
liquid, but large in relation to the molecules. SI Units: pascal (pa) = newtons per
 221

square metre (N m- 2 ) = kg m- 1 s-z. Dimensions ML -IT- Z •

RETAINED WATER. Water in the saturated zone that will not drain by gravity (cf.field
capacity, specific yield).
REYNOLDS NUMBER. Ratio of the products of two lengths and velocities,
L1L1T;1/ LzLzT;l. Ingeneral,itis the number Vd/v, where Vis a velocity, dis a
characteristic length, and v is the kinematic viscosity. It is interpreted as the ratio
of inertial to viscous forces acting on the system.
RUNOFF. Rain water that flows on the surface of the ground to streams, rivers, etc,
without passing through porous and permeable soil or aquifers. River runoff, or
discharge, however, includes the ground-water component. Stormwater runoff is
synonymous with rain water runoff because this usually has the quality of a storm
since light rain leads to evaporation and infiltration, not runoff, unless the water
table is already at the surface.
SATURA TION. The saturation of a rock with respect to a fluid is the proportion of the
pore space filled with that fluid. Dimensionless.
SLUG. Unit of MASS in American usage, as distinct from the pound, which is the unit
of weight or force. A slug is the unit of mass that acquires an acceleration of 1 foot
per second per second when acted upon by a force of one pound weight. From
Newton's second law of motion, acceleration = force/mass. Thus, unit accele-
ration is acquired by 32.174 lb mass, which equals one slug. This is usually
rounded to 32.2 lb mass. For conversions, 1 slug = 14.594 kg mass.
SPECIFIC RETENTION. The proportion of soil or rock occupied by water adhering to
solid surface after gravity drainage. Dimensionless (volume/volume). Cf. specific
yield.
SPECIFIC STORAGE. See storage coefficient.
SPECIFIC SURFACE. Surface area of solids in bulk volume of porous material. Dimen-
sions L -1.
SPECIFIC VOLUME. The volume of unit mass of substance, reciprocal of mass density.
Dimensions M- 1 L 3 .
SPECIFIC WEIGHT. Weight divided by volume of substance or material, pg. Dimen-
sions ML -ZT- Z .
SPECIFIC YIELD. Volume of water that will drain by gravity from saturated soil or
rock, divided by the gross volume of the soil or rock. What is left behind is
retained water, hence sp~cific retention. Cf. storage coefficient, storativity.
STATIC LEVEL. The level of the water surface in a well when it is not producing and
has stabilized. This level may be given relative to the ground surface or the
bottom of the well; but if given relative to some horizontal datum surface becomes
synonymous with total head.
STOKE. Unit of kinematic viscosity, which is the absolute or dynamic viscosity
divided by the mass density of the fluid, IJ/p. Units: square centimetres per
second. Common practical unit, centistoke (cSt). Dimensions L Z T- 1 .
STORAGE COEFFICIENT, STORATIVITY. When the total head of an aquifer changes,
222

the volume of water in the aquifer changes. The storage coefficient is the volume
of water released from (or taken into) storage in unit volume of the aquifer per
unit decline (increase) in total head. It consists of two parts: one related to
the compressibility of the water, the other related to the compressibility of the
rocks skeleton (elastic storage). For an unconfined aquifer, storage coefficient
specific yield. Dimensions L 31 L 4 = L - 1.
It is also defined in terms of a vertical column through the aquifer, with unit
basal area. In this case, it is dimensionless.
SUBARTESIAN. Said of wells or aquifers in which the energy of the water is sufficient to
raise it above the water-table, but not above the ground surface. Cf. artesian.
There is merit in calling all wells in an artesian aquifer artesian, but there is merit
also in distinguishing those that need pumping by calling them subartesian.
SURFACE TENSION. In a capillary tube, or fine-grained porous material, the air Iwater
interface acts as if it were an elastic membrane in a state of tension, supporting a
pressure differential across it. This is surface tension. But it is defined in terms of
the work required to separate unit area of interface. Cf. interfacial tension.
Dimensions MT- 2.
TRANSMISSIVITY. For horizontal flow in an aquifer, the product of hydraulic con-
ductivity and the thickness of the aquifer. Dimensions L T - 1 L = L 2T - 1 .
VADOSE WATER, ZONE. Water in the zone of aeration descending to the water table.
(Derivation variously ascribed to latin vado = go, walk, rush, and vadosus =
shallow.)
VOID RATIO. Engineers prefer void ratio to porosity. The void ratio is the volume of
pores divided by the volume of solids. It is related to porosity by s = fl(1 - f).
WATER-TABLE. In practice, the free water surface of an unconfined aquifer in a well.
Strictly, it is the level in the water in the zone of saturation at which the pressure is
atmospheric, i.e., the base of the capillary ji-inge.
WEIGHT DENSITY. The weight of a substance or material divided by its volume, i.e.,
pg. Synonym: specific weight. Dimensions ML -2T- 2 .
WET, WETTED, WETTING. When two immiscible fluids saturate a porous material, one
preferentially adheres to the solid surfaces, excluding the other. The adhering
fluid is the wetting fluid or wetting phase, the other is the non-wetting fluid or
phase. Sands are usually water-wet.
ZONE OF AERATION. The subsurface zone above the zone of saturation in which the
voids are partly filled with air, partly with water and water vapour.
ZONE OF SATURATION. The subsurface zone below the level at which the voids are
entirely filled with water at hydrostatic pressure, i.e., strictly, below the capillary
fringe.
 POSTSCRIPT

POROSITY ESTIMATION FROM SONIC LOGS

Work done while this book was in press indicates that fractional porosity of
mudstones (with no carbonate) can be estimated from the formula

1=10 (Atsh - Atma)

Ato - Atma
where Ato is the transit time when porosityisio (both obtained from plots of Atsh and
porosity against depth, extrapolated to z = 0), Atma is the transit time when I = 0
- the so-called matrix transit time.
Reasonable values of the material constants are:fo = 0.5; Ato = 165 Ils/foot, 540
Ils/metre; Atma = 55 Ils/ft, 180 Ils/m. The formula therefore reduces to
1= (Atsh - 55)/220 for sonic logs in Ils/foot;
1= (Atsh - 180)/720 for sonic logs in Ils/metre.
Pore-water expulsion by compaction from porosity 11 to 12 can then be estimated,
as a proportion of bulk volume of mudstone at porosity 12, from
q= (f1 - 12 )/(1 -11 )
(see Chapman, 1972; reference on p. 204).
For sandstones and carbonates, the fractional intergranular porosity can be
estimated from the formula

This formula results from the assumption that the sonic path is through solids only,
and that porosity affects the length of this path (a sort of tortuosity in solids). It is
quite different in form and in theory from those in use at present, but it gives good
results and can be used when there is no core data. The following values of Atma are
suggested: sandstones, 55Ils/ft, 180 Ils/m; carbonates, 47! Ils/ft, 155 Ils/m. Local
experience may indicate other values.
It is a simple matter to make scales so that porosity can be read directly from the
sonic log.
 SUBJECT INDEX

Aach spring, Germany 98,101 dominant wavelength 157, 160, 168

abnormal pressures 124ff drawdown 85ff
-, association with growth faulting 156 drilling rate 134
-, shallow depth 143-144, 170 drillpipe fallacy 10-11
-, usually below sea-level 175 Dupuit assumptions 88-89
ambient fluid effect 137-138, 187-189
aquifers 73ff, 93ff effective compaction depth, equilibrium d.,
-, artesian 73, 79, 97, 108, 114-120 136-137
-, coastal 105-109 energy 28ff
-, leakage 114-115 equations, formulae
Archimedes principle 9 -, bulk density 11
-, capillary pressure 202-203
Bernouilli's theorem 30-32, 61 -, channel flow 33-34
Brunei 155-157 -, Darcy's 50
Buckingham's 11 theorem 37, 57-58 Drabbe-Badon Ghijben 105
buoyancy 7-11,14-15,105-109,169,177,187-189 -, drawdown 83-84, 88-89
-,0137
capillary pressure 202-204 -, Ghijben-Herzberg relationship 105
cement, effect on permeability 62 -, Hagen-Poiseuille 38
cementation factor 62, 196 -, harmonic mean 54
Chezy formula 34 -, heads 34, 73
clay mineral diagenesis, effect 143 -, hydraulic equivalence 45
clays, shales, mudstones 15, 126-131, 134, 136, -, hydraulic radius 27
138, 142, 143, 153, 155-170 -, liquid pressure 5
compaction of sediment 17, 124-131 -,A 134
-, Manning 33
Danube water to Rhine 98-101 -, mudstone compaction 126-131, 158
darcy, unit of intrinsic permeability 64, 207, 210 -, pipe flow 38, 46
-, poorly defined 207 -, Poiseuille 38, 46
Darcy-Weisbach formula 38 -, relative permeability 195-197
Darcy's law 49-70, 74, 81, 83,195-196,205-211 -, Reynolds number 36
-, expanded 56-57,60-61 -, settling tube calibration 40-43
-, limits of 64-66, 209-210 -, sliding slope 177
-, real critical Reynolds number 65-67 -, specific surface 54
deformation, models 157 -, Stokes' law 40
-, role of pore water 149-170, 180-189 -, Thiem 84
density 4, 13, 34, 105-108, 131, 158-159 equilibrium, stable 28
-, bulk II, 13-14
-, inversion 158-168 fields of flow 76-82
-, mean grain II fish, in wells, qanat 112
dimensional analysis 22-24, 37-39, 40-41, 57-59 flow, laminar 26
dimensionless numbers 39 -, steady 26
dimensions 24 -, turbulent 26, 36,66-67, 209-210
 225

-, uniform 26 Ohm's law analogous to Darcy's 61

flow net 74-75, 87, 90-92 osmosis, 101, 144
flow rate 27 overburden pressures 132
fluid potential 62-63, 73, 76, 78-80, 85, 101-106, overthrust fault, length of 184-187
113, 115-117, 136,208
fluids, ideal 2 Papua New Guinea 162-168, 182-189
fluids, properties of 1-4 pascal 4
fluidization 46 permeability, coefficient of 50,52-55, 57, 60, 62,
Formation Resistivity Factor 57, 60-62 65, 74, 78, 84
friction factor 38 -, - , dimensional analysis 57-59
Froud€: number 39, 41, 58 -, -, expanded 56-57, 60-61
permeability, effective 194-201
Ghijben-Herzberg lens, relationship 105-109, 196 -, intrinsic 60, 74, 195ff, 210
-, artesian 108 -, relative 195-201
Great Artesian Basin, Australia 114-120 poise 3
-, potentiometric surface maps 116-117 pore pressure, abnormally high 124-145,161,
-, water age 120 168-170,175
-, water velocity 120 -, -, other possible causes 140-145
growth faults, cause of flattening 152-153 -, - , related to stratigraphy 133-134, 146, 156
growth structures 149-160 pore pressure, abnormally low 101-104, 131
-, stratigraphic associations 152, 157 pore pressure, normal hydrostatic 11-18, 131
-, structural style 155-157 pore pressure to overburden p. ratio 134
pore water migration 70, 73ff, 124ff, 139-140,
head 30-32,39,61,63,73-74, 85ff, 106, 114-115 145-146, 191-193
hydraulic conductivity 50, 52-55, 57, 60, 62, 74, porosity lJ-J2, 53ff, 62, 66, 125-131, 134, 158
78,84 -, surface 56
-, dimensional analysis 57-59 potentiometric surface 74ff, 101-104
-, expanded 56-57, 60-61 pressure 4-7, 14ff, 28, 30, 39, 63, 73, 76, 80, 89,
hydraulic equivalence 43-46 124-145
hydraulic gradient 28,61-62, 73ff, 81, 83, 115 -, build-up 89
-, threshold 65 -, capillary 202-204
hydraulic mean diameter 43
hydraulic radius 27, 53, 62, 203 qanat 110-114
hydrological cycle 93-96
regressive sequences 134, 145-146, 149, 155ff, 160,
interfacial tension 12,202-204 168
-, associations 146, 157
Kalimantan, Indonesia 168-169 Reynolds number 35-36, 37, 39-41, 58-60, 65,
209-210
Kutter's n 33 Rhine, water from Danube 98-101
land subsidence 18-21 rivers and ground water 93-101
loading 17-18
sea-level, impediment to sliding 177, 187-189
Manning equation, M.'s n 33 settling tube, calibration 40-43
mean grain diameter 53, 54 sink 82, 90-92
-, harmonic 54, 62, 206-207, 211 sliding, critical length of block 180
-, hydraulic 43-46 -, field examples 162-170, 178-189
Mohr-Coulomb criterion 176, 183 -, gravity 175-189
mudlumps 160 -, lubricated 161-170, 180-189
mudstone, bulk flow 160 -, obstacles to 167, 183-189
mudstone compaction 126ff -, role of pore pressure 174ff
-, formulae 126-131 -, unlubricated 175-180
-, Hedberg's conclusions 126-128 sonic log 129-131, 138
-, undercompacted 158-161 source 81
specific discharge 52, 78
New Zealand, East Coast basin 169 specific gravity 4
226

specific storage 120 velocity, mean 28-29

specific surface 53, 61 viscosity 2-4
specific volume, weight 4 -, absolute, coefficient of, dynamic 3-4, 34
springs, causes of 93-96 -, kinematic 4,34,37,69
-,mound 115 -, newtonian 3-4, 180
Stoke 4 -, non-newtonian 3-4
Stokes' law 40-43 viscosity ofrocks, apparent v., equivalent v.
stream tube 82 157-158,161-170,180
stress, stress field 1,4, 14-15, 154-157, 176, 191 -, mantle 182
stress, effective 15-17, 120, 133-134, 136, 142 void ratio 11
-, neutral 15-17, 131-145 VYCOR, permeable glass 64-65
subartesian 114
surface tension 12,202-204 wall-sticking 10
water, connate, meteoric 124, 13(}
Tenerife, Canary Islands 109 water, fresh 105ff
Terzaghi's relationship 16, 18-21, 120, 133-134, -, -, deep 108, 196
136, 142 -, -, washing 124
thrust faults, length of 184-187 water, isotopic dating 120
tortuosity 55-57, 61, 206 water table 88-92, 105, 111
tracers in ground water 99-10 1 -, inclined 91, 106-108, III
transgressive sequences 145-146 water well, producing 82-89
turbidity currents, possible mechanism 180 water wells, fluctuating levels 18, 95
West Irian 170
undercompaction, electrical logs 134, 158 Wilmington oil field, California 19,21
 AUTHOR INDEX

Ackers, P. 34,47 Fair, G.M. and Hatch, L.P. 53, 59, 70, 206, 208,
Airy, P.L., Calf, G.E., Campbell, B.L., Hartley. 211,212
P.E., Roman, D. and Habermehl, M.A. 115, Fertl, W.H. 136, 147
120 Findlay, A.L. 162, 165,171
Archie, G.E. 57, 60, 61, 70 Fowler, W.A., Boyd, W.A., Marshall, S.W. and
Athy, L.F. 126-128,147 Myers, R.L. 153, 171
Australasian Petroleum Co. 162,170 Fiichtbauer, H. 62, 70

Bakhmeteff, B.A. and Feodoroff, N.V. 66, 70 Gabrysch, R.K. 18,20,25

Barker, C. 141, 147 Gibbs, R.J., Matthews, M.D. and Link, D.A. 40,
Barrer, R.M. and Barrie, J.A. 65, 70 42, 44-45, 47
Bear,J. 196, 198,201,204 Gorgy, S. and Salah, M.M. 95, 121
Beckett,P. 110, 112, 121 Graves, W. 113,121
Bett, K.E., and Cappi, J.B. 69, 70 Gray, D.E., 69, 70
Biot, M.A. and Ode, H. 157, 160,170
Blake, F.C. 53, 70, 205, 212 Habermehl, M.A. 115-120,121
Bottcher, H. 150,170 Halley, E. 94-95, 121
Bredehoeft, J.D. and Hanshaw, B.B. 64, 70 Hansbo, S. 64, 70
Breuninger, F.W., 98,121 Harkins, K.L. and Baugher, J.W. 134, 146, 147,
Bridgeman, M. 194,204 153,171
Brinkworth, B.1. 23, 25 Harza, L.F. 15,25
Brownell, L.E. and Katz, D.L. 66, 70 Hedberg, H.D. 126-128,147,201,204
Bruce, C.H. 152,170 Heezen, B.c. and Drake, c.L. 175, 178-179,190
Buchan, S. 97, 121 Herzberg, B. 105,121
Buckingham, E. 23, 25, 37, 47 Hill, G.A., Colburn, W.A. and Knight, J.W. 101,
Burst, J.F. 143,147 103-104,121
Butler, M.A. 110, 112, 121 Horner, D.R. 89, 92
Hsii, K.J. 177,190
Carman, P.C. 207-208, 211, 212 Hubbert, M.K. 49, 53, 61, 63, 66, 71, 78, 90, 92,
Chapman, R.E. 125, 128, 134, 137, 142,147, 154, 106,121,155,171,176,190,191,204,208-209,
168-170, 175, 177-178, 185,189,201,204 211,212
Cowin, S.c. 12,25 Hubbert, M.K. and Rubey, W.W. 10, 15-16,25,
Crittenden, M.D. 182,189 134,147,152,171,174,176,190
Hubbert, M.K. and Willis, D.G. 191,204
Dailly, G.c. 124,147, 160, 170, 180,189 Ineson, J. and Downing, R.A. 97-98, 121
Darcy, H. 34, 36, 39,47,49,70,210,211
Davis, S.N., and De Wiest, R.J.M. 89,92· Jacob, C.E. 18, 19,25
De Wiest, R.J.M. 89,92 Jenkins, D.A.L. 162, 165,171,182,185-187,190
Debye, P. and Cleland, R.L. 64, 65, 70
Dennis, J.G. 152,171 K1iss, W. 99-100, 121
Dickinson, G. 132-133, 140,147,153,171 K1iss, W. and Hotzl, H. 99, 121
Drabbe, J. and Badon Ghijben, W. 105-106,121 Katz, H.R. 143, 147, 170, 171
Dron, R.W. 150,171 Kehle, R.O. 161,171, 180, 182, 190
228

Keller, G.V. 61, 71 Riley, S.J. and Bryant, T. 41-42, 48

King, F.H. 18,25,70,71,90,92,95,122 Rittenhouse, G. 45, 48
Kozeny, J. 53, 59, 71,205-206,211,212 Roberts, I., see Cowin, S.C.
Krumbein, W.e. and Monk, G.D. 209, 210, 212 Roberts, J.L. 144,148
Kugler, H.G. 21,25 Rose, H.E. 209, 212, 213
Rose, W.D. 62, 71
Laing, A.e.M. 169, 171 Rubey, W.W. 40, 48
Latham, B. 91, 92 Rubey, W.W. and Hubbert, M.K. 128, 136, 148,
Lemcke, K. 101, 102, 122 174-175, 190
Lepine, F.H. and White, J.A.W. 167,171
Leverett, M.e. 195, 197-198,204 Sandford, K.S. 110, 111,122
Levorsen, A.I. 198, 204 Schaub H.P. and Jackson, A. 155-156,172
Liechti, P., Roe, F.W. and Haile, N.S. 155, 157, Scheidegger, A.E. 65, 71
171 Schneebeli, G. 53,66-67, 71,209-210,213
Lindquist, E. 53, 66, 71, 196, 204, 212 Schriever, W. 58-60, 65, 67, 71,210,213
Lutz, J.F. and Kemper, W.D. 64, 71 Sibson, R.H., Moore, J.McM. and Rankin, A.H.
Lyell, e. 112,122, 160,171 144, 148
Slichter, e.S. 59, 71,210,213
Magara, K. 64, 71, 128, 141-144,147 Smith, A. 112, 122
Mariotte, E. 94, 122 Smith, N.E. and Thomas, H.G. 141, 148
Maxwell, J.e. 125, 147 Smoluchowski, M.S. 173, 175, 180, 190
Mayuga, M.N. 19,21,25 Spinks, R.B. 143,148
Miller, R.J. and Low, P.F. 64, 71 Stephenson, L.P. 125,148
Morgan, J.P., Coleman, J .M. and Cagliano, S.M. Stokes, G.G. 4, 40, 48
160,171 Street, N. 57, 71
Murray, G.E. 153,171 Stutzer, O. 150, 154, 172

Nielsen, R.F. 65, 71 Tanner, W.F. and Williams, G.K. 157,172

Noel, E. 110, 112, 122 Terzaghi, K. 14, 15, 18,24,25, 133, 148, 174, 176,
Nordberg, M.E. 64, 65, 71 190
Norton, W.H. 62, 71,112,122 Thesiger, W. 113,122
Thiem, G. 84, 92
Ode, H. 183, 190 Thorsen, e.E. 153-154,172
Ogilvie, e. 115, 122 Tiddeman, R.H. 150,172
Ohrt, F. 108, 122 Todd, D.K. 89,92
Olsen, H.W. 64, 71 Tornebohm, A.E. 173,190

Pankhurst, R.e. 23, 25 Van Bemmelen, R.W. 168, 172

Parker, T.J. and McDowell, A.N. 157,171 Veatch, A.e. 18,25,95,122
Perreault, P. 94, 122 Versluys,J. 61-62,72,105-106,108,122,141,148,
Perry, E. and Hower, J. 143,147 196,204
Plain, G.J. and Morrison, H.L. 65, 71 Visser, W.A. and Hermes, J.J. 143,148, 170,172
Poiseuille, J.L.M. 3, 34, 36, 38-40,46-47,48 Von Engelhardt, W. and Tunn, W.L.M. 64, 72
Powers, M.e. 143,148
Watson, R.L. 40, 48
Queensland, Artesian Water Investigation Com- Weaver, e.E. and Beck, K.e. 143, 148
mittee 114, 122 Weeda, J. 169,172
Weller, J.M. 128,148
Ramberg, H. 157,171,180,190 Wentworth, e.K. 108, 122
Reyer, E. 173, 174, 190 Whitaker, W. and Thresh, J.e. 97,122
Reynolds, O. 34-35, 37-39, 48 Whitehouse, F.W. 115, 123
Rhys Price, see Kugler, H.G. Wyckoff, R.D., Botset, H.G., Muskat, M. and
Ridd, M.F. 162, 169,171 Reed, D.w. 207, 210, 213