CONTROL OF SEEPAGE THROUGH FOUNDATIONS

AND ABUTMENTS OF DAMS*

by
ARTHUR CASAGRANDE

” To pass judgement on the quality of a dam foundation is one of the most difficult and responsible tasks.
It requires both careful consideration of the geological conditions and the capacity for evaluating the hydrau-
lic importance of the geological facts which can only be obtained by a thorough training in the hydraulics of
seepage.” Karl Tertaghi, 1929

I am deeply grateful to the members of the British National Committee for the great
honour of inviting me to be your first Rankine Lecturer. The name Rankine was, of course,
familiar to me in my student days. In my mind I had classified him in the same category
with such eminent German engineers and teachers as Otto Mohr, Mtiller-Breslau, and Foppl.
But until recently I had no conception of the enormous breadth and depth of Rankine’s con-
tributions in several areas of engineering as well as in pure science. And all this he accom-
plished in his short life span that I have exceeded already by 6 years. I could easily use the
entire hour to talk about my impressions when reviewing Rankine’s books and scientific
papers. But may I mention just one item that concerns the conflict between theoretical
science and engineering, a topic often discussed at the present time particularly in the United
States where we are undergoing a period of critical review of engineering education and are
groping for something new that nobody seems to be able to define clearly. As always in
periods of uncertainty, there is a tendency to be over critical of past efforts; there is a danger
of “throwing the baby out with the bath water”-at least that is the impression I have about
certain changes which are being attempted in the teaching of civil engineering at some of our
schools. Let me read two short paragraphs that bring out in essence what Rankine thought
about this conflict that seems to have already existed in his days. My quotation is from his
inaugural address when he accepted the professorship at the University of Glasgow, in 1856.
This is what he said more than 100 years ago:
“In theoretical science, the question is-What are we to think?-and when a doubtful
point arises, for the solution of which either experimental data are wanting, or mathematical
methods are not sufficiently advanced, it is the duty of philosophic minds not to dispute
about the probability of conflicting suppositions, but to labour for the advancement of
experimental inquiry and of mathematics, and await patiently the time when these shall be
adequate to solve the question.
“But in practical science the question is-What are we to do?-a question which in-
volves the necessity for the immediate adoption of some rule of working. In doubtful cases,
we cannot allow our machines and our works of improvement to wait for the advancement
of science; and if existing data are insufficient to give an exact solution of the question, that
approximate solution must be acted upon which the best data attainable show to be the
* Lecture delivered on 25 January, 1961, at the Institution of Civil Engineers.
161

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 162 PROFESSOR ARTHUR CASAGRANDE

most probable. A prompt and sound judgement in cases of this kind is one of the character-
istics of a PRACTICAL MAN, in the right sense of that term.”
This is, indeed, the best definition of the difference between theoretical science and engineer-
ing I have been able to find. Also, it serves as an admirable introduction to the subject of
my Lecture in which I will make use of theory to supplement empirical knowledge and to
enhance “sound judgement”.

CONTROL OF SEEPAGE THROUGH ROCK

IlYTRODUCTION

By control of seepage I refer to all measures which the design engineer has at his disposal
in order to protect a dam, its foundation, and its abutments against any undesirable or danger-
ous effects of seepage. The seepage losses per se are more often than not of secondary con-
sideration. In fact, our control measures may cause a substantial increase in seepage rather
than a decrease.
The design assumptions for uplift beneath concrete dams 30 years ago were very con-
servative. Gradually, on the basis of uplift measurements on numerous concrete dams, one
began to realize that with a combination of a grout curtain and a line of drainage holes the
uplift could be reduced to much smaller values than those assumed in the design. Today the
largest builders of concrete dams in the United States, the U.S. Bureau of Reclamation, the
Tennessee Valley Authority, and the U.S. Corps of Engineers, use similar assumptions which
are a straight-line drop from reservoir level at the heel of the dam to a certain fraction of the
difference in head between reservoir and tailwater along the line of the drains; and from there
another straight line to tailwater elevation at the downstream toe. The value of that fraction
at the line of drains was gradually reduced during the past 20 years and is now one-third for
the Bureau of Reclamation,1 * and one-fourth for the Tennessee Valley Authority.s* These
uplift values are assumed to act over lOOo/oof the base area. While several independent
organizations have arrived on an empirical basis at similar design assumptions, there is no
agreement among designers on the relative merits of the grout curtain and the line of drains.
I have noted in recent years a growing awareness that a grout curtain consisting of a single
line of holes may be very unreliable. Leading engineers have expressed to me such views
privately, but they seem to be reluctant to state so publicly. It seems as if they were afraid
of attacking something that is believed by a majority in the profession almost like a religious
dogma. Others cloak their doubts in statements such as: “We consider the grout curtain
good insurance; but in our design we rely only on the drainage.”
In 1952 the Tennessee Valley Authority published a book entitled “Civil and structural
design”,3 summarizing their design practice at that time. From its chapter on “Foundation
cut-off and drainage” is reproduced Fig. 1 which shows typical uplift observations for a con-
crete gravity dam. From a foundation gallery, drainage wells extend to a depth of about
40 ft, on 8ft centres. Not shown on the original Figure in that book is the grout curtain. I
obtained that information from another source and added it on that drawing. It consists of a
single line of 3-in. grout holes, 4-8 ft inside the heel of the dam, spaced at an average 5 ft, with
every third hole extending to a depth of 80 ft, the others to about 40 ft.
In the cross-section reproduced in Fig. 1, the measured uplift plots as a straight-line drop
from full reservoir level at the heel to tailwater elevation at the line of drain holes. Down-
stream of the line of drain holes the uplift rises only very slightly above tailwater, such that
there is only an insignificant magnitude of uplift in the zone downstream from the line of
* Superior figures refer to list of references on p. 181.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 163
drain holes in excess of the tailwater level. Later I will show that this type of uplift distribu-
tion can be readily explained by means of a theoretical analysis, provided that the drainage
wells penetrate through the more pervious zone of the rock. If the grout curtain would be
reasonably effective, then the observed uplift between the heel and the line of drains should
not be the practically straight line as indicated in this cross-section.
Although in the published design assumptions for uplift, the existence of a grout curtain
is not even mentioned and seems to be ignored, that does not mean that the &signers con-
sider the grout curtain of secondary importance; on the contrary, in most publications primary
stress is placed on the discussion of the
HW
grout curtain. By comparison, discussion
of the drain holes is usually very brief.
When considering, in addition, that the
drain holes are always spaced much
farther apart than the grout holes, and
that their depth is generally only about
one-half of the depth of the grout curtain,
I get the impression that the drain holes
are treated like a step-child. Certainly,
the cost of such a line of drain holes is
small as compared to the cost of the grout
curtain. Obviously there is a contra-
diction which is perhaps most easily
explained by the fact that the relative
merit of grout curtains and drainage holes
has been a highly controversial subject
for a long time. The most frequent
statement one finds in publications, and
on which there seems to be some measure
of agreement, is that the purpose of
grouting is to control the rate of seepage
beneath the dam and the purpose of
drainage is to relieve uplift. However, I
cannot see how these two effects can be
separated in this simple manner. Any
substantial reduction in seepage by means
of a grout curtain must also reduce the
Fig. 1. Foundation uplift pressures, Hiawassee
DtllXl (Reproducedfrom Fig. 54 of Ref. (3)) Iuplift pressures downstream of the grout
, curtain. If the piezometric surface be
tween the heel and the drain holes is practically a straight line, as in Fig. 1, then the grout
curtain is obviously not doing much good, while the drainage is doing an excellent job in
controlling the uplift pressures in the area between the line of drains and the down-
stream toe.

ANALYSIS OF DRAINAGE

The only theoretical analysis of the effect of drainage of rock beneath concrete dams with
which I am familiar is that published by the late Dr Brahtz* of the Bureau of Reclamation,
and which was later extended by the Tennessee Valley Authority. 5 Fig. 2, reproduced from
T.V.A. Technical Monograph No. 67, shows the theoretical results for various locations of a
horizontal drain upon the uplift. The rock is assumed semi-infinite in extent and of isotropic
permeability, with Darcy’s law valid. The location of the drain closest to the heel gives, of

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 164 PROFESSOR ARTHUR CASAGRANDE

'Lineof drain wells

Fig. 2. Theoretical uplift pressure reduction caused by a drainage system which is effective at
the foundation level
(Reproduced from Fig. 36 of Ref. (5))

course, the largest reduction in uplift; but even for that case the reduction is far less and of
different shape than the majority of actual uplift observations on concrete dams. One should
‘not really expect good agreement with this theoretical solution, because replacement of deep
[penetrating wells by a surface drain, and assumption that the rock is of uniform permeability
lto great depth, are excessive deviations from the usual conditions.
A more realistic analysis should not neglect the fact that the wells are intended and should
penetrate through the zone of relatively pervious rock. In addition, one should take into
account that the hydrostatic pressure against the dam, which is transmitted into the founda-
tion, may cause joints along the heel of the dam to open up and full hydrostatic pressure to
extend more or less throughout the pervious zone of the rock along a vertical extension of the
upstream face. The joints need not be parallel to the dam axis; even an irregular system may
develop such an effect. If we make these assumptions, we arrive at the case illustrated in
Fig. 3, consisting of a pervious foundation layer of limited depth with a vertical entrance face
below the heel of the dam, and with a row of drainage wells located at a distance d from the
heel. One can readily see from this picture that the drainage wells should be effective in
controlling uplift downstream from the line of wells, provided of course that the wells are
deep enough to really penetrate the pervious zone.
In Fig. 4, I summarize my analysis on the basis of the assumptions just stated, namely a
vertical entrance face, such as along a waterfront, into a confined pervious stratum of thick-
ness D, with an infinite row of wells parallel to this entrance face at the distance d, with a well-
spacing a, and a well radius rW. Basically, the solution of this problem is contained in
~Muskat’s monumental book published in 1937.6 In terms of the drawdown s at any point
(x, y) it can be expressed by the long equation at the top of Fig. 4. From this general
equation * one can derive several results which are of particular significance to our problem as
discussed in the following paragraphs.
For any given difference in elevation h, between reservoir and tailwater levels, and given
* Extensive use of the Muskat solution has been made by the Corps of Engineers in connexion with
investigations of relief wells for earth dams and levees. In that problem, as will be discussed in a later
Paper, the main interest lies in the shape of the piezometric surface upstream of the wells and in the maximum
head between the wells, because relatively large well-spacings are used. However, beneath concrete dams,
we are principally interested in the shape of the piezometric surface downstream of the line of drains and
in the relative water levels in the drain holes and of the tailwater.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 165
dimensions d, a, and rho,there is a definite water level in the wells which will ensure that all
seepage entering the pervious stratum is removed by the wells, and that the piezometric
surface downstream of the wells will be horizontal. Only near the wells there is some flow
also downstream of the wells, but which merely curves back into the wells. For practical
purposes the horizontal piezometric surface is established beyond a distance of about one-half
of the well-spacing downstream of the wells. The water level in the wells is 12, below the
tailwater level, and halfway between the wells the piezometric surface is h,,, above the tailwater
level. *

’Well 1 Well i Area1h...t h..

Fig. 3. Numerical examples

for full control of uplift by
single line of drainage
wells

lmperwour rock

The piezometric surfaces plotted in Fig. 4 were computed for a = d. If the well-spacing a
is much smaller than d, as is generally the case in the foundation treatment of concrete dams,
then h, and 12, become very much smaller in relation to h, than shown in Fig. 4. It is really
the ratios h,/hc and h,/h, (for convenience expressed in per cent.) which are of primary interest
to us. Working formulas for these ratios are given in Fig. 4. In their derivation certain
mathematical simplifications were introduced, but which still ensure results of slide-rule
accuracy, provided we keep the well-spacing smaller than 3d, and the ratio of well radius to
well-spacing smaller than 0.1. Both conditions are generally futilled for the foundation
drainage below concrete dams. The “well-level ratio” h,/hc is a function of the dimensionless
ratios a/d and a/2r,. However, the “midwell-level ratio” h,/he is only a function of the ratio
a/d. It is independent of the well diameter because the well diameter only influences the
shape of the drawdown surface close to the well and the water level in the well. Once Iz, is
given or assumed, the piezometric surface is determined, except close to the wells and in the
wells, and the rate of seepage qw toward each well is also determined. The thickness of the
water-bearing stratum D has no effect on the piezometric surface, and only enters in the rate
of seepage which can be expressed simply as the flow through the layer of thickness D under a
hydraulic gradient of he/d. As shown by the equations in Fig. 4, the rate of seepage may be
expressed either as rate of flow per unit width 7, or as rate of flow toward each well qco= aij
* For convenience, tailwater is used as reference plane for measuring h values. The positive sign is
used for h values above tailwater elevation, and the negative sign for h values below tailwater elevation.
The signs for s are opposite those of h.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 166 PROFESSOR ARTHUR CASAGRANDE

In the Table in Fig. 3 are listed numerical results for a 400-ft-high concrete dam, for the
line of wells arranged 50 ft from the heel, and computed for two cases of well-spacings, namely
10 ft and 5 ft, and for well diameters of 3 in., 6 in., and 12 in. For a well-spacing of 10 ft and
well diameter of 3 in.-which is probably the most common combination for concrete dams
in the United States-the water level in the drain holes must be kept about 8% of h, below

Known or assumed: hc, a, d, rw, D, k

Determine: hm, hw, qw, or 4

Working formulas (slide rule accuracy) for h, and

h, when$<3and? <O.l:

h, _ --ed 1% ..& = -0.366 ; log,, ;

h,- 7r Trw w
hm
a 1122 = 0.110;
h, = 2nd

ij = kh,; Shape factor = i

T(perunit qw = kh,a :
k width)

Fig. 4. Theoretical solution for line of drainage ~011s pandIe to vertical entrance face and no
seepage downstream of wells

tailwater level in order to eliminate all excess uplift downstream of the line of drains. How-
ever, if for the same well diameter we reduce the well-spacing to 5 ft, then the water level in
the drain holes needs to be only about 3% of h, below tailwater level. If we maintain the
lo-ft spacing, but increase the well diameter from 3 in. to 12 in., the water level in the drains
has to be depressed about 4%. In other words, the 3-in. drain holes at 5ft spacing are not
only cheaper but somewhat more efficient than the lBin.-dia. drain holes at IO-ft spacing.
In all these examples the rise of the piezometric surface halfway between the drain holes is
only about 1% or 2% of h,, thus negligible.
In the above and subsequent numerical examples the head loss in the wells was disregarded.
Under certain conditions this factor should not be neglected.
The mathematical solution of Fig. 4 can be expanded to include the case when the water
level in the wells is at the same elevation as tailwater, or even higher. Then a certain amount
of seepage will continue beyond the line of drainage holes and some uplift in excess of tail-
water will develop downstream of the line of drains. This combination of flow can be solved
by direct mathematical superposition of seepage through the pervious stratum under a uni-
form gradient and the solution presented in Fig. 4. Such direct superposition is mathemati-
cally correct because both systems of flow are solutions of the Laplace differential equation.
The results of the analysis, for the water level in the drain holes at tailwater level, and again

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 167

Known or assumed: ht,--a, d, b, r,, D, k,

Determine: Uplift u,, 4, qt
(1) Compute first auxiliary quantity 12,:

(2) Compute hydraulic gradients it and i,:

h - hc
Downstream of wells: it = b + d
allwater
Upstream of wells: i, = 2 + it

(3) Uplift at line of wells:

uw = b.it
(4 s=IeP%& flow
4 = kDi,
Flow passing wells: 4t = kDit
Ratio passing wells: it/g = it/i,
Fig. 5. Theoretical solution for water level in drainage wells at tailwater level

with the limitations a/d < 3, and y,/a < 0.1, to permit mathematical simplifications, are
shown in Fig. 5. It is convenient to compute first a fictitious tailwater Iz, for which there is
no flow downstream of the drains, and such that if one superposes a constant gradient i,, the
new tailwater elevation and the water level in the wells will coincide. The total uplift down-
stream of the drainage line is the triangular hatched area, with the maximum uu, at the line
of wells. In Fig. 6 are shown the computed results for the same cross-section used before, for
drain-hole diameters of 3 in., and the two well-spacings of 10 ft and 5 ft. At the line of drains
the uplift is 6.8% and 4.5% of the triangular uplift area for a straight-line drop from head to
tailwater. Thus, if the drainage gallery is located at the elevation of tailwater, the magnitude
of the uplift downstream of the wells is still very modest.
In Fig. 7 is presented the solution for the case when the drainage gallery is at an elevation
higher than tailwater. This solution is useful for analysing some of the observed uplift data.

Fig. 6. Numerical examples for water

level in drainage wells at tailwater
level

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 168 PROFESSORARTHURCASAGRANDE

Known or assumed: ht, Ah,, a, d, b, rw, D, k

Determine: uplift uw, 4, qt
(1) Compute first auxiliary quantity h,:

ht- Ahwv
h, =
l a.bdJn.a
‘+z;;‘;i b 2nrw
(2) Compute hydraulic gradients it and i,:
ht - hc
Downstream of wells: it = ~
b+d
h,
Upstream of wells: i, = - + it
d
(3) Uplift at line of wells:
uw = b.it + Ah,
(4) Se;eIg flow
? = kDi,
Flow passing wells: it = kDit
Ratio passing wells: ;it/? = it/i,
Fig. 7. Theoretical solution with water level in drainage wells higher than tailwater level

ANALYSIS OF IMPERFECT CUT-OFFS

In an attempt to illuminate the question of efficiency of a grout curtain consisting of a

single line of holes we make use of the theory of imperfect cut-offs which was published by
Dachler (1936) in his book “Grundwasserstromung”. 7 The solution shown in Fig. 8 expresses
the efficiency of a very thin cut-off wall which has narrow slits. Whether the slits are hori-
zontal or vertical is of no significance. If D is the total thickness of a pervious layer, d that
portion which is made positively impervious by the cut-off, and W = D - d = total width
of open spaces, then d/D is called the cut-off ratio and W/D the open-space ratio. For example,
a cut-off ratio of 0.99 would mean that 99% of the area does not let a drop through, and when
we add up all the small openings that are not sealed, we obtain a total of 1o/o of the area, or an
open-space ratio of 0.01. The cut-off efficiency E, can be expressed as the ratio (4 - ~J/Q,
where q is the rate of flow without a cut-off, and qc the rate of flow with the cut-off. When
qc is zero, we have a perfect cut-off, i.e. a cut-off efficiency of unity or 100%. For qe = q, the
cut-off efficiency is zero.
Dachler’s equation for the cut-off efficiency E, can be simplified for open-space ratios of
W/D < 0.1, with slide-rule accuracy, to the working equation shown in Fig. 8 and plotted for
B/D = 1, and for the number of openings n = 20 and n = 120. If we further consider only
cases with large numbers of openings (say n > 20), and equal open-space ratio, it can be seen
that the cut-off efficiency decreases approximately inversely proportional with the number of
openings. (This paradoxical conclusion loses its validity when dealing with thick walls.)
THEORETICAL CONSIDERATIONS CONCERNING EFFICIENCY OF A SINGLE-LINE
GROUT CURTAIN

A theoretical analysis of the hydraulic efficiency of a single line of grout holes in rock would
require so many assumptions that the results would be of little value. However, the pre-
ceding analysis is useful to make plausible the reasons why single lines of grout cut-offs have
proved inefficient in the majority of cases for which reliable observations are available. For
example, we could resort to the following question: Is it reasonable to assume that a single-line
grout curtain could be considered as good a cut-off as a steel membrane which has a slit &-in.
wide spaced every 5 ft? This would correspond to an open-space ratio of 0.1%. Offhand,

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 169

Cut-off efficiency E, = y

In sin g
E, =
In sin br - g

(Dachler 1936)
Simplified for W/D < 0.1 (slide rule
accuracy) :

0’
0
’
I
’
2
’
3
’
4
’
5
’
6
’
7
’
8 9 IO

OPEN SPACE RATIO : W/D IN PER CENT

Fig. 8. Theoretical efficiency of imperfect cut-off

one might well question whether such perfection could ever be achieved by means of a single-
line grout curtain. Assuming B = D = 100 ft, the 5-ft spacing would make n = 20, and
this would correspond to the upper curve in Fig. 8. For these assumptions the theoretical
cut-off efficiency is 297;. In other words, 71% of the rate of flow without cut-off would still
be going through these thin slits. If we divided the same 0.1% of the area into a larger
number of smaller slits, the cut-off efficiency would drop to even lower values.
Based on a review of instances where reliable observations of the piezometric surface were
made both sides of a single-line of grout holes in rock and not supplemented by additional
grouting after filling of the reservoir, I have come to the conclusion that a 30% efficiency has
rarely been exceeded in the past for most conditions. Could then a designer dare to rely on
a single-line grout curtain, when analysing the safety of the dam? And when he considers
seepage losses, the cases when a reduction of seepage by the order of 30% would warrant the
high cost of a grout curtain would be a small minority. To achieve the same result in the
case of a concrete dam with a wide base, one could simply move the line of drains farther
downstream by one-third of its distance from the heel of the dam; e.g. instead of arranging the
drains 30 ft from the heel, this distance could be made 40 ft, which would achieve the
desired reduction in seepage in a more positive manner, and merely at the expense of a small
increase in uplift. Or, one could provide an impervious zone or blanket upstream of the
dam, as discussed later.

DISCUSSION OF HYPOTHETICAL EXAMPLES

Before proceeding to a presentation of a number of uplift measurements on concrete dams,
it will be helpful to consider in Fig. 9 hypothetical uplift diagrams for different assumptions
concerning the grout curtain, the line of drains, and the geologic conditions.
(a) A perfect grout cut-off through the pervious zone would result in full head upstream
of the cut-off, then, at the cut-off line a sharp drop to the elevation of tailwater, and from the
cut-off a horizontal line extending to the downstream toe.
(b) A reasonably effective grout cut-off through the pervious zone, but without drainage,

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 170 PROFESSOR ARTHUR CASAGRANDE

would be reflected by an uplift diagram such as shown in (b). It is what I would hope to
achieve under the most favourable conditions from a single line of grout holes. But it is not
really “reasonably effective” from the standpoint of reduction in seepage, because with the
slopes as shown in this sketch, the reduction of seepage would be only about 50%, and that I
consider not good enough to justify the high cost of a grout curtain.
(c) If we add to (b) also a line of drain holes, then we obtain an uplift diagram which still
shows a well-defined drop in head along the grout cut-off.
In Fig. 9 (d) and (e) are repeated the uplift diagrams for a line of drain holes of appropriate
depth, but without a grout curtain. We are already familiar with these patterns from the
preceding discussion, namely in (d) with water level in holes below tailwater, and in (e) above
tailwater. The majority of the observed uplift diagrams show, in fact, good similarity with
these two patterns.
Patterns (c), (d), and (e) are based on the assumption that the drain holes extend through
the pervious rock stratum, and that there are no unusual geological complications. It is a
surprising fact that the shallow drain holes which have been used on most dams are quite
effective, as will be shown later. It indicates that at the majority of sites the pervious founda-
tion zone is relatively shallow. This may not be true for abutments for which unfortunately
very few observations are available.
In Fig. 9 (f), (g), and (h) are shown three examples of the deviations from the ideal pattern
that one may encounter.
In (f) the drain holes are assumed to penetrate only partially into the pervious zone. That
would cause a large bulge in uplift pressures downstream of the drains, similar to the Brahtz
solution illustrated in Fig. 2, p. 164.
In (g) stratified rock is assumed dipping upstream, with a pervious layer sandwiched
between impervious strata, with the single-line grout curtain extending through the pervious

(a) Perfect (d) Effective drainage

L
grovr cut-off
T

below tailwater
(g) Geologic conditions
cause excessive uphlt
lnruffic~enr depth
I, of dram holes

I
I,,, “,,,,,,,,,,,,”

(e) Effective drainage

(h) Geologic conditmnr

ca”se dangerous
pore pressures on rock

Fig. 9. Hypothetical examples

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 171

layer, but the drain holes stopping short of it. As in the other cases, we also assume that
entrance of water is gained just upstream of the dam by means of joints that open up under
the effect of the hydrostatic pressure acting against the dam; and that the pervious layer
meets the base of the dam in such a manner that drainage is prevented. Such a combination
may cause an uplift diagram which shows beneath a downstream zone of the dam uplift
pressures almost equal to reservoir level. If uplift measurements are made at a sufficient
number of points in a cross-section, we might be warned in time of the development of this
dangerous condition. We could lower the reservoir, investigate the cause, and then extend
the drain holes through the pervious layer. We might also drill supplementary drainage holes
near the downstream toe through the dam into the pervious stratum.
In (h) we assume a somewhat similar geological situation, except that the pervious layer
does not come to the surface, and that downstream of the dam it is cut off by an impervious
fault. The drain holes will control the uplift along the base of the dam at approximately
tailwater, particularly if a slightly more pervious surface zone should exist. But such uplift
observations along the base would create a false sense of security. Actually, there exist
dangerously high uplift pressures in the pervious stratum not far below the base of the dam,
and which extend downstream of the dam. Here again, drilling the drain holes much deeper
could readily eliminate all danger. Effective relief of the dangerous pressures could also be
achieved by drainage wells drilled along the downstream toe of the dam through the pervious
stratum. To discover such conditions we would need not only uplift measurements along the
base of the dam, but numerous piezometers extending to various depths into the rock. A
similar example was discussed by Terzaghi in the paper which he presented in 1929 before
the American Institute of Mining and Metallurgical Engineers.13
From these hypothetical examples we may draw the following conclusions:
(I) Considering the low cost of drain holes, it seems logical that they should be drilled to a
depth at least half the height of the dam, instead of 40 or 50 ft, as was used for most concrete
dams in the past. Local geologic conditions may indicate the need for still greater depths.
(2) Where the geologic conditions create the slightest doubt as to the control of hydro-
static pressures in various rock zones below the dam and in the abutments, not only uplift
measurements along the base, but piezometer observations at many locations in the rock are
needed.
For anyone with a good knowledge of engineering geology and of solving seepage problems
by means of plotting flow nets, it is easy to invent combinations of geological details which
can cause serious pressures in the rock foundation and abutments downstream of a dam. I
consider it such an instructive exercise that I encourage my students to indulge in this
pastime.

UPLIFT OBSERVATIONS FOR TESXESSEE VALLEY AUTHORITY DL\MS

The uplift diagram for the Hiawassee Dam, Fig. 1, consisting of a straight line from reser-
voir level to tailwater at the line of drains, and then practically a horizontal line at tailwater
level for the area downstream of the drains, is an ideal example confirming the preceding
theoretical analysis for the performance of drainage wells. The Authors of T.V.A.‘s book on
Design3 selected it as a representative uplift diagram. The foundation rocks at this dam are
steeply dipping, intensely jointed quartzites, and schists.
In Fig. IO,10 are plotted the average uplift measurements for four T.V.A. dams (Fontana
480 ft, Hiawassee 307 ft, Cherokee 202 ft, and Douglas 175 ft). Upstream of the drains the
mean of the averages drops off from reservoir level in form of a slightly concave curve, with
no indication of the location of the grout curtains. The average water level in the drains is
approximately at tailwater level. The average uplift downstream from the drains is about
1Oqg of a linear theoretical drop for the condition without drains. This is certainly very safe

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 172 PROFESSOR ARTHUR CASAGRANDE

PERCENTAGE OF BASE WIDTH

Fig. 10. Composite foundation uplift pressures for four dams of the Tennessee Valley Authority
r (Reproduced from Fig. 17 of Ref. (10))

when compared with the original design assumptions for these dams. However, it also
indicates that the depth of the drainage holes (a modest 40-50 ft) has not been quite deep
enough, and that even better control could have been achieved easily and economically by
carrying the drainage holes deeper.
Uplift diagrams on most of T.V.A.‘s concrete dams are published in separate reports on
each project. From the report on the Fontana Dam, published in 1953,s are reproduced the
two diagrams in Fig. 11. The Fontana Dam is T.V.A.‘s highest concrete dam. The founda-
tion rock consists chiefly of intensely jointed quartzite. When I first saw the diagram in
Fig. 11 (a), I thought that I had at last discovered one clear-cut case of evidence for an effective
grout curtain. From the heel to the first line of observation at A, the uplift along the base
of the dam is almost equal to the full reservoir head. Then it drops abruptly to about one-
fourth a short distance away, at point B; from there a further drop to the line of drains; and
downstream of the drains a small rise which might indicate that the drains were not deep
enough for fully effective control of uplift downstream of the drains. When examining this
uplift diagram more carefully, I noticed that the grout curtain, which was carried out from the
foundation gallery angling slightly in upstream direction, intersects the base of the dam in
such a manner that observation points A and B are both located upstream of the grout curtain,
and that between points B and C, where the grout curtain intersects the base, there is only a
small drop in head. Therefore, the grout curtain can hardly be responsible for the sharp
drop between A and B, but rather some local irregularity at the concrete-rock interface.
In the uplift diagram in Fig. 11(b), f or another section of Fontana Dam, one can see a
different pattern for two of the three observation dates. There is a sharp drop from the
heel to point A, a very small drop from point A to point B, both points being located upstream
of the grout curtain. However, for one set of observations there is a large continuous drop
from the heel to point B, a small drop from B to C, and then a fairly large drop to the line
of drains.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 173
UPLIFT OBSERVATIONS FOR BUREAU OF RECLAMATION DAMS

Fig. 12 is a plot reproduced from the Paper by Keener,9 which summarizes the averages of
all measured uplift pressures for eight Bureau dams. The heavy line is the mean of all these
averages and shows downstream of the drains about 25% of the theoretical straight-line
uplift. This relatively high average is due to the high uplift pressures at Hoover Dam and
one other dam before corrective measures were undertaken. When allowing for these changes
(compare Fig. 14 (a) and (b)) th e average uplift downstream of the drains reduces to the order
of lo%, which is comparable to the average for the T.V.A. dams shown in Fig. 10.
In Fig. 13, reproduced from “Design criteria for gravity and arch dams “, 1 typical
uplift measurements at Grand Coulee and Shasta dams are compared with the original
uplift design assumption for these dams and other design assumptions used more recently by
the Bureau. At the Grand CouIee Dam the drainage gahery is about 50 ft below tailwater.
Thus there is actually a negative uplift on the base of this dam, and some of the seepage
pumped from the drainage gallery is being pulled through the foundation from downstream.
(This case could also be investigated theoretically in a similar manner as shown in Fig. 5.)
At the Shasta Dam there is practically no uplift downstream of the drains and this case may
6-6.46 El ’ EL170Q

b-7.$9)G ?-

l-21-48%

OEI 1400

1265

for design
+

E
I (b)

Fig. 11. Uplift pressure observations for Eontana Dam, Tennessee Valley Authority
(Reproduced from Ref. (8))

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 174 PROFESSOR ARTHUR CASAGRANDE

DlSTANCE IN PERCENTAGE OF BASE WIDTH

Fig. 12. Average uplift pressures at the base of eight Bureau of Reclamation Dams
(R,eproduced from Fig. 10 of Ref. (9))
be considered a good example for the theoretical analysis presented in Figs 3 and 4. The
observed uplift upstream of the drains shows no indication that the grout curtain has an
important effect.
In Fig. 14, reproduced from Keener’s Paper,9 are shown the measured uplift pressures for
the 726-ft-high Hoover Dam before (1938), and after (1947) extensive additional drainage and
grouting operations were carried out. The original grout curtain with holes on 5-ft centres,
went to a maximum depth of 150 ft, sloping from the drainage gallery 15” from the vertical
in upstream direction. The drain holes were drilled vertical from the same gallery to a
maximum depth of about 100 ft. After the large uplift pressures developed, the following
supplementary work was carried out. A new grout curtain was made in the vertical plane
of the drain holes, using and extending the existing drain holes to 400 ft deep. A new line
of drain holes was drilled from the gallery, sloping downstream at an angle of 15” to an
average depth of 200 ft, with a spacing in part of 5 ft and in part of 10 ft. Drain holes were
also drilled in the powerhouse area.
From a comparison of the measured uplift pressures before and after the supplementary
treatment, Fig. 14(a) and (b), one can see that the corrective measures have been fully success-
ful. Simondsrr credits this to both the new grout curtain and the new lines of drains. From
the published details it is not possible to determine how the credit should be divided. The
low point of the pressures in Fig. 14 (b) seems to lie between observation pipes 1 and ‘2 which
is located just about in a vertical plane placed through the lower ends of the new drainage
holes. The uplift observations just upstream and downstream of the new grout curtain, which
lies in the vertical plane of the drainage gallery, show no significant effect of the grout curtain.
On the other hand, Simonds also gives data on seepage measurements which would indicate
that the new grout curtain has helped to reduce the rate of seepage. From additional data
supplied to the Author by the Bureau, the following information was extracted. For approxi-
mately equal reservoir elevations (El. 1180), the rate of foundation seepage from the drainage

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 175
gallery in 1938 was about 200 gal/min. Starting late in 1938 and extending to 1944, the pro-
gramme of supplementary work was carried out intermittently. During a few months in
1939, the rate of seepage increased sharply to about 2,000 gal/min and then dropped just as
sharply to about 600 gal/min at the beginning of 1940. Much of this additional flow entered
through new, deep grout holes. Inflows of 200-300 gal/min were encountered in a number of
holes. Grouting against such inflow proved difficult and was finally accomplished by means
of a new type of packer. From early 1940 to early 1944 the discharge diminished steadily
when it reached about 200 gal/min. From 1944 to 1948 (when measurements were discon-
tinued) the rate of discharge increased steadily to about 400 gal/min. The decrease in rate of
flow between 1940 and 1944 was the result of the extensive additional grouting operations.
The steady increase in flow since 1944 is not explained. It might be loss in efficiency of the
grout due to leaching caused by hot alkaline waters which were frequently encountered in the
drilling operations.
The foundation conditions at the site for the Hoover Dam were exceptionally unfavourable
from the standpoint of control of seepage. Experience on this dam showed that with suffici-
ently deep drain holes the uplift can be controlled satisfactorily even for such conditions. But
control of rate of seepage required much more extensive grouting than the equivalent of a
single row of grout holes.

MISCELLANEOUS COMMENTS ON GROUT CUT-OFFS

My overall impression from the data presented herein, and many similar cases of which I
had included a number in my oral presentation, is that for the great majority of dam sites,
,..W.S. II- IS-49 El 1285.12
’ ,-Normal W.S. .W.S. 11-8-49 El9783
rule*- I l,w,
::‘;--Normal
: W.S. El 1065.0

%x SO0

SHASTA DAM
GRAND COULEE DAM

(a) (b)
Fig. 13. Uplift pressure observations for Grand Coulee and Shasta Dams, Bureau of
Reclamation
(Reproduced from Fig. 1 of Ref. (1))
-- Measured uplift pressure.
-.- Uplift pressure based on a gradient varying from full reservoir at the face of the dam to
30% of the full reservoir at the line of the drains, and from there to normal tailwater.
--- Uplift pressure based on gradient varying from full reservoir pressure at the face of the dam
to one-half the differential of normal water surface and normal tailwater at the line
drains, and from there to normal tailwater.
-. .- The original uplift design assumption.
“*

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.
 FIRST RANKINE LECTURE 177
wise and pound-foolish” by spending much money on grouting and then little or nothing on
the kind of observations which would clearly establish whether the grout curtain is effective.
He said that if one observes small seepage, grouting is automatically given credit, whereas, in
fact, one does not know whether the same flow would have occurred also without grouting,
and one does not know whether one has wasted money for grouting. But when one observes
excessive flow, then one knows that one has wasted one’s money. Therefore, seepage
observations are not sufficient to determine the effectiveness of a grout cut-off. He then
described five projects for which he had insisted on accurate piezometer observations and which
showed in each case that the grout curtain had little or no effect.
Terzaghi emphasized that the first requirement is to engage the most experienced and
reliable grouting contractor one can find; but that one must not let him work alone. He must
be constantly and closely supervised by a competent and independent engineer who also has
a broad experience in grouting operations. Terzaghi pointed out that the trouble with
grouting contractors, and often also engineers, is that they believe that the success of grouting
can be measured by the amount of grout, or cement, that one succeeds in injecting, and he
suggested that this reasoning is just as logical as when people believe that a medicine must be
good because it tastes so awful. He expressed the opinion that even if all conditions are ful-
filled for a satisfactory grouting job, one can still not be certain that the grouting will accom-
plish the intended purpose. Finally, he stated that on every project for which he had
recommended grouting, the safety of the project did not hinge on the success of the grouting.
He recommended grouting merely for the purpose of trying to reduce seepage losses; and he
admitted that in every case it was a gamble and that some of these failed to accomplish the
purpose.
In his lectures, Terzaghi discussed among a number of examples the following four cases:
Case I.---In 1932, in connexion with his investigations for the Bou Hanifia Dam in North
Africa, Terzaghi wanted to demonstrate the necessity for designing the dam such that it
would be safe in case a proposed grout curtain in rock should prove ineffective. He performed
laboratory tests with a diaphragm containing openings equal to 5% of the area. This dia-
phragm caused only very little reduction in flow and very little drop in head. Upon Ter-
zaghi’s suggestion, Dachler supplemented this investigation by a theoretical analysis7 which
is used in Fig. 8.
Case 2.-In connexion with his association with the Sasumua Dam in Africa, Terzaghi
proposed to replace a grout curtain in volcanic rocks by a row of drainage wells. This sug-
gestion was adopted, and after filling of the reservoir the total rate of seepage from the drainage
wells was insignificant.*
Case 3.-The original design of an earth clam in South America included a grout curtain in
jointed gneiss. Grouting tests, performed during the early construction stage, showed that
the total quantity of grout which would be needed would be prohibitive in cost and time.
Since loss of water by seepage would have been of relatively small importance, Terzaghi
replaced the proposed grout curtain by a row of drainage wells. The total measured discharge
from these wells is a small fraction of a cubic foot per second.
Case 4.-On another dam in South America, seepage developed through jointed gneiss,
causing slides in downstream abutment slopes. In an effort to stop the seepage, 55 tons of
cement were injected into more than 3,000 m of grout holes; but the effect of these grouting
operations on the discharge of the springs was negligible. Later, when Terzaghi became
connected with this project, he relied exclusively on drainage in order to cure the slide
conditions.
One instructive case with which I am familiar is the lOO-ft-high earth clam shown in Fig. 15.

* Karl Terzaghi, “Design and performance of the Sasumua Dam”. Proc. Instn civ. Engrs, vol. 9
(Apr. 195S), p. 369.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 178 PROFESSOR ARTHUR CASAGRANDE

It was built as a homogeneous compacted dam, of a residual sandy clay soil derived from
underlying gneiss, with liquid limits ranging between 40 and 80. The foundation consists of a
similar residual soil, but because of its undisturbed structure it is several times more pervious
than the compacted material in the dam. The upper zone of bedrock, on an average 20 ft
thick, is badly fractured and much more pervious than the overlying residual soil; but the
gneiss below the fractured zone is practically impervious. Therefore, the designer decided
to make a grout cut-off through the badly fractured gneiss, continuing the cut-off well into
the impervious gneiss. For this purpose a trench, about 15 ft deep, was first dug into the
residual soil and then a concrete wall was constructed which served as grouting cap, and
which penetrated through the remaining thickness of the residual soil into the fractured
gneiss. The grouting was carried out by one of the world’s leading grouting firms and was
done with competence. Detailed records were kept of all operations. The holes were spaced
as close as 2 ft on centres, but all in a single line. The grout-take was high and erratic, as was
to be expected. In spite of a thorough job of grouting, we find that the piezometers which
extend into the fractured rock, show a practically straight-line drop from reservoir level, at a
location coinciding approximately with the upstream toe of the dam, to the downstream toe
of the dam, and without the slightest indication that there might be an obstacle to seepage at
the location of the grout curtain. Similar results were observed in other cross-sections of this

ompacted
randy
dry

Fig. 15. Piezometer observations in pervious rock underlying earth dam

dam. These observations were, of course, a great shock to the designer who was convinced
that the grout curtain would achieve a satisfactory impervious cut-off of the fractured rock.
Those who had implicit confidence in grouting then suggested that the seepage was pro-
bably passing through the relatively narrow, clay-filled cut-off trench, and then back into the
residual soil and the fractured rock; in other words, that the seepage was bypassing the grout
curtain by flowing over it. If that were the case, the piezometric surface should be flatter on
both sides of the grout curtain, and should show a steep gradient over that cIay-filled trench
where the flow lines would crowd through the much less pervious clay of the trench. The
observations do not support such a hypothesis. Fortunately, the seepage losses are very small
and in my opinion they do not in the slightest endanger the stability of this dam. Also, there
is not the slightest doubt in my mind that omission of this grout curtain would have resulted
in a piezometric surface practically identical with the surface that has developed.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 179
COMMENTS ON CONTROL OF SEEPAGE THROUGH ABUTMENTS OF THIN-ARCH DAMS

Unquestionably the most difficult problems of seepage control in rock are those which
arise in connexion with the combination of a modern, very thin arch dam and steep rock
abutments. In contrast, the wide contact areas of concrete and rock for gravity and arch-
gravity dams constitute a substantial margin of safety against development of dangerous
water pressures at critical locations in the rock downstream of the dam. Even if a fall-out
of rock should occur on a steep abutment slope due to excessive water pressures, this would
not necessarily endanger a concrete dam with a wide base. However, for a thin-arch dam
the concrete-rock contact area is not wide enough to accommodate both a reliable grout cut-off
consisting of three lines of grout holes, and also a satisfactory drainage system which should be
located at least 20 ft downstream from the nearest line of grout holes. The stability of the
rock immediately downstream of a thin-arch dam is of critical importance to the safety of such
a dam.
Grouting of steep abutment slopes poses great difficulties and may easily do more harm
than good. Even if the grouting is carried out after completion of the dam, the contact area
of a thin-arch dam is not wide enough to really protect the rock against displacements. The
removal of all undesirable rock along the abutment slopes should be done with much greater
thoroughness for a thin-arch dam than for any other type of dam; but unfortunately, blasting
of rock on steep slopes, no matter how carefully done, is liable to cause a system of tight joints
to open up. On one dam project with which I am well acquainted, one abutment
consists of intensely jointed quartzite. Test holes sloping into the abutment at 45” were
water-tested without pressure, using only gravity flow; and for that condition were found to
be tight. Just as soon as the customary water pressures were applied, joints opened up and
large water losses were observed. After very light blasting was started, to prepare the contact
surface with a clay core, all test holes which were originally tight began to develop large water
losses. After these observations all removal of rock from this abutment by blasting was
prohibited.
Perhaps there is no satisfactory way of making a reliable grout cut-off in steep abutments
for a thin-arch dam. In that case the second line of defence, drainage, becomes the only line
of defence. How can we make certain that drainage will provide positive protection against
all possible combinations of geologic details and prevent the build-up of dangerous water pres-
sures in the abutments downstream of the dam? We can, of course, provide several lines of
drainage holes downstream of the dam. But then the question arises how to prevent freezing
of the outlets of the drain holes, and of the entire rock face for that matter, during cold
winters? Rock-falls are a common occurrence in spring when the infiltration of thawing
water builds up high pressures in joints behind the frozen zone. For protection against such
conditions we would be forced to resort to a system of “internal drainage “, by means of tunnels
and shafts, but which would require blasting which in turn may be very harmful.
Another possibility would be to widen substantially the thickness of the thin-arch dam as
it approaches the rock. This would go a long way towards improving the stability of the
abutments. In any case, I believe that I am not exaggerating the difficulties of achieving
positive control of the water pressures in the steep abutments for thin-arch dams. When the
geologic conditions are of a character which cannot ensure positively safe control of water
pressures in the rock downstream of the dam, then a thin-arch dam should not be considered
for that site. Surely, if the structural designer of such a dam had to deal with similar un-
certainties concerning the integrity of the arch itself, he would never dream of building such a
structure.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 180 PROFESSOR ARTHUR CASAGRANDE

CONCLUSIONS AND RECOMMENDATIONS

(1) The effects of seepage through rock foundations and abutments on the safety for most
types of dams can be controlled reliably and economically by means of comprehensive drainage
measures. For most concrete gravity and arch-gravity dams, even the usual relatively shallow
line of drain holes has been reasonably effective for controlling uplift pressures. This control
could be improved and made more positive, also for very unfavourable geologic conditions,
by using much deeper drain holes. The effect of a deep line of drain holes on the uplift can
be predicted with reasonable accuracy by means of formulas given in the Paper.
(2) Probably a great majority of the single-line grout curtains in rock foundations and
abutments of dams that were carried out in the past were relatively ineffective in reducing
seepage losses, and they could not be relied upon for purposes of stability analysis. There is
an urgent need for comprehensive observations on the effectiveness of such grout curtains.
Some of the existing concrete dams would offer an opportunity to carry out such investiga-
tions at relatively small cost, by installing numerous piezometers between the heel of the dam
and the line of drain holes, both sides of the grout curtain, along the base of the dam, and in
the rock at various depths below the base.
(3) There is a need for more reliable methods to determine in advance whether and where
grouting is needed. The usual geological investigations and “water tests” in drill holes may
be actually misleading. They are of value only in conjunction with comprehensive testing
for mass permeability of the rock by pumping from or feeding water into a series of drill holes
and by observing the effects in numerous observation wells. For concrete gravity and arch-
gravity dams one should test in this manner systematically the upstream third of the founda-
tion area, and to a depth at least equal to one-half the height of the dam. Excessively
pervious zones for which grouting will be considered necessary or desirable, may require three
rows of closely spaced grout holes, or equivalent clusters of grout holes, in order to create the
necessary width of grouted rock mass. (Note: Depending on their location, all exploratory
holes should also be utilized either as permanent observation wells, as drainage holes, or as
grout holes.)
(4) On some projects the rate of seepage could be effectively reduced by more economical
means than by grout curtains. For a concrete dam with a reasonably wide base and for
abutments with gentle slopes, one could construct an impervious earth fill against the lower
portion of the upstream face of the dam. To prevent opening up of a crack between earth
and concrete, it would be helpful to slope the lower part of the upstream face of the dam such
as, for example, at Shasta Dam, Fig. 13(b). If the dam is already in operation, one could try
blanketing an area adjacent to the dam by hydraulic dispersion of suitable soils in the reservoir
just upstream of the dam. Another possibility is to move the line of drainage holes further
downstream. It is customary to locate the drains approximately 10% of the base width from
the upstream face of the dam. This distance could readily be doubled. When the depth
and spacing of the drain holes is so selected that the uplift will be positively controlled, the
designer would be justified in reducing his assumptions for design uplift pressures below those
currently assumed, so that moving the drains further downstream would not require an
increase in volume of the dam.
(5) In contrast to gravity and arch-gravity dams, it is an extremely difficult problem to
develop seepage control measures in steep abutments for a thin-arch dam which will ensure
the safety of such a structure for the worst conceivable combination of geologic details. Of
those dam sites which would be entirely satisfactory for a high gravity or arch-gravity dam,
only a small percentage could be developed with the same assurance of safety for a modern
thin-arch dam. The treatment of steep abutments poses special difficulties also because
blasting and grouting operations can easily cause more harm than good. Even a comprehen-
sive drainage system may provide insufficient protection against development of dangerous
hydrostatic pressures in the rock downstream of a thin-arch dam.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 FIRST RANKINE LECTURE 181

REFERENCES

1 U.S. BUREAU OF RECLAMATION, DECEMBER 1960. “Design criteria for concrete gravity and arch dams.”
Denver, Colorado. Engng Monograph. No. 19, p. 4.
t EMMONS, W. F. Chief Design Engineer. Tennessee Valley Authority, stated in a letter to the writer
dated 13 January, 1961, T.V.A.‘s present design assumptions as follows:
” I. The intensity of uplift pressure is assumed as full headwater pressure at the upstream face, full
tailwater pressure at the downstream face, full tailwater pressure plus one-fourth the difference
between headwater and tailwater pressure at the line of drains, with intensity varying uniformly
between these points.
“2. The uplift pressure is assumed to act on 100% of the base area.”
3 TENNESSEE VALLEY AUTHORITY TECHNICAL REPORT No. 24 (1952). “Civil and structural design,”
T.V.A., Washington, vol. 1.
4 BRAHTZ, J. H. A., 1936. “ Pressures due to percolating water and their influence upon stresses in hydrau-
lic structures.” Trans. 2nd Congr. Large Dams, Washington, 5 : 43-71.
5 TENNESSEE VALLEY AUTHORITY TECHNICAL MONOGRAPH No. 67 (1950). “Measurements of the struc-
tural behaviour of Norris and Hiawassee Dams.” T. V.A., Knoxville, Tennessee, Fig. 36.
@MUSKAT, M., 1937. “The flow of homogeneous fluids through porous bodies.” McGraw-Hill, N.Y.,
Section 9.8
7 DACHLER, R., 1936. “ Grundwasserstriimung ” (“The flow of water in ground “). Springer, Vienna, p. 82.
* TENNESSEE VALLEY AUTHORITY TECHNICAL MONOGRAPH No. 69 (1953). “Measurements of the struc-
tural behaviour of Fontana Dam.” T. V.A., Knoxville, Tennksee,’
g KEENER, K. B., 1951. “ Uplift pressures in concrete dams.” Trans. Amer. Sot. civ. Engrs, 116 : 1128.
~~RIEGEL, R. M., 1951. Discussion of Ref. 9. Trans. Amer. SOG. civ. Engrs, 116 : 1239, Fig. 17.
r1 SIMONDS, A. W., 1953. “Final foundation treatment at Hoover Dam.” Trans. Amer. Sot. civ. Engrs,
118 : 78-99.
r2 TERZAGHI, K., Discussion on paper by Julian Hinds on “Upward pressures under dams,” p. 1563, 1929.
ASCE Transactions of the American Society of Civil Engineers.
r3 TERZAGHI, K., 1929. “Effect of minor geological details on the safety of dams.” American Institute of
Mining and Metallurgical Engineers. Technical Publication 215, p. 30.

[It is hoped to publish an article by Professor Casagrande in due

course elaborating the control of seepage through pervious alluvium
for earth dams, which was also presented in the Lecture. Sec. I.C.E.]

In proposing a vote of thanks to Professor Casagrande, Dr L. F. Cooling, Deputy Chief

Scientific Officer, Building Research Station, said that it was indeed a great pleasure and
privilege for him to propose a vote of thanks to Professor Casagrande for his very interesting
and instructive Lecture. The applause that it had received made it unnecessary for him
(Dr Cooling) to say any more than that it had given an extremely good send-off to the series of
Rankine Lectures.
In his Introduction, Professor Skempton had given one or two reasons, with which Dr
Cooling personally would heartily agree, as to why the choice of Professor Casagrande for this
Lecture had been such an appropriate and happy one. Dr Cooling felt that there were other
reasons also. Members would recall that the late Professor Cook* had begun his article on
Rankine with the following phrases:
“The name of Macquorn Rankine is among the most honoured and renowned in the
annals of engineering science. He is acknowledged as one of the great pioneers in the
movement to bring the powerful resources of mathematics and physical science to the
practical problems of the engineer. It is to his efforts chiefly that we owe the success
achieved in the struggle, in the 19th century, to give engineering an honourable place among
the studies of our universities. . . .”

l COOK, GILBERT, 1951. “ Rankine and the theory of earth pressure.” Gbotechnique, 2 : 4 : 271.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.

 182 VOTE OF THANKS

On both these counts, Dr Cooling was sure they would agree that Professor Casagrande was a
worthy follower in the Rankine tradition. Through his writings and his prowess as a teacher,
Professor Casagrande had done much to establish the status of soil mechanics and foundation
engineering as a subject for university study. Through his many outstanding Papers, and
his consulting work, he had helped substantially to bring about the acceptance among present-
day engineers of the value, when applied to practical engineering problems, of those powerful
resources of mathematics and physics which formed an essential part of modern soil mechanics.
The Lecture that they had been privileged to hear was a typical example of his powerful
combination of theory and practice. They all knew him as a recognized authority on the
subject of seepage, but his talk had shown just how many of these problems he had dealt with.
They were all very indebted to him for giving such a clear and interesting exposition of this
fascinating subject.

In seconding the vote of thanks, Mr Henry Grace said that he had had the privilege of
knowing Professor Casagrande for more than 20 years, first as one of his students, then for a
short period as one of his assistants, but always as a friend to whom one could turn for
information and advice.
The Lecture had covered a much wider field than he had anticipated, and any remarks that
he himself might make would be directed purely to earth dams. As a soil mechanics specialist
he had expected Professor Casagrande to confine his remarks to that aspect, but he had not
done so. Mr Grace could not recall hearing a lecture that was so packed with information.
It had been outstanding in every way, and they would long remember it and refer to it time
and time again.
In order to illustrate the great progress that had been made in regard to filters and seepage,
he would like briefly to recall the methods of dam design that had prevailed in this country up
to the early ‘30’s. At that time dam design was based almost entirely on empirical rules.
The only sites used were those where a cut-off wall could be taken down to rock or some other
impervious strata. The term “seepage” was not referred to. To mention it to a dam
designer was like speaking to the Chancellor of the Exchequer about leakages in the Budget !
The only criticism that he had to make of the Lecture was that Professor Casagrande had
omitted to mention the very great part that he had played in this work. Mr Grace had had
the privilege of being at Harvard while a great deal of the research work was being carried out.
Under Professor Casgrande’s direction research work had established the criteria for rational
filter design, for example. It was also shown that it was absolutely essential to use de-aerated
water for permeability tests.
At the time Professor Casagrande had acted as consultant for the Franklin Falls Dam. The
dam was about 140 ft high and built on a fine sand foundation where there was no cut-off at
all. Mr Grace believed this to be the first rationally designed dam built on a pervious founda-
tion. It could, therefore, be appreciated that Professor Casagrande’s work had opened up
new possibilities for dam designers by greatly increasing their understanding of the behaviour
of soils affected by seepage.
Mr Grace said that he could dwell for a long time on the great help that it had been to him
personally, but he would be brief and say that all present, and especially engineers concerned
with the design and construction of dams, would like to express their appreciation of the work
that had been done by Professor Casagrande. They would wish to thank him also for sparing
the time to come to England and deliver this most important Lecture.

Downloaded by [] on [11/05/20]. Copyright © ICE Publishing, all rights reserved.