INTERNATIONAL SOCIETY FOR

SOIL MECHANICS AND

GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of

the International Society for Soil Mechanics and
Geotechnical Engineering (ISSMGE). The library is
available here:

https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands

of papers published under the Auspices of the ISSMGE and
maintained by the Innovation and Development
Committee of ISSMGE.

Quality Control of Earth Embankments
Contrôle qualitatif de remblais en terre
by F. J. D a v is , Engineer, Bureau of Reclamation, Design and Construction Division, Denver, Colorado, U.S.A.
Summary
Quality control of earth embankment implies: selection o f cer­
tain properties o f the placed embankment to be brought under
control; measurement of these properties; analysis of the control
tests of these properties; practical limitation o f these properties
within an acceptable range compatible with design criteria.
The average value of a set of data has certain disadvantages as a
control criterion which are shown by actual examples o f control test
data. Statistical methods are applied to control test data to permit
better interpretation of results. The methods o f analysis are shown
by examples of control data from some Bureau of Reclamation
Earth Dams.
The control data are assumed to be obtained under essentially
stable conditions. The assumptions for these conditions are strength­
ened by: separate analysis of tests from each borrow area, some­
times also separated by periods of time; separate analysis of tests
from material compacted under different compactive effort; elimina­
tion of all tests indicated by the inspectors to be nonrepresentative;
special emphasis on the importance o f representative sampling;
standardized testing and sampling procedures.
Practical limitations of placement moisture and density control
are discussed for rolled impervious zones. The discussion is based
on results obtained on a number of large earth dams constructed by
the Bureau of Reclamation.
The cumulative frequency concept of data presentation is developed
for measuring the quality of control based on relatively few tests.
This method is believed to be more applicable to periodic control.
Introduction
The term “quality control” has become a familiar phrase in
recent years, particularly in referring to manufacturing pro­
cesses. Industry has almost unanimously turned to the sta­
tistician and his methods for obtaining strict control necessary
for its continued profitable operation.
The application of statistical methods to quality control of
construction materials is not new. The American Society for
Testing Materials published a “Manual on Presentation of
Data” in 1933, which dealt with the application of statistical
218
Session 3/2
Sommaire
Par le contrôle qualitatif de remblais en terre nous entendons:
Le choix des propriétés du matériau de remblai qui doivent être con­
trôlées; la mesure de ces propriétés; l’analyse des essais de contrôle
portant sur ces propriétés; la limitation pratique de ces propriétés
dans des limites compatibles avec les conditions du projet.
La valeur moyenne d’une série de données a certains désavan­
tages en tant que critériums de contrôle ainsi qu’il est montré par
l’exemple d’essais de contrôle. Les méthodes statistiques auxquelles
les auteurs ont recours pour la vérification des essais permettent
une meilleure interprétation des résultats. Les méthodes d’analyse
sont illustrées par les essais de contrôle de quelques barrages en
terre construits par le «Bureau of Réclamation».
On suppose que les essais de contrôle sont faits dans des con­
ditions essentiellement stables. Les hypothèses sur lesquelles s’ap­
puient ces conditions sont: l’analyse séparée des essais de chaque
zone de prélèvement, tenant compte parfois des intervalles écoulés
entre les prélèvements; l’analyse séparée d’essais sur des matériaux
compactés à différents degrés; l’élimination de tous les essais con­
sidérés non représentatifs par les inspecteurs; la prééminence accor­
dée à la valeur des échantillons représentatifs; la standardisation des
essais et des procédés de prélèvement d’échantillons.
Le présent rapport a pour objet la limitation pratique de la teneur
en eau des matériaux mis en place et le contrôle de la densité après
compactage pour des zones imperméables. Il est basé sur les résultats
obtenus lors de la construction de plusieurs grands barrages en terre
par le «Bureau of Réclamation».
Il est montré que le concept fréquence cumulative, utile pour la
présentation des essais, permet de chiffrer la qualité des contrôles
basés sur peu d’essais. Cette méthode est plus applicable aux con­
ditions usuelles des essais de contrôles périodiques.
methods to test data. Since then, this manual has undergone
constant revision and reprinting, the latest edition being pub­
lished in January 1951 (ASTM, 1951).
In contrast to the emphasis placed on quality control of
industrial processes and of other construction materials, quality
control of earthwork has, in general, been largely neglected or
entirely ignored. Perhaps the lag in quality control technique
for earthwork is to be expected as a normal consequence of the
lag of soil mechanics behind the other sciences. Standardized
construction procedures and control methods now in use per­ Table 1 Placement Moisture Tests (24 tests representing 58,750
mit control of placement soil properties within definite, rela­ cubic yards o f material placed)
tively narrow, ranges and statistical methods applied to the Essais sur la teneur en eau du matériau mis en place
control tests permit accurate analysis of the quality of the (24 essais représentatifs de 58,750 cu.yds. de matériau mis
en place)
control.
Placement Laboratory Optimum Variation of
Scope and Purpose Moisture Moisture Placement from
% % Laboratory Optimum
Quality control of earth embankments implies:—
12.0 13.3 — 1.3
A. Selection of certain properties of the placed embankment
13.1 12.0 + 1.1
to be brought under control; 13.5 14.4 —0.9
B. Measurement of these properties of the placed embank­ 10.9 12.3 — 1.4
ment; 9.3 11.0 — 1.7
C. Analysis of the control tests of these properties; 15.5 13.0 +2.5
D. Practical limitation of these properties within an acceptable 11.1 12.5 — 1.4
range compatible with design criteria. 11.1 12.0 —0.9
This discussion is limited mainly to the latter two aspects of 13.1 13.5 —0.4
the subject, with only brief mention made of the properties cho­ 13.8 15.5 — 1.7
15.8 12.8 + 3.0
sen for control purposes and of the methods of measuring these
14.5 14.0 —0.5
properties. The analysis of control test data is based on an 9.0 12.5 —3.5
application of statistical methods. The practical limitation of ' 12.0 13.4 — 1.4
the properties under control will be seen from a presentation 13.0 12.4 +0.6
of factual data, and control charts will be demonstrated which 14.9 12.6 +2.3
serve as a guide for controlling placement conditions during 13.5 12.5 + 1.0
construction. 8.9 12.7 —3.8
11.0 12.1 — 1.1
Placements Limits 15.5 13.7 + 1.8
15.9 12.0 + 3.9
The properties chosen for quality control tests are placement 15.6 12.4 + 3.2
moisture and density, or rather, their variation from the Bureau 10.6 12.7 —2.1
of Reclamation laboratory standard.1) 13.8 13.5 —0.3
Initial attempts at earthwork control were centered primarily Average 12.8 12.9 —0.1
on the unit dry weight obtained in the embankment. The
Bureau of Reclamation recognizes both an upper and a lower Table 2 Relative Density Tests (10 tests representing 94,000 cubic
placement moisture limit as well as a minimum density require­ yards of material placed)
ment for rolled impervious fill. The basis of control of com­ Essais sur la densité relative (10 essais représentatifs de
pacted pervious zones is primarily a consideration of relative 94.000 cu.yds. de matériau mis en place)
density. This discussion is therefore limited to moisture- Relative Density o f Zone 2
density relationships although the methods herein described 51.2
are applicable to the analysis of other soil properties which 89.9
may be of importance in certain cases. 70.4
84.1
The Average Value 65.5
67.4
While the average value of a series of test data is easily 82.1
understood and readily computed, it has certain disadvantages 63.3
as a reliable measure of quality control which can best be 53.7
shown by actual example. Consider the data in Table 1 taken 54.9
from a Bureau of Reclamation earth dam. These tests, 24 in Average 68.3
number, represent the placement of 58,750 cubic yards of
impervious material during a period of one month. While the material was 70 percent relative density. In this case, the
average placement water content is almost exactly equal to the average value is near the minimum requirement, and would
average laboratory optimum, a careful review of individual probably be accepted as satisfactory placement. It is easily
tests reveals that no samples were placed at a water content recognized from the tabulated values, however, that only 40
equal to the laboratory optimum. Assuming all of the samples percent of the tests are at or above acceptable density. Again
to be representative, we have no material placed in the embank­ assuming the tests to be representative of the total material,
ment at the water content indicated by the average value. 60 percent was placed at less than the minimum density
Another example of the inadequacy of the average value as requirement.
a reliable measure of quality control is shown in the case of
Basis o f Analysis
ten relative density tests of compacted pervious material sub­
mitted from a different project. The tests shown in Table 2 The inadequacy of the average values of soil properties as
represent the placement of about 94,000 cubic yards of ma­ a reliable measure of the placement condition of an earth
terial. The minimum requirement for compaction of this embankment and the resulting misinterpretation of control
data were recognized several years ago, and statistical methods
1 T h e B ureau o f R eclam ation standard com paction test is m ade in a
¿-cu b ic fo o t cylinder 6 inches high in 3 layers a t 25 blows per layer with were applied to test data to measure the quality of the control.
a 5£-pound ham m er d ro p p ed 18 inches. Statistical methods may be defined as “methods for dealing

219
Table 3 Moisture Control: Bonny Dam—North Borrow Area— special interest centered on the number o f operations in each
Zone I group, the data are said to be in the form o f a frequency d is­
Contrôle de la teneur en eau: Barrage de Bonny - Zone tribution. Under this condition, there is a mathematical m odel
de prélèvement nord - Zone I which can be em ployed to study the variation o f the measured
property.
Fill Moisture Content—% by Dry Weight
446 Tests—1950 Table 3 consists o f a set o f data from the control tests o f a
large earth dam. The mass o f figures com piled in this manner
16.2 17.5 16.0 15.2 13.9 14.5 15.8 15.5 16.0 16.5
has little significance. The arithmetical mean o f 15.5 percent
16.2 15.3 14.0 16.5 14.9 16.0 14.9 14.0 13.2 16.5
16.7 may be com puted rather easily and, by visual perusal, the
16.0 16.9 18.5 17.5 17.5 15.5 14.7 14.8 17.5
14.3 16.9 13.9 14.5 17.7 17.2 16.4 16.2 17.0 15.6 maximum and minimum values o f 21.6 and 11.8 percent can
15.6 15.9 16.7 14.8 14.1 13.6 15.9 12.9 13.1 14.4 be detected. This set o f data may be conveniently grouped in
14.0 15.8 16.3 13.1 16.0 16.0 16.2 16.3 17.3 15.6 the manner o f a frequency distribution as show n in Fig. 1.
15.6 13.3 12.3 17.1 17.3 15.2 15.1 14.5 14.3 14.3 Values o f the variable are grouped into classes and the f r e ­
15.0 14.3 14.0 15.3 14.2 16.0 16.7 15.3 14.0 14.3 quency o f occurrence o f each value is tabulated in the proper
12.1 15.6 15.2 15.2 16.3 16.5 13.5 12.4 13.4 14.3 group. One measure o f the dispersion or scatter o f the values
14.3 16.3 18.6 17.5 15.2 15.8 16.0 14.0 14.8 14.1 from the mean is called the standard deviation. This measure
16.6 14.5 15.1 15.4 14.5 14.5 14.6 12.7 16.7 15.4 is defined as the root-mean-square o f the deviation o f the values
14.5 13.9 14.3 13.1 13.9 13.3 15.0 12.7 15.1 14.7 from their arithmetical mean. The form ula for standard devi­
13.1 12.7 13.5 14.1 12.9 14.6 14.5 14.4 15.2 14.9 ation in this case in which an arbitrary or guessed m ean is
14.1 14.6 15.5 14.8 14.7 14.5 14.0 15.2 16.1 15.3 chosen is :
15.2 15.2 16.1 15.2 15.3 15.9 15.8 15.1 15.1 14.8
16.6 15.6 14.9 16.2 15.7 13.4 15.7 14.8 13.6 14.8 I £ fd 2 (2 fd ) 2
17.3 16.1 15.2 16.1 14.7 16.5 16.0 16.2 16.1 15.7 a = / ------------------------------- C (Arkin and Colton, 1946)
N N N
15.5 15.3 17.1 16.3 14.9 14.6 16.2 13.6 15.1 15.6
15.6 14.5 15.4 15.5 13.8 14.2 17.0 17.3 14.3 12.6 w here:
16.1 13.9 16.0 13.5 15.4 13.9 12.0 14.4 12.6 17.9 a Standard deviation
=
16.9 17.5 15.0 15.3 17.8 16.7 17.5 16.5 13.6 17.7 2 Summ ation
=
15.9 17.6 15.9 14.8 14.2 16.3 15.9 18.8 13.5 15.9 / =
N um ber o f operations occurring in each group o f values
17.4 15.9 14.9 17.6 17.3 16.9 15.3 17.2 21.6 15.7 d D eviation o f mid-value o f each group o f values from the
=
16.3 13.6 15.6 16.3 15.1 14.3 16.2 16.3 15.0 15.9 guessed mean o f the set o f data
13.8 17.0 13.8 16.3 16.0 12.4 15.6 17.7 14.8 15.1 N = Total number o f operations
15.8 12.4 16.3 16.3 17.0 17.1 15.0 15.8 16.4 17.0 C = A correction for grouping the values.
15.4 16.4 18.7 17.1 17.6 16.9 15.2 15.5 15.1 17.5
17.6 17.1 16.7 15.4 15.3 15.4 16.0 15.8 15.6 15.3 The standard deviation is determined from this set o f data to
16.6 17.6 13.7 18.1 17.1 15.9 16.6 15.1 16.9 15.5 be ± 1 - 4 percent.
15.4 15.0 15.3 14.5 14.2 14.9 15.4 14.7 15.6 16.8 The mathematical m odel m ost com m only used for analyzing
frequency distributions o f data obtained from repetitive opera­
15.3 13.6 16.4 15.3 15.9 14.2 16.8 15.4 14.6 15.2
14.2 16.7 14.6 14.9 14.2 16.2 12.3 14.4 14.5 16.6 tions, is the “ N orm al D istrib u tion ” , as it has been found that
15.5 16.0 15.6 16.4 15.3 17.3 15.2 18.0 15.9 15.4 many data o f this type are approxim ated well by the theoretical
16.7 15.0 15.9 16.0 15.2 16.7 14.8 14.9 17.5 16.5 normal curve. There are a number o f characteristic properties
14.1 13.3 15.1 15.1 16.5 16.1 18.8 17.1 16.7 17.3 o f this curve which can be defined mathematically, one o f
15.6 15.3 14.4 17.6 15.3 15.3 14.7 16.1 14.9 14.7 w hich is the standard deviation. This property, described in
17.2 13.5 14.9 16.5 12.4 15.3 17.2 18.7 16.7 16.6 terms o f area under the curve (total number o f operations), has
16.5 17.1 15.5 17.1 17.1 17.3 19.7 15.8 16.9 17.7 the follow ing characteristics as show n in Fig. 2:—
19.6 16.2 15.3 14.9 16.0 15.3 16.5 17.1 15.2 15.6
18.7 15.9 16.5 16.2 17.1 17.7 17.0 16.7 16.4 15.9 m ± a = 68.26 percent o f the area
m ± 2 a = 95.46 percent o f the area
12.0 16.4 15.5 15.1 13.9 14.8 15.8 14.7 15.2 15.8 m ± 1 a = 99.73 percent o f the area.
16.6 14.4 11.8 15.3 16.8 16.4 14.4 14.8 13.4
13.9 14.7 13.3 14.8 15.8 16.7 15.5 16.0 18.5 Thus, the standard deviation property is a valuable to ol in
12.6 15.4 15.3 14.2 15.5 14.2 14.7 14.6 14.3 describing the frequency o f occurrence o f a variable within any
14.4 14.0 14.3 15.8 15.7 14.6 14.4 14.8 12.4 arbitrarily chosen limits. It has been used as a criterion o f con ­
average = 15.5 trol, keeping always in mind the follow ing considerations:—
A . Each test is considered to be representative o f equal am ounts
with data that have been obtained by repetitive operation” . o f em bankm ent material;
Experience indicates that many repetitive operations behave as B. The operation o f sam pling and testing is considered to be
though they occurred under essentially stable conditions, i.e., repetitive;
there is only one variable to be measured by the operation. C. The value is used for approxim ating limits embracing j o f
The field density tests from which the placem ent moisture- the tests.
density relationship is determined, and the standard laboratory The assum ption o f repetitive operation (B, above) is strength­
com paction tests from which maximum density and optim um ened by:
m oisture content for a given com pactive effort are determined, A . Separate analysis o f tests from each borrow area, som etim es
are considered to be repetitive operations fitting this condition. also separated by periods o f time;
In other words, variations o f test results due to testing or B. Separate analysis o f tests from material com pacted under
sam pling techniques are ignored and the soil property under different com pactive efforts;
test is assumed to be the only variable to be measured. W hen C. Elim ination o f all tests indicated by the inspectors to be
data from this type o f operation are classified into groups with nonrepresentative;

220
 M O ISTURE CONTROL

B onny DAM NORTH BORROW AREA 20HE O ne

W > T e s t a - 1950

F re q u e n c y o f o c c u rre n c e f d fd f d*
21.5-21.9 / -------------------------------------------------------------------------------------------- 1 13 13 169
21.0-2 1 ./* --------------------------------------------------------------------------------------------------- 12
20.5-20.9 11
20.0-20. 4 10
19.5-19.9 / / -------------------------------------------------------------------------------------------- 2 9 18 162
19.0-19.4 6
18.5-18.9 UU I I I —---------------------------------------------------------------------------------- 8 7 56 392
18.0-18./* / / -------------------------------------------------------------------------------------------- 2 6 12 72
110 550
124 ¿*96
126 378
110 220
59 59
äs -56 56
-86 172
-69 207
u l o ^ ^ — ---------------------------------------------------- 12 -t -48 192
-1*0 200
12.0-12./* fJfl / / / / -------------------------------------------------------------------------------------10 -6 -60 360
11.5-11.9 / -------------------------------------------------------------------------------------------- 1 -? - 7 49

TOTALS ¿*46 262 3734

Av e r a g e o p t i m u m m o i s t u r e 16 .2

Av e r a g e fil l mois t ur e 15 .5
Fig. 1 D istribution o f Fill M oisture
Me a n v a r i a t i o n fr om o p t . mo i s t ur e -0 .7 D istribution de la teneur en eau
St a n da r d deviat ion du m assif

D . Special emphasis on the im portance o f representative sam ­

/( x ) where u -
. /ï(x -m )2 and N is the total
number of values pling;
V2 TT
E. Standardized sampling and testing procedures.
The analysis shown in Fig. 1 gives a measure o f the dispersion
or distribution o f the placement moisture content, but does not
give any hint to the placem ent-optimum relationship which is
really the im portant control criterion. It is, therefore, more
advantageous to tabulate the variation o f the placem ent m ois­
ture from the laboratory optim um in each case, as in Fig. 3,
than to consider individual values. Interpreting the data in this
figure in terms o f standard deviation o f the variation, it may
be said that the placement water content o f approximately I
o f the tests fall within the limits o f 0.6 percent above and 2.0
percent below the laboratory optimum condition.
The above data are only one example o f the approxim ation
o f control test data to the normal distribution curve. A study
o f earthwork control data shows this tendency toward normal
Fig. 2 T ypical N o rm al D istribution distribution for all o f the usual control tests, i.e., densities,
D istribution norm ale typique moisture contents, penetrations-resistance needle readings,

EARTHWORK CONTROL WORK SHEET

KDISTURE CONTROL
Bonny NORTH BORROW AREA
446 T ee te - 1950

Frequency o f O ccurrence f d fd fd 2

1 9 9 81
8
1 7 7 49
2 6 12 72
5 5 25 125
8 4 32 128
23 3 69 207
20 2 40 80
28 1 28 28
64
-1 -5 7 57
//// //// y /// / , / 57
-2 -174 348
1111 / / i I JJ/ / i l l ! I / / / / / / / _________________________1 60 -3 -180 540
^ w m m / 7/ — ------------------------ 53
14
-4
-5
-212
- 70
848
350
13 -6 - 78 468
4 -7 - 28 196
4 -8 - 32 256
-9
2 -10 - 20 200

T O T A LS /./.A -629 i.033

Average optim nn m o istu re Fig. 3 D istribution o f Fill |M oisture ab o u t

Average f i l l m olature O ptim um
Mean v a r ia t io n from o p t, m o iatu r« D istribution de la teneur en eau du
S ta n d ard d e v ia tio n o f v a r ia t io n
m assif référé à l’optim um

221
 PLATORO DAM placed at an average o f 0.5 percent or less below the optim um .
ZONE 2
115 TESTS
-f This group has, with the exception o f Boysen D am , a very small
tí f -S i standard deviation. Green M ountain, Platoro, and Granby
1 D am s all had borrow pits which were naturally conditioned to
S;

%J.
-r

"P
1
70 -a
T ?
V about this mean placem ent moisture content. Apparently,
VARIATION OF PLACEMENT WATER CONTENT RELATIVE DENSITY-PERCENT natural conditioning o f pits results in a very uniform dispersion
FROM LAB OPT % BY DRY WEIGHT
TREN " .
o f m oisture. In the case o f A ngostura D am , the borrow pit
NORTH E.--------
AREA 1951 consisted o f a stratum o f lean clay overlying sand and gravel.
6 5 3 TESTS
A well-planned system o f pre-irrigation, initiated well in ad­
vance o f excavation in this case, resulted in uniformity o f m ois­
ture content which approxim ated the naturally conditioned
pits noted above. A t A nderson Ranch D am , the occasional
líM± addition o f small am ounts o f water on the fill was effected by
0 »I *2
VARIATION OF PLACEMENT WATER CONTENT VARIATION OF FILL DRY DENSITY FROM MAX
FROM LAB OPT X BY DRY WEIGHT LAB DRY DENSITY- LBS PER CU FT application o f water to the material as it dropped from the
conveyor used for transportation o f borrow material to the
Fig. 4 T ypical F requency D istribution Curves
C o u rb es typiques de la fréquence de distribution de la teneur embankm ent. B oysen D am , with a mean placem ent condition
en eau near the optim um , is the exception to this group o f very sm all
standard deviations. In this case, the borrow areas were, for
specific gravities, and relative densities. Fig. 4 show s som e the m ost part, very dry. Considerable water had to be added as
typical frequency distributions o f soil properties selected from other considerations required placem ent at near optim um co n ­
Bureau o f R eclam ation control data. dition. Som e pre-irrigation was performed in the pits but
apparently n o t far enough in advance to obtain uniform dis­
R esu lts O b tain ed persion o f the added water. In m ost cases it was a matter o f
m ixing very w et and very dry com ponents o f soil. Som etim es,
Fig. 5 show s the placem ent m oisture control achieved on a excessively wet material from the bottom o f the pit w as pur­
number o f Bureau o f R eclam ation dams. The projects are posely included in the cut in an attempt to bring the placem ent
listed in order o f the magnitude o f the variation between m ean m oisture o f the total material to a desirable mean. In addition,
placem ent condition and m ean laboratory optimum. Several about 2 percent m oisture had to be added on the fill. Extensive
trends can be readily observed. There is one group which was attempts to obtain a uniform dispersion o f moisture by mixing
soil com ponents which are both too dry and too wet were not
VARIATION OF PLACE­
M A T E R IA L NO. OF M ENT WATER CONTENT effective.
SOURCE TE S TS FROM OPT IN Below this group is a large number o f dams in which the
863
standard deviation varies from about 1.5 to 2.0 percent. All
B O Y S EN A ll A re o s
G R EEN M O U N TA IN A ll A r e a s 1066 o f this group had dry borrow areas o f varying degree. M ost
PLA TO RO A ll A r e a s o f the projects attempted pre-irrigation o f areas. T he smallest
G RA N BY A re a I deviations for the group are Shadehill and M edicine Creek
A N G O ST U RA A re a C
2139 D am s. Borrow areas for these tw o projects were characterized
A N D ERSO N RA N CH D ix ie P it
BO N N Y N o rth A re a 1069 by relatively shallow depths o f fine-grained material overlying
H EA R T B U T T E A ll A re a s pervious strata. The effectiveness o f pre-irrigation in such
SH A D EH I L L M a in E m ban km en t borrow areas is evidenced by the relatively small standard
M ED I C I N E C R E E K A ll A re a s 1347
2692 deviations. Larger values o f standard deviation occurring in
C ED A R B L U F F A ll A r e a s
BO N N Y S o u th A r e a this group may also be explained by a knowledge o f the borrow
D A V IS A re a E areas. The required excavation at Bonny D am , consisting o f
S H A D EH I L L W in q D am
very fine silt, was 6 to 8 percent dry o f the optim um condition.
EN D ER S A ll A re a s
O ' SU L L IV A N A ll A re a s 1739 N o attem pt was m ade to pre-irrigate this material. Theaddition
D I C K I N SO N A ll A re a s o f 4 to 6 percent o f water on the fill could n ot be effectively
S H A D EH I L L D ik e s handled even by m ixing in thin layers. In the south borrow
BO N N Y Required Excavation
area at this project, sprinkling operations were barely kept
LO N G L A K E A re o “ 5
abreast o f excavation, but were kept well in advance o f ex­
LO N G L A K E C u t - o f f T re n c h
SO U T H C O U L E E A re a “ 2 cavation in the north area. These differences in moisture appli­
H 0 R SET 0 0 T H A ll A re o s cation operations are reflected in the standard deviation values
LO N G L A K E A re o “ I
o f 2.0, 1.7, and 1.5 for the respective sources.
S O L D I E R CA N YO N A ll A re a s
N O RT H C O U L E E A ll A re a s
Other variations in the standard deviation values can be cor­
JA C K S O N G U LCH A re a s F8 G related with borrow area conditions and m oisture application
D IXO N CA N YO N A ll A re a s operations. From the evidence in the chart, it seems clear that
S P R I N G CA N YO N A ll A re a s
for close moisture control o f an embankm ent, preconditioning
EX P L A N A T I O N o f borrow areas is mandatory. The period o f irrigation re­
M ean Valve
quired depends on the character o f the borrow area. Areas
M ean t o n e st d naturally conditioned to approximately optim um water con ­
d e v iat io n
tent result in a very narrow range o f co n tro l; however, for very
Fig. 5 M o istu re C ontrol. V ariation o f M ean Placem ent W ater C o n ten t high dams, it m ay be necessary to reduce the water content to
fro m L ab o rato ry O ptim um and S tandard D éviation o f V ariation
a considerably drier range o f placement.
C o n trô le de la ten eu r en eau. V ariation d e là ten eu ren eau m oyenne
du m atériau mis en place par ra p p o rt à la valeur o p tim a en Fig. 6 presents fill-dry unit weight data for the sam e projects
lab o rato ire e t déviation sta n d ard de cette variation shown in Fig. 5. This figure again calls attention to the fallacy

222
 VA RIA TIO N O F F I L L D RY U N IT
uring the dispersion o f the data about its central value. H ow ­
W EIGH T FRO M LA B O RA T O RY ever, unless the control limits happen to coincide with the limits
M A T E R I A L M A X IM U M - PERC EN T O F M A XIM U M
DAM
SO U R C E D RY U N IT W EI G H T
defined by m ± cr, the curve does not reveal information about
_ O — co fO ^ in the percentages o f occurrence beyond the desired limits. For
rO^J-invDr^cOCTiooOOOO
— — —
CT>CT>(T>Cr>(T><T>(T>-
^
periodic control, it is more desirable to use the “ cumulative
B O Y SEN A l l A re o s
G R EEN M O U N TA I N A ll A r e o s frequency” concept o f analysis. Figs. 7 and 8 are work sheets
PLA TO RO A ll A r e a s containing data relative to control tests from a borrow area for
GRA N BY A rea I
one particular reporting period. Figs. 9 and 10 are graphical
A N G O STU RA A rea C
A N D ER SO N RA N C H D i xie Pit
presentations o f data com piled in this way. Similar control
BO N N Y North Area charts m ay be constructed for other soil properties. For con ­
H EA R T B U T T E A ll A r e a s struction control o f earth embankm ent, these charts and similar
SH A D EH I L L M ain Em b .
A ll A r e a s
ones for relative density o f pervious zones, are the principal
M ED I C I N E C R E E K
C ED A R B L U F F A ll A r e a s ones in use. From these charts, the degree, or quality o f the
BO N N Y So u t h A r e a control can be readily and accurately judged.
D A V IS A rea E
SH A D EH IL L W in g D am
EN D ER S A ll A r e a s ZONE.
O ' SU L L I V A N A ll A r e a s Borrow Areo
D I C K I N SO N A ll A r e a s s T H IS P ER IO O TO D A TE
s Cu m CU M CU M .
SH A D EH I L L D ik e s u.
F R EQ U EN C Y O P O C C U R R EN C E F
f %
F
F
CU M

BO N N Y Re q ’d Exc . > 1 2 .6
I I . 6- 12.6
LO N G L A K E A rea 5 10.5*11.5
Cut- off Trench 1 i
LO N G L A K E * 9 4-10.4 1 1 1.2 2 2 1.5
o 8 3- 9 3 1 1 2 2 .5 1 3 2 3
SO U T H C O U L EE A rea 2 7 2-82 II 2 4 4 9 2 5 3 8
œ
H O R S ET O O T H A ll A r e a s 6 .1 - 7 1 1 III 3 7 8 6 4 9 6 .8
5 .0 - 6 0 5 1 1 8 9 9 6 15 11.4
LO N G L A K E A rea * I 39-49 5 IH Í mi 9 17 2 1 .0 14 29 22 0
A ll A r e a s 2 8-38 12 1 II 34 6 23 39 4
S O L D I E R C A N YO N It f l IH1 28 52
*X 1 7 - 2 .7 6 IHT IH1 1 II 39 4 8 .1 17 69 5 2 .3
N O RT H C O U L EE A ll A re as 8 IH1 II 7 46 56 8 15 84 63 6
S I ♦ 0 .5 0T6O- - 01 .56 10 97 7 3 .5
JA C K S O N G U L C H A r e a s FÔ G 3 IH1 IKt 56 69 1 13
4 7
D IXO N A ll A r e a s IS 06- 1 6
1 7 - 2 .7 2
Ufl
IK I
II
IK1 10
63
73
77 8
90.1
II
12
108
1 20
81 8
90 9
SP R I N G CA N YO N A ll A re as 2 8-38 2 mi 4 77 95.1 6 126 95 4
ll J 3 9 - 4 .9 1 u 2 79 97 5 3 1 29 97 7
> 5 .0 - 6 0 1 u 2 1 00 1 00
EX P L A N A T I O N Î .I g
81 3 132
6.1 - 7 1
¡Mean Valve <
72-82
m 83-9 3
9 4- 1 04
M ean * one std.
10.5- I l 5
d eviat io n I l 6 - 12 .6
> 12.6
Fig. 6 D ry U n it W eight C ontrol. V ariation o f M ean Fill D ry U n it TO TA LS 51 81 132

W eight from L ab o rato ry M axim um and S tan d ard D eviation o f P R EV

P « ¡i n D A TE
Avero g e m ox lo b y 0 114 .0 112 8 113.3
V ariation Avero g e f i l l yo 112 0 I I I .4 III 7
C o n trô le d u p oids sec unitaire. V ariation du poids sec unitaire M eon v o r io f i o n f ro m m o x lo b y» - 2 .0 - 1 .4 -16
du m assif p a r ra p p o rt à la valeur m axim a en lab o rato ire e t dé­ A vero g e ro ck c o n t e n t (% o f p lu s No 4 b y d ry w e ig h t ) 21.0 2 6 .1 1 24. 1

viation sta n d ard de cette variation PERIODOFREPORT 12-l-Sl TO12-31-51

TESTS H-29-A-I-R TOI2-27-A-I-R
Fig. 7 E arth w o rk C o n tro l W ork Sheet. D ry U n it W eight C ontrol
o f the average or m ean value as a reliable measure o f control.
F o rm u le p o u r le contrôle du terrassem ent. C ontrôle du poids sec
The usual Bureau o f R eclam ation requirement for dry unit unitaire
weight in an im pervious rolled fill is 98 percent o f the laboratory
maximum. U sually the design is predicated on a minimum ZONE .
density o f at least this m agnitude. W hile the m ean placem ent Borrow Areo
condition is, in m ost cases, above the minim um requirement, THIS PERIOD
FREQUENCY OFOCCURRENCE T
it is clearly evident that much em bankm ent has been placed at
lesser densities.
Som e o f the low densities and w ide variations can be readily
explained by a know ledge o f the project conditions. In the case
o f Shadehill W ing D am , for instance, the relatively low values
o f fill density result from a lack o f hom ogeneity. The material
was excavated in the borrow area by tractor-drawn scrapers 2 TO- 0 . 2
b 3- 0 7
w orking on a sloping cut in an attem pt to mix tw o distinct
strata o f material, i.e., a top layer o f clay with an underlying
layer o f gravel. Even with extensive m ixing operations on the 28-32
fill, a lack o f hom ogeneity is reflected by low com paction. In “3Í-17
contrast, similar material was excavated for the main em bank­
ment by Euclid loader equipped with a vertical cutting arm.
A hom ogeneous material resulted from this excavation which
accounts in large part for the superior com paction control o b ­ PREV. Wf
Averogeoptimum moisture 14.8 15. 4 p15.2
tained on this part o f the embankment. Averoge fill moisture 13.6 14. 5 14.1
Meon voriotion from opt moisture -1.2 -0.9 -1 .1
PERIOOOFREPORT_J2-I-51 .TO12-31-SI
Control Charts TFÇTÇ II-29-A-I-R TOI2~27-A-l-R
Fig. 8 E arthw ork C ontrol W ork Sheet M oisture C ontrol
A s previously explained, data presented in the usual form F o rm u le p o u r le contrôle du terrassem ent. C ontrôle de la ten eu r
o f a frequency distribution afford an excellent means o f meas­ en eau

223
 Figs. 11 and 12 are typical cumulative-frequency curves de­
monstrating good control on two types o f com pacted material.

C o n clu sio n s

Engineers are being called on to design and construct earth

embankments o f increasing height at increasingly difficult sites.
The degree o f control o f placem ent properties o f earth em bank­
ments has, along with advanced knowledge o f soil m echanics,
an im portant role in the engineering o f these structures. Cor­
rect evaluation o f the quality control is necessary if the control
is to be properly considered as a factor in this design and co n ­
struction.
It has been demonstrated that the average value applied to VA RIA TIO N O F F I L L D RY D EN SI T Y FRO M M A X L A B D RY D EN SIT Y
L B S P ER . C U F T
earthwork control tests is not a reliable means o f evaluating
these data. Statistical m ethods can be readily applied to such Fig. 11 T ypical F requency D istribution C urve D ry D ensity—N o rth
data and afford an excellent measure o f the quality o f the Borrow A rea—T ren to n D am 653 T ests— 1951
C o u rb e de la cloche typique - D ensité à sec - Z one de prélève­
control. m ent nord - B arrage de T ren to n - 653 essais - 1951

Fig. 12 Typical F requency D istribution C urve R elative D ensity o f

Z one 2— P lato ro D am 115 T ests
•6 -5 -4 -3 -2 -I 0 «I »2 »3 *4 *5 "
C ourbe typique de la fréquence de d istribution de la ten eu r en
VARIATION OF F IL L WATER CONTENT FROM LABORATORY OPT eau - D ensité relative de la Z one II - B arrage de P lato ro -
PERCENT BY DRY WEIGHT 115 essais
Fig. 9 M oisture C o n tro l C h art. C achum a D am
D iagram m e de co ntrôle de la teneur en eau. B arrage de C achum a
The evidence resulting from statistical analysis o f a large
am ount o f control data suggests:—
A . R elatively narrow ranges o f placem ent m oisture and den­
sity can be obtained by practical and ordinary construction
.95% of Max. lab dry procedures.
‘ unit weight is within
: this range for all B. D esigns m ay be based on anticipated quality o f control.
; material having max Quality control may be estimated from previous results ob­
• lab dry density of
: no to 120 lbs per cu.ft. tained under similar borrow area conditions and construction
•Mean| placement i.6 lbs
- j below lab m ax.-|-|-j-j-
methods.
II |j-L-LLl_LlpRlFviQus C. Critical designs may be based on predetermined limits o f
i f c l j : t ------------DEC.,1951 placem ent moisture and density. The control required to m ain­
T ~ T , — — -t o t a l
tain these limits may be o f higher order than previously ob ­
tained by normal procedures. M odifications to ordinary pro­
cedures, or supplementary operations may be required to m ain­
D R Y U N IT W EI G H T
CO N TRO L CH A RT
tain rigid control within these limits. It is believed that such
CA CH U M A DAM m odifications, or supplementary operations, can be reasonably
and econom ically required if such rigid control is imperative
for a resulting stable structure.
im a m
VARIATION OF F IL L ORY UNIT WEIGHT FROM LABORATORY M AX
References
LBS PER CU. FT
A rk in , H. and Colton, R. R. (1946): A n O utline o f S tatistical M ethods.
Fig. 10 D ry U n it W eight C ontrol C h a rt. C achum a Dam R eprinted M ay 1946, p. 36, ch ap ter IV.
D iagram m e d e co n trôle du poids sec unitaire. B arrage de C a­ A S T M (1951): M anual on Q uality C ontrol o f M aterials. Special T ech­
chum a nical Publication 15-C, January.

224