See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/311779578

An empirical method for predicting post-construction settlement of concrete

face rockfill dams

Article  in  Canadian Geotechnical Journal · December 2016

DOI: 10.1139/cgj-2016-0193

CITATIONS READS
14 846

3 authors, including:

Mohammad Kermani Jean-Marie Konrad

WSP Laval University
6 PUBLICATIONS   174 CITATIONS    134 PUBLICATIONS   4,666 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modelling the fundamental behaviour of sands View project

Prediction of post-construction settlements of rockfill dams based on construction field data View project

All content following this page was uploaded by Mohammad Kermani on 10 November 2018.

The user has requested enhancement of the downloaded file.

 755

ARTICLE
An empirical method for predicting post-construction
settlement of concrete face rockfill dams
Mohammad Kermani, Jean-Marie Konrad, and Marc Smith

Abstract: In this study, employing a database of 19 concrete face rockfill dam (CFRD) cases, two prediction methods for
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

post-construction settlement of CFRDs are presented. In the first method, post-construction settlements are estimated using
height of the embankment. In the second method, characterization of the stress–strain behavior of the compacted rockfill layers
during construction allows prediction of the subsequent stress–strain–time behavior of the embankment. Knowledge of rock
particles strength is necessary in both methods. In the presented methods, settlements are estimated separately for each of the
three life-cycle phases: before, during, and after impoundment. The presented results show that, in addition to addressing some
limitations of previous methods, the proposed approach is precise and highly practical. It also allows a better understanding of
rockfill deformation mechanisms. Apart from using this method for predictive purposes, the presented graphs can be used to
distinguish unexpected settlement behavior of a CFRD during its post-construction lifespan.

Key words: concrete face rockfill dam (CFRD), rockfill, time-dependent, post-construction, settlement, prediction.

Résumé : Dans cette étude, à partir d’une base de données de 19 barrages en enrochement avec un masque amont en béton
(« CFRD »), deux méthodes de prédiction de tassement post-construction de CFRDs sont présentées. Dans la première méthode,
les tassements post-construction sont estimés en fonction de la hauteur du remblai. Dans la deuxième méthode, le comporte-
ment contrainte–déformation de l’enrochement des remblais lors de la construction permet de prédire le comportement
For personal use only.

contrainte–déformation–temps subséquent du barrage. La connaissance de la résistance des particules rocheuses est nécessaire
avec les deux méthodes. Pour ces deux méthodes, les tassements en crête sont estimés séparément pour chacune des trois phases
du cycle de vie d’un barrage : avant, pendant et après la mise en eau du réservoir. Les résultats présentés montrent qu’en plus de
surmonter les difficultés reliées aux méthodes actuelles, les approches proposées sont précises et très pratiques. Elles per-
mettent, également, une meilleure compréhension des mécanismes de déformation de l’enrochement. En plus de servir d’outil
prédictif des tassements, les relations présentées peuvent être utilisées afin de détecter un comportement atypique d’un CFRD
au cours de sa durée de vie.

Mots-clés : barrage en enrochement avec masque amont en béton (CFRD), enrochement, en fonction du temps, post-construction,
tassement, prédiction.

Introduction the post-construction behavior of CFRDs constructed during

the modern period.
The origin of rockfill dams can be traced back to the mining
CFRDs are widely used because of their cost-effectiveness and
region of California Sierras, in the 1850s. Rockfill dams were of the
the similarity of their constituent materials. Like other types of
impervious face type, constructed by dumped rockfill, until earth
rockfill embankments, CFRDs are not highly prone to instability.
core dams began to be developed in about 1940. The period from
However, attention must be paid to their settlement as excessive
1850 to 1940 is called the early period by Cooke (1984). The dumped
displacement may cause cracking of the concrete face, which can
(in 18–60 m lifts) concrete face rockfill dams (CFRDs) performed
result in severe leakage. Some examples of modern CFRDs con-
safely, but the leakage problem became more serious as dams structed in the 21st century that suffered severe leakage are Cam-
became higher, due to high compressibility of dumped rockfill. pos Novos and Barra Grande dams in Brazil, and the Mohale dam
During the years 1940 to 1970 (called the transition period by Cooke in Lesotho (Marulanda 2013).
1984), earth core rockfill dams became more popular due to Rockfill consists of rock particles having a large diameter (aver-
higher flexibility of earth materials, and consequently more com- age size of at least 5 cm and maximum size of up to 2 m) with a
patibility with dumped rockfill. Additionally, mostly during the variety of particle sizes, shapes, and minerals. As a consequence,
period 1955–1965, a transition developed to the use of sluiced full scale experimental study of rockfill material is almost impos-
rockfill in thinner layers of 3–3.6 m subsequently compacted by sible. Scale effects are shown to influence the representativeness
construction truck and dozer traffic. The period after 1970, of the results of laboratory experiments on rockfill (Marachi et al.
when compacted and sluiced rockfill was used in construction 1969; Marsal 1973; Ramon et al. 2008; Alonso et al. 2012). Also,
of rockfill dams of both concrete face and earth core type, is limited laboratory samples are not perfect representations of
called the modern period (ICOLD 2010). This study addresses in situ conditions considering the variety of materials used in

Received 2 April 2016. Accepted 9 November 2016.

M. Kermani and J.-M. Konrad. Department of Civil and Water Engineering, Laval University, Pavillon Adrien-Pouliot, 1065, avenue de la Médecine,
Quebec City, QC G1V 0A6, Canada.
M. Smith. Hydro-Quebec, 75 René-Lévesque Blvd. West, Montréal, QC H2Z 1A4, Canada.
Corresponding author: Mohammad Kermani (email: [email protected]).
Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.

Can. Geotech. J. 54: 755–767 (2017) dx.doi.org/10.1139/cgj-2016-0193 Published at www.nrcresearchpress.com/cgj on 20 December 2016.
 756 Can. Geotech. J. Vol. 54, 2017

Fig. 1. Post-construction crest settlement of Bastyan dam (adapted from Hunter and Fell 2002). ␣, relative settlement rate in semi-log scale.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

different zones or even in a single zone of a dam. When studying suggested that the rate of long-term crest settlement, ␣, as well as
long-term deformations, load application periods are also limited the crest settlement attributed to first filling can be estimated
in the laboratory compared to reality. Moreover, no laboratory considering dam height, the rockfill placement method and the
experiment can apply the same stress path to the sample as in the unconfined compressive strength (UCS) of rockfill materials.
field. In this study, first, the empirical method for prediction of the
Numerical models must be calibrated using laboratory experi- post-construction settlements of CFRDs by (Hunter and Fell 2002)
ments, which are prone to the above-mentioned uncertainties. is modified by changing the time datum for the settlements after
Also, there are uncertainties about using concepts from classical impoundment and considering different relative crest settlement
soil mechanics to evaluate the strength and deformation behavior rates for the periods before and after impoundment. Also, using a
For personal use only.

of rockfill. Accordingly, so far, the design of rockfill dams has data base of 19 CFRD cases, a novel prediction method is presented
been mostly based on engineering judgment and experience. using behavior of the embankment during construction. Then, compar-
There are several studies in the literature in which settlements ing the relative settlement rates before and after impoundment, some
of different typical rockfill embankments, including CFRDs, are aspects of long-term mechanical behavior of compacted rockfill
compared to find a settlement prediction method. These studies material are put into perspective. Finally, an example application
are categorized into two groups. The first category consists in of the prediction methods is presented.
studies in which an implicit equation is suggested to calculate
dam crest settlement. These equations are mainly of three types: Post-construction crest settlement prediction
logarithmic strain versus time (Sowers et al. 1965), power type methods
(Soydemir and Kjaernsli 1979), or strain rate versus time (Parkin
1977) relationships. In the method presented here, it is assumed that the post-
The second category includes more recent studies, which sug- construction deformations take place in three phases: (i) time-
gest a qualitative judgment procedure to predict settlements. For dependent deformations after construction and before impoundment;
instance, Clements (1984), conducting a study on post-construction (ii) deformations during impoundment; and (iii) time-dependent
crest deformations of 68 rockfill dams, suggested predicting the deformations after impoundment (Fig. 2a). The settlements attrib-
crest settlements by comparing the dam with previously moni- uted to each phase are calculated separately.
tored dams with similar characteristics. The major drawback of For the time-dependent deformations after construction and
this approach is that there are many influencing factors that can before impoundment, the time datum is the end of construction.
lead to a high degree of subjectivity. Milligan and Coyne (2005) However, for the long-term crest settlements after impoundment,
prepared a database of 95 rockfill dams, with which they ques- the time datum is the end of impoundment. This has two major
tioned the statements of previous researchers who claimed that advantages.
rockfill dams settle a maximum 0.02% of their height per year First, as the stresses inside the dam body change during the
after 3 years from the end of construction. They also concluded impoundment period and stay almost constant after the end of
that because of the considerable amount of variables involved, it impoundment, it is reasonable that the time datum for the time-
is unreliable to use the empirical formulas to predict deforma- dependent deformations be the end of impoundment. Therefore,
tions. the settlement rate values (␣2 in Fig. 2b) are more representative
One of the most comprehensive and practical studies on post- values, when comparing to settlement rate, ␣, values in Fig. 1. This
construction settlement of CFRDs was carried out by Hunter and is the same concept as setting the time to zero for every load
Fell (2002). They suggested some empirical methods to predict the increment within an oedometer test.
rockfill modulus during construction, crest settlement due to im- Second, as it is shown in Fig. 2a for Bastyan dam and observed in
poundment, and the rate of time-dependent crest settlement. almost all of the case studies, it takes almost 1 year after the
They divided the post-construction deformations into two parts: impoundment for the settlement data points to fall onto a line in
time dependent crest settlements and crest settlement under semi-log scale. In the method presented by Hunter and Fell (2002),
stresses from first filling. For the post-construction settlements, it is assumed that deformations that take place from the begin-
they suggested that the time datum begin at the end of main ning of impoundment to the end of this time lag are attributed to
rockfill zone construction. The crest settlement attributed to first the effect of first filling. Also, no indication was given as to how
filling takes place from the beginning of impoundment to a lim- the length of this delay can be calculated. Because of the nature
ited time after the end of impoundment. The two components are of graphs drawn by logarithmic scale, this delay changes the crest
shown in Fig. 1 for the Bastyan rockfill dam. Hunter and Fell (2002) settlement calculations significantly. Therefore, making settle-

Published by NRC Research Press

 Kermani et al. 757

Fig. 2. Post-construction crest settlement of Bastyan dam (a) time Table 1. Classification of unconfined com-
datum at the end of main rockfill construction; (b) time datum at pressive strength (UCS) of rock according
the end of impoundment (data from Hunter and Fell 2002). [Color to Australian code AS 1726–1993 (data
online.] from Hunter and Fell 2002).
UCS range
Strength descriptor (MPa)
Extremely high >240
Very high 70–240
High 20–70
Medium 6–20

zones as classified by the Australian code (Table 1). This means

that the rockfill used in the construction of the dam has been
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

extracted from quarries, where the intact rock has been charac-
terized to have the average strength classified as high or very high
strength within Table 1.
In this study, no case with dumped rockfill was studied because of
the lack of quality and quantity data in such cases. Therefore, here,
the placement method is not directly considered as an influencing
factor. Also, as the rockfill used in all of the studied cases is of quar-
ried origin, the influence of shape of particles could not be studied.
The database used in this study is mostly adapted from Hunter
and Fell (2002), but also from (Kermani 2016). Nevertheless, the
cases with high quality data on post-construction deformations as
well as behavior during construction are chosen. Table 2 presents
a summary of embankment and rockfill properties for the case
studies. The CRFD cases constructed of very high strength rockfill
materials (above) are separated with a line from the cases of high
For personal use only.

strength rockfill (below).

A typical cross section of a modern CFRD is shown in Fig. 3. The
main rockfill zones are zones 3B and 3C. Usually, rockfill is coarser
and is placed in thicker layers in zone 3C compared to zone 3B
(layer thickness and maximum rock diameter are usually limited
to 1 m for zone 3B, and 2 m for zone 3C). The central variable zone
in this figure shows variations in the size of these two zones in
design practice around the world. Zones 1A and 1B are earthfill
zones, and zones 2B and 3A support the concrete face slab. The
data presented in Table 2 are mostly associated with rockfill placed
in zone 3B and also in zone 3C.
ment predictions using the method presented by Hunter and Fell
(2002) is complicated. Post-construction crest settlement prediction method using
On the contrary, in this study, by changing the time datum, the dam height and rock strength
data points after the impoundment will lie on a line in semi-log In the following, the three phases of post-construction settle-
scale almost right after the end of impoundment (Fig. 2b). There- ment are compared with dam height for different rock strength
fore, if the time datum is set to the time after which stresses categories. The adopted concept is the same as Hunter and Fell
remain almost constant in the embankment, i.e., end of impound- (2002); however, here, the time datum for the third phase is moved
ment, the settlement of the crest will fall on a line in a semi-log to the end of impoundment and different rates are considered for
graph. This is in line with the generally observed behavior of the relative crest settlements before and after impoundment. Ad-
rockfill materials, which show logarithmic creep deformations ditionally, the crest settlement attributed to first filling is consid-
with time (e.g., Sowers et al. 1965; Marsal 1973; Clements 1981; ered to finish by the end of impoundment period. These changes
Alonso et al. 2005). It also allows using this approach to predict the yield more consistent results and facilitate the application of the
post-construction deformations. prediction method.
To develop a prediction method, for the two time-dependent
phases, the crest settlement rates per height of embankment (rel- Crest settlement rate before beginning of impoundment versus
ative crest settlement rates, ␣1 after main rockfill construction dam height
and ␣2 after impoundment in Fig. 2a and Fig. 2b, respectively), and Figure 4 presents relative crest settlement rates after the end
for the reservoir filling phase the maximum crest settlement at- of construction and before start of impoundment versus correspond-
tributed to impoundment per height of the embankment (relative ing dam heights. Despite the lack of data for dams of high strength
crest settlement attributed to impoundment, Simp in Fig. 2a) are rockfill, the graph shows an increase in the relative crest settlement
considered. The crest settlements, as well as dam heights, were rate with height. The effect of dam height on the relative crest set-
normally measured at cross-sectional locations with maximum tlement rates in this phase is associated with the fact that the rate of
height for each case. For all three phases, the crest settlement is creep deformation of rockfill material is a function of stress level.
shown to be predictable, knowing UCS of intact rock and one of Extrapolating from the trendlines drawn in Fig. 4, the settle-
the following: dam height or the rockfill mechanical behavior ment rate is negligible for dams with heights of less than around
during construction. To take strength of rockfill particles into 25 to 40 m for high strength and very high strength rockfills,
account, the data are presented separately for the dams constructed respectively. In other words, at stress levels smaller than certain
of high strength rockfill zones, or, of very high strength rockfill values, rockfill time-dependent deformations are not significant.

Published by NRC Research Press

 758 Can. Geotech. J. Vol. 54, 2017

This is consistent with the observations in laboratory oedometer

Compaction specifications†

5%–10% W; 1.0 m; 4p; 10 t

15% W; 0.9 m; 4–6p; 10 t

and isotropic loading experiments on granular materials that

15% W; 1.0 m; 4–6p; 10 t

7.5% W; 0.45 m, 4p; 10 t
100% W; 0.9 m; 4p; 10 t
25% W; 0.8 m; 4p; 10 t

10%W; 1.0 m; 4p; 10 t?

15% W; 0.9 m; 4p; 10 t
20% W; 1.0 m; 8p; 10 t

20% W; 1.0 m; 8p; 10 t

30% W; 1.0 m; 4p; 15 t

10% W; 1.0 m, 8p; 10 t
No W; 0.9 m; 4p; 15 t
No W; 0.9 m; 4p; 15 t
indicate a threshold stress for stresses lower than which, creep

No W; 1.0 m; 4p; 10 t

? W; 1.7 m; 4p; 13.5 t

? W; 1.0 m; 6p; 15 t
? W; 0.6 m; 4p; 9 t
strains are negligible (Marsal 1973; Clements 1981; Colliat-Dangus
et al. 1988; Oldecop and Alonso 2007).

20% W; ?; ?
Crest settlement attributed to impoundment versus dam height
In CFRDs, the deformations during the impoundment period
are associated with an increase in stresses inside the dam body
due to the reservoir load on the upstream surface. Simple finite
element modelling (further explained in the section titled “Com-
Dry density Rockfill void

parison between time-dependent deformations before and after

impoundment”) shows that during impoundment of CFRDs, both
mean and deviatoric stresses increase on the upstream side of the
ratio

0.30
0.29

0.26
0.33

0.20

0.24
0.23

0.23

0.28
0.22
0.27

0.18

0.21
0.21
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

dam body. However, mean stress increases more significantly and

—

—
—

—
consequently, the shear stress ratio decreases. Therefore, the set-
tlement of the crest in this phase is considered to occur mostly
due to increase in the mean stress.
UCS (MPa) (t/m3)

2.287

2.03 In addition to the stress level increase, collapse due to wetting

2.34

2.24
2.07

2.07
2.27
2.12

2.15
2.3

2.3
2.2

2.2
2.2

2.2
2.1
—

—
—

might also contribute to the settlements during impoundment.

Wetting collapse occurs in rockfill structures due to the strength
loss caused by increasing the amount of water in contact with
185.5*

rock particles. The rockfill moisture content might increase due

225*

49*

44*
148

118
118
25

45

66

to malfunctions of the concrete face, high downstream water

—
—

—
—

level or penetration of water to downstream rockfill due to exces-

Millstone grit sandstone

sive precipitations. However, the effect of rockfill wetting collapse

Greywacke, some slate
Basalt, basaltic breccia

Charnockitic–gneissic

Limestone, mudstone

is not studied in detail in this paper.

Siltstone, sandstone

In this study, to obtain the deformations due only to stress

increases caused by reservoir loading, the time-dependent com-
Gneiss, biotite

Granite gneiss
completed height (m) length (m) strength category Rockfill source
For personal use only.

Quartz schist

ponent of settlements was subtracted from the total settlements

Quartzite

during impoundment. To do so, the rate of time-dependent settle-

Siltstone
Rhyolite

Rhyolite

Dolerite

ment during impoundment period (second phase in Fig. 2a) was

Gneiss
Gneiss
Schist

Tuff

assumed to be the same as that in the period before impoundment

(first phase in Fig. 2a); i.e., ␣1.
?

Note: MS, HS, VHS, and EHS denote medium, high, very high, and extremely high strength, respectively.

The relative crest settlements attributed to impoundment val-

†Volumetric added water (W); layer thickness; number of passes (p); and weight of roller, respectively.

ues, Simp, are plotted against dam height for the studied CFRDs in
Fig. 5. It can be seen that the relative dam crest settlement due to
EHS to VHS

EHS to VHS

HS to VHS?

HS to VHS?
Max. dam Dam crest Intact rock

HS to VHS

HS to VHS
MS to HS

MS to HS

MS to HS

impoundment generally increases with embankment height. The

linear trend lines can be used as a guide for a predictive purpose.
VHS?
VHS

VHS
VHS

VHS
VHS

HS?

The Toulnustouc main dam does not follow the general trend in
HS

HS

HS

Fig. 5. Although the relative settlement rates before impound-

Table 2. List of the studied CFRDs (data mostly from Hunter and Fell 2002).

ment (see Fig. 4) and those after impoundment (presented in the

next section) are in the expected range, the relative crest settle-
347.5

ment due to impoundment is much higher than other similar

430
213
828
200

374
675
400
535
178
560
465
380
134
1168
146
1050
520
850

cases. This unusual behavior can be attributed to the fact that

according to Fig. 6, the impoundment of the dam took place mostly
in February 2005, but the reservoir reached its maximum level
during the snowmelt season, mid-April to mid-June 2005. Al-
45.7
75
110
160
94

122
233
46
76
60
90
75
80
38
178
43
85
53
140

though the reservoir level increased much less during the snow-
melt season comparing to the previous months (20 versus 50 m),
the crest settled much more during this period. The reason might
be that because the main rockfill zone of the dam was compacted
Tasmania 1983
Tasmania 1971
1979
Tasmania 1982
1978
Tasmania 1986
2007
2004
2004
Australia 1969
Sri Lanka 1984
Tasmania 1981
1981
Tasmania 1971
1999
Tasmania 1989
Australia 1978
1974
1993
Year

without watering, the effects of snowmelt along with the subse-

quent precipitations increased the wetting collapse in the down-
stream rockfill zones during the impoundment period.
*Average of different zones of rockfill.
Australia

‡Asphaltic concrete face rockfill dam.

Location

England
Canada

Canada
Canada
China

China
Brazil

Brazil

Crest settlement rates after impoundment versus dam height

Relative crest settlement rates after impoundment, ␣2, are drawn
versus the embankment heights in Fig. 7. For the dams of very
Toulnustouc south dyke

high strength rockfill zones, the relative crest settlement rate

Toulnustouc main dam
Outardes 2 main dam

increases slightly with dam height and is almost similar for all of
the cases. The value is less than 0.05 for all of the studied dams.
Mangrove Creek
Kangaroo Creek

Tianshengqiao I

This means that for a 100 m high dam, the crest settlement attrib-
Foz Do Areia

uted to time-dependent deformations after impoundment is less

Mackintosh

White Spur
Dam name

Serpentine
Murchison

than 14 cm in 50 years of operation.

Shuibuya

Winscar‡
Winneke
Kotmale
Cethana
Bastyan

In the CFRDs constructed of high strength rockfill, there is a

Xingo
Reece

clear tendency of increasing relative long-term crest settlement

rates with dam height. Although Fig. 7 generally correlates with

Published by NRC Research Press

 Kermani et al. 759

Fig. 3. Typical zoning of concrete face rockfill dams (adapted from ICOLD 2010).
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Fig. 4. Relative crest settlement rate before impoundment versus that also takes into account the shape and gradation of parti-
dam height. [Color online.] cles and the compaction method.
Therefore, a novel approach is introduced here, in which the post-
construction settlements are determined using the dam’s deforma-
tion behavior during construction. This method is adapted from the
well-known concept that the secondary compression index C␣ and
compression index Cc are related for different types of earth ma-
terials. In other words, the time-dependent mechanical behavior
of a material in constant stress can be estimated through its be-
havior during loading. The material can be a fine-grained soil
(Mesri and Godlewski 1977), sand (Mesri and Vardhanabhuti 2009)
or rockfill (Oldecop and Alonso 2007; Cormier and Konrad 2015).
In the case of rockfill dams, secondary compression index corre-
sponds to post-construction settlement behavior and compres-
For personal use only.

sion index corresponds to the rockfill construction period.

In this study, the secant modulus of the main rockfill zone at
the end of construction (eq. (1) and Fig. 8) is used to illustrate the
behavior during construction. The rockfill secant modulus, for a
certain rockfill layer, at end of construction can be defined by
the findings of Hunter and Fell (2002), when comparing the data dividing the calculated vertical stresses by the strains inferred
with an adjusted time datum with the same data presented by from readings of settlement gauges
Hunter and Fell (2002), more consistent values are obtainable for
Kotmale and Winneke dams. Yet, for Winneke, the impoundment ␴z
(1) ERC ⫽
began 1.6 years after end of rockfill construction and took 3.8 years to ␦s /d1
end. Consequently, a large amount of time-dependent deforma-
tion had already taken place before and during the impound-
where ␴z is the stress at the mid-height of the layer, as shown in
ment. Therefore, the rate of settlements relative to dam height is
Fig. 8; ␦s is the vertical displacement of the considered layer; and
lower than expected for this dam. Additionally, because at the
d1 is thickness of the considered layer. To calculate the moduli, the
Mangrove Creek dam, the impoundment was still ongoing at the
stress at the mid-height of the layer ␴z is considered in this paper
end of studied period, i.e., 15 years after end of construction, this
and is calculated using a simple linear elastic numerical analysis.
dam does not yield any data point in this figure.
In this way, the effect of embankment shape on vertical stress is
Similar to the data presented for post-construction settlements
taken into account. Other than Shuibuya, Tianshengqiao, and
before impoundment, the dependence of relative settlement rates
Winscar dams for which the moduli are calculated by the authors,
after impoundment on embankment heights tends to decrease with
in this section, the values of modulus during construction, ERC,
dam height. In other words, the evolution of relative settlement rates
are taken from Hunter and Fell (2002). These values are calculated
with dam height is nonlinear. It is also worth noting that Shuibuya
and Winscar dams fall off the general trend in this figure. for the lower half to one third of the dam height. Note that, this
Despite the similarity of the general trend of relative settlement modulus calculation method differs from that of Fitzpatrick et al.
rate versus dam height curves before and after impoundment, the (1985), Pinto and Marques Filho (1998), and Giudici et al. (2000), in
relative settlement rate values before impoundment are generally the sense that first, here, the shape of embankment is considered;
smaller compared to those after impoundment. This aspect will i.e., the vertical stress is not simply taken as the multiplication of
be further examined in this paper. fill height to fill density. Secondly, to determine the modulus, the
vertical stress is calculated in this paper for the mid-height, in-
Post-construction crest settlement prediction method using stead of the top, of the layer. The calculated stress at mid-height is
rockfill modulus and rock strength supposed to yield the average vertical stress of the layer. An ex-
The previously presented method has the following limitations: ample of modulus calculation is presented further in this paper.

• The embankment height does not perfectly represent the dam’s Crest settlement rate before beginning of impoundment versus
geometry. Embankment slopes and valley shape are also influ- rockfill secant modulus
encing factors. The relative crest settlement rates before impoundment are
• The UCS of intact rock is not the only material parameter that shown versus rockfill secant modulus at end of construction in
affects the post-construction settlement of CFRDs. To reach a Fig. 9. For the dams where the main rockfill zone is constructed of
more reliable prediction method, a parameter shall be defined very high strength rockfill, there is a slight decrease in the relative

Published by NRC Research Press

 760 Can. Geotech. J. Vol. 54, 2017

Fig. 5. Crest settlement per dam height attributed to impoundment versus dam height for studied CFRDs. [Color online.]
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Fig. 6. Evolution of crest settlement per dam height and reservoir creases. Note that, because in the case of CFRDs deformations
water level history of Toulnustouc main dam in the highest section. during both construction and impoundment represent stress–
[Color online.] strain behavior of their materials due to gradual increases in
stress level, the existence of a relationship between the two
phases is quite rational.

Crest settlement rate after impoundment versus rockfill secant

modulus
For personal use only.

Figure 11 shows the long-term relative crest settlement versus

rockfill secant modulus for the studied CFRDs. The same trend as
settlement rates before impoundment (Fig. 9) is observed here.
The only difference is that, after impoundment, the relative set-
tlement rates are generally higher than those in Fig. 9 and the
difference between the trend lines, for the two strength catego-
ries, is more evident.
It is interesting to note that despite very high strength of rock-
fill in Foz Do Areia and Reece dams, the rockfill has low secant
modulus at the end of construction, even less than some of the
high strength rockfills. Additionally, as can be seen in Table 2,
crest settlement rate, ␣1, with secant modulus at the end of con- rockfill void ratio is high for the two cases. Therefore, it can be
struction. For the dams where the main rockfill zone is con- inferred that the void ratio affects the rockfill stress–strain behav-
structed of high strength rockfill, the relative crest settlement ior during construction more than the intact strength of rock
rate decreases more clearly with rockfill secant modulus. In other particles. However, for the time-dependent deformations, the in-
words, the stiffer the rockfill layers are during construction, the tact strength is more influential.
less settlement they undergo in the post-construction period. A Winscar and Shuibuya dams, which were outside the trends
power law curve is fitted to the curve to allow its use for predictive observed in the graphs of ␣2 versus dam height, fall well inside the
purposes. trend in Fig. 11. This shows the efficiency of the current method. It
The Mackintosh dam, which was considered as an outlier in is also worth noting that the Winneke dam, which was outside the
Fig. 4 falls inside the generally expected trend in Fig. 9. This shows trend observed for the long-term relative crest settlement (after
the reliability of this predictive approach. As the in situ behavior impoundment) of dams constructed of high strength rockfill, is
of the same structure is being used for prediction, it reduces the well inside the expected trend in the other two sections (time-
shortcomings of the other prediction approaches. dependent settlements before impoundment and settlements at-
tributed to impoundment). This validates the hypothesis that its
Crest settlement attributed to impoundment versus rockfill secant
unexpected behavior during the long-term settlement phase is
modulus
due to the delay in completion of impoundment.
The concept of linking the load–deformation behavior of the
A summary of the parameters calculated for the studied cases
dam during construction to its behavior afterwards is also used
and used in the previous three sections is presented in Table 3.
here. Figure 10 presents the relative crest settlement attributed to
impoundment versus average rockfill secant modulus at the end
Comparison between time-dependent deformations
of construction for different CFRDs. Once again, the higher the
secant modulus of the rockfill at the end of construction, the before and after impoundment
lower the dam’s relative settlement attributed to impoundment. In this section, the relative crest settlement rates before and
However, the effect of intact rock strength is not significant here. after impoundment are compared to study the effect of changes in
Except for one case, CFRDs constructed of both very high and high stress distribution and properties of materials on time-dependent
strength rockfill zones lie along the same general trend line. In deformation of CFRDs. To compare the stress state inside an em-
other words, by comparing the settlements caused by impoundment bankment before and after impoundment, a generic 100 m high
with rockfill secant modulus, the effect of rock intact strength de- embankment having slopes of 1 vertical to 1.4 horizontal was

Published by NRC Research Press

 Kermani et al. 761

Fig. 7. Relative crest settlement rate after impoundment versus dam height. [Color online.]
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Fig. 8. Calculation of modulus during construction (adapted from Hunter and Fell 2002).
For personal use only.

Fig. 9. Relative crest settlement rate before impoundment versus rockfill secant modulus at end of construction. [Color online.]

modeled using the Plaxis 2D Hardening soil model. This model, (2) E ≈ Eref(␴ /pref)m
which uses theory of plasticity to reproduce hyperbolic axial
strain – deviatoric stress in a drained triaxial test, and, includes where Eref is the stiffness at a reference stress, ␴ is “minimum prin-
soil dilatancy, is quite common in modeling rockfill embankments. cipal stress” for Eref ref nc
50 and Eur and “minimum principal stress over K0 ”
The mechanical parameters used for the model are presented in ref nc
for Eoed. K0 is the K0 value for normal consolidation, ␯ur is Poisson’s
Table 4. In the table, Eref ref
50 and Eur are the secant modulus in 50% of ratio for unloading–reloading, and Rf is the failure ratio (ultimate
maximum shear stress and unloading–reloading stiffness in a stan- over asymptotic deviatoric stress). c, ␸, and ␺ are cohesion, friction
ref
dard triaxial test, respectively; Eoed is the tangent stiffness for pri- angle, and dilation angle, respectively.
mary oedometer loading at a reference stress, pref; and m is the power The construction of the generic embankment was modeled us-
for stress-level dependency of the stiffness moduli: ing the software’s staged construction option in which 10 m lifts

Published by NRC Research Press

 762 Can. Geotech. J. Vol. 54, 2017

Fig. 10. Relative crest settlement attributed to impoundment versus rockfill secant modulus at end of construction. [Color online.]
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Fig. 11. Relative crest settlement rate after impoundment versus rockfill secant modulus at end of construction. [Color online.]
For personal use only.

Table 3. Calculated parameters for the studied CFRDs.

␣1 (% height Simp ␣2 (% height Time from
ERC per log time (% dam per log EoC to
Dam name (Mpa) cycle) height) time cycle) EoI (years)
Bastyan 130 0.022 0.016 0.017 1.05
Cethana 137 0.027 0.036 0.038 0.48
Foz Do Areia 47 0.059 0.031 0.048 0.90
Murchison 190 0.022 0.013 0.030 1.09
Outardes 2 main dam — — 0.015 0.017 0.15
Reece 86 0.026 0.045 0.039 1.46
Shuibuya 96.5 — 0.055 0.028 1.07
Toulnustouc south dyke — 0.009 0.015 0.007 1.66
Toulnustouc main dam — 0.028 0.079 0.037 0.53
Kangaroo Creek — — — 0.082 —
Kotmale 61 0.059 0.090 0.113 1.04
Mackintosh 45 0.093 0.115 0.134 2.90
Mangrove Creek 55 — — 0.285 —
Serpentine 92 — 0.051 0.076 2.90
Tianshengqiao I 43 — — 0.140 —
White Spur 180 0.023 0.013 0.056 0.24
Winneke 55 0.064 0.084 0.026 5.40
Winscar 17 — — 0.175 —
Xingo 34 — 0.203 0.163 1.46
Note: EoC, end of construction; EoI, end of impoundment.

Published by NRC Research Press

 Kermani et al. 763

Table 4. Hardening soil model parameters used for modeling a generic embankment.
␥ (kN/m3) ref
E50 (MPa) ref
Eoed (MPa) ref
Eur (MPa) ␯ur pref (kPa) Rf m c (kPa) ␸ (°) ␺ (°) K0nc
23 100 100 300 0.2 100.0 0.9 0.5 0.0 45 10 0.5

Fig. 12. Stress state inside a generic CFRD (a) before and (b) after impoundment. [Color online.]
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17
For personal use only.

Fig. 13. Relative crest settlement rates before impoundment, ␣1, Fig. 14. The ratio of relative crest settlement rates after and before
versus after impoundment, ␣2. [Color online.] impoundment versus time elapsed from the end of rockfill
construction, EoC, to the end of impoundment, EoI. [Color online.]

were considered. After construction modelling, the reservoir load to the failure envelope and varies between 0 for isotropic stress
was applied on the impervious upstream face in a single stage. The state and 1 for failure state. As can be seen, by applying reservoir
computed mean effective stress, deviatoric stress, and relative load, both mean and deviatoric stresses increase on the upstream
shear stress, ␶mob/␶max, before and after application of reservoir side and centerline of the embankment. However, the mean stress
load are shown in Fig. 12. ␶mob is the mobilized shear strength or increases more significantly and the relative shear stress de-
the radius of the Mohr circle and ␶max is the maximum value of the creases in the upstream side of the embankment. While stresses
shear stress, for the case when the Mohr circle expands and generally increase inside the embankment, if considering a Cou-
touches the Coulomb failure envelope considering a fixed center. lomb failure criterion, the material moves away from failure state
Therefore, ␶mob/␶max is an indicator of the proximity of an element due to the induced stresses. This is an advantage of CFRDs com-

Published by NRC Research Press

 764 Can. Geotech. J. Vol. 54, 2017

Fig. 15. Typical cross section of La Yesca dam (adapted from Marengo-Mogollón and Rivera-Constantino 2012).
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Fig. 16. Contours of settlements (cm) at La Yesca Dam for the end of construction (modified from Marengo-mogollón and Rivera-Constantino
2012). [Color online.]
For personal use only.

pared to other types of embankment dams, where impoundment the voids. Additionally, some of the asperities of the particles
increases the risk of slope instability. Additionally, it can be im- break. The change in properties throughout time, and normally
plied that the instantaneous as well as long-term deformations stiffening of the material, is called the aging effect. The first as-
caused by impoundment in a CFRD are mostly due to increase in pect results in larger time-dependent deformations while the sec-
confining pressure rather than shear stress. ond generally leads to smaller deformations. As can be seen in
Returning to the data from studied CFRDs, Fig. 13 shows how Fig. 13, in most of the studied cases, the first mechanism prevails
the settlement rates before and after impoundment compare for (right side of ␣2 = ␣1 line) and in some, the second one is dominant
different cases. Despite some exceptions, the relative settlement (left side of ␣2 = ␣1 line).
rates become larger after impoundment. Generally, the ␣2/␣1 ratio To study aging effect, the ratios of relative crest settlement
lies between 0.8 and 2.5. The relative crest settlement rate after rates, after and before impoundment, ␣2/␣1, are drawn versus the
impoundment, ␣2, is very small for Winneke dam because it has
elapsed time between end of construction and end of impound-
undergone a very long settlement period before the end of im-
ment (start of the third phase) (Fig. 14). A clear trend can be seen in
poundment (5.4 years). Neglecting this case, generally, the dams
this figure for the dams constructed of high strength rockfill. As
constructed of high strength rockfill zones show higher ␣2/␣1 val-
the time interval increases, the ratios of settlement rates decrease.
ues in comparison with those constructed of very high strength
rockfill. It can be inferred that for the rockfill of lower strength, In other words, the longer the time before start of the third phase
the strain rate is more influenced by the stress level than the (the end of impoundment), the more significant the effect of aging
rockfill of higher strength. This is in accordance with the results and the less the dam’s crest settlement in the period afterwards.
of the previous sections where the relative crest settlement rates The Winneke dam, which was considered an outlier in Fig. 13, falls
of dams constructed of very high strength rockfill zones show less very well inside the trend in Fig. 14. For the dams constructed of
dependency on the dam height. very high strength rockfill, in contrast, no clear trend is observed.
Reservoir impoundment has two influences on the behavior of This may indicate that the aging effect is more influential in the
CFRDs: first, as discussed previously, the stress distribution inside less stiff rockfill, where the particles break more frequently and
the dam changes due to the load application on the upstream face; the voids fill more often between the rock particles. Therefore, the
second, the rockfill material undergoes changes, some of the number of interparticle contact points increases and less breakage
cracks inside the particles propagate, and particles break and fill takes place in the subsequent phase.

Published by NRC Research Press

 Kermani et al. 765

Fig. 17. Contours of vertical stress for a generic embankment with 1 Fig. 18. Post-construction crest settlement prediction of La Yesca
horizontal to 1.4 vertical slope at end of construction. H, dam height dam (a) before and (b) after the end of impoundment. [Color online.]
in the section; ␥, fill density. [Color online.]
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

Example of post-construction crest settlement

prediction
The La Yesca dam in Mexico, a 208 m high and 628 m long
structure, was chosen to permit an actual prediction of the post-
construction settlement of a CFRD. This can be considered as a
“class B” prediction according to (Lambe 1973) because the results
For personal use only.

are not known at the time prediction is made. Construction of the

dam body began in 2008 and ended in April 2012. As can be im-
plied from the existing data, the impoundment started right after
end of construction and ended in September 2013. However, no
data on post-construction settlements of the dam were found. The
material properties and deformation of the dam during construc-
tion are presented in Marengo-Mogollón and Rivera-Constantino
(2012). The typical cross section of the dam includes three major
zones (Fig. 15). The main embankment zones (3A and 3B) are com-
prised of rockfill and coarse-grained alluvial soil. Therefore, the La
Yesca dam is different from the studied cases in this respect. De-
spite the lack of detailed data, the UCS of the rockfill (that is
originated from ignimbrites) is assumed to be in the very high
strength category.

Using dam height and rockfill strength

The height of the maximum section of the dam is 195 m at the
centerline from the crest to the foundation. Therefore, using Fig. 4,
relative crest settlement rate after construction, ␣1, extrapolates
0.055% per log cycle of time. Although the impoundment of the dam
started right after end of construction and the first phase does not estimation of settlements in each phase, t1, is considered to be
exist in this case, ␣1 must be used to calculate the time-dependent 0.1 years in the calculations (e.g., in eq. (3)) because it yields ac-
deformations during impoundment. So, considering that impound- ceptable results for the studied cases. Considering smaller values
ment finishes 1 year and 4 months (1.33 years) after end of construc- of beginning of the time interval tends to overestimate the settle-
tion, settlement of ments.
(3) S(% dam height) ⫽ 0.055 × log(1.33/0.1) ⫽ 0.06% or 12 cm Using rockfill secant modulus and strength
The vertical deformation contours of La Yesca dam measured
must be summed with the settlements attributed to first filling to with hydraulic levels at end of construction are shown in Fig. 16.
estimate total settlement during impoundment period. The crest As it is indicated, for a layer located in the bottom one third of the
settlement attributed to stress increase caused by impoundment dam, the vertical settlement, ␦s, is approximately 80 cm and the
can be calculated extrapolating from Fig. 5. Considering the x layer thickness, d1, equals 65 m. The vertical stress contours rela-
value (dam height = 195 m), the Simp amount is estimated as 0.05%
tive to dam height, computed from linear elastic analysis of a
of dam height or 10 cm. Likewise, extrapolating from Fig. 7 (x value =
dam height = 195 m), the relative crest settlement rate after im- generic homogeneous, symmetrical embankment with the same
poundment is estimated as 0.05% per log cycle of time. Therefore, embankment slopes as La Yesca dam, are presented in Fig. 17.
for instance, in 50 years of operation after impoundment, the Using this figure, the vertical stress at the mid-height of the con-
relative crest settlement would be 0.052 × log(50/0.1) = 0.14% dam sidered layer (supposedly, the average vertical stress of the layer)
height or 27.5 cm. Note that the beginning of the time interval for can be calculated as

Published by NRC Research Press

 766 Can. Geotech. J. Vol. 54, 2017

Table 5. Summary of predicted parameters for La Yesca dam.

Settlement attributed Settlement after
to impoundment impoundment
Time-dependent
Parameter used ␣1 (% per log settlement during Simp (% dam ␣2 (% per log (cm in
for prediction cycle of time) impoundment (cm) height) (cm) cycle of time) 50 years)
Dam height 0.055 12 0.05 10 0.052 27.5
Rockfill modulus 0.02 4.5 0.02 4 0.025 13.5

␴z It must also be noted that the presented method is a phenom-

(4a) ⫽ 0.68 ⇒ ␴z ⫽ 0.68 × 20.1 × 195 ⫽ 2665 kN/m2
␥H enological approach. While all of the estimations and calculations
in this study depend on the definition of the time datum, accord-
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

ing to Leroueil (2001), any change in the time frame datum must
where ␥ is the unit weight of the overlaying fill layers and H is dam
not change results of the response of a system due to the axiom of
height. Thus, the modulus will be
objectivity. Nevertheless, the guidelines presented here yield a
␴z representative engineering estimation of the post-construction
2861
(4b) ERC ⫽ ⫽ ⫽ 216 550 kN/m2 behavior of the CFRDs.
(␦s /d1) (0.8/65)
Conclusions
Conducting the same procedure for two other layers adjacent to In the presented study, the post-construction crest settlement
the foundation with different thicknesses, an average modulus of of CFRDs is linked to the in situ mechanical properties of rockfill
210 MPa was calculated for prediction. during construction and the geometry of the embankment through
Now, using the calculated rockfill secant modulus, extrapo- an empirical method. The presented methods can be used to over-
lating from Fig. 9, ␣1, equals 0.02% per log cycle of time (time- come some limitations of numerical models and laboratory experi-
dependent settlement during impoundment period would equal ments, such as scale effects, difference between densities and
0.02% of dam height using the same method as in eq. (3)). Likewise, moisture contents in the field and in the laboratory, diversity of
For personal use only.

implementing Fig. 10, the crest settlement attributed to first materials in a rockfill embankment, and complexity of stress
filling equals 0.02% of dam height, i.e., 4 cm. Also, using Fig. 11, paths in the field. Additionally, after assessment of the time-
the relative crest settlement rate after impoundment, ␣2, equals dependent settlements of the studied cases, aging is shown to
0.025% per log cycle of time i.e., 0.07% (or 13.5 cm) in 50 years significantly influence the settlement rates for compacted rockfill
after impoundment. The relative settlements calculated using within the high strength category.
this method are shown in Fig. 18 for the periods before and
after impoundment. Acknowledgements
Table 5 presents a summary of estimated parameters for post-
The authors gratefully acknowledge the financial support of the
construction settlements of La Yesca dam. The second method
Natural Sciences and Engineering Research Council of Canada
yields smaller settlement parameters compared to the first method.
(NSERC) – Hydro-Québec Industrial Research Chair for life cycle
The settlements of La Yesca dam during construction were small
optimization of embankment dams. The authors also extend their
compared to similar dams. This yields very large modulus for rockfill
and consequently, small values for post-construction settlement. It is appreciation to their industrial partners: Hydro-Québec, SNC-
expected that the values from the modulus method will yield more Lavalin, Qualitas, WSP, Golder Associates, Klohn Crippen Berger,
reliable estimations. It must be noted that as the annual rainfall ConeTec, and Hatch.
histogram of the dam indicates, the impoundment period of the
dam includes two rainy seasons with maximum monthly precip-
References
Alonso, E.E., Olivella, S., and Pinyol, N.M. 2005. A review of Beliche Dam. Géo-
itation of 300 mm. Therefore, it can be anticipated that the settle-
technique, 55(4): 267–285. doi:10.1680/geot.2005.55.4.267.
ments during impoundment would be higher than the estimated Alonso, E.E., Tapias, M., and Gili, J. 2012. Scale effects in rockfill behaviour.
values due to wetting collapse, especially if the rockfill had not Géotechnique Letters, 2(3): 155–160. doi:10.1680/geolett.12.00025.
been sluiced during construction. Charles, J.A. 2008. The engineering behaviour of fill materials: the use, misuse
and disuse of case histories. Géotechnique, 58(7): 541–570. doi:10.1680/geot.
Limitations of the presented methods 2008.58.7.541.
Clements, R.P. 1981. The deformation of rockfill; inter-particle behaviour, bulk
In this study, the UCS of intact rock was used to characterize properties and behaviour in dams. King’s College, London.
rock quality. However, the susceptibility of the rock to lose its Clements, R.P. 1984. Post-construction deformation of rockfill dams. Journal of
strength while in contact with water, which leads to collapse due Geotechnical Engineering, 110(7): 821–840. doi:10.1061/(ASCE)0733-9410(1984)
110:7(821).
to wetting in rockfill, must also be considered. This aspect is of
Colliat-Dangus, J.L., Desrues, J., and Foray, P. 1988. Triaxial testing of granular
essential importance in areas with high precipitation and where soil under elevated cell pressure. In Advanced triaxial testing of soil and rock.
rockfill is not watered during construction. The Roadford asphal- ASTM STP977. ASTM. pp. 290–310. doi:10.1520/STP29082S.
tic concrete face rockfill (sandstone and mudstone) dam (Charles Cooke, J.B. 1984. Progress in rockfill dams. Journal of Geotechnical Engineering,
2008) and the Martin Gonzalo concrete face rockfill (slate and 110(10): 1381–1414. doi:10.1061/(ASCE)0733-9410(1984)110:10(1381).
Cormier, V., and Konrad, J.-M. 2015. Short and long term compressibility of rock
greywacke) dam (Justo and Durand 2000) are two examples of
particle assemblages. In Proceedings of the 68th Canadian Geotechnical
impervious upstream face dams where significant wetting col- Conference; GéoQuébec, Que., 20–23 September.
lapse settlement has taken place. Compared to the other cases Fitzpatrick, M.D., Cole, B., Kinstler, F., and Knoop, B. 1985. Design of concrete-
evaluated in this study, the moduli at end of construction values faced rockfill dams. In Proceedings of the Symposium on Concrete Face Rock-
are very low for Roadford and Martin Gonzalo (32 and 16 MPa, fill Dams - Design, Construction and Performance, Detroit, Michigan. Edited
by Cooke and Sherard. ASCE, New York. pp. 410–434.
respectively). Therefore, low rockfill secant modulus at end of Giudici, S., Herweynen, R., and Quinlan, P. 2000. HEC experience in concrete
construction can also be an indicator of susceptibility to wetting faced rockfill dams - past, present and future. In Proceedings of the Interna-
collapse. This aspect needs further investigation. tional Symposium on Concrete Faced Rockfill Dams, Beijing. pp. 29–46.

Published by NRC Research Press

 Kermani et al. 767

Hunter, G., and Fell, R. 2002. The deformation behaviour of rockfill. The Univer- Marulanda, A. 2013. Recent successful very high CFRD’s. In Proceedings of the
sity Of New South Wales, School Of Civil And Environmental Engineering. Annual Canadian Dam Association (CDA) Conference, Montréal, Que.
UniCiv Report No. R-405. [January.] Mesri, G., and Godlewski, P.M. 1977. Time- and stress-compressibility interrela-
ICOLD. 2010. Concrete face rockfill dams: concepts for design and construction. tionship. Journal of the Geotechnical Engineering Division, ASCE, 103(GT5):
Bulletin 141. ICOLD. 417–430.
Justo, J.L., and Durand, P. 2000. Settlement-time behaviour of granular embank- Mesri, G., and Vardhanabhuti, B. 2009. Compression of granular materials.
Canadian Geotechnical Journal, 46(4): 369–392. doi:10.1139/T08-123.
ments. International Journal for Numerical and Analytical Methods in
Milligan, V., and Coyne, L. 2005. Review of factors influencing the settlement of
Geomechanics, 24(3): 281–303. doi:10.1002/(SICI)1096-9853(200003)24:3<281:: rockfill dams. In Proceedings of the K.Y. Lo Symposium. The University of
AID-NAG66>3.0.CO;2-S. Western Ontario. pp. 1–31.
Kermani, M. 2016. Prediction of post-construction settlements of rockfill dams, Oldecop, L.A., and Alonso, E.E. 2007. Theoretical investigation of the time-
based on construction field data. Ph.D. thesis, Université Laval. dependent behaviour of rockfill. Géotechnique, 57(3): 289–301. doi:10.1680/
Lambe, T.W. 1973. Predictions in soil engineering. Géotechnique, 23(2): 151–202. geot.2007.57.3.289.
doi:10.1680/geot.1973.23.2.151. Parkin, A.K. 1977. The compression of rockfill. Australian Geomechanics Journal,
Leroueil, S. 2001. Natural slopes and cuts: movement and failure mechanisms. G7: 33–39.
Géotechnique, 51(3): 197–243. doi:10.1680/geot.2001.51.3.197. Pinto, N.D., and Marques Filho, P.L. 1998. Estimating the maximum face deflec-
Marachi, N.D., Chan, C.K., Seed, H.B., and Duncan, J.M. 1969. Strength and de- tion in CFRDs. International Journal on Hydropower and Dams, 5: 28–32.
formation characteristics of rockfill materials. University of California, Ramon, A., Alonso, E.E., and Romero, E.E. 2008. Grain size effects on rockfill
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by 69.157.157.233 on 08/27/17

constitutive behavior. In Unsaturated Soils. Advances in Geo-Engineering:

Berkeley. Report No. TE-69-5.
Proceedings of the 1st European Conference on Unsaturated Soils, Durham,
Marengo-Mogollón, H., and Rivera-Constantino, R. 2012. Geotechnical design for
UK, 2001. pp. 341–347. ISBN 978-0-415-47692-8.
“La Yesca” dam: description of its behavior during the construction stage and Sowers, G.F., Williams, R.C., and Wallace, T.S. 1965. Compressibility of broken
first filling. In Proceedings of the CDA Annual Conference, Saskatoon, Sask., rock and the settlement of rockfills. In Proceedings of 6th ICSMFE, Montréal,
22–27 September 2012. Que. Vol. 2, pp. 561–565.
Marsal, R.J. 1973. Mechanical properties of rockfill. In Embankment dam engi- Soydemir, C., and Kjaernsli, B. 1979. Deformations of membrane-faced rockfill
neering (Casagrande Volume). Edited by R.C. Hirschfeld and S.J. Poulos. dams. In Proceedings of the 7th European Conference on Soil Mechanics and
pp. 109–200. Foundation Engineering, Brighton, England. pp. 281–284.
For personal use only.

Published by NRC Research Press

View publication stats