Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Statistical Model for Dam-Settlement Prediction

and Structural-Health Assessment


Fjóla G. Sigtryggsdóttir 1; Jónas Thór Snæbjörnsson 2; and Lars Grande 3

Abstract: This paper considers a hydrostatic–seasonal–time (HST) statistical model for predicting the settlement behavior of a concrete-
faced rockfill dam (CFRD) during operation. The constituents of the model are innovatively related to viscoelastic–plastic material model
components, with the novel addition of considering unloading–reloading behavior when determining the model parameters. The statistical
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

model developed considers both load-related and time-related deformation behavior of rockfill and CFRDs. The discussion compares differ-
ent functions from the literature describing time-dependent deformation of rockfill dams and CFRDs and emphasizes cautious selection of
predictors to enhance any statistical model’s credibility. The model is applied to the case of the 200-m-high Kárahnjúkar CFRD. The aim of
the statistical analysis is first to create a prediction model for short- and long-term settlements of the dam, and second to extract behavioral
patterns for structural-health monitoring applications. This is important for the operational safety of large dams, and thus the resilience and
sustainability of related civil infrastructure systems. DOI: 10.1061/(ASCE)GT.1943-5606.0001916. This work is made available under the
terms of the Creative Commons Attribution 4.0 International license, http://creativecommons.org/licenses/by/4.0/.
Author keywords: Concrete-faced rockfill dam; Rockfill; Statistical model; Settlement; Viscoelastic–plastic; Creep; Structural-health
monitoring.

Introduction monitoring data from concrete dams. Thus, there is a need to


establish this field in the case of geotechnical structures like
Monitoring and surveillance of important infrastructures is an embankment dams.
integral part of civil engineering practice. This has evolved along The focus of this paper is on statistical prediction modeling of
with technology and is referred to, in its most advanced form, as monitoring data from measurements in a concrete-faced rockfill
structural-health monitoring (SHM). SHM is the process of imple- dam (CFRD). In recent years, large dams of this type have gener-
menting a monitoring system along with a strategy to identify un- ally been monitored extensively, some with more than 300 meas-
favorable change (damage) in the normal condition of a structure uring locations. Each measuring location results in a time series
(Worden et al. 2007). An SHM management system embraces, first, of monitoring data for the dam surveillance to consider. This large
damage diagnosis, integrating statistical prediction models with amount of data theoretically enables comprehensive application of
numerical mechanics models [e.g., using finite-element (FE) mod- an SHM. There are reports on the usage of monitoring data from
els], and second, damage prognosis for making decisions on main- such dams for calibrating mechanics models. However, reports on
tenance or giving alarms for emergency preparedness (e.g., Farrar statistical prediction models relating to SHM of a CFRD or other
and Lieven 2007). SHM is important for the operational safety of geotechnical structures are, by the authors’ experience, rare and
large dams. incomplete.
Fig. 1 demonstrates the use of statistical and mechanical models The deformations of CFRDs govern their behavior during nor-
in dam engineering considering dam operation. This relates to mal operation and induce strains and stresses in the concrete face.
an SHM procedure for prognosis of the health of a dam or its Thus, it is logical to initiate the SHM process by analyzing moni-
parts. However, judging from scientific publications on this topic tored deformations, such as the settlements. Hence, this paper
for dams, statistical models have mostly been used in analyzing proposes a multivariate linear regression (MVLR) statistical model
to simulate the settlement of a large CFRD based on records of
1
Associate Professor, Dept. of Civil and Environmental Engineering, full-scale settlement data.
Norwegian Univ. of Science and Technology, S.P. Andersens veg 5, The statistical model proposed considers the settlement
Trondheim 7491, Norway (corresponding author). Email: fjola.g behavior of rockfill and relates the components of the model to a
.sigtryggsdottir@ntnu.no
2 viscoelastic–plastic (VEP) material model. Additionally, the calcu-
Professor, School of Science and Engineering, Reykjavík Univ.,
Menntavegur 1, Reykjavik IS-101, Iceland; Professor II, Dept. of lation procedure considers the unloading–reloading behavior of
Mechanical and Structural Engineering and Material Science, Univ. of the dam material, which is a novel addition in such models.
Stavanger, Stavanger 4036, Norway. Email: jonasthor@ru.is; jonas.t This paper briefly reviews general statistical models for estimat-
.snebjornsson@uis.no ing dam response as well as multivariate linear regression method-
3
Professor Emeritus, Dept. of Civil and Environmental Engineering, ology for obtaining the statistical parameters. Subsequently, rockfill
Norwegian Univ. of Science and Technology, Høgskoleringen 7, material behavior is reviewed and the statistical model for the
Trondheim 7491, Norway. Email: lars.grande@ntnu.no
settlement behavior of a CFRD introduced, followed by relation
Note. This manuscript was submitted on July 1, 2017; approved on
February 7, 2018; published online on June 27, 2018. Discussion period of its components to a VEP material model. Discussion on the
open until November 27, 2018; separate discussions must be submitted different parameters of the proposed model follows and includes
for individual papers. This paper is part of the Journal of Geotechnical a comparison of different functions from the literature on time-
and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. dependent deformation of rockfill dams and CFRDs. Finally, the

© ASCE 04018059-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Herein, MVLR methods were used. MVLR methods were applied
to a set of simultaneous observations to determine the relationship
between the predictor (independent) variables and the response
(dependent) variables. Predictions can then be made from the es-
tablished relationship between the response variables or samples,
Y, versus one or more predictor variables X. The general relation-
ship can be expressed as (Rencher 2002)
Y ¼ Xb þ Ξ ð2Þ

Here the predictors are represented with the matrix


X ¼ ½1; x1 ; x2 ; : : : ; xn  of n variables, where 1 is a column vector
of ones and each variable xi has p dimensions [xi ¼
ðx1i ; x2i ; : : : ; xpi ÞT ] or observations (T is transpose). The matrix
b contains the regression coefficients, and Ξ is a matrix of resid-
uals. Conversely, the matrix Y ¼ ½y 1 ; y 2 ; : : : ; y m  of m variables
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

represents the response variables, where each variable y i has p


dimensions [y i ¼ ðy1i ; y2i ; : : : ypi ÞT ]. Thus, the matrix X is p by
(n þ 1) [p rows by (n þ 1) columns], and the matrix Y is p by
m (p rows by m columns). Expansion of Eq. (2) results in
2 3 2 32 3
y11 y12 · · · y1m 1 x11 x12 · · · x1n β 01 β 02 · · · β 0m
6 7 6 76 7
6 y21 y22 · · · y2m 7 6 1 x21 x22 · · · x2n 76 β 11 β 12 · · · β 1m 7
6 7 6 76 7
6 . . . 7¼6 76 7
6 .. .. . . ... 7 6 ... ... ... . . . ... 76 ... ... . . . ... 7
4 5 4 54 5
Fig. 1. Scheme for engineering control using numerical models during 1 xp1 xp2 · · · xpn
yp1 yp2 · · · ypm β n1 β n2 · · · β nm
dam operation. (Adapted from ICOLD 2012.) 2 3
ε11 ε12 · · · ε1m
6 7
6 ε21 ε22 · · · ε2m 7
6 7
þ6 . . . 7 ð3Þ
test case involving the Kárahnjúkar CFRD is introduced, and the 6 .. .. . . ... 7
application of the proposed model is demonstrated. 4 5
The aim of the study is first to create a prediction model for the εp1 εp2 · · · εpm
short- and long-term settlements, and second to extract behavioral
patterns for structural health monitoring applications, for example, In the case where there is only one response variable, the
to distinguish between normal and abnormal behavior. Addition- regression methodology is called multiple linear regression
ally, the objective is to demonstrate the possibility and advantages (Rencher 2002) and the matrix reduces to the vector y 1 ¼
of considering physical factors and their relation to dam response in ðy11 ; y21 ; : : : ; yp1 ÞT . In that case b is a vector of the regression
the development of a statistical model for a rockfill dam. coefficients, b ¼ ðβ o ; β 1 ; β 2 ; : : : ; β n Þ, and Ξ reduces to a vector
e containing the error terms. This method assumes that the errors
are uncorrelated, and the expected value of the error terms is zero,
General Multivariate Hydrostatic–Seasonal–Time i.e., EðeÞ ¼ 0; but the variance is VðeÞ ¼ sd 2 , where sd is the
Regression Model standard deviation.
The general HST model, first proposed by Ferry and Willm
The general statistical model for the recorded response of a monitor- (1958) and Willm and Beaujoint (1967), has been used for analyz-
ing instrument in a dam is often called a hydrostatic–seasonal–time ing the deformation monitoring data of dams, mainly concrete
(HST) model (Chouinard et al. 1995) after the key components dams (Chouinard and Roy 2006; Léger and Leclerc 2007;
generally defined for the statistical model. Such models have been De Sortis and Paoliani 2007). The same model approach has also
widely applied to concrete dams. been adopted for analyzing settlements of earth-rockfill dams
The components of a general HST model represent the loading (Hu et al. 2011) as well as for a CFRD (Wu et al. 2009).
and environmental conditions influencing the recorded response, This statistical model approach applies to the normal operational
δ, as well as irreversible time-dependent effects such as creep. The conditions and is generally valid for prediction within the limits
HST model assumes that the various response properties can be of the hydrostatic loading and environmental conditions defined
studied separately, and can be formulated as follows (Chouinard by a set of prediction or design variables used for the regression
et al. 1995; Léger and Leclerc 2007; De Sortis and Paoliani (Chouinard et al. 1995).
2007; Mata 2011):

δðh; t; sÞ ¼ δ H ðh; tÞ þ δ S ðsÞ þ δ T ðtÞ þ ε ð1Þ Rockfill Material Behavior


where δðh; t; sÞ = recorded response; δ H ðh; tÞ = effect of hydro- The HST model has mainly been used for concrete dams. In
static thrust; h = reservoir elevation; t = time; δ S ðsÞ = seasonal the case of rockfill dam settlements, an alternative to the philosophy
environmental effect; s = seasonal parameter; δT ðtÞ = irreversible behind the traditional presentation of the HST statistical model
time-dependent effects; and ε = residual error. is to relate the different components of a statistical model to
The parameters of statistical behavior models as presented the deformation behavior, which for rockfill material generally
in Eq. (1) can be obtained using different regression techniques. depends on both loading and time.

© ASCE 04018059-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Load-related deformation, plotted as stress against strain, pro- X 
m
ðH̄ f;i − H̄f;0i Þ
vides information on material stiffness and its dependence on the δ ¼ δH þ δT þ ε ¼ a1i
i¼1
Href
stress-strain level induced in the material. Triaxial testing of granu- ff
lar soils has shown that the unloading and reloading modules are X  
k
ðH̄ave;j − H̄ ave;0j Þ
higher than the modulus on primary loading (Stewart 1986; Byrne þ a2j;u=r
j¼1
Href
et al. 1987). It has also been shown that the unloading and reloading
X q  
moduli may differ slightly (Duncan et al. 1980; Saboya and Byrne ΔH̄p;i
1993). Furthermore, the unloading process provides only partial þ a3i;l=u þ a4 ½ln θ − ln θ0  þ a0 ð4Þ
Href
recovery of the initial deformation. i¼1 σ>σMP

Considering time-dependent deformation, several theories


have been set forth to explain the time-dependent behavior of soils The symbols used in Eq. (4) are described in detail sub-
sequently; however, for an overview, ai are regression coefficients;
under load. Bjerrum (1967) proposed the concept of delayed com-
the symbols H̄ f;i , ΔH̄ p;i , H̄ ave;j , and Href all relate to the water
pression under constant effective stress, and Janbu (1969, 1985)
head; σ is the current stress, while σMP is the previously experi-
introduced the time resistance concept. Similar phenomena are
enced maximum stress; θ represents accumulated total days (t)
observed for granular materials, such as rockfill (e.g., Augustesen
divided by 100 (θ ¼ t=100); and θ0 is a corresponding base value.
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

et al. 2004; Lade et al. 2009; Justo and Durand 2000). Plotting
The symbols i and j are indexes of summation, while m, k, and q
deformation versus time for a rockfill sample subjected to step-
are the upper limit of summation.
wise loading up to a certain maximum has been shown to contain
The left-hand side of Eq. (4) represents the response on each
initial instantaneous deformation (compaction) and transient or observation day, i.e., at one point in time. Furthermore, the predic-
delayed deformation for each load increment (the viscoelastic re- tors on the right-hand side of the equation are calculated relating to
sponse), as well as steady-state (long-term) creep (e.g., Dolezalova that same observation day. The subsequent discussion of Eq. (4)
and Hladik 2011). assumes the resulting time series of settlement (δ) since the end
Similarly, the deformations induced by the water pressure on of the first filling. For proper consideration of delayed response
the first filling of a CFRD include (1) the near-instantaneous re- to hydrostatic loading, the time series of the reservoir elevation
sponse to the primary loading represented by the added water must extend from nd days (nd ¼ 110–190 for the test case) prior
pressure, and (2) the delayed response. The instantaneous re- to the end of the first filling. The time series should have time steps
sponse is largely irrecoverable, e.g., due to particle breakdown as of 1 day. The regression assumes a sample set of size p for the
well as rearrangement of the rock particles during compression. responses and predictors. However, the calculation requires time
During operation of a CFRD the particle breakdown continues, series of the reservoir elevation with p þ nd time steps. The calcu-
and creep dictates the behavior, with decelerated settlements ap- lation procedure starts by calculating all the predictors on the left-
proaching a horizontal asymptote during the operational lifespan. hand side for each observation day, at time step ip (ip ¼ 1; : : : ; p).
Additionally, variations in the reservoir elevation are likely to af- The reservoir elevation at the observation day is at time step nd þ ip
fect the settlement behavior. Unloading during a dam’s operation of the reservoir time series. The routines within the calculation
(from lowering the reservoir) results in partial recovery, mainly of procedure use the observed values of reservoir elevation and the
the viscoelastic response, whereas reloading (raising the reservoir) time in days as well as the constraints described for each predictor.
induces this again. The stress condition depends on the location Subsequently the predictors are regressed to the observed settle-
within the dam. The settlement response at the upper levels may ments to obtain the regression coefficients.
be mainly due to viscoelastic response, while at the lower levels, In Eq. (4), the subscript ff on the first term refers to first filling
where the state of stress is higher, plastic deformations may have and denotes that this term accounts for the settlements induced by
occurred. Reservoir elevations within the limits already experi- the first filling. The parameters a1i , a2j;u=r , and a3i;l=u are regression
enced by the dam are likely to result in viscoelastic response, coefficients for the hydrostatic component, a4 is a regression
mainly in the upper region of the dam. Conversely, hydrostatic coefficient for the time effect component, and a0 is a constant.
pressure above the limits already experienced by the dam may The subscript u=r for regression coefficient a2;u=r denotes that dif-
induce further permanent settlement. ferent coefficients are calculated, depending on whether the settle-
ment occurs under unloading (u) or reloading (r) conditions.
Similarly, the subscript l=u for regression coefficients a3i;l=u
denotes that different coefficients are calculated, depending on
Settlement Prediction Model Proposed for CFRDs
whether the settlement occurs under loading (l) or unloading (u)
This section explains the calculation procedure for the prediction conditions. The component accompanying a3i;l=u is only used if
model proposed here for a CFRD settlement during operation, the reservoir loading exceeds previous loadings.
i.e., from the end of the first filling. The inclusion of material In the model, H ref is the dead water level and H is the water
behavior as described previously was an essential part of the model head calculated as the difference of the water level at the relevant
development, as further explained in the next section. The predic- observation day and the dead water level. The different factors de-
tion model was developed by considering material behavior and fined for the hydrostatic component account for delayed response
stress state, but is expressed in Eq. (4) in the form of an HST model, to the hydrostatic loading. The predictor ðH̄f;i − H̄ f;0i Þ=H ref ac-
with δ as the recorded response, δ H as the load-related deformation counts for delayed response to the first reservoir impounding.
considering effect of hydrostatic thrust, δT as the time-dependent The delay depends on location of the instrument within the
settlement, and ε as the residual error. The seasonal component dam, with the delay somewhat increasing toward the downstream
from temperature variations is generally insignificant for rockfill side. At time step ip of this predictor, H̄ f is the average value of the
dam settlements and thus is not included. However, the model water heads observed at time step ip to [ip þ ðna − 1Þ] of the res-
developed considers unloading–reloading (u=r) from variations ervoir time series (i.e., at nd to [nd − ðna − 1Þ] days prior to the
in the reservoir elevation as well as the effect of the first filling observation day). The number na is the number of observations
P a −1
(ff). The statistical model is presented as follows: included in the average. Thus, H̄f ¼ ð ni¼0 Hip þi Þ=na . For the

© ASCE 04018059-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Table 1. Overview of the coefficients (a0 , a11 , a2j;r=u , and a4 ) used in the statistical model for the test case with grouping based on an HST model, along with
material behavior and loading conditions (Fig. 2)
Residuals (ε) Hydrostatic Seasonal Time
Transient response (viscoelastic) Loading above previous
to subsequent unloading maximum loading and subsequent
Response prior to the end Transient response Time
and reloading unloading and reloading
of first filling (other than (viscoelastic) to dependent
transient response, see a11 ) first filling Unloading Reloading Loading Unloading Reloading Not included creep
a0 a11 a21;u , a22;u , a23;u , a21;r , a22;r , a23;r , Not relevant for the Not included a4
a24;u , a25;u a24;r , a25;r test case (a3i;r;u )

test case, na ¼ 7 was generally used, and nd had a value of 110– Relation to a Viscoelastic–Plastic Material Model
160 days. Similarly, H̄f;0 is a corresponding average of the water VEP constitutive model components (e.g., Oyen and Cook 2003)
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

heads considering the first na days of the reservoir elevation time can be used to describe rockfill deformation behavior, and they
series. The summation in Eq. (4) denotes that more than one pre- have been used in a finite-element analysis of CFRDs (e.g., Gan
dictor of this type may be required to capture this effect. In the case et al. 2014). In Fig. 2, the components of the statistical model pro-
study presented herein, m ¼ 1 was used. posed through Eq. (4) are related to elements of a VEP material
In the summation associated with a2j;r=u , k is a value that should model. The relation considers that the true material behavior is
be defined for each case studied. For instance, the test case uses inherent in the recorded data used for the regression.
k ¼ 5. In the ratio ðH̄ave;j − H̄ave;0j Þ=H ref , H̄ave;j is the average In Fig. 2, EM;I is a deformation modulus describing the instanta-
water head observed on Days 1–3, 4–10, 11–30, 31–60, and neous compression of CFRDs when loaded (which replaces
61–110 before the observation day, whereas H̄ave;0j is the average the elastic modulus of the Maxwell model) and ηM;t represents a
of water head observed on Days 1–3, 4–10, 11–30, 31–60, and time-dependent viscous function to describe the time-dependent
61–110 before the first date of settlement observation series. creep of CFRDs (which replaces a constant viscous coefficient of
The factor ΔHp;i accounts for the loading induced when the the Maxwell model). Furthermore, EK and ηK are respectively the
reservoir elevation rises above an elevation to which the dam elastic modulus and viscosity coefficients for the transient elastic
has previously been subjected, and thus the current stress (σ) is
creep, i.e., the viscoelastic behavior (Kelvin-Voigt model). Finally,
larger than previously experienced maximum stress (σMP ). This
Ey and ηy are respectively the plastic hardening modulus and
component was not activated in the model because this loading
plastic viscosity coefficient, and σy is the yield stress.
condition was not relevant for the particular case studied. However,
The VEP model in Fig. 2 is essentially as presented by Gan et al.
it is included in the model description for the sake of completeness.
(2014) for the rockfill in an FE analysis of a CFRD, with the ex-
The parameter q in the last summation of Eq. (4) is a value that
ception of the different possible loading conditions indicated with
should be defined for each case studied. Additional components
the shaded boxes. For example, there is the possibility of a different
may be required to describe the actual behavior under different
loading conditions.
Other symbols used in Eq. (4) are θ ¼ t=100, where t represents
the number of days passed since the end of construction (or when
the embankment elevation surpassed the level of the instrument
location) until the observation day; and θ0 ¼ t0 =100, where t0 is
the number of days from the end of construction to the end of
the first filling. Thus t ¼ 0 corresponds to the end of dam construc-
tion, or, as may apply, when the embankment elevation reached
a level that was above the instrument location.
Table 1 summarizes the coefficients used for the test case.

Components of the Proposed Statistical Model


The fitting accuracy of a statistical model and its representation of
actual behavior depends on the choice of influence factors (predic-
tors) and their physical relation to the recorded data. In Eq. (4), the
predictors of the proposed model are grouped as components of a
HST statistical model, whereas this section introduces a different
grouping basing on material behavior as introduced in Table 1. This
entails relating the predictors to the components of a viscoelastic–
plastic material model. This approach was important in the devel-
opment of the model to ensure that each of the traditional HST
components actually spans and captures the settlement behavior
of a CFRD. For this purpose, this section first discusses rockfill
deformation and then the different components of the statistical
Fig. 2. Components of the statistical model for dam operation related
model from the consideration of material properties and deforma-
to a VEP material model.
tion behavior.

© ASCE 04018059-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Table 2. Functions for time-dependent creep
Function Expression Explanation Reference
F1 c1 θ þ c2 θ2 þ c3 e−Dθ ci are regression coefficients and D is obtained from triaxial tests Wu et al. (2009)
In Fig. 3,
F 1.1: c1 is positive, c2 and c3 are negative
F 1.2: c1 and c2 are positive, c3 is negative
F 1.3: c1 is positive, c2 ¼ 0, and c3 is negative
F 1.4: c1 ¼ 0, c2 ¼ 0, and c3 is negative
F2 c1 θ þ c2 ln θ ci are regression coefficients Hu et al. (2011)
In Fig. 3,
F 2.1: c1 and c2 are positive
F 2.2: c1 is negative, c2 is positive
c1 ¼ 0 essentially results in Function F6
 t b
F3 a a and b are obtained from creep tests Haifanga and Yinqi (2012) [a general
t0 t0 is a point in time of the creep test duration version of the initial Ohde’s equation
In Fig. 3, a ¼ 1 and b ¼ −0.08 (Ohde 1939; Janbu 1963)]
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

F4 θ a and b are regression coefficients Zhou et al. (2009)


a þ bθ In Fig. 3, a ¼ −0.1 and b ¼ −0.003
F5 εc ð1 − t−λ Þ λ is obtained from triaxial tests Zhou et al. (2011)
εc is proportional to confining stress
In Fig. 3, εc ¼ 3 and λ ¼ −0.08
F6 a4 ½ln θ − ln θ0  a4 is a regression coefficient Various authors have regressed this
In Fig. 3, a4 ¼ 0.2 formula with monitoring data
θ0 ¼ t0 =100, where t0 is a base value of accumulated total number
of days from a selected reference date
Note: θ ¼ t=100, where t is the accumulated total days (unit of time is days).

settlement behavior if the reservoir loads the dam to a higher stress investigate the relationship between strain, stress, and time for
level. This may result in stresses exceeding the yield stress or in long-term creep. They found that a linear strain-log(time) function
deformation behavior related to primary loading. However, once gave the best fit and argued that this was a feature also supported
the full supply level is reached in reservoir impounding, any addi- by settlement records from rockfill dams. Hunter and Fell (2003)
tional reservoir loading is generally limited to a few meters. When also found that for CFRDs constructed of well-compacted
effects of the first filling no longer have to be accounted for, and rockfill, a log(time) function reasonably describes the overall set-
while the loading (stress state σ) does not exceed previous loading tlement. The results were established by considering the start of the
[and the stress state is within the yield stress state (σy )], the model postconstruction deformation at the end of construction of the main
reduces to a viscoelastic model. rockfill body.
The remainder of this section explains the selection of the func- Various researchers have presented other formulations of the
tions used in the statistical model in Eq. (4), aiming at capturing the long-term creep of CFRDs. Table 2 presents a summary of six
material behavior. different formulations.
In Fig. 3, the functions of Table 2 are plotted against time
Time-Dependent Creep to visualize potential shapes of the curves within a timeframe of
200 years. It is clear that some time constraints are required for
The selection of a predictor for creep is important because this Functions F1 and F2 since they do not provide a natural description
generally governs the overall settlement during normal operation of the settlement behavior. On the other hand, the relations num-
of a dam. The long-term creep of CFRDs has been studied using bered F3–F6 may be considered further. Of those functions, F3–F5
recorded data and rockfill creep behavior from triaxial tests. For use regression coefficients from triaxial tests. In this respect, F3 is
example, Alonso and Oldecop (2007) used experimental data to particularly sensitive to the parameters used since two parameters
must be fitted simultaneously in addition to the linear regression
coefficient. Function F5 assumes convergence to a finite value
εc , which is dependent on the stress state. This stress dependence
of the creep limit, εc , is an important feature of this expression con-
sidering that the state of stress at any point in a CFRD changes
during its different life-cycle phases.
At the initiation of prediction model development, it is desirable
to start with a relation that describes the settlement behavior appro-
priately through a function that can be regressed linearly with other
predictors. In this study, the functional form F6 was selected since it
has been shown to regress reasonably well to the actual settlement
behavior of CFRDs, particularly those of well-compacted rockfill.
Function F6 is, in fact, the simplest version of the Janbu time
resistance concept (Janbu 1969, 1985). Function F6 increases in-
finitely to an asymptote, with the increment decreasing at each time
step, which is in agreement with the general settlement behavior of
Fig. 3. Different functions from Table 2 plotted against time.
a CFRD.

© ASCE 04018059-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Hydrostatic Effect
In HST models for concrete dam displacements, the hydrostatic
effect is usually expressed as a third-degree polynomial function
(Chouinard and Roy 2006), similar to what has been used for
earth-rockfill dams (Hu et al. 2011). Some researches (Wu et al.
2009) have, through back and forth shifts between physical and
empirical relationships, derived an expression for the hydrostatic
loading, resulting in functions with the depth of water as a power
of the elastic modulus number. However, the modulus number can
take on a wide range of values from negative to positive (Soroush
and Jannatiaghdam 2012), which makes the model impractical.
Furthermore, it is important to maintain a basis for physical rela-
tionships and to bear in mind that a statistical model assumes that
different response properties can be studied separately. The trans-
parent relationship of actual behavior is advantageous for statistical Fig. 4. (Color) Simplified scheme to define artifact settlement defor-
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

model application and was preferred in the development of the pro- mation moduli for a rockfill dam due to hydrostatic pressure during
posed model. operation [see similar approach and figure in Hunter and Fell (2003)
Fitzpatrick et al. (1985) presented simplified schemes to for the definition of simplified modulus during reservoir filling].
estimate the rockfill modulus, during construction on one hand
and during first filling on the other [see also Hunter and Fell
(2003)]. These moduli are widely used in the empirical design incorporated into the statistical model to account for the delayed
of CFRDs. The modulus for the construction phase is calculated response.
from the measured settlements of the rockfill dam body, whereas
the first filling deformation modulus is calculated from the mea-
sured deflections of the face slab. The two moduli capture the
Test Case
overall deformation behavior, i.e., no distinction is made between
the two different sources of deformation, the near-instantaneous The case used for testing the prediction models studied in this paper
response to loading, and the delayed contribution. Still, they re- was the 198-m-high and 700-m-long Kárahnjúkar CFRD in Iceland
present a clear physical relation between deformation and loading (Fig. 5). The reservoir elevation since the start of the first
as well as strains and stresses. The hydrostatic loading at the impounding is shown in Fig. 6. The dam is constructed of palagon-
relevant location represents the stress increment and the strain is ite tuff rockfill with dam zoning based on the state of the practice
derived from dividing the measured deformation by the thickness at the time of design and construction.
of the dam layer underneath (with the same alignment as the de- Instrumentation of the dam related to settlement monitoring
formation). A similar artifact modulus can be obtained considering includes benchmarks at the dam crest (CS) and on the downstream
only the vertical deformation, i.e., the dam settlements, during side (TS), as well as hydraulic settlement (HS) gauges within the
operation. This is shown in Fig. 4, explaining the expression for dam body. The locations of the crest stations labeled CS-1 to CS-12
the artifact modulus, rewritten in Eq. (5) to extract the settlement are shown in Fig. 5. The HS gauges are located within three main
sections of the dam, as shown in Fig. 7 (labeled HSA-, HSB- or
γ w d1 HSC-, with a number). The main benchmarks on the downstream
δs ¼ h ð5Þ side, labeled TS, serve as reference stations for the HS gauges.
Es
Fig. 8(a) presents all the available monitoring data series, with
a starting point (all settlements set to 0) at the end of June 2007,
where Es = artifact settlement modulus; γ w = unit weight of water; i.e., during the first impounding. The starting point selection
δ s = settlement at depth h from the reservoir surface; and d1 = depth
of rockfill column.
In a statistical model the settlements δ s at different times are
known at measuring locations within the dam. For each measuring
location, the value of d1 is known from the dam geometry, the value
of h is obtained for each time step from monitoring data of the res-
ervoir elevation, and the density of water γ w is a constant value. The
settlement deformation during operation at a certain location within
the dam body can thus be related in a simple way to the water pres-
sure, i.e., the reservoir elevation h, using Eq. (5).
It seems reasonable in a regression model to relate the deforma-
tion directly to h and consider that the regression is in a way
providing values representing relation to an artifact moduli Es at
different measuring locations, although within the limitations of
the regression model. For operational conditions, it is additionally
possible to account for unloading and loading conditions arising
from variations in the reservoir level.
This applies to both the instantaneous response as well as the
delayed response, both of which have to be accounted for in the
Fig. 5. (Color) Downstream side of the dam and the location of crest
statistical model. Once the time lag between the loading and re-
stations (CS-1 to CS-12). (Image courtesy of Emil Thor.)
sponse has been identified from the acquired data, this can be

© ASCE 04018059-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Fig. 6. Reservoir water elevation since the start of impounding.
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. (Color) Standardized settlements at stations HSA-5, HSA-6,


and HSA-7.

However, it is clear that there are details in the deformation history


that require further attention.

Analysis
The stations HSA-5 to HSA-7 captured reasonably well the settle-
ment response induced by the variations in the reservoir elevation.
Similar details were also observed at other locations, including at
the crest, where stations CS-7 to CS-9 are of special interest be-
cause they are located at the maximum dam section. Hence, these
stations were selected for the analysis.
Fig. 7. Plan and section sketches of the dam showing location of Fig. 9 displays the standardized settlement at stations HSA-5 to
hydraulic settlement (HS) gauges and terminal structures (TS). HSA-7, extending from the end of the first impounding. A delay in
the response to the reservoir impounding was observed, with the
gauges closer to the upstream side showing the earliest response.
The delay in the response to the first filling was about 110 days for
considers the CS data series, which have the latest starting point HSA-5, 160 days for HSA-6, and 190 days for HSA-7. This time
in time. Fig. 8(a) presents processed time series. The processing delay was accounted for in the statistical model.
(Sigtryggsdóttir et al. 2013) included interpolation between read- Fig. 10(a) presents the settlement at station HSA-5. The settlement
ings taken at uneven time intervals to obtain synchronized time time series starts when the reservoir elevation rises above the station
series with equal time steps. The actual readings were conducted location at 585 m above sea level. Fig. 10(b) shows how these set-
frequently during construction and the first impounding. However, tlements plot against an estimate of a proportional value of the de-
the interval between readings has gradually increased, resulting in viatoric stress. This proportional value can be expressed as follows:
some loss of detail in the monitored settlement behavior.
In Fig. 8(b), all the time series shown in Fig. 8(a) are standard- ðσ1 − σ3 Þt
Δσdp ¼ ð6Þ
ized. A standardized variable has a 0 mean and a standard deviation ðσ1 − σ3 Þt0
of 1 (Rencher 2002) and retains the shape properties of the original
variable. Thus, standardization can be used to put data sets on the where ðσ1 − σ3 Þt = estimate of the deviatoric stress at time t (the
same scale for further analysis or comparison. The high correlation current monitoring time); and ðσ1 − σ3 Þt0 = estimate of the deviatoric
in the overall settlement behavior can be observed from Fig. 8(b). stress at time t0 (start of the settlement time series). The estimated

Fig. 8. (Color) Settlement at HS, CS, and TS stations (total of 48 time series) reset to zero at a common starting point: (a) the settlement time series;
and (b) standardized value of the time series in (a).

© ASCE 04018059-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Fig. 10. Settlement at station HSA-5, starting at the time when the reservoir rose above the location of the settlement gauge at elevation 585 m above
sea level: (a) settlement versus time; and (b) settlement versus proportional value of a deviatoric stress estimation.
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. (Color) Reservoir elevation and loading function used for
changing between unloading–reloading conditions (station HSA-5).

deviatoric stress was calculated considering stresses within the dam at


the instrument location. These comprise earth pressure as well as
pressure from hydrostatic loading. The resulting stress-deformation
relationship in Fig. 10(b) for HSA-5 resembles the general stress-
strain relationship presented, e.g., by Byrne et al. (1987) for initial
and repeated loading.
Fig. 12. (Color) Results from multiple linear regression model for a
The importance of incorporating the delayed response to the res-
response variable represented by settlement data at station HSA-5 since
ervoir loading in the model is evident from Fig. 9, as well as from
the end of the first filling: (a) model values (MLR) compared with the
similar results from other locations within the dam. It is also of interest
response variable (HSA-5); (b) hydrostatic component of the model;
to alternate between the unloading and reloading behavior. The pro-
and (c) time-dependent component of the model.
posed model in Eq. (4) therefore accounts for these effects. The stress
state was calculated for each time step. Different regression coeffi-
cients were calculated for the unloading versus the reloading condi-
tions. This was determined from a loading function, as illustrated in
Fig. 11 for station HSA-5, with 0 representing the reloading condi-
tions, and 1 the unloading conditions. The stress level was also calcu-
lated and compared with previous stress levels. As mentioned
previously, reloading above the previously experienced stress state
has not been of significant influence for the dam during its operation.
Multiple regression was performed on the time series from
HSA-5, HSA-6, and HSA-7 using the processed settlement time
series from the end of the first filling as the response variable. Sub-
sequently, a multivariate multiple regression with the time series
recorded at CS-7, CS-8, and CS-9 as response variables was con-
ducted using both a training data subset as well as a prediction Fig. 13. (Color) Residuals resulting from the multiple regression mod-
check data subset. el, predicting the interpolated response variable HSA-5: (a) considering
unloading–reloading; and (b) not considering unloading–reloading.
Analysis Results
The results from the multiple regression analysis on the time series cycles in the reservoir elevation, and Fig. 12(c) extracts the time-
HSA-5 are presented in Fig. 12. Fig. 12(a) compares the model dependent settlement in the period considered from the model.
results (MLR) with the processed time series (HSA-5), Fig. 12(b) Residuals from comparison of the model and the processed
highlights the details of the settlement behavior due to annual response time series are plotted in Fig. 13(a). The shape of the

© ASCE 04018059-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. (Color) Comparison of instrument readings versus results from multiple regression model of settlement data since the end of the first filling
for (a) HSA-5; (b) HSA-6; and (c) HSA-7.

for stations close to the middle of the dam or at the downstream side
than for those at the upstream face.
Fig. 15 presents model results from multivariate multiple linear
regression, simultaneously considering three response variables at
the crest, i.e., at stations CS-7, CS-8, and CS-9. These response
variables were obtained from geodetic surveys on benchmarks,
and there were not as many measurements available as there were
for the HS stations. The response variable data set was divided
into a training sample subset and a prediction check sample sub-
set. In Fig. 15, the model results considering the training sample
subset are presented with an unbroken line, while the model re-
sults considering the prediction check sample subset are presented
with a dotted line. The actual survey readings are also presented in
the figure with a circle denoting a survey data point. It can be seen
Fig. 15. (Color) Results from multivariate multiple linear regression of that the prediction check sample subset agrees reasonably well
settlement data at stations CS-7, CS-8, and CS-9 since the end of the with the actual response. Furthermore, the prediction model com-
first filling. Prediction from the model (starting May 2012 and ending plements the readings and brings out details that are lost in the
September 2017) is compared with the actual readings for model va- actual data since there are only two surveys conducted annually
lidation. The settlements are reset to zero at the end of the first filling. in this period.
The residuals for station CS-9 are plotted in Fig. 16. Two
residual plots are provided: (1) one for the regression results shown
residual plot resembles a normal probability distribution, and the in Fig. 15 from a model that considers stress state from unloading–
expected value of the residuals is zero [EðeÞ ¼ 0. Fig. 13(b) plots reloading conditions, and (2) another for a regression model that
the residual for station HSA-5 from a regression model that does does not consider the stress state. Comparison of the distribution
not take the stress state into account. Comparison of the distribution of the residual in Figs. 16(a and b) indicates a better fit when
of the residuals in Figs. 13(a and b) indicates that a better fit is the stress state is considered, which corresponds to the results pre-
achieved when the stress state is considered. viously seen for data from HSA-5. The correlation of the training
In Fig. 14(a) the model is compared with the actual readings data subset with the model results was very high in both cases (both
from station HSA-5. The figure demonstrates how intervals be- ∼0.997), but slightly higher for the model considering the stress
tween readings have gradually increased since the first state, indicting a better fit. Similar results were obtained for other
impounding, as previously mentioned. It is thus to be expected that locations. This is in agreement with observations from triaxial tests
the accuracy of the processed time series used for the regression (Stewart 1986; Byrne et al. 1987), which indicated a slightly differ-
had simultaneously gradually decreased. Figs. 14(b and c) show ent modulus for unloading and reloading.
similar analysis results for stations HSA-6 and HSA-7, respec- Fig. 17 presents a prediction of time-dependent settlement from
tively. However, the estimate of the deviatoric stress is less accurate the multivariate multiple regressions of variables for stations CS-7,

© ASCE 04018059-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


The proposed model requires daily values of the reservoir eleva-
tion and actual observations for the period used in the regression.
The application of the model for prediction with detailed pattern
identification due to changes in reservoir filling is mainly depen-
dent on the hydrostatic component, and thus also on accurate pre-
diction of the hydrostatic loading from the reservoir, e.g., based on
reservoir inflow and outflow for reservoir elevation forecasting.
Conversely, monthly values of the settlement can be considered
to provide a minimum amount of data points for processing to a
data set suitable for obtaining the regression parameters of the pre-
diction model. More closely spaced observations in time will en-
hance the regression and improve the reliability of the model,
Fig. 16. (Color) Residuals for CS-9 from a regression model consider- particularly for identification of details in the behavioral pattern
ing (a) loading-reloading; and (b) a regression model that did not con- of the settlement. Preferably, additional observations should be col-
sider this. lected during periods of sudden and large changes in the reservoir
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

elevation. The number of available data points for the case studied
herein and the varying time interval between observations for
the crest stations at the top of the test case dam is portrayed by the
readings shown in Fig. 15. The original settlement time series were
processed to produce daily values (Sigtryggsdóttir et al. 2013).
This study indicates that monitoring data from three to five
cycles of seasonal variations in the reservoir elevation after the ini-
tial impounding is required to achieve reasonable prediction of the
overall settlement behavior during normal operation. A part of the
available data should be used for a validation check of the model.
The model parameters can be updated, if required, based on past
performance as additional data are gathered from repeated reload-
ing cycles. As the time span of the training data set increases, the
prediction of the long-term settlement will become more reliable.

Measured versus Predicted Performance


The benefits of the predicted response from a statistical model as
presented here relate to both the long- and short-term prediction.
If the predicted long-term response indicates unreasonably large
Fig. 17. (Color) Prediction of the time-dependent settlement for settlements compared with design estimates, there are mainly two
stations CS-7, CS-8, and CS-9 since the end of the first filling, along focus points for further evaluation. First, the influence on the re-
with a plot of the survey readings available up to September 2017. quired freeboard of the dam, considering that rockfill dams, includ-
Periods I, II, and III are defined on the figure. The model is regressed ing CFRDs, are vulnerable to overtopping. Second, the increase
to the settlement data set within Period I since the end of first filling, in settlement-induced stresses in the concrete face slab and the po-
validated with the data set in Period II (Fig. 15), and predicts the time- tential for excess cracking leading to unacceptable leakage. The
dependent deformation in Period III. stresses induced in the slab due to the long-term settlements can
be estimated, e.g., in an FE model.
Under normal operation, the actual reservoir elevation is often
recorded automatically with daily values available for large
CS-8, and CS-9. The survey readings at these stations are plotted on CFRDs, while a survey of the settlement is often manual and only
the figure and start at the end of the first filling. conducted a few times a year. Thus, the actual reservoir elevation
can be used to predict the short-term settlements in between actual
observations as well as at the observation day. In this case, one
Discussion relevant application of the prediction model would be to compare
the surveying measurements to the predicted response. If the varia-
Application of the proposed model in Eq. (4) to the different set- tion between the two is larger than a prespecified deviation, the
tlement time series of the case study demonstrates that the model measurement should be repeated. This could reduce or even elimi-
adequately captures the response and gives a credible comparison nate outliers from the surveying data and enhance the accuracy of
to the response variables used. The prediction of the crest settle- the observations collected. This suggestion is based on the fact that
ments in the validation period (Fig. 17) is also convincing. The set- strange outliers were observed in the surveying data used in this
tlement prediction will improve as the model is updated as new data study. However, if a repeated measurement still does not follow
are collected. the prediction, more frequent observations should be conducted
along with a site inspection, followed by a detailed evaluation
of the change in settlement behavior and analysis of the effects
Required Performance History
on structural integrity.
The methodology used in developing the prediction model requires The patterns observed in the settlement behavior can be inter-
information on both the hydrostatic loading, i.e., the reservoir preted to signify the dam’s structural health and normal state of
elevation, and the response variable, i.e., the settlements. operation. It is difficult to envision causes for changes in this

© ASCE 04018059-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


behavior, unless possibly if the dam would be subjected to extreme Bjerrum, L. 1967. “Engineering geology of Norwegian normally-
loading. Still, any changes in these details, as well as sudden de- consolidated marine clays as related to settlements of buildings.” Géo-
viations from the long-term settlement behavior, should call for fur- technique 17 (2): 83–118. https://doi.org/10.1680/geot.1967.17.2.83.
ther investigation and analysis to assess the dam’s structural health Byrne, P. M., H. Cheung, and L. Yan. 1987. “Soil parameters for deforma-
tion analysis of sand masses.” Can. Geotech. J. 24 (3): 366–376. https://
and safety.
doi.org/10.1139/t87-047.
Chouinard, L. E., D. W. Bennett, and N. Feknous. 1995. “Statistical analy-
sis of monitoring data for concrete arch dams.” J. Perform. Constr.
Concluding Remarks Facil. 9 (4): 286–301. https://doi.org/10.1061/(ASCE)0887-3828
(1995)9:4(286).
In this paper the development of statistical models of a rockfill Chouinard, L. E., and V. Roy. 2006. “Performance of statistical models
dam deformation behavior has been described and the importance for dam monitoring data.” In Proc., Joint Int. Conf. on Computing
of a cautious selection of predictors for the credibility of a statis- and Decision Making in Civil and Building Engineering. Montréal,
tical model demonstrated. A statistical model was proposed and Canada.
tested for the case of Kárahnjúkar CFRD. Such models are required De Sortis, A., and P. Paoliani. 2007. “Statistical analysis and structural
identification in concrete dam monitoring.” Eng. Struct. 29 (1):
for SHM management, as well as being consistent with Inter-
110–120. https://doi.org/10.1016/j.engstruct.2006.04.022.
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

national Commission on Large Dams–recommended (ICOLD Dolezalova, M., and I. Hladik. 2011. “Constitutive models for simulation of
2012) dam engineer’s practice on the usage of numerical models. field performance of dams.” Int. J. Geomech. 11 (6): 477–489. https://
Thus, the paper presents important information, considering the doi.org/10.1061/(ASCE)GM.1943-5622.0000039.
limited and incomplete reports hitherto available for rockfill dam Duncan, J. M., P. Byrne, K. S. Wong, and P. Mabry. 1980. Strength, stress-
settlements. strain and bulk modulus parameters for finite element analysis of
The proposed statistical model [Eq. (4)] has elements of a stresses and movements in soil masses. No. UCB/GT/80-1. Berkley,
general HST model, but without the seasonal component [δ S ðsÞ], CA: Dept. of Civil Engineering, Univ. of California.
which is irrelevant. At the same time, components of the statistical Farrar, C. R., and N. A. J. Lieven. 2007. “Damage prognosis: The future of
model are innovatively related to a viscoelastic–plastic constitutive structural health monitoring.” Philos. Trans. R. Soc. Math. Phys. Eng.
Sci. 365 (1851): 623–632. https://doi.org/10.1098/rsta.2006.1927.
material model, with the novel addition of considering the unload-
Ferry, S., and G. Willm. 1958. “Méthodes d’analyse et de surveillance de
ing–reloading behavior of the dam settlement. For rockfill dams, déplacements observés par le moyen de pendules dans les barrages.” In
this consideration is clearly an option and aids in defining predic- Proc., 6th Int. Congress on Large Dams. Paris: International Commis-
tors that relate to actual physical factors influencing the response sion on Large Dams.
variable. The algorithm developed for the regression has aspects Fitzpatrick, M. D., B. A. Cole, F. L. Kinstler, and B. P. Knoop. 1985.
related to FE analysis in terms of stress state and varying loading “Design of concrete-faced rockfill dams.” In Proc., Symp. on Concrete
functions. The incorporation of a more accurate stress state infor- Face Rockfill Dams: Design, Construction, and Performance,
mation, e.g., from an FE model, would enhance the resulting stat- 410–434. New York: ASCE.
istical model, and vice versa, the data regression would provide Gan, L., Z. Shen, and L. Xu. 2014. “Long-term deformation analysis of the
valuable reference data for FE modeling. Such joint consideration Jiudianxia concrete-faced rockfill dam.” Arab. J. Sci. Eng. 39 (3):
1589–1598. https://doi.org/10.1007/s13369-013-0788-6.
of a statistical model and mechanics model is inherent in a com-
Haifanga, L., and Z. Yinqi. 2012. “Creep rate and creep model of rockfill.”
prehensive SHM process. Procedia Eng. 28: 796–802. https://doi.org/10.1016/j.proeng.2012.01
The statistical analysis of the settlement monitoring data .812.
has brought forth behavioral patterns, including details in the dam Hu, D., Z. Zhou, Y. Li, and X. Wu. 2011. “Dam safety analysis based on
settlement induced by a viscoelastic response to variations in the stepwise regression model.” Adv. Mater. Res. 204–210: 2158–2161.
reservoir hydrostatic loading. The behavioral pattern can be inter- https://doi.org/10.4028/www.scientific.net/AMR.204-210.2158.
preted to signify the dam’s structural health and normal state of Hunter, G., and R. Fell. 2003. “Rockfill modulus and settlement of concrete
operation. The statistical model is well suited for an online predic- face rockfill dams.” J. Geotech. Geoenviron. Eng. 129 (10): 909–917.
tion process, with regular updates based on past performance. https://doi.org/10.1061/(ASCE)1090-0241(2003)129:10(909).
Comprehensive SHM management incorporating statistical models ICOLD (International Commission on Large Dams). 2012. Guidelines for
use of numerical models in dam engineering. Bulletin No. 155. Paris:
as discussed herein is important for the operational safety of large
ICOLD.
dams, and thus the resilience and sustainability of the related civil Janbu, N. 1963. “Soil compressibility as determined by oedometer and
infrastructure systems. triaxial tests.” In Vol. 4 of Proc., 3rd European Conf. on Soil Mechanics
and Foundation Engineering, 19–25. Wiesbaden, Germany: German
Society for Soil Mechanics and Foundation Engineering.
Acknowledgments Janbu, N. 1969. “The resistance concept applied to deformations of soils.”
In Vol. 1 of Proc., 7th Int. Conf. on Soil Mechanics and Foundation
The support and input by the late Professor Ragnar Sigbjörnsson Engineering, 191–196. Mexico City: Sociedad Mexicana de Macanica.
is gratefully acknowledged. The authors thank Landsvirkjun for Janbu, N. 1985. “Soil models in offshore engineering.” Géotechnique
access to the monitoring data. 35 (3): 241–281. https://doi.org/10.1680/geot.1985.35.3.241.
Justo, J. L., and P. Durand. 2000. “Settlement-time behaviour of granular
embankments.” Int. J. Numer. Anal. Methods Geomech. 24 (3): 281–
303. https://doi.org/10.1002/(SICI)1096-9853(200003)24:3<281::AID
References -NAG66>3.0.CO;2-S.
Lade, P. V., C. D. Liggio, and J. Nam. 2009. “Strain rate, creep, and stress
Alonso, E. E., and L. A. Oldecop. 2007. “Theoretical investigation of the drop-creep experiments on crushed coral sand.” J. Geotech. Geoen-
time-dependent behaviour of rockfill.” Géotechnique 57 (3): 289–301. viron. Eng. 135 (7): 941–953. https://doi.org/10.1061/(ASCE)GT
https://doi.org/10.1680/geot.2007.57.3.289. .1943-5606.0000067.
Augustesen, A., M. Liingaard, and P. V. Lade. 2004. “Evaluation of time Léger, P., and M. Leclerc. 2007. “Hydrostatic, temperature, time-
dependent behavior of soils.” Int. J. Geomech. 4 (3): 137–156. https:// displacement model for concrete dams.” J. Eng. Mech. 133 (3):
doi.org/10.1061/(ASCE)1532-3641(2004)4:3(137). 267–277. https://doi.org/10.1061/(ASCE)0733-9399(2007)133:3(267).

© ASCE 04018059-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059


Mata, J. 2011. “Interpretation of concrete dam behaviour with artificial Stewart, H. E. 1986. “Permanent strains from cyclic variable-amplitude
neural network and multiple linear regression models.” Eng. Struct. loadings.” J. Geotech. Eng. 112 (6): 646–660. https://doi.org/10
33 (3): 903–910. https://doi.org/10.1016/j.engstruct.2010.12.011. .1061/(ASCE)0733-9410(1986)112:6(646).
Ohde, J. 1939. “Zur theorie der druckverteilung im Baugrund.” Der Bauin- Willm, G., and N. Beaujoint. 1967. “Les méthodes de surveillance des bar-
genieur. 20 (33–34): 451–459. rages au service de la production hydraulique d’Électricité de France.
Oyen, M. L., and R. F., Cook. 2003. “Load-displacement behaviour during Problèmes anciens et solutions nouvelles.” In Vol. 3 of Proc., 9th ICOLD
sharp indentation of viscous-elastic plastic-plastic materials.” J. Mater. Congress, 529–50. Paris: International Commission on Large Dams.
Res. 18 (01): 139–150. https://doi.org/10.1557/JMR.2003.0020. Worden, K., C. R. Farrar, G. Manson, and G. Park. 2007. “The funda-
Rencher, A. C. 2002. Methods of multivariate analysis. 2nd ed. Hoboken, mental axioms of structural health monitoring.” Proc. R. Soc. Math. Phys.
Eng. Sci. 463 (2082): 1639–1664. https://doi.org/10.1098/rspa.2007.1834.
NJ: Wiley.
Wu, H., R. Li, and X. Wu. 2009. “Rockfill dam deformation monitoring
Saboya, F., and P. M. Byrne. 1993. “Parameters for stress and deformation
model considering rheology effect and stress path.” In Proc., 1st Int.
analysis of rockfill dams.” Can. Geotech. J. 30 (4): 690–701. https://doi
Symp. on Rockfill Dams. Beijing: Chinese National Committee on
.org/10.1139/t93-058. Large Dams and the Brazilian Committee on Large Dams.
Sigtryggsdóttir, F. G., J. T. Snæbjörnsson, R. Sigbjörnsson, and L. Grande. Zhou, F., Y. Wu, and M. GuliMire. 2009. “Analysis of settlement deforma-
2013. “Rockfill dam settlement data: Processing and statistical analy- tion of concrete face rockfill Wuluwati Dam.” In Proc., 1st Int. Symp. on
sis.” In Proc., 3rd Int. Symp. on Rockfill Dams, Hydropower, CHIN- Rockfill Dams. Beijing: Chinese National Committee on Large Dams
Downloaded from ascelibrary.org by 5.2.189.221 on 07/02/21. Copyright ASCE. For personal use only; all rights reserved.

COLD 2013. China: Chinese National Committee on Large Dams and the Brazilian Committee on Large Dams.
and China Society for Hydropower Engineering. Zhou, W., G. Ma, and C. Hu. 2011. “Long-term deformation control theory
Soroush, A., and R. Jannatiaghdam. 2012. “Behavior of rockfill of high concrete face rockfill dam and application.” In Proc., 2011
materials in triaxial compression testing.” Int. J. Civ. Eng. 10 (2): Asia-Pacific Power and Energy Engineering Conf. (APPEEC), 1–4.
153–183. New York: IEEE.

© ASCE 04018059-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2018, 144(9): 04018059

You might also like