The Deformation Behaviour of Rockfill CFRD

ISSN 0077-880X
THE UNIVERSITY OF
NEW SOUTH WALES
STUDIES FROM
SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING
THE DEFORMATION BEHAVIOUR OF ROCKFILL
G. HUNTER and R. FELL
UNICIV REPORT No. R-405 JANUARY 2002

THE UNIVERSITY OF NEW SOUTH WALES
SYDNEY 2052 AUSTRALIA
ISBN : 85841 372 8

The University of New South Wales
SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING
BIBLIOGRAPHIC DATA SHEET ISSN 0077-880X
1. UNICIV Report No. R-405 2. ISBN : 85841 372 8 3. Date : 23rd January 2002
4. Title THE DEFORMATION BEHAVIOUR OF ROCKFILL
GAVAN HUNTER
5. Author(s)
ROBIN FELL
6. Keywords Concrete face rockfill dam, rockfill, deformation
7. Abstract This report presents the findings from analysis of the deformation behaviour
of rockfill in concrete faced rockfill dams.
The deformation behaviour of rockfill during construction, on first filling and

post first filling has been analysed from a data set of thirty-six mostly
concrete faced rockfill dams. Factors influencing the stress-strain properties
of the rockfill and applied stress conditions have been evaluated. Guidelines
are presented for prediction of rockfill modulus and the deformation
behaviour during construction, on first filling and post construction that
incorporate the significant factors influencing the rockfill deformation
behaviour.
Text = 58 pages
8. Number of pages Appendix A = 10 pages
-i-
UNICIV Report No. 405 – Deformation Behaviour of Rockfill January 2002
TABLE OF CONTENTS
1.0 INTRODUCTION.........................................................................................................................1
1.1 USE OF ROCKFILL, AND OBJECTIVES OF THIS STUDY ..............................................................................1
1.2 DEFINITIONS ..............................................................................................................................................1
1.2.1 PROPERTIES OF ROCKFILL ..................................................................................................................1
1.2.2 ROCKFILL MODULI .............................................................................................................................2
1.2.3 ZONING OF MAIN ROCKFILL IN CONCRETE FACE ROCKFILL DAMS......................................................5
2.0 LITERATURE REVIEW OF ROCKFILL DEFORMATION................................................6

2.1 HISTORICAL SUMMARY OF ROCKFILL USAGE IN EMBANKMENT DESIGN ................................................6
2.1.1 DEFORMATION PROPERTIES OF ROCKFILL ..........................................................................................6
2.2 PREDICTIVE METHODS FOR ROCKFILL DEFORMATION ............................................................................9
2.2.1 FINITE ELEMENT ANALYSES...............................................................................................................10
2.2.2 EMPIRICAL PREDICTIVE METHODS OF DEFORMATION BEHAVIOUR DURING CONSTRUCTION AND ON
FIRST FILLING .................................................................................................................................................12
2.2.3 PREDICTIVE METHODS FOR DEFORMATION BEHAVIOUR POST CONSTRUCTION.................................15
3.0 ANALYSIS OF DEFORMATION BEHAVIOUR OF CFRD................................................18

3.1 CASE STUDY DATABASE .........................................................................................................................18
3.2 DEFORMATION DURING CONSTRUCTION ................................................................................................21
3.2.1 BASE PLOTS OF DATA ........................................................................................................................21
3.2.2 EFFECT OF COMPACTION/PLACEMENT METHOD ..............................................................................21
3.2.3 EFFECT OF CONFINING STRESS .........................................................................................................23
3.2.4 EFFECT OF VALLEY SHAPE ................................................................................................................24
3.2.5 EFFECT OF PARTICLE SIZE DISTRIBUTION .........................................................................................27
3.2.6 MODULI DURING CONSTRUCTION OF DUMPED ROCKFILL ................................................................29
3.3 DEFORMATION OF THE FACE SLAB OF CFRD ON FIRST FILLING ...........................................................30
3.3.1 SIMPLIFIED METHODS OF ESTIMATION OF MAXIMUM FACE SLAB DEFORMATION .............................30
3.3.2 SHAPE OF THE FACE SLAB DISPLACEMENT ........................................................................................34
3.3.3 PREDICTION OF FACE SLAB DEFLECTION ON FIRST FILLING .............................................................36
3.4 POST CONSTRUCTION CREST SETTLEMENT ............................................................................................37
3.4.1 CASE STUDY RECORDS OF POST-CONSTRUCTION CREST SETTLEMENT ..............................................37
3.4.2 POST-CONSTRUCTION INTERNAL SETTLEMENT ..................................................................................46
4.0 DISCUSSION AND RECOMMENDED METHODS FOR PREDICTION .........................48

4.1 GUIDELINES ON DEFORMATION PREDICTION DURING CONSTRUCTION .................................................48
4.2 GUIDELINES ON DEFORMATION PREDICTION DURING FIRST FILLING ...................................................50
4.3 GUIDELINES ON DEFORMATION PREDICTION POST-CONSTRUCTION .....................................................51
5.0 CONCLUSIONS .........................................................................................................................52
6.0 ACKNOWLEDGEMENTS........................................................................................................53
7.0 REFERENCES............................................................................................................................53
Appendix A: CFRD Case Study Information

- ii -
LIST OF TABLES
Table 1.1: Classification of unconfined compressive strength or rock (AS 1726-1993).........................................1

Table 2.1: Historical summary of rockfill usage in embankment design (Galloway 1939; Cooke 1984; Cooke
1993). ...............................................................................................................................................................7
Table 2.2: Parameters for deflection prediction (Soydemir and Kjaernsli 1979) ..................................................16
Table 2.3: Rates of post-construction crest settlement of dumped and compacted rockfills in CFRDs (Sherard
and Cooke 1987) ............................................................................................................................................17
Table 3.1: Summary of embankment and rockfill properties for CFRD case studies ...........................................19
Table 3.2: Assessment of cross-valley influence on arching for case studies analysed ........................................26
Table 3.3: Stress conditions representative of the data sets from Figure 3.7 ........................................................29
Table 3.4: Summary of crest settlement during the period of first filling .............................................................42
Table 3.5: Values of the coefficient, m, in the strain rate – time power function (Equation 3).............................43
Table 3.6: Estimates of long-term crest settlement rates for dumped rockfills .....................................................45
Table 3.7: Location of internal post-construction vertical settlement in the CFRD..............................................47
Table 4.1: Approximate stress reduction factors to account for valley shape .......................................................49
LIST OF FIGURES
Figure 1.1: Determination of rockfill moduli E rc and E rf (Fitzpatrick et al 1985) ..............................................3

Figure 1.2: Vertical stress contours for embankments during construction. Slope = 30 degree slope, Poisson’s
ratio (ν) = 0.3. (Poulos and Davis 1974).........................................................................................................4
Figure 1.3: Typical design of dumped rockfill CFRD (adapted from ICOLD (1989)) ...........................................5
Figure 1.4: Typical zoning of main rockfill in current CFRD design practice for construction with sound
quarried rockfill (adapted from Cooke 1997; zoning designators after Fell et al (1992))................................5
Figure 2.1: Compression curves for dry and wet states, and collapse compression from dry to wet state for
Pyramid gravel in laboratory oedometer test (Nobari and Duncan 1972)........................................................8
Figure 2.2: Settlement versus time curves for laboratory oedometer tests on rockfill (Sowers et al 1965) ............9
Figure 2.3: Typical stress-strain relationship of rockfill from a triaxial compression test (Mori and Pinto 1988)
........................................................................................................................................................................11
Figure 2.4: Finite element analysis of Foz do Areia CFRD (Saboya and Byrne 1993), (a) model and (b) stress
paths during construction and first filling. .....................................................................................................12
Figure 2.5: Deformation modulus during construction ( E rc ) versus void ratio (Pinto and Marques Filho 1998)13
Figure 2.6: Ratio of deformation modulus on first filling to during construction ( E rf E rd ) versus valley shape
factor (Pinto and Marques Filho 1998) ..........................................................................................................13
Figure 2.7: Deformation modulus during construction ( E rc ) for HEC CFRDs (Giudici et al 2000) ...................14
Figure 2.8: Results of 3-dimensional finite element analysis of CFRD, to give vertical displacement during
construction versus valley shape (Giudici et al 2000)....................................................................................15
Figure 2.9: Settlement rate analysis of Cedar Creek dam; (a) determination of time to and (b) settlement rate
versus time (Parkin 1977) ..............................................................................................................................16
Figure 2.10: Post-construction crest settlement of membrane faced compacted rockfill dams (Clements 1984). 17
Figure 2.11: Settlement index versus time for well-compacted rockfills (adapted from Public Works Department
NSW 1990).....................................................................................................................................................18
Figure 3.1: Stress-strain relationship for rockfills observed during construction..................................................22
- iii -
Figure 3.2: Secant modulus versus vertical stress from monitoring during construction......................................22
Figure 3.3: Tangent modulus versus vertical stress from monitoring during construction ...................................23
Figure 3.4: Idealised model for two-dimensional finite difference analysis of cross-valley influence .................24
Figure 3.5: Results of two-dimensional finite difference analysis of the effect of cross-valley shape on vertical
stresses in the dam. (a), (c) and (e) represent construction in 5 m lifts (to 100 m) and (b), (d) and (f)
construction in a single 100 m lift. Vertical stresses, at the end of construction, are normalised against
those obtained for the zero abutment slope analysis. .....................................................................................25
Figure 3.6: Indicators of cross valley arching effects from variations in the secant modulus with vertical stress.
........................................................................................................................................................................27
Figure 3.7: Representative secant modulus (mostly Zone 3A rockfill) at end of construction ( E rc ) versus D80
from average grading of the rockfill...............................................................................................................28
Figure 3.8: E rf E rc ratio versus embankment height ..........................................................................................30
Figure 3.9: Stress paths during construction and first filling for nominal 100 m embankment with 1.3H to 1V
upstream slope angle; (a) monitoring point locations, (b) stress paths for points normal to face slab at 30%
of the embankment height. .............................................................................................................................33
upstream slope angle; (a) monitoring point locations, (b) stress paths for points normal to face slab at 30%
of the embankment height. .............................................................................................................................33
Figure 3.11: Face slab deflection during first filling (4/2/71 to 25/4/71) of Cethana dam (Fitzpatrick et al 1973).
........................................................................................................................................................................34
Figure 3.12: Face slab deformation during first filling of Aguamilpa dam (Mori 1999) ......................................35
Figure 3.13: Face slab deformation during first filling of Ita dam (Sobrinho et al 2000) .....................................35
Figure 3.14: Scotts Peak dam (courtesy of HEC Tasmania) .................................................................................35
Figure 3.15: Mackintosh dam (a) embankment section (courtesy of HEC Tasmania) and (b) face slab
deformation on first filling (Knoop and Lack 1985) ......................................................................................36
Figure 3.16: Examples of derivation of zero time for post-construction settlement .............................................38
Figure 3.17: Post-construction crest settlement versus time for dumped rockfill CFRD, to at end of main rockfill
construction. ...................................................................................................................................................39
Figure 3.18: (a) and (b) Post-construction crest settlement versus time for CFRD constructed of compacted
rockfills of medium to high intact strength, to at end of main rockfill construction.......................................39
Figure 3.19: (a) and (b) Post-construction crest settlement versus time for CFRDs constructed of well-
compacted rockfills of very high intact strength and of well-compacted gravels, to at end of main rockfill
construction. ...................................................................................................................................................40
Figure 3.20: Post-construction crest settlement versus time for dumped rockfill CFRD, to at end of first filling.40
Figure 3.21: Post-construction crest settlement versus time for CFRD constructed of compacted rockfills of
medium to high intact strength, to at end of first filling. ................................................................................41
Figure 3.22: Post-construction crest settlement versus time for CFRDs constructed of well-compacted rockfills
of very high intact strength and of well-compacted gravels, to at end of first filling. ....................................41
Figure 3.23: Post construction crest settlement of Bastyan dam ...........................................................................44
Figure 3.24: Long-term crest settlement rate (as a percentage of embankment height per log cycle of time)
versus embankment height for compacted rockfills.......................................................................................45
Figure 3.25: Crest settlement attributable to first filling (excluding time dependent effects)...............................46
-1-
1.0 INTRODUCTION
1.1 USE OF ROCKFILL, AND OBJECTIVES OF THIS STUDY

Rockfill, including other coarse granular materials such as gravels and granular colluvium, have since the late
19th century and continue to be an integral part of embankment dam construction. Recent and current dam
projects incorporating rockfill as a key component in embankment design include concrete faced rockfill dams
(CFRD) now reaching heights in excess of 200 m and earth and rockfill dams (mostly central core earth and
rockfill dams) reaching heights of 250 to 300 m. The key attributes of rockfill relate to high shear strength,
allowing for stable embankment slopes up to 1.3 to 1.4H to 1V (horizontal to vertical), and relatively low
compressibility when placed in a well-compacted condition.
The stability of rockfill embankments is evidenced by the lack of historical records of slope instability
associated with embankments incorporating rockfill zones. Foster et al (1998) present statistical data on dam
failure and accident incidents (ICOLD definitions) for large embankment dams from around the world
constructed up to 1986. Of 124 incidents involving slope instability, embankments incorporating rockfill
zoning were involved in only 7 cases. Hunter and Fell (2001), from analysis of 54 cases of slope instability in
embankment dams (excluding hydraulic fills and sloughing type slope instability), reported five cases that
incorporated rockfill in the embankment design. Only in one case, a zoned earth and rockfill dam with very thin
outer rockfill shoulders, did slope instability incorporate bulging of the outer rockfill slope. In the remaining
four cases the surface of rupture preferentially passed through the foundation rather than the rockfill.
This report presents the results of a study on the deformation properties of rockfill. It is based on a review of
published data and analysis of the deformation behaviour of concrete faced rockfill dams (CFRD). A method
for predicting the modulus from the rock type, unconfined compressive strength, particle size distribution and
placement method is presented. Also presented are methods for estimating the modulus during reservoir filling
and long-term crest settlement of concrete faced rockfill dams.
1.2 DEFINITIONS
1.2.1 Properties of Rockfill

Qualitative descriptors have been used to describe some of the properties of rockfill and the placement methods.
The classification system from Australian Standard AS 1726-1993 is used to define the unconfined compressive
strength (UCS) of the rockfill. The descriptors and the UCS range they represent are given in Table 1.1.
Table 1.1: Classification of unconfined compressive strength or rock (AS 1726-1993)
Strength Descriptor UCS Range (MPa)
Extremely High > 240
Very High 70 to 240
High 20 to 70
Medium 6 to 20
For the purposes of the deformation analysis the strength of rockfill has been simplified into two classes, very
high strength and medium to high strength. In very few cases has rock of extremely high strength been used in
embankment construction. The basis of the division is associated with the likelihood of breakdown of the
rockfill on compaction. It is recognised that there is likely to be “grey zone” rather than a division based only
on UCS strength, as the type and weight of roller, compaction procedure and strength of cementing of the rock
will affect the amount of breakdown on rolling. Particle size distribution analyses (from the literature) carried
-2-
out after compaction generally indicate that rock borrow sources of medium to high UCS show significant
breakdown on compaction as reflected in the high percent passing 19 mm and the generally high Cu (coefficient
of uniformity, D60/D10) values.
The method of placement of the rockfill has a significant influence on the modulus and deformation behaviour
of rockfill. The definitions by Cooke (1993, 1984) for dumped and compacted rockfill have be used as a basis
for categorisation of the method of placement:
• Dumped Rockfill – rock placed in lifts ranging from several to tens of metres thickness, with or without
sluicing, and without formal compaction.
• Compacted Rockfill – rock placed in layers up to 2 m thickness (generally 0.9 to 2.0 m thick) and
compacted by smooth drum vibrating roller. Accepted practice is typically 4 to 6 passes of a minimum 10
tonne (possibly up to 15 tonne) deadweight vibrating roller, with variation in layer thickness, added water
and number of passes depending on the quality and type of the rockfill, amount of fines and location within
the embankment. Three classifications for compacted rockfill have been used:
− Well-compacted – layer thickness typically less than about 1.0 m (depending on intact strength) and
compacted with a minimum of four passes of a 10 to 15 tonne deadweight smooth drum vibrating roller
(SDVR). Generally placed with the addition of water, except for some very high strength rockfills.
− Reasonable Compaction – layer thickness typically 1.5 to 2 m (depending on intact strength), generally
placed without the addition of water and compacted with typically four passes of a 10 tonne SDVR.
− Reasonably to Well Compacted - layer thickness typically 1.2 to 1.6 m (depending on intact strength),
placed with the addition of water and compacted with typically 4 to 6 passes of a 10 to 15 tonne SDVR.
For dumped rockfill, sluicing had a significant influence on the deformation behaviour of the rockfill as
evidenced by the large collapse deformations of dry dumped or poorly sluiced rockfills when wetted (Cogswell
dam (Bauman 1958), Strawberry and Dix River dams (Howson 1939)). Terminology used to define the level of
sluicing of dumped rockfill has been broadly defined into three classes; dry dumped, poorly sluiced and well
sluiced. A well-sluiced rockfill is described (Steele and Cooke 1958) as rockfill sluiced at a water ratio of 2 - 3
to 1 (water to embankment fill volume) using high-pressure jets that are directed on the tipped material.
Watering may be important for rolled rockfills if the rock is susceptible to weakening on wetting, or if it breaks
down under the action of the roller (for medium to high strength rockfills), or if the rockfill contains large
quantities of fines. It has generally been found (Cooke 1993) that watering is not overly important for
compaction of very high strength rockfills that are not susceptible to weakening on wetting.
1.2.2 Rockfill Moduli

Fitzpatrick et al (1985) defined two moduli for assessment of the deformation behaviour of rockfill (Figure 1.1),
the rockfill modulus during construction, E rc , and the rockfill modulus on first filling, E rf . Both are used
extensively throughout the report.
E rc is determined from the settlement of hydrostatic settlement gauges (HSG) installed during construction,
generally from under the embankment centreline. The modulus is a secant modulus and, for HSGs under the
embankment centreline, very closely approximates the confined modulus. Where possible the calculation of
E rc has taken into account the distribution in applied vertical stress with depth due to embankment shape
(Figure 1.2) from the elastic solutions by Poulos and Davis (1974). This is different to the method of Fitzpatrick
et al (1985), shown in Figure 1.1a, who did not allow for the stress distribution effect.
-3-
E rf is calculated from the deformation of the face slab normal to the upstream face on first filling as shown in
Figure 1.1b. Fitzpatrick et al (1985) comment that the modulus calculated is only an indicative value of the
modulus of the rockfill on first filling, but that the face slab deformations calculated using this method
reasonably (and simply) approximate those from finite element analysis. The method is used very broadly
throughout the concrete face rockfill dam (CFRD) community. In reality, E rf is only an index of performance
useful for the estimation of face slab deflection. This is discussed further in Section 3.3.
Figure 1.1: Determination of rockfill moduli E rc and E rf (Fitzpatrick et al 1985)
Where possible, Figure 1.2 has been used in this report to estimate vertical stresses or changes in the vertical
stress during embankment construction, and therefore take into account the effect of embankment shape in the
calculation of the rockfill modulus, E rc . Where detailed data was available on internal settlements (from
hydro-static settlement gauges or HSGs) as construction proceeds it has been possible to calculate the rockfill
modulus (both secant and tangent modulus) with increasing vertical stress, allowing for the effects of
embankment shape. In summary, the method used to calculate the vertical stress and the rockfill moduli
between say two HSGs or between a HSG and the foundation, was:
• Take the mid point between the HSGs as representative of the vertical stress of the layer of rockfill (the
HSGs were usually a vertical distance of 10 to 20 m apart).
• On placement of the upper HSG, estimate the initial vertical stress at the mid point of the layer from Figure
1.2. This is the stress at the zero settlement reading of the upper HSG.
• For subsequent fill heights above the upper HSG determine the vertical stress (at the layer mid-point)
appropriate to each fill height from Figure 1.2.
• The secant modulus (or E rc ) at each fill height can then be determined by dividing the change in vertical
stress (from the initial vertical stress) by the vertical strain in the rockfill zone of interest (i.e. the settlement
between the HSGs) appropriate to the change in vertical stress.
Figure 1.2 is for embankments with slopes of 30 degrees (1.75H to 1V, horizontal to vertical) constructed of
materials with a Poisson’s Ratio of 0.3 on a rigid foundation. From finite difference analysis it was found that
Figure 1.2 could be used for approximation of vertical stresses for embankments with slopes as steep as 1.3H to
1V without introducing significant errors. Where the HSGs were located off the centreline, but within the
central half of the embankment, the vertical stress was estimated by applying a percentage offset from the
centreline from the case study in Figure 1.2. A Poisson’s Ratio of 0.3 is considered a reasonable estimate for
rockfill.
In comparison to the vertical stresses under the embankment centreline estimated by the Fitzpatrick et al (1985)
method, equivalent stresses are obtained in Figure 1.2 up to 60 to 70 percent of the embankment height. Above
this height (i.e. construction of the upper 30 to 40%) the embankment shape influences the vertical stresses and
the Fitzpatrick et al 1985 method will over-estimate the actual stresses and under-estimate the rockfill moduli.
-4-
Figure 1.2: Vertical stress contours for embankments during construction. Slope = 30 degree slope, Poisson’s ratio (ν) = 0.3. (Poulos and Davis 1974).
-5-
1.2.3 Zoning of Main Rockfill in Concrete Face Rockfill Dams

Zoning descriptors used for the main rockfill zones varies widely throughout the community. A uniform system
of classification of the main rockfill zones has been applied to the CFRD case studies analysed as part of this
report. It broadly follows the classification systems of Cooke and Sherard (1987), ICOLD (1989) and Fell et al
(1992).
For CFRDs of dumped rockfill construction (typical of the CFRD design of the 1920’s to 1960’s) the typical
embankment design incorporates the main body of dumped rockfill (designated Zone 3 in this report) and an
upstream zone of derrick placed rockfill supporting the facing membrane (Figure 1.3).
Figure 1.3: Typical design of dumped rockfill CFRD (adapted from ICOLD (1989))
Following the resurgence of CFRD usage in dam design with the change to compaction of rockfill from about
the mid to late 1960’s, current design practice is dependent on a number of factors including the minimisation of
deformation of the face slab on first filling and control of any leaks which may occur through the face slab or
seepage through the foundation. For rockfills sourced from quarried sound rock the design of the main rockfill
(Figure 1.4) typically incorporates two zones; Zone 3A and Zone 3B. Zone 3A, the main support zone for the
upstream face, is typically placed in 1 m layers and well compacted to achieve a high modulus for minimisation
of face slab deformation. For Zone 3B, the downstream rockfill zone, a high modulus is not as critical as for
Zone 3A and the rockfill is typically coarser and placed (and compacted) in 1.5 to 2 m thick layers.
Figure 1.4: Typical zoning of main rockfill in current CFRD design practice for construction with sound
quarried rockfill (adapted from Cooke 1997; zoning designators after Fell et al (1992))
-6-
The central variable zone in Figure 1.4 represents variations in the sizing of Zones 3A and 3B in current design
practice around the world. HEC Tasmania typically design with a dominant Zone 3A forming 75 to 100% of
the main rockfill (i.e. Zone 3B is only present in the outer portion of the downstream shoulder). The typical
design of large CFRDs in Brazil (of sound quarried rockfill) consists of Zone 3A comprising the upstream third
of the main rockfill and Zone 3B the downstream two thirds.
Design of the main rockfill zones becomes more complex when the rockfill is to consist of potentially low
permeability materials, such as dirty gravels or medium to high strength quarried rockfills that breakdown
significantly on compaction. Drainage zones behind the face slab and above the foundation are typically
incorporated into the design for control of seepage. The zoning classification of the main rockfill used in this
report for designs other than typical is based on similar principles as for the typical design with Zone 3A
designated the main upstream zone and Zone 3B, if used, the main downstream zone. To be designated Zone
3A the rockfill zone must comprise at least 25% of the main rockfill.
In the case of Aguamilpa dam (refer Figure 3.12 in Section 3.3) the main rockfill incorporated three zones
designated Zone 3A for the upstream zone of compacted gravels, Zone 3T for the central/downstream rockfill
zone and Zone 3B for the outer downstream rockfill zone.
For details on current design features for the concrete face slab, plinth slab, jointing details, upstream facing
zones, upstream toe zoning and non-typical CFRD designs for low strength and low permeability rockfills refer
to Cooke and Sherard (1987), ICOLD (1989), Fell et al (1992), Cooke (1999) and Marulanda and Pinto (2000).
2.0 LITERATURE REVIEW OF ROCKFILL DEFORMATION
2.1 HISTORICAL SUMMARY OF ROCKFILL USAGE IN EMBANKMENT DESIGN

An historical summary of the use of rockfill in embankment design and construction (Galloway 1939; Cooke
1984; Cooke 1993) is given in Table 2.1.
From the 1930’s to the mid 1960’s the height of construction of CFRDs was limited to about 80 to 100 m due to
extensive face slab cracking and high leakage rates as a result of excessive deformation of the dumped and
sluiced rockfill. In terms of stability though (Cooke 1984) the embankment performance was excellent. The
transition from dumped and sluiced to compacted rockfill in embankment design (Cooke 1984 and 1993)
occurred in the late 1950’s to 1960’s and resulted in proliferation in the use of CFRD from the late 1960’s.
2.1.1 Deformation Properties of Rockfill

From the mid 1930’s (Galloway 1939; Morris 1939) it was recognised that a significant component of rockfill
deformation was associated with particle breakage at point contacts under increasing loads during construction
and on reservoir filling. Petersen (1939) commented on the “critical importance” of sluicing in obtaining
settlements during the construction period. Terzaghi (1960) surmised that the particle breakage that occurred
under increasing stress conditions resulted in the rearrangement in the granular structure to a more stable
position, giving the large deformations during construction and impounding and the subsequent significant
reduction in the rate of deformation post initial impounding.
-7-
Table 2.1: Historical summary of rockfill usage in embankment design (Galloway 1939; Cooke 1984; Cooke
1993).
Approximate Method of Placement and Comments
Time Period Characteristics of Rockfill
Concrete Faced Rockfill Dams
Mid to late Dumped rockfill with timber facing Early embankments constructed with timber facing. Typically of
1800’s to early very steep slopes (up to 0.5 to 0.75H to 1V). First usage of
1900’s concrete facing in the 1890’s. Height limited to about 25 m.
1920’s to 1930’s Dumped in high lifts (up to 20 to 50 Rockfill typically sound and not subject to disintegration. Dam
m) and sluiced, hand or derrick placed heights reaching 80 to 100 m. For high dams cracking of facing
upstream rockfill zone. Sluicing slab and joint openings resulted in high leakage rates (2700 l/sec
Dix River, 3600 l/sec Cogswell, 570 l/sec Salt Springs)
relatively ineffective.
Late 1930’s to High pressure sluicing used. Rockfill Cracking of face slab, particularly at the perimeter joint, and high
1960’s still very coarse. leakage rates a significant issue with higher dams (3100 l/sec at
Wishon, 1300 l/sec at Courtright).
From late 1960’s Rockfill placed in 1 to 2 m lifts, Significant reduction in post-construction deformations due to
watered and compacted. Reduction in low compressibility of compacted rockfill. Significant reduction
particle size. Usage of gravels and in leakage rates; maximum rates typically less than 50 to 100
l/sec. Continued improvement in plinth design and facing details
lower strength rock.
to reduce cracking and leakage.
Earth and Rockfill Dams
1900 to 1930 Dumped rockfill Use of concrete cores with dumped rockfill shoulders at angle of
repose. Limited use of earth core. Dam heights up to 50 to 70 m.
1930’s to 1960’s Earth core (sloping and central) with Use of earth cores significant from the 1940’s due to the
dumped rockfill shoulders. difficulties with leakage of CFRD. Increasing dam heights up to
150 m.
From 1960’s Use of compacted rockfill. Typically Improvements in compaction techniques. Early dams compacted
placed in 1 to 2 m lifts watered and in relatively thick layers with small rollers. Gradual increase in
compacted with rollers. roller size and reduction in layer thickness reduced the
compressibility of the rockfill. Significant increase in dam
heights in the mid to late 1970’s, up to 250 to 300 m.
With the increased usage of rockfill in embankment construction of high dams from the 1960’s came the use of
large scale laboratory testing (mainly oedometer, and triaxial and plane strain shear testing) to better understand
the strength and deformation properties of rockfill. The testing confirmed the significance of particle breakage
on the deformation behaviour of rockfill under increasing applied stresses and on saturation. In summary, it was
found that the deformation (and modulus) of rockfill was predominantly affected by:
• The compactive effort in placement of the rockfill. Increased modulus was observed with increased
compactive effort (Marsal 1973).
• The applied stress level. Increased particle breakage and decreasing modulus were observed with
increasing deviatoric stress levels in triaxial compression tests (Marsal 1973; Marachi et al 1969). In
oedometer tests relatively high moduli were observed for compacted rockfill samples up to normal stresses
in the order of 800 to 1000 kPa, thereafter the modulus was observed to decrease with increasing normal
stress (Marsal 1973).
• The stress path. Significantly higher modulus was observed on un-loading and re-loading at stress levels
less than previously experienced by the rockfill as shown in Figure 2.3 (Mori and Pinto 1988).
• The particle shape and grading of the rockfill. Greater deformation (and lower modulus) was observed for
angular (in comparison to rounded), more uniformly graded (i.e., lower coefficient of uniformity, Cu) and
coarser (similar Cu but higher maximum particle size) rockfills (Marachi et al 1969; Marsal 1973; Bowling
1981).
• Intact rock strength. Reduced modulus, greater deformation and reduced strength were observed for weaker
strength rockfills (Marsal 1973).
-8-
A significant aspect of rockfill deformation behaviour is its propensity to collapse on wetting. Probably the
most infamous case of collapse settlement was observed during the construction of Cogswell dam (CFRD) in
1933 (Bauman 1958). The rockfill was dumped in high lifts, up to 46 m, without sluicing. Following a very
heavy rainstorm in December 1933 the embankment crest was observed to settle up to 4.6 m (or 5.4%) and
bulging on the mid to lower slopes caused significant damage to the facing. High pressure sluicing after the
rainstorm resulted in further crest settlements of up to 1.7 m. Similar, but not as dramatic, collapse
deformations during construction on flooding of poorly sluiced rockfills were observed at Strawberry and Dix
River dams (Howson 1939). Terzaghi (1960) attributed the deformation on saturation to a loss in strength of the
rockfill, mainly in the outer surface of the particles and commented that it was more likely to occur in weathered
rockfills.
The results of laboratory testing investigating the collapse deformation behaviour of rockfill indicate:
• The stress-strain curve of rockfill in a dry state has lower compressibility than for similar rockfill (similar
grading and density) in a saturated state.
• Collapse deformation on wetting occurs for dry rockfill when the stress state (in stress-strain space) is above
the normal compression line for wetted or saturated rockfill (Alonso and Oldecop 2000).
• On wetting, the collapse strain is equivalent to the difference in strain (at a given confining stress) between
the dry and wetted states (Figure 2.1) (Nobari and Duncan 1972; Alonso and Oldecop 2000).
• The initial water content at placement has a significant influence on the amount of collapse deformation on
wetting (Nobari and Duncan, 1972), the higher the moisture content at placement the lesser the collapse
deformation on wetting.
• There is a time delay between flooding of the sample and collapse deformation (Marsal 1973; Alonso and
Oldecop 2000). Martin (1970), as reported by Justo (1991), using water and organic liquids for sample
flooding, found that the amount of collapse deformation and time for collapse deformation to occur was
dependent on the liquid used for saturation of the sample.
• Collapse deformations of similar magnitude to that occurring on flooding were obtained by imposing 100%
relative humidity on the rockfill sample (Alonso and Oldecop 2000), indicating that flooding or wetting of
the voids between the rock particles was not required for collapse deformation.
Figure 2.1: Compression curves for dry and wet states, and collapse compression from dry to wet state for
Pyramid gravel in laboratory oedometer test (Nobari and Duncan 1972)
Alonso and Oldecop (2000) concluded that the collapse deformation occurred within the individual rock
particles, and that the amount of deformation was controlled by the imposed stresses and initial moisture content
within the outer exposed voids or pores of the individual rock particles. They hypothesised that the collapse
-9-
mechanism was associated with crack propagation of the rock pores, and that the rate of propagation was
significantly affected by the total suction in the rock pores. They further concluded that “any situation leading
to a change in moisture content in the rock pores is enough to cause collapse deformation”, which is consistent
with the observation of collapse deformations induced by flooding, such as on reservoir filling, or rainfall.
This assessment seems to assume that the reduction in compressive strength on saturation observed for many
rocks, which leads to failure at the highly stressed point contacts in an angular rockfill mass, is entirely
dependent on suction effects in the outer pores of the rock pieces at the highly stressed point contacts. This
seems unlikely.
One further property of rockfill observed from oedometer testing (Sowers et al 1965; Marsal 1973; Alonso and
Oldecop 2000) was that deformation on application of an increase in applied stress occurred as a large, almost
instantaneous component followed by a much smaller time-dependent strain component. Sowers et al (1965)
commented that this secondary time-dependent component of strain approximated a straight line when plotted
against the log of time (Figure 2.2), and that the log rate of strain increased with increasing applied stress for dry
rockfill samples and on saturation (post the collapse deformation). Alonso and Oldecop (2000) observed similar
findings and hypothesised that this time dependent strain was due to the on-going process of crack propagation
and particle breakage.
Figure 2.2: Settlement versus time curves for laboratory oedometer tests on rockfill (Sowers et al 1965)
2.2 PREDICTIVE METHODS FOR ROCKFILL DEFORMATION

For CFRD and the rockfill zones of earth and rockfill dams the major components of rockfill deformation are
deformations that occur due to load application (during construction and impoundment), on saturation or
wetting and on-going time dependent or creep type deformations. Predictive methods are typically divided into
the three components deformation during construction, on first filling and long-term post construction (or post
first filling).
Most predictive methods cover the deformation behaviour of one or two of these components. Finite element
analyses have been used (Saboya and Byrne 1993; Kovacevic 1994; amongst others) for analysis of rockfill
deformation during construction and on first filling treating the events sequentially, and depending on the
embankment type, consider the deformation on saturation only during first filling. Empirical methods, usually
based on historical performance of embankments, are typically available for prediction of deformation during
- 10 -
construction and post-construction. The deformation on first filling and on wetting or saturation implicitly
incorporated in the post-construction deformation component for these methods. In the case of CFRD, methods
are available that specifically consider face slab deformation during first filling, and in these cases, the rockfill
is assumed to remain un-wetted or unsaturated on impoundment.
From field observations the significant factors affecting the deformation behaviour of rockfill are reported to
include:
• Degree of compaction of the rockfill.
• Applied stress conditions and stress path (Mori and Pinto 1988).
• Geometric shape of the valley. Pinto and Marques Filho (1998) and Giudici et al (2000) consider valley
shape to have a significant influence on deformation behaviour due to cross valley arching resulting in
reduction in applied stresses.
• Particle shape and size distribution.
• Intact strength of the rockfill.
• Wetting or saturation of the rockfill causing collapse deformation.
• Time dependent or creep type deformations
Available predictive methods incorporate several of these factors as discussed in the following sub-sections.
Finite element analyses may incorporate most of these factors in the stress-strain relationships on which the
constitutive models are based, but cannot model the time-dependent deformations.
2.2.1 Finite Element Analyses

Duncan (1992, 1996) and Kovacevic (1994) are recent references covering the state of the art in finite element
analyses applicable to the deformation behaviour of embankments constructed of rockfill during construction
and on first filling. They discuss the methods of analysis, their limitations, available constitutive models of the
stress-strain relationship and areas of uncertainty. As Duncan points out, most analyses from the literature are
Class C1 (Lambe 1973), i.e. after the event, which may account for the generally good agreement between
predicted and observed deformation behaviour.
Two important points raised by Duncan (1992, 1996) are; that the choice of stress-strain constitutive model used
in the analysis is a balance between simplicity and accuracy, and that the choice of constitutive model will
depend on the purpose of the modelling. Duncan commented that more simplistic models (e.g. linear elastic)
might be suitable if the purpose of the analysis is to analyse stresses and/or the trend of deformations and the
moduli are relatively uniform between different embankment materials. For more accurate modelling of
deformations a more complex stress-strain model is required that more realistically approximates the real
behaviour of the rockfill.
Several other important aspects of finite element analysis with respect to modelling dam embankments (more
specifically toward embankments constructed of rockfill) raised by Duncan (1992, 1996) and Kovacevic (1994)
include:
• The importance of modelling the construction as a series of incremental layers
• The type of stress-strain relationship used for modelling materials. Linear elastic models are not suitable for
accurate modelling of deformation behaviour due to the non-linear stress-strain relationship of rockfill
(Figure 2.3). Hyperbolic or multi-linear elastic models (determined from laboratory oedometer tests) can
reasonably model deformations during construction given that the stress paths in oedometer testing are
similar to the stress paths under the centreline of the embankment and that plastic deformations under the
shear stress conditions are generally not significant for rockfill embankments.
- 11 -
• A limitation of hyperbolic and multi-linear elastic models is that the model parameters are typically derived
from triaxial compression or oedometer laboratory tests and are therefore limited by the narrow range of
stress paths covered in comparison to the broader range of stress paths imposed in the field. This is
particularly evident for modelling reservoir impoundment of CFRDs. Kovacevic (1994) highlighted the
limitations of multi-linear elastic models to accurately model the deformation on reservoir impoundment.
He found that elasto-plastic models (that model pre-peak plasticity) were more suited for modelling the
deformation behaviour of the upstream face during reservoir impoundment due to the ability of these
constitutive models to more realistically account for the rockfill deformation under the stress path
conditions imposed (Figure 2.). He further commented that multi-linear elastic models based on triaxial
compression tests tended to under estimate the deformation of the face due to the initial reduction in
deviatoric stress on impounding whilst those based on oedometer test results tended to over-estimate the
face slab deformation.
• A significant factor of uncertainty associated with Class A type predictions based on laboratory testing is
the difference in material properties between the laboratory test results and those in the field due to
limitations on the maximum particle size that can be tested and variations in stiffness due to differences in
material strength, compacted density and moisture content.
Figure 2.3: Typical stress-strain relationship of rockfill from a triaxial compression test (Mori and Pinto 1988)
A further problem with laboratory test results is the limitation of representing the layering and segregation that
occurs within any single layer of rockfill. The upper part of the layer is broken down to a greater degree under
the action of the roller than the lower part of the layer and segregation occurs during placement. As a result of
the field compaction process this layering effect results in density, modulus and grading variations within a
single layer.
An important component of the modelling of embankment dams with upstream rockfill is the potential for
collapse deformation that occurs on initial impoundment. This is also potentially an important factor for rockfill
downstream of the core zone and in CFRDs where collapse deformation may occur on wetting due to rainfall or
leakage. Incorporation of collapse deformation into constitutive models adds further complexity and greater
uncertainty in estimation of material parameters between laboratory and field conditions due to the dependency
of collapse deformation on compaction moisture content, compacted density, applied stress conditions and
rockfill properties (intact strength, mineral composition and grading). Justo (1991) and Naylor (1989) propose
methods for incorporation of collapse deformation in constitutive models, and the analysis of Beliche Dam, a
central core earth and rockfill dam, by Naylor et al (1997) is an example where collapse settlements were
considered.
- 12 -
Figure 2.4: Finite element analysis of Foz do Areia CFRD (Saboya and Byrne 1993), (a) model and (b) stress
paths during construction and first filling.
2.2.2 Empirical Predictive Methods of Deformation Behaviour During Construction

and on First Filling
As previously discussed, Class A predictions of deformation behaviour of rockfill embankments during
construction and first filling using finite element analysis are often based on stress-strain relationships of
rockfills from laboratory testing or from empirically derived properties. Laboratory testing requires specialist
laboratory equipment, particularly for rockfill, to accommodate particle size distributions that are at least
somewhat representative, although still scaled down, of the placed rockfill. Available facilities to undertake this
testing are also limited and the testing is expensive. Depending on the proposed embankment design,
construction materials and compaction specifications justification for laboratory testing of rockfill may not be
warranted.
Several methods are available for estimation of the rockfill modulus from historical records of performance of
CFRD for well-compacted rockfills. Poulos et al (1972) provide a simplistic procedure, based on linear elastic
analysis, for estimation of the deformation during construction from non-dimensional factors.
Most of the methods based on historical performance serve merely to highlight the significant factors associated
with the field deformation behaviour, although, Pinto and Marques Filho (1998) proposed a simple method to
provide a “rough estimate” of the modulus for estimation of the face slab deflection of CFRD. From historical
records of the performance of CFRDs constructed typically of well-compacted angular rockfills they conclude
that void ratio and valley shape (Figure 2.5) are the dominant influences on the deformation modulus during
- 13 -
construction for narrow valleys with a valley shape factor (SF) of less than 3.5 (SF = A/H2, where A = area of
the upstream face slab measured normal with the upstream face in square metres and H = embankment height in
metres). For wide valleys (shape factor, SF, greater than 4) the effect of the valley shape can be disregarded and
the deformation modulus (ranging from 30 to 60 MPa) is dependent on the void ratio. For well-compacted
rockfills the void ratio is dependent on the rockfill gradation. Higher void ratios are representative of poorly
graded rockfills and lower void ratios of well-graded rockfills. For CFRD constructed of rounded gravels Pinto
and Marques Filho (1998) indicate that the deformation modulus during construction is very high, in the order
of 200 MPa.
Figure 2.5: Deformation modulus during construction ( E rc ) versus void ratio (Pinto and Marques Filho 1998)
Figure 2.6: Ratio of deformation modulus on first filling to during construction ( E rf E rd ) versus valley shape
factor (Pinto and Marques Filho 1998)
- 14 -
Pinto and Filho Marques (1998) recommended that the maximum face slab deformation, D, on first filling be
estimated from Equation 1, based on empirical correlation of the ratio of the deformation modulus on reservoir
filling ( E rf ) to during construction ( E rc ) with the valley shape factor (Figure 2.6).
æ 0.003 ö æ H 2 ö
D = çç ÷⋅ç ÷ Equation 1
0.21(1+ A H 2 ) ÷ ç E ÷
è e ø è rc ø
where D = face slab deflection in metres, A = area of the upstream face slab in m2, H = embankment height in
metres and E rc = deformation modulus during construction in MPa. E rc is a secant modulus determined at the
end of construction from internal settlements. It is usually determined from measurements under the
embankment centreline and in the lower half of the maximum section and therefore closely approximates a
confined modulus. Pinto and Marques Filho (1998) comment that the deformation estimate is a rough estimate,
but that it is sufficient for design of CFRDs.
Guidici et al (2000), based on field observations of well-compacted rockfill from HEC (Hydro Electric
Commission, Tasmania) CFRDs, also considered that valley shape had a significant effect on the deformation
modulus measured during construction (Figure 2.7) and on that measured during first filling ( E rf ) (both
calculated using the methods shown in Figure 1.1). Higher deformation moduli were generally measured for
narrower valleys. They commented that the scatter is associated with other factors affecting the deformation
moduli including intact rock strength and stress levels (as a function of embankment height). Guidici et al
(2000) further investigated the effects of valley shape by undertaking 3-dimensional finite element analysis of
an idealised CFRD and commented that the results confirm the influence of valley shape on deformation
behaviour during construction (Figure 2.8) and on first filling due to arching effects.
Figure 2.7: Deformation modulus during construction ( E rc ) for HEC CFRDs (Giudici et al 2000)
- 15 -
Figure 2.8: Results of 3-dimensional finite element analysis of CFRD, to give vertical displacement during
construction versus valley shape (Giudici et al 2000)
2.2.3 Predictive Methods for Deformation Behaviour Post Construction

For CFRD and other rockfill embankments post-construction time dependent deformations are observed for
more than 30 years after construction (Sherard and Cooke 1987), typically at a gradually reducing strain rate.
Empirical predictive methods of the post-construction deformation from the literature are typically based on
historical deformation curves for similar embankment types or empirical relationships derived from historical
performance of deformation behaviour. Several of the methods incorporate the deformation during first filling
whilst others only consider the prediction of deformation post first filling.
Sowers et al (1965), based on review of the post construction crest settlement of 14 dams (CFRD and sloping
and central core earth and rockfill dams) and results of laboratory oedometer tests (refer Section 2.1.1),
proposed an empirical relationship (Equation 2) to describe the post construction crest settlement with time.
They identified the method of placement of rockfill as the only factor contributing to the crest settlement of
rockfill and suggested values of α range from 0.2 %/log cycle of time for compacted and well sluiced rockfills
up to 1.05 for dumped, poorly sluiced rockfill. It should be noted that at the time of publication of this paper
rockfill placement procedures were in the transitional phase from dumped to compacted and therefore the level
of compaction was dissimilar to what would currently be considered well-compacted.
∆H = α (log t 2 − log t1 ) Equation 2
where ∆H = percent vertical crest strain (crest settlement as a percentage of dam height), α = slope of the
settlement-time curve in units of percent strain per log cycle of time (time in months), and t1 and t2 are time in
months from the date when construction was half completed.
Parkin (1977) points out that the logarithmic relationship of the form proposed by Sowers et al (Equation 2) is
obtained by integration of the phenomenological power law model for primary creep (Singh and Mitchell 1968)
derived from rate process theory with a power function of 1 (i.e. Equation 3, with m = 1). As indicated by
Parkin (1977), usage of Equation 2 can lead to errors in forward prediction of crest settlements due to incorrect
assumptions with respect to; (i) the zero time (to) for initiation of post-construction time-dependent deformation,
and (ii) assumption that the power function m (Equation 3) equals 1. Parkin proposed that a compression rate
analysis based on the power law relationship derived from rate process theory (Equation 3), incorporating
determination of to, would more accurately describe the post-construction crest settlement behaviour of rockfill.
Figure 2.9a shows the estimation of time to (from the plot of the inverse of settlement rate versus time) and
Figure 2.9b the settlement rate versus time (t – to) plot with slope of m for Cedar Creek dam.
- 16 -
.
ε = a(t − t o ) − m Equation 3
where ε& = settlement rate in percent/month, a = constant, to = initial or origin of time, t = time in months after
to and m = slope of the settlement rate versus time plot (Figure 2.9b).
Figure 2.9: Settlement rate analysis of Cedar Creek dam; (a) determination of time to and (b) settlement rate
versus time (Parkin 1977)
Soydemir and Kjaernsli (1979) analysed the post construction deformation behaviour of 23 membrane faced
rockfill dams (mostly CFRD), and proposed an empirical correlation (Equation 4) for prediction of deformation
with respect to embankment height. Table 2.2 presents the coefficient values ( β and δ ) for maximum crest
settlement, crest horizontal displacement and deflection of the membrane normal to the upstream slope for
dumped and compacted rockfill on initial impounding and after 10 years service.
s = βH δ Equation 4
where s = deflection in metres, H = embankment height in metres, and β and δ are constants.
Table 2.2: Parameters for deflection prediction (Soydemir and Kjaernsli 1979)
Coefficients for β and δ (Equation 4)
Membrane Faced Dumped Membrane Faced Compacted
Rockfill Rockfill
β δ β δ
Maximum crest settlement, sV:
Initial inpounding 0.0005 1.5 0.0001 1.5
10 years service 0.001 1.5 0.0003 1.5
Maximum crest horizontal
displacement, sH:
Initial inpounding 0.00035 1.5 0.00005 1.5
10 years service 0.0006 1.5 0.00015 1.5
Maximum deflection normal to
upstream face, sN:
Initial inpounding 0.01 2 0.002 2
- 17 -
Clements (1984) studied the post-construction deformation behaviour of 68 rockfill dams, comprising
membrane faced (dumped and compacted), sloping core and central core embankments. Clements proposed
upper and lower bound limits of crest settlement and crest displacement for each type of dam differentiating
between dumped and compacted rockfill. The individual deformation curves and bounding limits for the crest
settlement of compacted rockfill for the membrane-faced dams are presented in Figure 2.10. In comparing the
predicted deformations of Soydemir and Kjaernsli (1979) with observed, Clements concluded that the errors
were too large (correlation coefficients of 0.274 to 0.897) to make any reasonable predictions based on their
method. He recommended the use of deformation curves of existing dams (e.g. Figure 2.10) of similar design
and construction methods for prediction of deformation behaviour. The zero time has been taken as the time of
the initial measurement after the end of construction.
Pinto and Filho Marques (1985) commented that the methods of historical comparison proposed by Clements
(1984) are useful, however, their application is not so straightforward as the deformation behaviour is strongly
dependent on the initial time of measurement in relation to the end of construction. They suggested that
comparisons could be based on the time from start of first filling.
Figure 2.10: Post-construction crest settlement of membrane faced compacted rockfill dams (Clements 1984).
Sherard and Cooke (1987) compared the post-construction crest settlement from nine CFRDs of dumped and
compacted rockfills. They observed that the crest settlement of dumped rockfill is 5 to 8 times greater than for
compacted rockfill and the rate of crest settlement (Table 2.3) for dumped rockfill 10 to 20 times that of
compacted rockfill.
Table 2.3: Rates of post-construction crest settlement of dumped and compacted rockfills in CFRDs (Sherard
and Cooke 1987)
Approximate Rate of Crest Settlement for 100 m High CFRD (mm/yr)
Type After 5 yrs After 10 yrs After 30 yrs
Compacted Rockfill 3.5 1.5 0.6
Dumped Rockfill 45 30 10
Public Works Department NSW (1990) presented a plot of settlement index (Equation 5) versus time (Figure
2.11) for well-compacted CFRDs, defining bounds for low compressive strength and high compressive strength
- 18 -
rockfill. The initial time is established after the end of construction. The data points have been updated where
data records have been available, several data points have been added and the rockfill has been categorised with
respect to intact UCS strength and coefficient of uniformity (Cu). Further discussion on these aspects is
discussed in Section 3.4.
Settlement Index = Crest Settlement / (Dam Height/100)2 Equation 5
where crest settlement is measured in millimetres from initial monitoring after the end of construction and dam
height in metres.
700 1. Mackintosh
Very high intact strength, Cu<15 Very high intact strength, Cu>15 2. Mangrove Creek
Medium to high intact strength Intact strength not known 3. Little Para
600 4. Kangaroo Creek
1
5. Serpentine
3 5 6. Tullabardine
500 low compressive 7. Winneke
Settlement Index .
strength rockfill 2 8. Foz do Areia

9. Reece
400
10. Cethana
11 11. Kotmale
23 4 12. Alto Anchicaya
300 6
18 7 13. Murchison
22
14. Shiroro
200 15. Bastyan
16. Crotty
8 9 10 17. Golillas
14 15 13
100 18. Aguamilpa
16 12 20
19. Outrades 2
high compressive strength rockfill
21 19 17 20. Fades
0 21. Salvajina
0 5 10 15 20 25 30 22. Xingo
Time (years) 23. White Spur
Figure 2.11: Settlement index versus time for well-compacted rockfills (adapted from Public Works Department
NSW 1990)
3.0 ANALYSIS OF DEFORMATION BEHAVIOUR OF CFRD

CRFDs are recognised for their stability, so one can be confident that measured deformations are not linked to
instability. Conventional limit equilibrium analyses are generally not undertaken for CFRD (Sherard and Cooke
1987) unless potential stability problems associated with the foundation are identified, such as shear along a
weak bedding seam in the foundation.
The deformation behaviour of rockfill during construction, on first filling and long-term post-construction (post
first filling) has been analysed from the case study database of thirty-six mainly CFRD. The factors likely to
affect the deformation behaviour of rockfill that have been considered in the analysis include placement
methods, rockfill UCS, particle size distribution (grading and size), particle angularity, wetting or saturation,
embankment geometry and valley geometry.
As part of the analysis, current methods (mainly empirical methods) for prediction of deformation are evaluated
and, where appropriate, what are considered improved methods have been developed.
3.1 CASE STUDY DATABASE

The database of case studies (Table 3.1) includes thirty-five CFRDs and one central core earth and rockfill
embankment (El Infiernillo). A summary of the embankment details and rockfill properties of the case studies
is presented in Table 3.1 with further details given in Table A1.1 (Appendix A).
- 19 -
Table 3.1: Summary of embankment and rockfill properties for CFRD case studies (data is for Zone 3A or 3 rockfill unless otherwise noted)
Embankment Dimensions Material Parameters / Properties of Rockfill
Time For First Filling
Zone *4 Strength*2 dmax % finer
Dry Layer
Erc (MPa), Erf (MPa)
Dam Name Height, Upstream 1 Particle Void Level of (from end of main
L/H Rockfill Source* Classification Cu Density Thickness Placement
H (m) Slope Class Auth Shape (mm) 19 mm Ratio, e Compaction average rockfill construction)
UCS (MPa) (t/m3) (m)
Aguamilpa 4p 10t SDVR, in moist
185.5 2.6 1.5 to 1 3A 3B dredged alluvium (Very High) rounded 85 600 34 2.22 0.18 0.6 Good 305 (250 to 330) 770 0.48 to 1.83 years
(Zone 3A) condition
8-12p 6-10t SDVR,
Crotty 83 2.9 1.3 to 1 3A 3A Gravels (Pleitocene) (Very High) rounded 70 200 48 2.54 0.20 0.6 Good 375 (113 to 636) 470 0.87 to 2.0 years
watered
4p 10t SDVR, water 155 (145 to 165), 4 to 5.17 years (June 82 to
Golillas 125 0.9 1.6 to 1 3A 2 gravels, unprocessed (Very High) rounded 125 350 40 2.135 0.24 0.6 Good 250
added arching likely Aug 83)
Salvajina 4p 10t SDVR, water 205 (175 to 260) -
148 2.4 1.5 to 1 3A 2 gravels, unprocessed (Very High) rounded 9.2 400 32 2.24 0.25 0.6 Good 500 0.33 to >0.75 years (1985)
(Zone 3A) added arching likely
Aguamilpa UCS = 180
185.5 2.6 1.5 to 1 3T T ignimbrite angular 30 500 28 2.04 0.24 0.6 4p 10t SDVR Good 104 - -
(Zone 3T) Very High
4p 10t SDVR, 20% 138 (100 to 170), 375 (@ 30 to
Alto Anchicaya 140 1.9 1.4 to 1 3A D hornfels (Very High) angular 18 600 22 2.28 0.294 0.6 Good 0.18 to 0.2 (Oct 74)
water likely arching 40% height)
8p (10t ?) SDVR, 20%
Bastyan 75 5.7 1.3 to 1 3A 3A Rhyolite, SW to FR (Very High) angular 42 600 25 2.20 0.23 1.0 Good 130 (120 to 140) 290 0.86 to 1.05 years
water
160 (120 to 210)
4p 10t SDVR, 15%
Cethana 110 1.9 1.3 to 1 3A 3A Quartzite (Very High) angular 23 900 21 2.07 (0.27) 0.9 Good 105 (85 to 120) 300 0.26 to 0.48 years (1971)
water
likely arching
UCS = 80 6p 10t SDVR, 25% -0.5 to 4 years (12/88 to
Chengbing 74.6 4.4 1.3 to 1 3A 3B/3C Tuff lava angular 10.4 1000 - 2.06 0.277 1.0 Good 43 110
Very High water 1993).
3
Foz Do Areia * 160 5.2 1.4 to 1 3A 1B
basalt (max 25% basaltic UCS = 235
angular 6 600 10 2.12 0.33 0.8
4p 10t SDVR, 25%
Good 47 (38 to 56) 80 (65 to 92)
0.5 to 0.9 years (Apr to Aug
(Zone 3A) breccia) Very High water 80)
3 6p 9t SDVR, 10%
Ita (Zone 3A) * 125 7.0 1.3 to 1 3A E1/E3' basalt (Very High) angular 11 700 12 2.179 0.308 0.8
water
Good 48 87 (83 to 91) 0.7 to 0.9 years
UCS = 25 angular to 4p 10t SDVR, 100% 0 to 1.95 yrs (Sept 69 to Aug

Kangaroo Creek 60 3.0 1.3 to 1 3A 3 schist 310 600 44 2.34 0.201 0.9 to 1.8 Good - 140
Medium to High subangular water 71)
Khao Laem UCS < 190 4p 10t SDVR, 15% 0.6 to 1.9 years (June 84 to
130 7.7 1.4 to 1 3A 3A limestone angular - 900 - - - 1.0 Good 59 (43 to 79) 130 to 240
(Zone 3A) Very High water lower in 70 m Nov 85)
4p 15t SDVR, 30% 0.54 to 1.04 years (Nov 84
Kotmale 90 6.2 1.4 to 1 3A 3A charnockitic / gneissic (High to Very High) angular - 700 - 2.20 - 1.0 Good 61 (47 to 87) 145 (135 to 155)
water to Aug 85)
UCS = 8 - 14 angular, 4p 9t SDVR, no water 0.58 to 2.83 years (Aug 77

Little Para 53 4.2 1.3 to 1 3A 3A dolomitic siltstone 110 1000 35 2.15 0.223 1.0 Good 21.5 (19.5 to 23.5) -
Medium elongated ? lower half to Nov 79)
UCS = 45 angular, 8p 10t SDVR, 10% 45

Mackintosh 75 6.2 1.3 to 1 3A 3A Greywacke, some slate 52 1000 38 2.20 0.24 1.0 Good 63 1.92 to 2.93 years
High elongated water (35 to 60)
Mangrove Creek fresh siltstone & UCS = 45 - 64 angular to 4p 10t SDVR, 7.5% 0.6 to > 15 years (not
80 4.8 1.5 to 1 3A 1B 310 400 27 2.24 0.18 0.45 to 0.6 Good 55 to 60 -
(Zone 3A) sandstone High subangular water reached FSL by 1996)
Mangrove Creek weathered to fresh UCS = 26 - 64 angular to 3 4p 10t SDVR, dry of
80 4.8 1.5 to 1 3B 4 330 450 32 2.06 t/m 0.26 0.45 Good 46 (36 to 56) - -
(Zone 3B) siltstone & sandstone High subangular OMC
UCS = 148 8p 10t SDVR, 20% 190
Murchison 94 2.1 1.3 to 1 3A 3A Rhyolite (SW to FR) angular 19 600 22 2.27 0.234 1.0 Good 560 (485 to 640) 1.04 to 1.09 years
Very High water (170 to 205)
UCS = 80 - 370 4p 10t SDVR, 5 to 1.32 to 1.46 yrs (Apr to June
Reece 122 3.1 1.3 to 1 3A 3A Dolerite angular 10 1000 11 2.287 0.29 1.0 Good 86 (57 to 115) 190 (175 to 205)
Very High 10% water 1986)
Salvajina weak sandstone and 6p 10t SDVR, water
148 2.4 1.5 to 1 3B 4 (Medium ?) angular ? 45 600 32 2.26 0.21 0.9 Good 62 (likely arching) - -
(Zone 3B) siltstone added
UCS = 22 4-6p 10t SDVR, no 59 (Zone 3A) 0.42 to 2.58 years (June 72
Scotts Peak 43 24.8 1.7 to 1 3A 3A argillite angular ? 380 914 38 2.095 0.266 0.915 Good 20.5 (18.5 to 23.5)
Medium to High water 420 (Zone 2B) to Aug 74)
3
Segredo * 145 5.0 1.3 to 1 3A 1B
basalt (<5% basaltic UCS = 235
angular 7.4 - - 2.13 0.37 0.8
6p 9t SDVR, 25%
Good 55 (arching likely) 175 Approx. 0.6 years
(Zone 3A) breccia) Very High water
angular to 4p 9t SDVR, not sure 0.24 to 2.94 years (Dec 71
Serpentine 38 3.5 1.5 to 1 3A 3A Ripped quartz schist (Medium to High) 210 152 69 2.10 0.262 0.6 to 0.9 Good 92 (46 to 142) 97 (94 to 100)
subangular if water added to Aug 74)
6p 15t SDVR, 15% 1.44 to >1.8 years (from
Shiroro 125 4.5 1.3 to 1 3A 2 granite (Very High) angular 32 500 22 2.226 0.20 1.0 Good 66 (61 to 71) -
water 5/84)
Tianshengqiao - 1 UCS = 70 - 90 15 to 6p 16t SDVR, 20%
178 6.6 1.4 to 1 3A 3B Limestone, SW to FR angular 800 - 2.19 0.23 0.8 Good 49 (40 to 57) - -1.5 to >0.8 years
(Zone 3A) Very High 20 water
- 20 -
Table 3.1: Summary of embankment and rockfill properties for CFRD case studies (data is for Zone 3A or 3 rockfill unless otherwise noted)
Embankment Dimensions Material Parameters / Properties of Rockfill
Time For First Filling
Strength*2 Dry Layer
Dam Name Height, Upstream Zone *4 Particle dmax % finer Void Level of Erc (MPa), Erf (MPa) (from end of main
L/H Rockfill Source* 1
Classification Cu Density Thickness Placement
H (m) Slope Class Auth Shape (mm) 19 mm Ratio, e Compaction average rockfill construction)
UCS (MPa) (t/m3) (m)
Tianshengqiao - 1 UCS = 16 - 20 6p 16t SDVR, 20%
178 6.6 1.4 to 1 3B 3C Mudstone angular 40 600 20 to 35 2.23 0.21 0.8 Good 37 (32 to 42) - -
(Zone 3B) Medium water
UCS = 45 angular, 4p 10t SDVR, > 10% 2.05 to 2.35 years (5/81 to
Tullabardine 25 8.6 1.3 to 1 3A 3A Greywacke, some slate 28 400 30.5 2.22 0.23 0.9 to 1.0 Good 74 170
High elongated water 8/81)
(High to Very High (4p 10t ?) SDVR, > 180 0.18 to 0.24 years (June to
White Spur 43 3.4 1.3 to 1 3A 3A Tuff - SW to FR angular - 1000 - 2.30 0.18 to 0.25 1.0 Good 340
?) 10% water (160 to 200) July 1989).
UCS = 66 4-6p 10t SDVR, 15% 1.62 to 5.04 years (6/80 to
Winneke 85 12.4 1.5 to 1 3A 3 SW to FR Siltstone angular 33 800 28 2.07 0.302 0.9 Good 55 (50 to 59) 104
High water 11/83)
UCS = 240 8p 12t SDVR, 25 to
Xibeikou 95 2.3 1.4 to 1 3A 1 Limestone - FR angular - 600 - 2.18 0.284 0.8 Good 80 (60 to 100) 260 -1.5 to 6 years (June 1995)
Very High 50% water
(High to Very High 4p 10t SDVR, 15% 1.02 to 1.46 (mid to late
Xingo (Zone 3A) 140 6.1 1.4 to 1 3A 3 granite gneiss angular 18 650 4 to 33 2.15 0.28 1.0 Good 34 (30 to 39) 76 (73 to 80)
?) water 1994) years
3
Foz Do Areia * 160 5.2 1.4 to 1 3B
1C & mix basalt & basaltic UCS = 235
angular 14.2 - - 1.98 0.27
0.8 for 1D 4p 10t SDVR, 25% Good to
32 (29 to 38) - -
(Zone 3B) 1D breccia High to Very High 1.6 for 1C water Reasonable
Aguamilpa UCS = 180 Good to
185.5 2.6 1.5 to 1 3B 3C ignimbrite angular 22 700 - - - 1.2 4p 10t SDVR 36 (25 to 45) - -
(Zone 3B) Very High reasonable
3 (0.33 to
Ita (Zone 3B) * 125 7.0 1.3 to 1 3B E3 Breccia and Basalt (High to Very High) angular 13.3 750 15 2.066
0.39)
1.6 4p 9t SDVR, no water Reasonable 24 (14 to 46) - -
Khao Laem UCS < 190 4p 10t SDVR, 15%

130 7.7 1.4 to 1 3B 3B limestone angular - 1500 - - - 2.0 Reasonable 30 - -
(Zone 3B) Very High water lower in 70 m
3
Segredo * 145 5.0 1.3 to 1 3B
1C & basalt (<5% basaltic UCS = 235
angular 10.2 - - 3
0.43
1D - 0.8
4p 9t SDVR, no water Reasonable
28 (25 to 33),
- -
2.01 t/m
(Zone 3B) 1D breccia) High to Very High 1C - 1.6 likely arching
Sound and weathered (Medium to Very 6p 10t SDVR, no
Xingo (Zone 3B) 140 6.1 1.4 to 1 3B 4 angular 80 750 15 to 60 2.1 0.31 2 Reasonable 13 (12 to 14) - -
granite gneiss High ?) water
dumped and well
Courtright 97 2.8 1.15 to 1 3 3 granite (Very High) angular (<7) 1750 - (1.8) 0.47 8 to 52 Poor - - 0 to 9 years (from 1959)
sluiced
El-Infiernillo diorite & silicified UCS = 125
148 2.3 1.75 to 1 3B 3B angular 13 600 22 1.85 0.47 0.6 to 1.0 4p D8 dozer, no water Poor 39 (27 to 48) - -
(Zone 3B) conglomerate Very High
UCS = 100 - 140 2000 to dumped and well 0.07 to 0.58 years (Dec 52
Lower Bear No. 1 75 3.9 1.3 to 1 3 3 granodiorite angular 8 to 10 low - - max 65 Poor - 21
Very High 3000 sluiced to June 53)
UCS = 100 - 140 2000 to dumped and well 0.07 to 0.58 years (Dec 52
Lower Bear No. 2 46 5.7 1 to 1 3 3 granodiorite angular 8 to 10 low - - max 36.5 Poor - 40
Very High 3000 sluiced to June 53)
1500 to 8 to 52 dumped and well
Wishon 90 11.3 1.15 to 1 3 3 granite (Very High) angular (<7) - 1.80 0.47 Poor - - 0.4 to 0.45 years (May 1958)
2000 (variable) sluiced
dumped and poorly Poor to Very
Dix River 84 3.7 1.1 to 1 3 3 Limestone & Shale (?) (Very High) angular ? low - - - - 21 - - Started in 1925
sluiced Poor
UCS = 100 - 130 angular, low 2000 to dumped and poorly Poor to Very
Salt Springs 100 4.0 1.3 to 1 3 3 granite low (1.88) 0.41 5 to 52 - 20 0 to 1.48 years (1931/32)
Very High blocky (<10) 3000 sluiced Poor
UCS = 45 7.6 2.6 years. Filled in 71 hours
Cogswell 85.3 2.1 1.3 to 1 3 3 Granitic Gneiss angular 7 1300 5 2.05 0.37 dumped dry (no water) Very Poor - 44
High (46 m max) (Mar 1938)
El-Infiernillo diorite & silicified UCS = 125 Dumped and spread,

148 2.3 1.75 to 1 3C 3C angular <13 >600 - 1.76 0.54 2.0 to 2.5 Very Poor 22 (17 to 27) - -
(Zone 3C) conglomerate Very High no water
Legend:
H = dam height e)
Erc = secant modulus during construction (average *3 Brazilian CFRDs constructed with a mix of very high strength (…) in strength classification, Cu, dry density
L = crest length m
4p 10t SDVR = 4 passes of 10 tonne smooth drum basalt, vesicular basalt and high strength basalt breccia. and void ratio columns indicate estimation
Cu = uniformity coefficient (d60/d10) vibrating roller *4 Class = Classification system used in this report (refer % water = % by volume
dmax = average maximum particle size *1 FR = fresh, SW = slightly weathered Section 4.1.2.3); - indicates unknown
Erf = deformation modulus during first filling 2
* rock strength classification to AS 1726-1993 Auth = rockfill zoning used by author/s of referenced paper
- 21 -
Selection of case studies was limited, as far as practicable, to CFRDs on rock foundation or those with a limited
depth/area of alluvial gravels in the foundation, with good quality monitoring records and adequate information
on the properties of the rock in the rockfill. Case studies were targeted so as to assess the potential factors
affecting deformation behaviour, including dumped versus compacted rockfill, and very high strength versus
medium to high strength rockfill. The deformation during construction at El Infiernillo dam has been included
because of the quality of published internal deformation records of dry dumped and loosely compacted rockfill
for which little to no information was found in the literature for dumped rockfill in CFRDs.
3.2 DEFORMATION DURING CONSTRUCTION

The analysis of rockfill deformation behaviour during construction has been based on the observed deformation
behaviour from the monitoring of internal settlements mostly from hydrostatic settlement gauges. This is the
most frequently used method of deformation monitoring in CFRDs. For the most part deformations from
gauges installed under the embankment centreline have been used. For CFRDs internal horizontal movements
are generally monitored less frequently and are not considered here. Unless otherwise stated, the data is for
Zone 3A rockfill or the equivalent zone/s.
3.2.1 Base Plots of Data

Figure 3.1 and Figure 3.2 present the stress-strain and secant modulus-stress relationships from the internal
settlement monitoring records of eleven embankments (10 CFRDs and El Infiernillo dam). For the ten CFRDs
the monitoring records are from HSGs installed under the embankment centreline and in the lower third to half
of the embankment. For El Infiernillo the monitoring records are from internal vertical settlement gauges
(IVM) in the rockfill shoulders, and include monitoring records within the dry dumped (Zone 3C) and the
loosely placed (Zone 3B) rockfill zones. Vertical stresses have been estimated from the elastic solutions for
embankments by Poulos and Davis (1974) and assume that the embankment has been constructed in horizontal
layers across the full embankment width. Use of these solutions takes into consideration the effect of
embankment shape on vertical stress (refer Section 1.2.2).
Note that in Figure 3.1 and Figure 3.2 zero strain is equivalent to the vertical stress at the mid point of the layer
analysed when the upper HSG was placed. Therefore, in Figure 3.1 the stress-strain curve will not intersect the
origin. It will intersect the vertical stress axis at stresses in the range 30 to 150 kPa. The assumption of zero
strain at a small but finite value of vertical stress introduces a slight error in the secant moduli calculations.
3.2.2 Effect of Compaction/Placement Method

The stress-strain relationships (Figure 3.1) indicate that the degree of compaction has a significant influence on
rockfill modulus, confirming the results of large scale laboratory tests on rockfill (Section 2.1.1). The dry
dumped and dry, loosely placed rockfills of El Infiernillo dam have lower modulus than for the well-compacted
rockfills. The results also indicate:
• The very high modulus of well-compacted gravels (Crotty dam) in comparison to well-compacted quarried
(or angular) rockfills, although the smaller size and well-graded particle size distribution of the Crotty
gravel fill may also have influenced the modulus.
• The intact rockfill strength (or UCS) has an influence on the modulus for well-compacted quarried rockfills.
The very high strength rockfills (Cethana, Murchison and Reece dams in particular) have greater moduli
than the medium to high strength rockfill used at Tullabardine and Mackintosh dams.
Other factors will also influence the stress-strain relationship of the rockfill, such as particle size distribution
and valley shape, although they are not readily apparent from the small number of case studies represented in
Figure 3.1. The effect of these factors is addressed later.
- 22 -
2000
1800
Estimated Vertical Stress (kPa) .
1600
1400
1200
1000
800
600
400
200
0
0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5%
Measured Vertical Strain (%)
Cethana Crotty Mackintosh Murchison
Kotmale Reece Bastyan Tullabardine
Khao Laem White Spur El-Infiernillo (loose) El-Infiernillo (dry dumped)
Figure 3.1: Stress-strain relationship for rockfills observed during construction
600
Crotty
500
Secant Modulus (MPa) .
400
Murchison
Khao Laem
300
Cethana
Mackintosh
200
El Infiernillo - loose
100
0
0 200 400 600 800 1000 1200 1400 1600 1800
Estimated Vertical Stress (kPa)
Figure 3.2: Secant modulus versus vertical stress from monitoring during construction
- 23 -
3.2.3 Effect of Confining Stress

An important aspect of rockfill deformation behaviour identified from laboratory testing is the effect of
increasing applied stress. The non-linear stress-strain relationship, indicative of decreasing modulus with
increasing applied stress (either deviatoric, confining or mean normal stress), evident from laboratory oedometer
and triaxial compression testing is clearly evident in the field observations (Figure 3.2 and Figure 3.3). The
initial behaviour of increasing secant modulus at low stress levels (less than 400 to 600 kPa) for several dams in
Figure 3.2 (Murchison and Cethana for example) is considered to be due to seating of the HSG as a result of
initial incompatibility between the filling placed around the instrument and the predominant rockfill.
Plotting the field data against deviatoric, confining or mean normal stress has not been attempted due to the
introduction of errors in confining stress estimation associated with estimation of Poisson’s ratio and the
observation that Poisson’s ratio is also stress dependent (Duncan 1992, 1996). It is also noted that the stress
path represented by the field observations closely approximates that of the oedometer laboratory test (Kovacevic
1994) and the moduli (secant and tangent) would therefore approximate the confined modulus.
The plot of tangent modulus (Figure 3.3) shows a greater variation in modulus than for secant modulus plot due
to the incremental calculation of tangent modulus. However, the trend of decreasing modulus with increasing
vertical stress is clearly evident. Over the vertical stress range 500 to 1000 kPa, reductions in the tangent
modulus of approximately 40 to 50% are observed for the well-compacted rockfills.
The variation in modulus (both secant and tangent) for an individual case study is affected by the accuracy of
settlement measurements, the time-dependent or creep deformation of rockfill and therefore the rate of
construction, and the variation in rockfill height across the width and length of the embankment during
construction.
500
400
Murchison
Crotty
Tangent Modulus (MPa) .
300
200
Cethana
Mackintosh
El Infiernillo - loose
100
0
0 200 400 600 800 1000 1200 1400 1600 1800
Figure 3.3: Tangent modulus versus vertical stress from monitoring during construction
- 24 -
3.2.4 Effect of Valley Shape

The influence of valley shape is identified (Pinto and Filho Marques 1998; Giudici et al 2000) as a significant
factor affecting the calculated deformation modulus during construction, particularly for embankments
constructed in narrow valleys with relatively steep abutment slopes, because of arching across the abutment
slopes. To check these hypotheses finite difference analysis (FDA), using the FLAC program, was undertaken.
A two-dimensional FDA was undertaken assuming an idealised rockfill embankment of 100 m height
constructed in valleys with river widths of 20, 50 and 100 m, and uniform abutment slopes of 0, 26.5, 45 and 70
degrees. The rockfill was modelled as a linear elastic material with Young’s modulus of 100 MPa and
Poisson’s ratio (ν) of 0.27, and the foundation with Young’s modulus of 50 GPa (Figure 3.4). After establishing
initial stresses in the foundation, the embankment construction was modelled in 5 m lifts and stresses (major and
minor principal stresses, vertical stress and horizontal stress) monitored at seven locations (a to g in Figure 3.4)
in the centre of the gully. Grid element sizes of 1.25 m height and 1.25 to 2.5 m width were used. The analysis
was then repeated by construction of the embankment in a single 100 m lift (as was done in the analysis by
Giudici et al (2000)).
It is recognised that the analysis has several deficiencies, however, for the purpose of assessment of arching
effects on stresses within the embankment the results are considered reasonably representative. The greatest
deficiency in the two-dimensional analysis is not taking into consideration the significant three-dimensional
effect of embankment shape, which will therefore result in over-estimation of stresses in the latter stages of
construction.
Embankment
35.6 m E = 100 MPa, ν = 0.27
Foundation
a
E = 50 GPa 100 m
15 m
ν = 0.27 b Abutment
10 m
c slope
10 m
d
10 m
e
10 m
f
9.4 m
g
River
width
Figure 3.4: Idealised model for two-dimensional finite difference analysis of cross-valley influence
The results of the analysis are presented in Figure 3.5. Vertical stresses at the end of construction have been
normalised against those for the case with zero abutment slopes (i.e. no cross-valley arching effect) and plotted
against the abutment slope angle. Points a to g correspond to the locations shown in Figure 3.4.
- 25 -
(a) 80 River width = 20m, constructed in 5m lifts (b) 80 River width = 20m, constructed in 1 lift
70 70
Abutment Slope Angle (degrees) .

Point g
60 60 Point f
Point e
50 50 Point d
Point c
40 40 Point b
Point a
Point g
30 30
Point f
Point e
20 Point d 20
Point c
10 Point b 10
Point a
0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalised Vertical Stress Normalised Vertical Stress
(c) 80 River width = 50m, constructed in 5 m lifts (d) 80 River width = 50m, constructed in 1 lift
70 70

Point g
60 60 Point f
Point e
50 50 Point d
Point c
40 40 Point b
Point g Point a
Point f
30 Point e 30
Point d
20 Point c 20
Point b
10 Point a 10
0 0
0.75 0.8 0.85 0.9 0.95 1 0.75 0.8 0.85 0.9 0.95 1
80 River width = 100m, constructed in 5 m lifts 80 River width = 100m, Constructed in 1 lift
(e) (f)
70 70
Point g
60 60 Point f
Point e
50 50 Point d
Point c
Point b
40 40
Point g Point a
Point f
30 Point e 30
Point d
20 Point c 20
Point b
10 Point a 10
0 0
0.9 0.95 1 1.05 0.9 0.95 1 1.05
Figure 3.5: Results of two-dimensional finite difference analysis of the effect of cross-valley shape on vertical
stresses in the dam. (a), (c) and (e) represent construction in 5 m lifts (to 100 m) and (b), (d) and (f)
construction in a single 100 m lift. Vertical stresses, at the end of construction, are normalised against those
obtained for the zero abutment slope analysis.
- 26 -
The results of the analysis indicate:

• Cross-valley arching is only significant (greater than 20% reduction in vertical stress) for narrow valleys
(river width less than 30 to 40% of the dam height) with steep abutment slopes (greater than about 50
degrees, but also dependent on river width), and then only in the lower third to half of the embankment.
• Where the river width is approximately equal to half the embankment height, cross-valley arching has some
effect (10 to 20% reduction in vertical stress) for abutment slopes steeper than about 45 degrees, and then
only in the lower third to half of the embankment.
• Negligible arching is observed for river widths equal to about the embankment height regardless of the
abutment slope.
• Modelling embankments in single or very large lifts is likely to result in significant under-estimation of
stresses within the embankment, particularly for river widths less than about half the embankment height.
Table 3.2 presents an assessment of the effects of cross-valley arching on the embankments studied as part of
this analysis. Significant arching (greater than 20% reduction in vertical stress) is likely in the lower 25 to 50%
portion of the narrow, steep valley embankment sections of Segredo, Salvajina, Murchison, Cethana and Alto
Anchicaya. For Golillas cross valley arching is a significant factor given the narrow and very steep sided valley
in which the embankment has been constructed.
Table 3.2: Assessment of cross-valley influence on arching for case studies analysed
Influence of Cross-Valley Arching
Significant (> 20% reduction in Vertical Some (10 to 20% reduction in Limited to Negligible (< 10%
Stress) Vertical Stress) reduction in Vertical Stress
Segredo (narrow gully section, lower 15%) White Spur (lower quarter) Wishon, Xingo, Xibeikou, Winneke,
Salvajina (lower half in narrow gully) Shiroro (lower third) Tullabardine, Tianshengqiao,
Murchison (lower quarter) Lower Bear No. 1 (lower quarter) Serpentine, Scotts Peak, Salt Springs,
Golillas Little Para (lower 15%) Reece, Mangrove Creek, Mackintosh,
Cogswell (lower quarter) Kangaroo Creek (lower third) Lower Bear No.2, Kotmale, Khao
Cethana (lower quarter in steep gully) Courtright (lower third) Laem, Ita, Foz do Areia, Dix River,
Alto Anchicaya (lower half in steep gully) Bastyan (lower third) Crotty, Aguamilpa,
El Infiernillo (lower third)
The secant modulus versus vertical stress plot during embankment construction should be able to provide an
indication of cross-valley arching. The typical plot, assuming no arching effects, is for a small reduction in the
secant modulus with increasing vertical stress (Figure 3.2). An increase in the calculated secant modulus with
increasing vertical stress is considered to provide an indication of cross-valley arching. The deformation
behaviour for HSG 1 to foundation (lower 10 m or 15%) at Bastyan dam (Figure 3.6) is indicative of this
behaviour. The plot for HSG 1 to 2 (located 50 to 65 m below crest level) is more typical of the normal
response. Another indicator could also be a relatively high calculated secant modulus in the lowest section of
the embankment compared with the section above such as indicated by the monitoring results at Cethana dam
(Figure 3.6). However, this may be affected by variations in the rockfill properties or placement techniques.
The analysis shows that valley shape is not as significant as indicated from the empirical analyses of Pinto and
Marques Filho (1998) and Giudici et al (2000), Figure 2.5 and Figure 2.7 respectively, and the finite element
analyses by Giudici et al (2000), Figure 2.8. The differences in deformation moduli observed for the well-
compacted rockfills must be due to other factors in addition to valley arching, including the rockfill particle size
grading, strength of the rock and type of fill.
For the finite element analyses (Giudici et al 2000) significant errors are introduced into the vertical stress
profiles, particularly for embankments constructed in narrow valleys, by construction of the embankment in a
single stage rather than “building” the embankment in layers. The method used to calculate the rockfill modulus
- 27 -
during construction is also an important consideration for the empirical analyses. Pinto and Filho Marques
(1998) and Giudici et al (2000) (Figure 2.5 and Figure 2.7) used, as far as the authors are aware, the simplifying
assumption of Figure 1.1a, which would over-estimate the vertical stresses and therefore under-estimate the
calculated moduli.
250
Secant Modulus (MPa) .
200
150
100
Cethana (HSG 2)
50 Cethana (HSG 2 to 4)
Bastyan (HSG 1)
Bastyan (HSG 1 to 2)
0
0 200 400 600 800 1000 1200 1400 1600 1800
Figure 3.6: Indicators of cross valley arching effects from variations in the secant modulus with vertical stress.
3.2.5 Effect of Particle Size Distribution

The effect of particle size distribution (PSD) on the secant modulus was investigated by grouping the data into
sets with similar degree of compaction, type of fill (rockfill or gravel) and intact rock strength.
For each rockfill zone (mostly Zone 3A) of each case study the procedure to estimate a representative secant
modulus at the end of construction was:
• Estimation of the vertical stress and secant modulus at the end of construction for each pairing of HSGs
within the rockfill zone allowing for the effect of embankment shape (refer Section 1.2.2). The secant
moduli were, in general, determined from internal monitoring points in the region bounded by the lower
60% of the embankment and within the central half of the dam cross section.
• For case studies where valley shape was considered to having had some to significant influence (Table 3.2)
the secant moduli was adjusted by multiplication with an appropriate stress reduction factor determined
from Figure 3.5. Secant moduli values considered as being greatly affected by valley arching were omitted.
• A representative secant modulus for each rockfill zone was then calculated by averaging the estimates of
secant moduli from that rockfill zone corrected for valley shape.
The results of laboratory testing on rockfills (Section 2.1.1) indicated that the shape of the particle size
distribution curve (measured by the uniformity coefficient, Cu = D60/D10) and particle size itself affected the
deformation behaviour. To take into consideration both of these factors various combinations of particle size
(e.g. D50, D60, D80, etc.) and different definitions of uniformity coefficient (D60/D10, D80/D30, etc.) from the
average PSD for each rockfill zone were statistically analysed (using single and multiple variable methods)
against the representative secant modulus at end of construction, determined as outlined above. Further
statistical analyses were undertaken incorporating embankment height as a variable in an attempt to normalise
the data against vertical stress.
The analysis indicated that the best predictors of the secant modulus at the end of construction ( E rc ) were:
• D80 (diameter for 80% passing from average grading curve) for the very high strength well compacted
quarried rockfill data set (Figure 3.7) with a regression coefficient of 0.83 for the power function shown.
- 28 -
• D90 (diameter for 90% passing from average grading curve) for the medium to high strength rockfills with a
regression coefficient of 0.70 for the exponential function given as Equation 6. The fit to D80 provided a
reasonable correlation with coefficient of regression of 0.44 (Figure 3.7) for the exponential function shown.
In comparing of the two methods, the difference in calculated moduli in most instances is less than 5 MPa,
or statistically, a standard deviation of less than 7%. Higher moduli are obtained using the D90 method for
PSD with a steep coarse part of the curve (i.e. above D70) and higher moduli for the D80 method for PSD
with a relatively flat coarse part of the PSD curve typical of gravels. Either approach is suggested for
estimation of E rc for medium to high strength rockfills. For simplicity the authors have chosen to use D80
for predictive methods.
E rc = 153 ⋅ e (−0.0045⋅D90 ) for medium to high strength well-compacted rockfill Equation 6
From Figure 3.7 it is evident that the secant moduli of the medium to high strength rockfills shows less variation
with increasing D80 than for the very high strength rockfills. This is considered to reflect the broad, well-graded
shape of the particle size distribution of the medium to high strength rockfills as indicated by the high
uniformity coefficients (Cu values from 45 to 300) in comparison to steeper shape of the very high strength
rockfills (Cu values from 6 to 42).
400
Very high strength, well compacted rockfill
Crotty
Medium to high strength, well compacted rockfill
350
Gravels, well compacted
Aguamilpa 3A Reasonable compaction
Erc (MPa), at End of Construction .
300
250
(-0.0052 D80)
Erc = 113 e
200 2
R = 0.44
Murchison
5 -1.70
150 Erc = 9.0 x 10 . D80
Gollilas 2
R = 0.83
Cethana
100
Bastyan Reece
50
Salvajina 3B
Scotts Peak Little Para
0
0 50 100 150 200 250 300 350 400 450
D80 (particle diameter equivalent to 80 percent passing)
Figure 3.7: Representative secant modulus (mostly Zone 3A rockfill) at end of construction ( E rc ) versus D80
from average grading of the rockfill.
Other points to note are:
• The medium strength rockfills of the medium to high strength data set (Little Para, Scotts Peak and
Salvajina Zone 3B) plot below the trendline for the medium to high strength case studies.
• For the well-compacted gravels there is not enough data to establish a relationship between D80 and E rc .
• D80, and D90 in the case of medium to high strength rockfills, is a better predictor of E rc than uniformity
coefficient, Cu.
• The predictions are not improved by adding Cu to the equation. This is possibly related to the partial inter-
dependence of Cu and D80, with decreasing Cu generally observed with increasing D80, particularly for the
very high strength rockfills.
- 29 -
• For the very high strength rockfills embankment height (and hence stress level) appears to have some
influence. Apart from Bastyan dam (75 m height) the cases with embankment heights less than about
120 m plot above the trendline and embankment heights greater than about 130 m below the trendline. This
finding is not unexpected given the influence of stress level on the secant modulus (Figure 3.2).
• No similar correlation with height is evident for the medium to high strength quarried rockfill data set.
The E rc values in Figure 3.7 for each individual rockfill zone are representative of the applied vertical stress
range of the internal points considered in the estimation of E rc . The representative mean vertical stress for each
data set in Figure 3.7 is given in Table 3.3.
Table 3.3: Stress conditions representative of the data sets from Figure 3.7
No. Vertical Stress Characteristics of Data Points
Data Set
Cases Range (kPa) Mean (kPa) Comment
Very high strength, well-compacted 9 900 to 2200 1400 5 of 9 points with mean from
quarried rockfill 1300 to 1550 kPa
Medium to high strength, well- 9 400 to 1300 800 Even spread of data points
compacted quarried rockfill across the stress range
Well-compacted gravels 4 950 to 2200 1500 Spread of data points across
the stress range
3.2.6 Moduli During Construction of Dumped Rockfill

For dumped rockfill, the limited internal deformation data during construction is insufficient for analysis and
assessment of predictive methods for estimation of the moduli during construction. The available data records
are:
• At El Infiernillo very high strength diorite and silicified conglomerate were used for the rockfill. The
deformation records indicate E rc values (Figure 3.2) of:
− For the Zone 3B rockfill, placed dry in 0.6 to 1.0 m layers and compacted by bulldozer, E rc at end of
construction ranged from 25 to 49 MPa (average 39 MPa) at vertical stress levels in the range 550 to
1800 kPa.
− For the Zone 3C rockfill, dumped dry and spread in 2 m thick layers, E rc at end of construction ranged
from 17 to 27 MPa (average 22 MPa) at vertical stress levels in the range 400 to 800 kPa.
• At Ita dam (Figure 3.13), the base 10 m thickness of rockfill in the river section was dumped, presumably
through water. E rc at end of construction for the dumped rockfill (very high strength basalt) was estimated
at 15 to 19 MPa at vertical stress levels in the range 1400 to 2100 kPa.
The tangent moduli for the rockfill at El-Infiernillo (Figure 3.3) approached values as low as 15 to 20 MPa for
the Zone 3B rockfill and 10 to 15 MPa for the Zone 3C rockfill. Based on back-analysis of the deformations of
central core earth and rockfill dams (Fell et al 2000, Woodward Clyde 1998) the tangent moduli is likely to be
lower for dirty or weathered dumped rockfills, possibly down to as low as 5 MPa at stress levels exceeding
those previously experienced by the rockfill.
- 30 -
3.3 DEFORMATION OF THE FACE SLAB OF CFRD ON FIRST FILLING
3.3.1 Simplified Methods of Estimation of Maximum Face Slab Deformation

Fitzpatrick et al (1985) and Pinto and Marques Filho (1998) suggest that simple empirical methods are sufficient
to estimate face slab deformations. The most simple empirical estimates are based on ratios of the modulus on
first filling, E rf , to the deformation modulus during construction, E rc , calculated according to Figure 1.1a.
Cooke (1993) indicates that the ratio of moduli on first filling to moduli during construction ( E rf E rc ) is
typically in the range 1.3 to 3, a range generally confirmed by most authors discussing the variation.
The calculation of E rf (Fitzpatrick et al 1985) in Figure 1.1b does not take into consideration the stress path or
stress distribution within the upstream rockfill due to first filling. It is based on the assumption of a uniform
stress change, equivalent to the applied stress from the water load on the concrete face, acting over the distance
normal to the face slab between it and the foundation (Figure 1.1b). Fitzpatrick et al (1985) recognised that
these assumptions were incorrect; but commented that the method provides a simple and reasonable
approximation of the rockfill modulus and face slab deformation. They indicate that the method is applicable
for estimation of E rf (and therefore prediction of face slab deformation) over the range from 20% to about 60%
of the embankment height.
Analysis of the data set was undertaken in an attempt to improve on the method of prediction of the E rf E rc
ratio. Valley shape effects were not specifically taken into consideration given that valley shape is potentially
likely to affect both the secant moduli at end of construction and modulus during first filling. In the analysis the
representative secant moduli at end of construction and not corrected for valley shape for the Zone 3A rockfill
was used for each case study. As shown in Figure 3.8 there is a reasonable correlation between E rf E rc and
embankment height taking into consideration the upstream slope angle. A reasonable estimation of E rf E rc is
possible for upstream slope angles of 1.3 to 1.4H to 1V. For flatter upstream slopes a lower ratio is considered
appropriate, however, there are insufficient data points for estimation of a correlation, although a very
approximate estimation of the trendline is presented in Figure 3.8.
4.0
Khao Laem
3.5 Estimation of trendline for upstream

slopes of 1.3 - 1.4H to 1V
3.0
Scotts Peak
2.5
Erf / Erc
2.0
Mackintosh Ita
Foz do Areia
1.5 Golillas
Crotty
1.0 1.3H to 1V
Serpentine
1.4H to 1V
0.5 Very approximate estimation of trendline for 1.5H to 1V
upstream slopes of 1.5H to 1V 1.6H to 1V
1.7H to 1V
0.0
0 25 50 75 100 125 150 175 200
Embankment Height (m)
Figure 3.8: E rf E rc ratio versus embankment height

- 31 -
A number of cases plot away from the trendline or are not in accordance with the general trend of decreasing
E rf E rc ratio with flattening of the upstream slope for the 1.3 –1.4H to 1V data set. The reasons for this are:
• For Mackintosh and Foz do Areia the face slab deflection profiles are different to the typical profile
(discussed in Section 3.3.2) with maximum deflection measured at 15 to 20% of the dam height above the
toe; i.e. lower than normal.
• For Ita dam the dumped rockfill in the river section (Figure 3.13), which is of low secant modulus in
comparison to the Zone 3A compacted rockfill, is considered to have influenced the face slab deflection.
• At Crotty, the facing zone between the main gravel rockfill and the concrete face comprised quarried
rockfill. The likely lower modulus of this facing zone is considered to have had an effect of the deformation
behaviour of the face slab resulting in the relatively low value of the E rf E rc ratio.
• At Scotts Peak the embankment zoning (Figure 3.14) is considered to have affected the calculation of E rf ,
which was based on the measured settlement in the HSGs adjacent to the upstream face slab.
• Serpentine dam plots at quite a low ratio. The reason for this is uncertain.
• At Khao Laem the high value of E rf , and hence the high E rf E rc ratio, was considered by Watakeekul et
al (1985) to possibly be due to arching effects at the maximum section within a localised deep section
upstream of the embankment centreline. Arching effects were not a factor in the calculation of the secant
modulus, E rc , at the end of construction.
Other factors considered to affect the E rf E rc ratio are:

• Fitzpatrick et al (1985) considered the time to completion of first filling a factor, the shorter the time of first
filling the greater the likely ratio due to time dependent strain effects.
• Cooke (1984) suggested that a possible factor in the E rf E rc ratio was that the compacted rockfill might be
substantially stiffer in the horizontal direction, due to the greater compaction of the upper compared to the
lower sections of a compacted layer. A two-dimensional finite difference analysis, using the computer
program FLAC, was undertaken on Reece dam to further evaluate the effect of intra-layer differential
moduli (differential in the both vertical and horizontal directions) on the face slab deformation behaviour.
The embankment was modelled in 1 m layers with a 3 to 1 ratio in moduli between the upper 0.5 m and
lower 0.5 m of each layer calculated such that the overall vertical modulus of any layer was equivalent to
that measured from actual observations during construction. The modelling indicated that the deformations
on first filling were similar, less than 5% variation, with or without consideration of variable inter-layer
modulus. In addition, finite element analysis by Kovacevic (1994), using and an elasto-plastic constitutive
model (modelling pre-peak plasticity), obtained reasonable predictions of face slab deformation on first
filling by for Winscar and Roadford dams (refer Section 2.2.1). It is concluded, therefore, that the effect of
intra-layer differential moduli has a negligible effect on the face slab deformation behaviour during first
filling.
• The valley shape may influence the E rf E rc ratio to a small extent. For the 1.3H to 1V cases the CFRDs
constructed in narrow valley (L/H < 3.2) tend to plot above the trendline whilst for the broader valleys tend
to plot below the trendline, although there are exceptions. For the 1.4H to 1V cases the trend with respect to
valley shape is not as evident.
Two-dimensional finite difference analysis was undertaken to evaluate the stress paths of the rockfill in the
upstream zone of the embankment for different upstream slopes. Analyses were undertaken on a nominal 100 m
high embankment with upstream slopes of 1.3H to 1V and 1.55H to 1V. The embankment was constructed in 5
m lifts and the reservoir impoundment in 6 stages to a maximum height of 96 m (i.e. assuming 4 m of
freeboard). Given that the interest was in the stresses within the embankment, the rockfill was modelled as a
linear elastic material with Young’s modulus of 100 MPa and Poisson’s ratio of 0.27. The concrete face slab
- 32 -
was not modelled and the upstream facing zone was modelled with similar modulus parameters to the main
rockfill. It is recognised that the assumption of linear elastic properties would lead to some errors in the stress
distribution, however, the purpose of the analysis was for comparison and errors were likely to be relatively
minimal.
Stresses were calculated at 16 locations in the upstream zone and under the embankment centreline (Figure 3.9
and Figure 3.10). The stress paths for the points normal to the face slab at a height 30% above the embankment
toe are presented in Figure 3.9 for the 1.3H to 1V upstream slope angle case and Figure 3.10 for the 1.55H to 1V
upstream slope angle case. The results indicate that:
• The stress paths during construction (Figure 3.9b and Figure 3.10b) show a similar trend, with some
variation, for both the 1.3H to 1V and 1.55H to 1V upstream slope angle. The general trend of the stress
path up to the end of construction is for increasing deviatoric and mean normal stress. In the early stages of
first filling the deviatoric stress decreases whilst the mean normal stress continues to increase. In the mid to
latter stages of filling the deviatoric stress starts to increase again reaching stress levels in excess of those at
the end of construction.
• The overall increase in deviator stress from the end of construction to the end of first filling is greater, at all
comparison points in the upstream shoulder, for the 1.55H to 1V upstream slope case, whilst the overall
increases in mean normal stress are relatively uniform.
• The difference in stress path from the end of construction to the end of first filling is likely to result in
greater strains in the 1.55H to 1V case on first filling and therefore a reduced deformation modulus on first
filling ( E rf ). Given that the secant modulus at the end of construction ( E rc ) will be similar between the
cases, the ratio E rf E rc is likely to be smaller for the 1.55H to 1V upstream slope. Estimation of E rf from
the deformation of the upstream face during first filling modelling confirms this.
The finite difference analysis confirms the findings from field observations (Figure 3.8) that the upstream slope
angle is likely to affect the E rf E rc ratio. Flatter upstream slope angles will result in a reduced E rf E rc ratio
all other factors being equal.
It is important to recognise that E rf is not a proper simulation of the rockfill modulus during first filling.
Rather it is an artefact of the method of calculation, which is shown in Figure 1.1b. This method uses
simplifying assumptions in regard to the geometry of the section analysed, and the assumed stress increase due
to the water is not properly modelling the stresses, as seen in Figure 3.9 and Figure 3.10. E rf should only be
used to calculate face slab deflections using the procedures shown in Figure 1.1b.
For numerical analyses the constitutive model defining the stress-strain relationship of the rockfill should be
based on the method for prediction of E rc (discussed in Section 4.1).
- 33 -
500
Principal Deviatoric Stress, q (kPa) .
Points normal to face slab at

(b) 30% of dam height
400
300
200 Point f
Point g
Point h
Point i
100
Point o
end of construction
0
0 200 400 600 800 1000 1200
Mean Normal Stress, p (kPa)
upstream slope angle; (a) monitoring point locations, (b) stress paths for points normal to face slab at 30% of the
embankment height.
500
Principal Deviatoric Stress, q (kPa) .
Points normal to face slab

(b) at 30% dam height
400
300
Point f
200 Point g
Point h
Point i
100 Point o
end of construction
0
0 200 400 600 800 1000 1200
Mean Normal Stress, p (kPa)
upstream slope angle; (a) monitoring point locations, (b) stress paths for points normal to face slab at 30% of the
embankment height.
- 34 -
3.3.2 Shape of the Face Slab Displacement

The deformed shape of the concrete face on first filling is strongly dependent on the zoning of rockfill in the
body of the embankment. For embankments constructed almost entirely of a single rockfill zone (Figure 3.11)
the typical pattern of deformation is for maximum deflection at about 30 to 50% of the embankment height
reducing to zero at the toe and with constant to gradual reduction in deformation toward the crest.
Where the embankment zoning comprises rockfill zones of differing modulus the deflected shape of the face
slab can reflect the modulus variations. In cases where large differences in rockfill modulus occur, such as the
use of gravels of very high modulus with quarried rockfills of significantly lower modulus, problems with
cracking of the face slab due to development of tensile stresses can occur resulting in leakage. Several
examples of the effect of rockfill modulus variation on face slab deflection are:
• Aguamilpa dam (Mori 1999), Figure 3.12. The face slab deflection (Figure 3.12) was significantly affected
by the modulus variation between the upstream gravel zone and downstream zone of quarried rockfill. The
maximum deflection was measured near to the embankment crest. And the resultant deformation profile
was considered by Mori (1999) to have caused cracking in the upper section of the face slab.
• The typical design detail for several large Brazilian CFRDs (Ita dam Figure 3.13) incorporates the better
quality quarried rockfill placed in relatively thin layers in the upstream third of the embankment and the
poorer quality rockfill placed in larger lifts in the central and downstream regions resulting in a modulus
variation. Ita dam incorporated a zone of dumped rockfill in the river section resulting in larger deflections
where the zone of influence incorporated these weaker rockfill zones.
• A similar pattern of face slab deflection was observed at Mangrove Creek dam due to modulus variation
between the fresh sedimentary rockfill close to the upstream face and the main body of weathered
sedimentary rockfill.
• At Scotts Peak dam (Figure 3.14) the use of rockfill zones with significantly different moduli contributed to
cracking of the asphaltic concrete membrane face. The embankment design incorporated a zone of well-
compacted gravels at the upstream toe and main body of argillite rockfill of medium to high intact rock
strength. On first filling, large differential deflections caused tensile cracking of the upstream face near to
the contact between the gravel and argillite rockfill zones resulting in leakage flows of 100 litres/sec.
Face slab deflection profiles different to the typical profile (Figure 3.11) were also observed at Mackintosh
(Figure 3.15) and Foz do Areia dams. In both cases the maximum deflection during first filling was measured at
15 to 20% of the dam height above the toe, a much lower level than typically observed. The reasons for this
observation are not clear, however, it is suspected that it is likely to be related to the rockfill modulus.
Potentially lower modulus zones may have been present in lower portion of these embankments due to the
initial production of rockfill of weaker intact rock strength or poorer gradation (low Cu or larger particle size), or
placement at a lower relative density.
Figure 3.11: Face slab deflection during first filling (4/2/71 to 25/4/71) of Cethana dam (Fitzpatrick et al 1973).
- 35 -
Figure 3.12: Face slab deformation during first filling of Aguamilpa dam (Mori 1999)
Figure 3.13: Face slab deformation during first filling of Ita dam (Sobrinho et al 2000)
Figure 3.14: Scotts Peak dam (courtesy of HEC Tasmania)

- 36 -
Figure 3.15: Mackintosh dam (a) embankment section (courtesy of HEC Tasmania) and (b) face slab
deformation on first filling (Knoop and Lack 1985)
3.3.3 Prediction of Face Slab Deflection on First Filling

For the more typical CFRD geometries (Figure 3.11) or where large variations in rockfill moduli are not likely
between rockfill zones, simple empirical correlations are considered sufficient for prediction of face slab
deflection on first filling, using E rf values estimated from Figure 3.8. Details on the suggested procedure for
empirical estimation of deformations are discussed in Section 4.2.
Mori (1999) considers that a more rigorous method of determination of the face slab deformation than the
method by Pinto and Marques Filho (1998), that takes into consideration the geometry of the structure, stress
path on impounding and the strength and deformation of the various rockfill zones within the embankment is
required.
For prediction of face slab deformation behaviour during first filling for complex rockfill geometries and/or
with large moduli variations between the rockfill zones, finite element analyses should be considered. The
detailed analysis would be justified considering the potential for face slab cracking as a result of possible
development of tensile stresses.
- 37 -
3.4 POST CONSTRUCTION CREST SETTLEMENT

Available methods for prediction of post-construction crest settlement of rockfill embankments (see Section
2.2.2) are empirical and are generally based on historical records of similar embankment types and similar
methods of construction.
An important aspect of the post construction deformation behaviour of rockfill related to its stress-strain
characteristics is that relatively large deformations occur on application of stresses above those not previously
experienced by the rockfill, such as on first filling or embankment raising. On un-loading or re-loading to stress
states less than those previously experienced (such as due to fluctuations in the reservoir level) the rockfill
moduli is very high and the resultant deformation therefore likely to be minimal. Collapse deformation on
initial wetting can also result in relatively large deformations. The following discussion will consider post-
construction deformation behaviour of rockfill (predominantly crest settlements) in CFRD due to:
• Events where stresses are likely to exceed those previously experienced, most notably first filling.
• Ongoing, time-dependent (or creep) deformation of rockfill.
• Deformations due to wetting from rainfall or leakage.
Two significant factors in using empirical methods for prediction are the base time of deformation, and
consideration of the timing of events such as first filling. If near full impoundment occurs before completion of
construction or before monitoring points are established, then a significant component of the deformation is
likely to have occurred and will not be captured by the post-construction monitoring. It is important that this be
considered in comparison between case studies and the forward prediction of future deformations. It is also
important that the base time of deformation is clearly defined in the predictive method being used and that the
same definition is used when forward predicting future deformations.
3.4.1 Case Study Records of Post-Construction Crest Settlement

The typical post-construction crest settlement pattern, ignoring first filling effects, is for crest settlement to
occur at a decreasing rate with time. Most equation based empirical methods for prediction of post-construction
deformation model this behaviour by logarithmic relationships of strain versus log time (Sowers et al 1965) or
power type relationships of strain (Soydemir and Kjaernsli 1979) or strain rate (Parkin 1977) versus time. Total
post-construction crest settlements will be dependent on when the base reading of monitoring points is
established after the end of construction. Differences in terms of even months to establish base readings after
the end of construction are significant as strain rates are greatest immediately after construction.
3.4.1.1 Zero Time, to, for Post Construction Crest Settlement
The method proposed by Parkin (1971) (refer Figure 2.7 in Section 2.2.3) was used for evaluation of the initial
or zero time, to, for start of post-construction crest deformation for each case study. The basis of Parkin’s
method, derived from rate process theory, is the assumption that strain rate versus time is a power law
relationship and therefore a plot of log strain rate versus log time (Figure 2.7b) will be linear provided to is
correctly established.
Figure 3.16 shows examples of the time versus inverse strain rate plots for four dams, showing that for three
cases to at the end of the main rockfill construction is the best fit of the data, and in one case (Crotty) to at the
end of first filling the best fit. It was found that the best fit to each case study was for a to at either the end of the
main rockfill construction or at close to the completion of first filling. In summary, the analysis indicated that:
• For CFRD constructed of dumped rockfill (excluding Salt Springs) and for well-compacted CFRD generally
less than 100 m in height (excluding Crotty) the best fit for to was at the end of construction.
• For well-compacted CFRD generally greater than 100 m in height as well as Salt Springs (dumped rockfill
CFRD of 100 m height) and Crotty (83 m high well-compacted CFRD constructed of river gravels) the best
fit for to was at the end of first filling.
- 38 -
The end of main rockfill construction is defined as the time of completion of the main rockfill zones (Zones 3A
and 3B). In some cases this may be a significant time (up to 1 to 2 years) before actual completion of the
concrete face and crest detail works.
20
Time (years since end of

16
construction)
12
Crotty
4 Murchison
Little Para
Reece
0
0 100 200 300 400 500 600 700
1/(Strain Rate), (year per % vertical crest strain)
Figure 3.16: Examples of derivation of zero time for post-construction settlement
For consistency it is considered necessary for the initial or zero time, to, to be at either the end of main rockfill
construction or at the completion of first filling. For zero time at the end of main rockfill construction (Figures
3.17 to 3.19) the power function of the slope of the log strain rate versus time log plot (Equation 3 and Figure
2.9) was closer to 1, and therefore the long-term slope of the vertical crest strain versus log time was close to
linear (with some exceptions) allowing for easier prediction of future deformation. However, visual
interpretation for comparative purposes is difficult due of the variation in time between end of construction and
start of monitoring for the cases and the deformations during first filling. In Figures 3.17 to 3.19 “ff” stands for
first filling.
For zero time, to, at the end of first filling (Figures 3.20 to 3.22) visually the comparison between case studies is
much clearer and upper and lower bounds for a specific set of historical monitoring records more readily
identified, allowing for easier evaluation of “abnormal” deformation behaviour. However, future predictions are
visually more difficult given the curved nature of the deformation behaviour in the plots of crest settlement
versus log time. This curvature is due to the power function, m, in Equation 3 being less than 1.
- 39 -
0.0%
Crest Settlement (% of dam height) .

0.2%
Acceleration due to raising of
0.4% full supply level by 3.05 m
0.6%
0.8%
Wishon
Dix River
1.0%
Lower Bear No. 1
1.2% Lower Bear No. 2
Salt Springs
1.4% Courtright
end of first filling
1.6%
0.1 1.0 10.0 100.0
Time (years since end of main rockfill construction)
Figure 3.17: Post-construction crest settlement versus time for dumped rockfill CFRD, to at end of main rockfill
construction.
a) 0.0%
0.2%
0.4%
0.6%
Scotts Peak (ff=0.42 to 2.6)
Mackintosh (ff=1.9 to 2.9)
0.8% Mangrove Creek (ff > 15)
Xingo (ff=1.02 to 1.46)
Tianshengqiao (ff > 0.8)
1.0%
1.2%
0.1 1.0 10.0 100.0
b) 0.00%
0.05%
0.10%
0.15%
Winneke (ff=1.6 to 5.4)

0.20%
Kangaroo Creek (ff=0 to 1.95)
Serpentine (ff=0.24 to 2.9)
0.25% Tullabardine (ff=2.0 to 2.35)
Little Para (ff=0.6 to 2.8)
White Spur (ff=0.18 to 0.24)
0.30%
0.35%
0.1 1.0 10.0 100.0
Figure 3.18: (a) and (b) Post-construction crest settlement versus time for CFRD constructed of compacted
rockfills of medium to high intact strength, to at end of main rockfill construction.
- 40 -
a) 0.00%
Foz Do Areia (ff=0.5 to 0.9)

Reece (ff=1.3 to 1.46)
0.05% Shiroro (ff > 1.8)
Kotmale (ff=0.54 to 1.04)
0.10% Aguamilpa (ff=0.48 to 1.83)

0.15%
0.20%
0.25%
0.30%
0.1 1.0 10.0 100.0
b) 0.00%
0.02%
0.04%
0.06%
0.08%
Alto Anchicaya (ff=0.18 to 0.2)
Bastyan (ff=0.86 to 1.05)
0.10%
Cethana (ff=0.26 to 0.48)
Murchison (ff=1.04 to 1.09)
0.12% Golillas (ff=4 to 5.2)
Crotty (ff=0.87 to 2.0)
0.14%
0.1 1.0 10.0 100.0
Figure 3.19: (a) and (b) Post-construction crest settlement versus time for CFRDs constructed of well-
compacted rockfills of very high intact strength and of well-compacted gravels, to at end of main rockfill
construction.
0.0%
Acceleration due to raising of
full supply level by 3.05 m

0.2%
0.4%
0.6%
Wishon
0.8% Dix River
Lower Bear No. 1
1.0% Lower Bear No. 2
Salt Springs
1.2% Courtright
1.4%
0.1 1.0 10.0 100.0
Time (years since end of first filling)
Figure 3.20: Post-construction crest settlement versus time for dumped rockfill CFRD, to at end of first filling.
- 41 -
0.00%
0.05%
0.10%
0.15%
0.20%
0.25%
Winneke
0.30% Kangaroo Creek
Scotts Peak
0.35% Serpentine
Tullabardine
Little Para
0.40% Mackintosh
Mangrove Creek
0.45% Xingo
White Spur
0.50%
0.1 1.0 10.0 100.0
Figure 3.21: Post-construction crest settlement versus time for CFRD constructed of compacted rockfills of
medium to high intact strength, to at end of first filling.
0.00%
Alto Anchicaya
Bastyan
0.02% Cethana
Foz Do Areia
Murchison
0.04%
Reece
Kotmale
0.06% Golillas
Crotty
Aguamilpa
0.08%
0.10%
0.12%
0.14%
0.1 1.0 10.0 100.0
Figure 3.22: Post-construction crest settlement versus time for CFRDs constructed of well-compacted rockfills
of very high intact strength and of well-compacted gravels, to at end of first filling.
3.4.1.2 Shape of the Vertical Crest Settlement Plots

For the vertical crest settlement versus log time plots for zero time at the end of the main rockfill construction
(Figures 3.17 to 3.19) two features are readily observed; the ‘S’ shape behaviour for a number of the case
studies, and the effect of the time difference between the end of rockfill construction and the initial crest
settlement reading. The ‘S’ shaped curve is a feature of those cases where first filling was commenced more
than about 0.3 to 0.5 years after the end of the main rockfill construction and the embankment height was
greater than about 50 m. The steep portions of the curve are associated with crest settlement during first filling
- 42 -
due to the increase in stress levels in the body of the rockfill beyond those previously experienced. Steeper
sections of the slope are generally observed for those case studies where the rate of filling was relatively rapid
and where first filling started at least several years after construction. At the end of first filling the strain rate
remains at a relatively high rate (on log time scale) due to time dependent effects and then decreases to a strain
rate typical of the long-term deformation rate (on a log scale).
From analysis of the deformation during first filling it was observed that the amount of crest settlement during
the filling period was dependent on a number of factors (Table 3.4). The settlement (as a percentage of
embankment height) was found to increase with/for:
• Increasing time of first filling
• Increasing embankment height.
• Dumped compared to compacted rockfills.
• Decreasing intact rockfill strength for compacted quarried rockfills. For embankments constructed
predominantly of gravels the crest settlements (as a percentage of embankment height) were similar to those
of the very high strength rockfills.
• The closer the start of first filling to the end of the main rockfill construction. This was considered to be
due to the concurrent process of time dependent deformation, the rate for which decreases with time after
the end of construction.
Table 3.4: Summary of crest settlement during the period of first filling
Rockfill Placement Method No. of Time of First Embankment Range of Crest Settlement
and Intact Strength Cases Filling (years) Height Range (m) (% of embankment height)
Dumped rockfill 6 Up to 1 year 43 to 100 0.15 to 0.48
> 1 year 0.23 to 0.34
Compacted gravels 2 1.13 to 1.2 83 and 125 0.017 to 0.024
< 0.1 yrs 43 to 140 0.010 to 0.014
Compacted very high strength 0.1 to < 0.5 year 75 to 110 m 0.020 to 0.027
10
rockfill > 110 m 0.053 to 0.20
> 0.5 year 90 and 166 0.030 to 0.13
< 0.5 year 25 0.008
Compacted medium to high
7 > 1 year 38 to 53 0.058 to 0.096
strength rockfill
60 to 85 0.058 to 0.129
The time between the end of rockfill construction and the initial crest settlement reading has a significant
influence on the measured crest settlement. Relatively large settlements are observed when initial readings are
taken shortly after the end of construction (Tianshengqiao in Figure 3.18a and Kotmale in Figure 3.19a). Where
the time delay to monitoring is relatively large (more than about four to six months after the end of rockfill
construction) the crest settlements (as a percentage of embankment height) are smaller than for similar type
CFRD (e.g. Gollilas in Figure 3.19a and Mangrove Creek in Figure 3.18a). As previously discussed, this effect
makes for difficulties when attempting to compare the deformation behaviour of CFRD.
These effects are not as evident for the crest settlement plots for zero time at the end of first filling (Figures 3.20
to 3.22) for which upper and lower bounds of deformation are clearer and can be more meaningfully defined.
The values of the coefficient, m, in the strain rate versus time power function (Equation 3) was investigated for
both to at the end of the main rockfill construction and at the end of first filling. Results are given in Table 3.5.
The value of m represents the slope of the log strain rate versus log time linear relationship.
- 43 -
The findings indicated some dependence of m on the method of placement and intact rockfill strength with
lower values for the poorly sluiced dumped rockfill. However, overall the degree of variability within any data
set was quite large and no readily identifiable correlation was apparent for prediction of m.
Table 3.5: Values of the coefficient, m, in the strain rate – time power function (Equation 3).
Rockfill Placement Method and Power Coefficient, m Comments
Intact Strength to at end of rockfill to at end of
construction first filling
Dumped and poorly sluiced 0.61 and 0.88 0.50 and 0.57 Limited number of cases (6), only two
Dumped and well sluiced 1.21 to 1.84 0.78 to 0.97 poorly sluiced.
Compacted medium to high 0.49 to 1.44 0.29 to 0.69 More consistent values for to at end of
strength quarried rockfill first filling; 6 of 7 cases in the range
0.42 to 0.69. Greater spread for to at end
of rockfill construction.
Compacted very high strength 0.92 to 1.84 0.62 to 1.04 Broad spread of values for to at end of
quarried rockfill rockfill construction. Better consistency
for to at end of first filling; 7 of 8 cases
in the range 0.82 to 1.09.
Compacted gravels 1.77 and 1.86 0.96 (both) Only two cases, but reasonably
consistent values.
3.4.1.3 Summary of the Factors Affecting the Post Construction Crest Settlement
Behaviour
In summary, the most dominant factors influencing the post-construction crest settlement (as a percentage of the
embankment height) are:
• The method of placement (dumped versus compacted). Significantly greater crest settlements are evident
for the dumped and sluiced rockfills compared with compacted rockfills. The rates of crest settlement (on a
log scale) post first filling are also significantly greater for the dumped rockfills.
• Intact rockfill strength for compacted rockfills. For embankments constructed of medium to high intact
strength rockfill, the total magnitude of settlement at 10 years is on average approximately twice that of
very high strength rockfills. The long-term, post first filling rates of crest settlement (on a log scale) are
also significantly higher for the medium to high strength rockfills, 2 to 10 times that of the very high
strength rockfills.
• Use of gravels as opposed to rockfill. For embankments constructed of predominantly well-compacted
gravels (Crotty and Golillas) the total post-construction settlement is less than that for the well-compacted,
very high strength rockfills. This is likely to be due to several reasons, but is considered to be mainly a
result of the rounded shape of the gravels. The point area of contact between particles of rounded gravel
will be significantly greater than that of angular quarried rockfill. Hence, the contact stresses will be
significantly less resulting is less particle breakage.
• Embankment height. The rate of crest settlement (on a log scale) is lower for the smaller embankments
within each data set; e.g. Lower Bear No. 2 of the dumped rockfill cases; Tullarbardine, Serpentine and
Little Para of the compacted medium to high strength rockfill cases; and Bastyan and Murchison of the
compacted very high strength rockfill cases. The effect of embankment height is probably not as significant
for the compacted very high strength rockfill cases.
Other factors that appear to affect the post construction crest settlement are:
• The poorly sluiced dumped rockfills (Salt Springs and Dix River) tend to have greater rates of crest
settlement (on a log scale) than for well-sluiced dumped rockfills. The values of the power coefficient m are
also lower for the poorly sluiced dumped rockfills.
• Particle size and uniformity coefficient are thought to have some influence on post-construction crest
settlement behaviour with greater crest settlement rates (as a percentage of embankment height) likely for
- 44 -
greater maximum particle sizes (such as D80). However, no statistical correlation is evident from the data
indicating that its influence is relatively minor in relation to other factors.
• Embankment zoning geometry of the main rockfill. Given that the Zone 3B rockfill is generally of lesser
quality compaction (and sometimes lesser quality rockfill materials) for typical CFRD designs, greater crest
settlement rates (per log cycle of time) would be expected for increasing size of Zone 3B.
• Amount of fluctuation in reservoir level. It is suspected, although the data does not indicate, that the greater
stress changes associated with larger fluctuations in reservoir level could result in greater overall
settlements.
• Climate. In higher rainfall areas greater amounts of water infiltration into the rockfill would be expected,
potentially resulting in greater settlements.
3.4.1.4 Prediction of Post Construction Crest Settlement
Prediction of the post-construction crest settlement is relatively difficult given the variable influence of first
filling and the significant effect of timing on the initial base reading. However, reasonable predictions are
possible.
The proposed method is to consider separately the time dependent deformation and the settlement due to first
filling, as shown in Figure 3.23 for Bastyan dam for example. During and shortly after the period of first filling
the time dependent deformations are assumed to occur concurrently with the crest settlements due to increase in
vertical stresses from the water load. Estimation of the settlement at some time after the end of main rockfill
construction is therefore by summation of the time dependent and first filling related components.
0.00%
Start of first filling
Crest settlement (% dam height) .
0.01%
0.02% Effect of first End of first filling

filling
0.03%
0.04%
0.05%
Time dependent
0.06%
deformation
0.07%
0.1 1.0 10.0
Figure 3.23: Post construction crest settlement of Bastyan dam

(a) Time dependent crest settlement
Figure 3.24 presents the data for long-term post first filling crest settlement rate (percent per log cycle of time)
for the compacted rockfill embankments, for to at the end of main rockfill construction. Trendlines with
reasonable to good regression coefficients have been derived for compacted high strength rockfills and for
compacted very high strength rockfills and gravels. The intact strength classification is that of the dominant
rockfill zone under the dam axis, which in most cases is the Zone 3A rockfill, but in several cases is the Zone
3B rockfill. A tentative trendline is also given for the medium strength rockfills.
Kotmale and Mangrove Creek stand out as outliers to the general trend. For Mangrove Creek this is thought to
be due to the fact that first filling to full supply level has yet to be reached after 15 years, and for which the
long-term settlement rate was estimated from the settlement data more than 3 years after the end of construction
(Figure 3.18). For Kotmale the reason is not clear, although the available crest settlement data is limited to
about 1 year after first filling, and it is possible that the rate (per log cycle of time) may have since reduced.
- 45 -
0.30
Kotmale
0.25
Settlement (%/log cycle) .

Long-term Rate of Crest .
Mangrove Creek
R2 = 0.79
0.20
0.15
0.10
R2 = 0.51
0.05
0.00
0 20 40 60 80 100 120 140 160 180
Gravels Very high strength rockfills
High strength rockfills Medium strength rockfills
Figure 3.24: Long-term crest settlement rate (as a percentage of embankment height per log cycle of time)
versus embankment height for compacted rockfills
For the dumped rockfill CFRD the data (Figure 3.17) indicates an increasing rate of vertical crest strain (per log
cycle of time) with time. Estimates of the long-term vertical crest strain (per log cycle of time) can be derived
from either Figure 3.17 or Table 3.6 based on the time period after the end of main rockfill construction.
Table 3.6: Estimates of long-term crest settlement rates for dumped rockfills
Post First Filling Crest Settlement Rate
Time Period (from
(settlement as a % of embankment height per log cycle
end of main rockfill
of time)
construction)
Range Mean
0.5 to 5 years 0.10 to 0.58 0.27
5 to 20 years 0.25 to 1.14 0.66
20 years plus 0.33 to 1.44 0.85
(b) Crest settlement under increased stresses from first filling

Whilst a reasonable estimate of the crest settlement rate per log cycle of time (as a percentage of embankment
height) can be made for the time dependent component of crest settlement from Figure 3.24 and Table 3.6,
estimates of crest settlement during first filling are difficult. Two alternatives methods are proposed:
(i) Estimation of the crest settlement attributable to first filling which covers the period of first filling and the
short time thereafter as shown in Figure 3.23. Figure 3.25 summarises the crest settlement (as a
percentage of embankment height) attributable to first filling and can be used for predictive purposes
based on the intact strength of the rockfill, placement method (either dumped or compacted) and
embankment height. The relatively broad variation for some groups, in particular for dumped rockfill,
reflects the high degree of variability from the case studies (refer Section 3.4.1.2). Trendlines are
provided for the compacted gravels and rockfill.
(ii) Estimation of the crest settlement during first filling from Table 3.4. The estimates from Table 3.4 are
different to those from Figure 3.25 in that they are inclusive of the time dependent settlement component
and only cover the period during first filling (i.e. they do not consider the higher rate of crest settlement
observed that occurs for some CFRDs after the end of first filling). When using Table 3.4 the time
dependent settlement component during the period of first filling should not be added to that from the
Table 3.4.
- 46 -
0.6
Crest settlement attributable to first .

compacted gravels
very high strength rock, compacted
0.5
filling, (% of dam height) .

medium to high strength rock, compacted
dumped rockfill
0.4
0.3
compacted medium to compacted gravels and
high strength rock very high strength rock
0.2
0.1
0
0 20 40 60 80 100 120 140 160 180
Figure 3.25: Crest settlement attributable to first filling (excluding time dependent effects)
There is a high degree of variability in using either method, particularly for the dumped rockfill case studies.
Interpretation of crest settlement attributable to first filling in Figure 3.25 from Figures 3.17 to 3.19 was
imprecise for a number of case studies, particularly those where first filling occurred soon after the end of
construction. For the dumped rockfill case studies the large variation may be attributable to high rates of
leakage and possible collapse settlement of the rockfill during first filling amongst other factors. The high rate
of settlement after the end of first filling for some case studies is an indication that the response of the rockfill
under changes in stress is not elastic. In addition, higher time dependent settlement rates are likely under the
higher stress conditions imposed at full supply level.
Method (i) is the considered the preferable method to use for prediction of crest settlements attributable to first
filling.
(c) Identification of “abnormal” crest settlement behaviour post construction

Figures 3.17 to 3.22 are useful in providing upper and lower bounds of deformation behaviour and identification
of potential “abnormal” deformation behaviour. In the case of Scotts Peak dam (Figure 3.21) the post first
filling deformation behaviour is distinctly different from the typical behaviour. In this case significant leakage
occurred during the latter stages of first filling due to cracking of the asphaltic concrete membrane face (Figure
3.14) resulting in wetting and saturation of the rockfill. Collapse type settlements occurred on wetting of the
medium strength argillite rockfill, which was prone to strength loss on saturation.
3.4.2 Post-Construction Internal Settlement

Post-construction internal settlements (under the embankment crest) were analysed for thirteen of the compacted
CFRD case studies, most of which are Australian CFRD. The settlements were determined from the records of
crest monitoring points and internal settlement gauges and divided into roughly four zones; the upper 25%, the
central upper 25%, the central lower 25% and the bottom 25%. The settlements were analysed from both the
start of monitoring (or first filling) and also from the end of first filling. The findings are summarised in Table
3.7 and indicate:
• For very high strength, compacted rockfills and compacted gravels a very high percentage of the total
settlement from initial monitoring occurs in the bottom 25% of the embankment and a negligible amount
within the upper 25 to 50%.
• For the medium to high strength rockfills this trend is evident for some case studies, whilst for others the
distribution is more evenly spread.
- 47 -
• After the end of first filling the percentage settlement in the bottom 25% of the embankment decreases
(compared to the percent settlement measured from the start of first filling), typically in the order of 10 to
20%.
Table 3.7: Location of internal post-construction vertical settlement in the CFRD
CFRD Name Intact *2 Height Start of No of Percent of Total Crest Settlement Below Crest
rockfill (m) Time Years Top 25% 25 to 50% 50 to 75% Bottom
strength Period *1 Data 25%
Crotty Gravel 83 Start M 2.2 4 6 6 84
Bastyan VH 75 Start FF 6.75 4 18 78
Cethana VH 110 Start ff 14.5 0 51 49
End FF 13.5 (heave) 64 36
Kotmale VH 90 Start M 1.5 (heave) 25 12 63
Murchison VH 94 Start M 11.5 (heave) 27 37 36
End FF 9.25 14 40 30 15
Reece VH 122 Start M 8 4 40 79
End FF 6.25 (heave) 21 49
White Spur H to VH 43 Start M 5.75 40 8 18 33
End FF 4.5 42 4 29 24
Tullabardine H 25 Start M 5 8 52 40
Winneke H 85 Start M 15.5 11 89
End FF 10.5 20 80
Mangrove Creek *3 H 80 Start M 8.3 5 43 26 26
Mackintosh M to H 75 Start M 7.5 (heave) 23 31 46
End FF 6 (heave) 23 32 45
Serpentine M to H 38 Start M 8 25 42 16 17
End FF 4 15 51 13 21
Scotts Peak M 43 Start M 17 15 7 29 49
End FF 14 37 9 15 39
*1 M = monitoring FF = first filling
*2 VH, H and M refer to very high, high and medium intact strength respectively
*3 For Mangrove Creek the central region (25 to 75%) is representative of the random fill zone (Zone 3B).
The reason for the observed distributions of small settlements in the upper part and very high settlements in the
lower part of the embankment can be explained by a combination of several factors:
• On first filling the increase in stress levels (above those previously experienced) will be greatest at the base
of the embankment and lowest at the crest.
• The stress-strain relationship of rockfill, which has a general tendency of decreasing tangent modulus with
increasing vertical stress (Figure 3.3). Therefore, combined with stress changes from first filling,
significantly greater settlements would be expected in the lower portion of the embankment.
• The tendency for time dependent strain rates to be dependent on stress level. Greater time dependent strains
would therefore be anticipated in the deeper sections of the embankment.
In Table 3.7 the medium to high strength rockfills have, in most cases, been sourced from sedimentary rock
types, and the very high strength rockfills from igneous or metamorphic rock types. The sedimentary rock types
typically show a greater loss in strength on wetting and therefore, may account for the greater percent of total
settlement in the upper portion of the embankment for several cases due to rainfall infiltration.
Zoning of the embankment will also have an affect as the results for Mangrove Creek dam indicate. A high
proportion of the post-construction crest settlement under the embankment centreline for Mangrove Creek is
- 48 -
from within the random fill zone (Zone 3B - weathered to fresh sedimentary rockfill) rather than in the better
quality Zone 3A rockfill (fresh siltstones and sandstones).
4.0 DISCUSSION AND RECOMMENDED METHODS FOR PREDICTION

Data on the deformation behaviour of 36 mostly CFRD (35 CFRD and 1 central core earth and rockfill
embankment) has been collected and analysed. The following discussion provides guidelines for prediction of
internal deformations during construction, face slab deformations during first filling and post-construction crest
settlements of CFRD based on the findings of the analyses undertaken. The proposed methods are then
compared to other existing methods.
4.1 GUIDELINES ON DEFORMATION PREDICTION DURING CONSTRUCTION

Empirical methods based on historical performance are useful for prediction of the deformation of rockfill in
CFRD. For relatively standard CFRD designs to be constructed with quarried rockfills the empirical methods
are considered to provide reasonable estimates of deformation, at least comparable, and possibly superior, to
those based on the results of triaxial and oedometer laboratory tests given the difficulties and limitations
associated with particle size, particle size distribution and sample preparation of laboratory samples.
Prediction of rockfill deformation (both numerical and empirical) is based on the quality of representation of the
stress-strain relationship of the rockfill, which is strongly dependent on the rockfill properties, in particular its
intact strength, method of placement and particle size distribution after compaction. The embankment height is
a significant factor as this predicates the level of applied stress within the embankment. Valley shape is a
significant factor for embankments constructed in narrow valleys due to the effects of cross-valley arching and
resultant reduction in applied stresses.
Typical stress-strain relationships of rockfill from field measurements during construction (Figure 3.1) show
that the relationship is non-linear and has a general trend of decreasing secant (and tangent) modulus with
increasing applied stress.
The proposed method of prediction of a stress-strain relationship for the rockfill is to firstly derive a
representative secant modulus at the end of construction, E rc , from Figure 3.7 and then apply correction factors
for layer thickness, stress level and valley shape.
The representative secant modulus at the end of construction, E rc , is for Zone 3A rockfill, placed in layers 0.9
to 1.2 m thick, compacted with 4 to 6 passes of a 10 tonne smooth drum vibratory roller and water added, and
applies to the stresses (Table 3.3):
• 1400 kPa for the very high strength, well-compacted rockfills
• 800 kPa for the medium to high strength, well-compacted rockfills
• 1500 kPa for the well-compacted gravels
To estimate the secant modulus at the end of construction:

• For the proposed rockfill source estimate the UCS within the broad ranges of Table 1.1 and the rockfill
particle size distribution, most likely from compaction trials. For existing dams (or those in the process of
construction) use the actual average particle size distribution.
• Estimate the representative E rc from the equations shown on Figure 3.7 using the estimated D80 value and
the appropriate equation for intact strength (either very high strength or medium to high strength).
- 49 -
• For medium strength well-compacted rockfills apply a correction factor of 0.7 to the E rc value determined
from the equation for medium to high strength rockfills.
• For Zone 3B or other thicker layer rockfills comprising similar rock type (and UCS range) to Zone 3A apply
a correction factor of 0.5 to obtain a representative E rc for rockfill placed in 1.5 to 2.0 m layers. This
correction factor is based on the ratio of E rc (of Zone 3A to Zone 3B) from six field cases.
• To account for the non-linearity of the stress-strain relationship for rockfill, estimates of E rc at stress levels
less than or greater than that representative for the corrected E rc can be estimated from historical records
(Figure 3.2). Although the relationship is likely to be exponential, the data (Figure 3.2) indicates a linear
approximation is reasonable over a defined vertical stress range. The following corrections are suggested:
− For very high strength rockfills apply a linear correction of ±7.5% per 200 kPa to the E rc estimated
from Figure 3.7 from a vertical stress of 1400 kPa (applicable range is 400 to 1600 kPa). Apply positive
corrections for decreasing stresses and negative corrections for increasing stresses.
− For medium to high strength rockfills apply a linear correction of ±6 % per 200 kPa to the E rc
estimated from Figure 3.7 from a vertical stress of 800 kPa (applicable range is 200 to 1200 kPa).
• For one-dimensional analyses appropriate correction factors for stress reduction due to the shape of the
embankment should be considered. These can be estimated from equations and charts such as those
published by Poulos and Davis (1974) and given in Figure 1.2.
• To correct for valley shape either:
− Model the valley and the dam using 3-dimensional numerical analysis, and a stress-strain relationship
for the rockfill based on the E rc versus vertical stress relationship derived above.
− For simplified one-dimensional analyses, to estimate the maximum settlement under the centreline in
the centre of the valley correct the vertical stress profile for embankment shape (refer Section 1.2.2 for
correction charts from Poulos and Davis (1974)) and then correct for valley shape using the appropriate
stress reduction factors given in Table 4.1.
Table 4.1: Approximate stress reduction factors to account for valley shape
Wr/H Ratio Average Abutment Stress Reduction Factor (embankment location)
(river width to Slope Angle Base Mid to Low Mid Upper
height) (degrees) (0 to 20%) (20 to 40%) (40 to 65%) (65% to crest)
0.2 10 to 20 0.93 0.95 0.97 1.0
20 to 30 0.88 0.92 0.96 0.98
30 to 40 0.82 0.88 0.94 0.97
40 to 50 0.74 0.83 0.91 0.96
50 to 60 0.66 0.76 0.86 0.94
60 to 70 0.57 0.69 0.82 0.92
0.5 < 25 1.0 1.0 1.0 1.0
25 to 40 0.93 0.95 0.97 1.0
40 to 50 0.91 0.92 0.95 0.95 – 1.0
50 to 60 0.87 0.88 0.93 0.95 – 1.0
60 to 70 0.83 0.85 0.90 0.95 – 1.0
1.0 All slopes 0.95 - 1.0 0.95 - 1.0 1.0 1.0
Wr = river width H = embankment height
Other relevant issues relating to the estimation of secant modulus and derivation of the correlation between
secant modulus and D80 (Figure 3.7) are:
• Cu, the uniformity coefficient for the particle size distribution curve, is implicitly allowed for in the D80
value. Generally, decreasing Cu is observed for increasing D80.
- 50 -
• For materials placed with larger rollers (i.e. 13 to 15 tonne deadweight vibrating rollers) the data does not
indicate an increase in moduli for the greater compactive effort. This may be due to greater material
breakdown under the heavier rollers, and a resultant reduction of D80, which can then be accounted for
directly in Figure 3.7.
• For weathered rockfills intact strength will be lower than for fresh rock, which will result in a decrease in
secant moduli, but the greater breakdown will give an increase in moduli associated with a reduction in D80.
• The E rc values in Figure 3.7 for each data point have been determined from published internal deformation
records, and generally from the lower 60% of the embankment and within the central portion of this lower
section. The E rc values have also been corrected for cross-valley arching effects using the normalised
vertical stress charts (Figure 3.5). It is recognised that these may not be a true representation of the actual
stresses due to assumptions made in the modelling process (refer Section 3.2). However, errors associated
with the assumptions in modelling are minimised by applying the same assumptions for prediction as is
proposed in the above.
• If testing on the proposed rockfill material indicates a significant reduction in UCS on wetting is likely and
only limited water has been, or is proposed to be, used in construction, then this strength reduction should
be taken into consideration. We do not have sufficient data to advise on how this should be done; however,
the rockfill is likely to be more susceptible to collapse settlements on wetting due to rainfall and flooding.
Limitations of the data and application of the proposed methodology include:

• For well-compacted gravels only a limited number of cases were available, insufficient to undertake
analysis. These materials generally have significantly greater secant moduli at any given stress level in
comparison to quarried, very high strength rockfill. Where the rockfill is of finer size (e.g. Crotty Zone 3A)
the use of large scale laboratory test that virtually encompass the field particle size distribution would
provide suitable estimates of moduli provided the layering and density in the field are reflected in the
laboratory testing.
• For the data presented in Figure 3.1 (vertical stress versus vertical strain) the stress-strain curves do not pass
through the origin. This is because zero strain is assumed at a vertical stress equivalent to the mid-height
between hydrostatic settlement gauges (HSGs). Zero strain is generally at vertical stresses in the range 50
to 200 kPa (depending on the vertical distance between HSGs).
• During the actual embankment construction monitoring should be undertaken as a check on the pre-
construction estimation of deformation. Deformation estimates can be reassessed, if required, as
construction proceeds using the stress-strain relationship/s extrapolated from the actual field records.
For estimating deformations due to embankment raising, it is recommended that the modulus be obtained from
monitoring during the earlier construction phase if that is available. If not, the methods outlined above can be
used to estimate the secant modulus versus vertical stress relationship. Simplistic estimates of deformation
could be obtained by modelling the embankment in a series of layers of varying tangent moduli (estimated from
the secant moduli vertical stress relationship). For a more rigorous analysis numerical modelling would be
appropriate.
For dumped rockfill there is not sufficient data from which to give guidelines on moduli during construction.
The data records from the limited number of cases is summarised in Section 3.2.6.
4.2 GUIDELINES ON DEFORMATION PREDICTION DURING FIRST FILLING

As discussed in Section 3.3, the modulus on first filling, E rf , estimated using the method by Fitzpatrick et al
(1985) is an artefact of the method used to calculate it and is not a true modulus of the rockfill, as Fitzpatrick et
al (1985) recognised. The high values of E rf compared to E rc are to be expected given the stress paths on first
- 51 -
filling in the upstream slope (Figure 3.9b and Figure 3.10b). The stress paths show an unloading then reloading
of deviatoric stress whilst compressive stresses continue to increase with increasing reservoir level.
The predictive method of face slab deformation comprises estimation of the ratio E rf E rc from Figure 3.8 and
therefore E rf given that E rc is already known or estimated, and use of the Fitzgerald et al (1985) equation
(Figure 1.1b) for prediction of face slab deflection. The method is considered applicable to the region of the
face slab between 20% and 60% of the embankment height, the general region of maximum face slab deflection
on first filling.
Guidelines for use of Figure 3.8 for estimation of E rf are:

• Estimate the ratio E rf E rc based on the embankment height and upstream slope angle. Trendlines are
given for upstream slopes of 1.3 – 1.4H to 1V, and 1.5H to 1V, which cover most CFRD designs.
• Estimate E rc , the representative secant moduli at end of construction, of the Zone 3A rockfill from the
methods outlined in Section 4.1. The E rc value should be adjusted for vertical stress such that it is
representative of the lower 50% of the rockfill in the central region of the embankment. The E rc values
used in the derivation of Figure 3.8 were not corrected for arching effects due to valley shape because valley
shape is potentially likely to affect both E rc and E rf , and would therefore be taken into consideration in
the E rf E rc ratio. Hence, E rc estimates derived from Figure 3.7 (corrected for valley shape effects) must
be un-corrected.
• E rf can then be estimated by multiplication of the E rf E rc ratio by E rc (not corrected for valley shape).
In using Figure 3.8 for estimating the E rf E rc ratio consideration should be given to the qualifications to this
figure as discussed in Section 3.3. The trendline for upstream slope angles of 1.5H to 1V is based on a limited
number of cases and it should therefore be recognised that the estimated E rf E rc ratio will be very
approximate.
The zoning geometry of the embankment has a significant effect on the deformation behaviour of the face slab
on first filling (Section 3.3) and the method of prediction of the face slab deformation is dependent on this
geometry and consideration of the differential moduli between the main rockfill zones. The empirical predictive
method developed (based on Figure 3.8) is appropriate for CFRD with relatively simple zoning geometries
comprising a significant Zone 3A (at least 50 to 60% of the main rockfill) and/or where large variations in
rockfill moduli are not likely between rockfill zones. For embankments with complex geometries (particularly
in the upstream half of the embankment) and/or large moduli variations between the rockfill zones finite
element analyses should be considered. The detailed analysis would be justified considering the potential for
face slab cracking as a result of possible development of tensile stresses.
4.3 GUIDELINES ON DEFORMATION PREDICTION POST-CONSTRUCTION

Guidelines for the prediction of post-construction crest settlements are given in Section 3.4.1.4. The methods
proposed are considered to provide an improvement on current available methods of crest settlement prediction
for CFRD.
- 52 -
The method of prediction comprises separate estimation of the time dependent deformation component and the
crest settlement attributable to first filling, and then summation to estimate the total post construction settlement.
In summary:
• The time dependent crest settlement rate (as a percentage of the embankment height per log cycle of time)
for compacted rockfill can be estimated from Figure 3.24. The estimate is based on the embankment height
and intact strength of the compacted rockfill under the embankment centreline. The zero time, to, has been
taken at the end of the main rockfill construction.
• For dumped rockfills, estimates of the long-term crest settlement rate (as a percentage of the embankment
height per log cycle of time) can be derived from either Figure 3.17 or Table 3.6 based on the time period
after the end of main rockfill construction.
• Two methods are provided for prediction of the crest settlement during or attributable with first filling.
Method (i) or Figure 3.25, the preferred method, is an estimation of the crest settlement attributable to first
filling based on the embankment height, intact rock strength and placement method. The degree of
variability is relatively large, particularly for the dumped rockfill cases, due to the number of factors
influencing the deformation behaviour.
• The time dependent deformations and deformations on first filling occur concurrently.
The plots of the historical records of case studies (Figures 3.17 to 3.22) are useful in providing upper and lower
bounds of deformation behaviour and identification of potential “abnormal” deformation behaviour based on the
method of placement and intact strength of compacted rockfill. As discussed in Section 3.4.1 the plots for
settlements after the end of main rockfill construction (Figures 3.17 to 3.19) should be used with caution due to
the influences of first filling and initial time of reading.
5.0 CONCLUSIONS
The proposed empirical predictive methods for estimation of the deformation of rockfill (summarised in Section
4.0), specifically targeted at the deformation of CFRD, are considered to provide improvements over currently
available empirical predictive methods because of the larger good quality data set available, and consideration
of significant factors which influence the deformation. The methods are based on historical records of
embankment performance and allow for estimation/prediction of the vertical deformation during construction,
deformation of the face slab and post-construction crest settlement. Where possible the significant factors
affecting the deformation behaviour of rockfill, evaluated from the analysis of historical records, were
incorporated into the empirical predictive methods.
The results of the historical data presented on the stress-strain behaviour of rockfill measured in the field is
considered to have advantages over rockfill properties estimated by laboratory test methods. Laboratory testing
of rockfill has the disadvantages of limitation on particle size and placement methods, both of which are shown
to have a significant effect on the prediction of the rockfill modulus. Methods, although relatively crude, are
proposed for estimation of the stress-strain relationship of a compacted rockfill sample and could therefore be
used as a basis for the constitutive model in finite element analyses.
The proposed empirical predictive methods of deformation are generally applicable to the more typical CFRD
designs. For relatively complex geometries, particularly those that incorporate high differential moduli between
rockfill zones (e.g. compacted gravels and rockfills such as Aguamilpa dam) numerical methods are considered
more appropriate. They are considered particularly appropriate for modelling potential tensile stress
development in the concrete face that could result in cracking.
- 53 -
6.0 ACKNOWLEDGEMENTS
This work forms part of a research project on the pre and post failure deformation behaviour of soil slopes,
being undertaken within the School of Civil and Environmental Engineering at the University of New South
Wales, Australia. The support of the Australian Research Council and industry sponsors of the project is
acknowledged:
• New South Wales Department of Land and Water Conservation
• Snowy Mountains Engineering Corporation
• Goulburn Murray Water
• Australian Water Technologies
• United States Bureau of Reclamation
• Dam Safety Committee of New South Wales
• ACTEW Corporation
• Queensland Department of Natural Resources
• Snowy Mountains Hydro-Electric Authority
• South Australian Water Corporation
• Water Authority of Western Australia
• Pells Sullivan Meynink Pty Ltd
• Roads and Traffic Authority, New South Wales
• New South Wales Dept of Public Works and Services
• Queensland Department of Main Roads
The support of Hydro-Electric Commission, Tasmania and Melbourne Water in providing case study data is
also acknowledged.
7.0 REFERENCES
Alonso, E.E. and Oldecop, L.A. (2000) Fundamentals of rockfill collapse. Proceedings of the Asian Conference on
Unsaturated Soils (Unsat-Asia 2000), (Rahardjo, Toll and Leong ed.) Singapore, Balkema, Rotterdam. pp. 3-13.
Amaya, F. and Marulanda, A. (1985) Golillas dam - design, construction and performance. Proceedings of the
Symposium on Concrete Face Rockfill Dams - Design, Construction and Performance, (Cooke and Sherard ed.)
Detroit, Michigan, ASCE New York. pp. 98-120.
Amaya, F. and Marulanda, A. (2000) Columbian experience in the design and construction of concrete face rockfill
dams. In Concrete Face Rockfill Dams, J. Barry Cooke Volume (Mori, Sobrinho, Dijkstra, Guocheng and
Borgatti ed.), Beijing, pp. 89-115.
Aphaiphuminart, S., Chanpayom, O., Mahasandana, T., Bhucharoen, V. and Pinrode, J. (1988) Design, construction
and performance - Khao Laem dam. Proceedings of the 16th International Congress on Large Dams, San
Francisco, ICOLD. pp. 95-114 (Q.61 R.6).
Australian Standard AS 1726-1993 (1993) Geotechnical Site Investigations. Council of Standards Australia,
Committee CE/15.
Baker, R. (1972) Report of inspections and tests for Scotts Peak dam. Laboratory Report 4164-1, HEC Tasmania,
Civil Engineering Laboratories.
Baumann, P. (1939) Discussion on Galloway paper: The design of rock-fill dams. Transactions, ASCE, Vol. 104, pp.
39-40.
Baumann, P. (1958) Rockfill dams: Cogswell and San Gabriel dams. A.S.C.E., Journal of the Power Division, Vol.
84, (No. PO3), pp. 1687-1 to 33.
Baumann, P. (1964) Limit height criteria for loose-dumped rockfill dams. Proceedings of the 8th International
Congress on Large Dams, Edinburgh, ICOLD. Vol. 3, pp. 781- (Q.31 R.13).
Blinder, S., Toniatti, N.B. and Ribeiro, M.S. (1992) Construction progress and behavioural monitoring of Segredo
dam. Water Power & Dam Construction (April), pp. 18-21.
Bodtman, W.L. and Wyatt, J.D. (1985) Design and performance of Shiroro rockfill dam. Proceedings of the
Boucaut, W.R.P. and Beal, J.C. (1979) The engineering geology of the Little Para dam site. Report Bk. 79/2,
Department of Mines and Energy, South Australia.
- 54 -
Bowling, A.J. (1978) Mackintosh dam rockfill control testing. Report 4301-18, HEC Tasmania, Civil Engineering
Laboratories.
Bowling, A.J. (1979) Tullabardine dam rockfill control testing. Report 4333-1 and -2, HEC Tasmania, Civil
Engineering Laboratories.
Bowling, A.J. (1981) Laboratory investigations into the suitability of rockfill for concrete faced rockfill dams.
ANCOLD Bulletin, Vol. 59, pp. 21-29.
Bowling, A.J. (1981-82) Bastyan dam - rockfill control testing. Report 4395-1 to -6, HEC Tasmania, Civil
Casinader, R. (1987) Discussion: Kotmale dam and observations on CFRD. A.S.C.E., Journal of Geotechnical
Engineering, Vol. 113 (No. 10), pp. 1198-1200.
Casinader, R. and Watt, R.E. (1985) Concrete face rockfill dams of the Winneke project. Proceedings of the
Clements, R.P. (1984) Post-construction deformation of rockfill dams. A.S.C.E., Journal of Geotechnical
Engineering, Vol. 110 (No. 7), pp. 821-840.
Cole, B.A. (1974) Gordon River power development - Stage 1, Scotts Peak Dam, design report. Report CDR 264,
HEC Tasmania, Civil Design Division.
Cole, B.A. (1981) Rockfill experiences at Mackintosh, Murchison and Lower Pieman dams. ANCOLD Bulletin, Vol.
59, pp. 32-35.
Cole, B.A. and Fone, P.J.E. (1979) Repair of Scotts Peak dam, Tasmania. Proceedings of the 13th International
Congress on Large Dams, New Delhi, ICOLD. pp. 211-231 (Q.49 R.15).
Cooke, J.B. (1958) Rockfill dams: Wishon and Courtright concrete face dams. A.S.C.E., Journal of the Power
Division, Vol. 84 (PO3), pp. 1746-1 to 33.
Cooke, J.B. (1984) Progress in rockfill dams (18th Terzaghi lecture). Journal of Geotechnical Engineering, ASCE,
Vol 110 (10), pp. 1383-1414.
Cooke, J.B. (1993) Rockfill and the rockfill dam. Proceedings of the International Symposium on High Earth-Rockfill
Dams, (Jiang, Zhang & Qin ed.) Beijing. pp 1-24.
Cooke, J.B. (1997) The concrete face rockfill dam. Non-Soil Water Barriers for Embankment Dams, 17th Annual
USCOLD Lecture Series, San Diego, California, USCOLD. pp. 117-132.
Cooke, J.B. (1999) The development of today's CFRD dam. Proceedings of the Second Symposium on Concrete Face
Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams. pp. 3-11.
Cooke, J.B. and Sherard, J.L. (1987) Concrete-face rockfill dam: II. Design. ASCE, Journal of Geotechnical
Engineering, Vol 113 (10), pp. 1113-1133.
Cooke, J.B. and Strassburger, A.G. (1988) Section 6: Rockfill Dams. In Development of Dam Engineering in the
United States (Kollgaard & Chadwick ed.), Permagon Press, pp. 885-1030.
Cribben, N.E. (1990) K.R.P.D., Crotty Dam, Field dry density testing on Zone 3A material. Report 4652-1, HEC
Tasmania, Engineering and Scientific Services Department.
Davies, T.J.G. (1993) King River power development, Crotty Dam, Record Design Report. Report CDR 614, HEC
Tasmania, Consulting Business Unit.
Davies, T.J.G., Smith, O. and Hancock, D.J. (1995) Reece dam, Surveillance Report No. 2 (1988 - 1995). Report
ENE-0010-SF-007, HEC Tasmania, Civil and Water Resources Engineering Group.
Duncan, J.M. (1992) State of the art: Static stability and deformation analysis. ASCE Geotechnical Special
Publication No. 31, (Seed and Boulanger ed.) Berkeley, California, Vol. 1, pp. 222-266.
Duncan, J.M. (1996) State of the art: Limit equilibrium and finite-element analysis of slopes. A.S.C.E., Journal of
Geotechnical Engineering, Vol 122 (7), pp. 577-595.
Eigenheer, L.P.Q.T. and Souza, R.J.B. (1999) Xingo concrete face rockfill dam. Proceedings of the Second
Symposium on Concrete Face Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams. pp. 271-284.
Engineering and Water Supply Department (1981) Kangaroo Creek dam. Historical account of Construction and
Operations. Ref. 81/41, South Australian Government.
Fell, R., Hungr, O., Leroueil, S. and Riemer, W. (2000) Keynote lecture - Geotechnical engineering of the stability of
natural slopes, and cuts and fills in soil. Proceedings of the International Conference on Geotechnical and
Geological Engineering (GeoEng2000), Melbourne, Technomic, Lancaster. Vol. 1, pp 21-120.
Fell, R., MacGregor, J.P. and Stapleton, D. (1992) Geotechnical Engineering of Embankment Dams. Balkema,
Rotterdam, pp. 675.
Fitzpatrick, M.D., Cole, B.A., Kinstler, F.L. and Knoop, B.P. (1985) Design of concrete-faced rockfill dams.
Proceedings of the Symposium on Concrete Face Rockfill Dams - Design, Construction and Performance,
(Cooke and Sherard ed.) Detroit, Michigan, ASCE New York. pp. 410-434.
Fitzpatrick, M.D., Liggins, T.B. and Barnett, R.H.W. (1982) Ten years of surveillance of Cethana dam. Proceedings
of the 14th International Congress on Large Dams, Rio de Janiero, ICOLD. pp. 847-866 (Q.52 R.51).
Fitzpatrick, M.D., Liggins, T.B., Lack, L.J. and Knoop, B.P. (1973) Instrumentation and performance of Cethana
dam. Proceedings of the 11th International Congress on Large Dams, Madrid, ICOLD. pp. 145-164 (Q.42 R.9).
- 55 -
Forza, S. and Hancock, D. (1995) Tullabardine dam, Surveillance Report No. 2 (1986 - 1994). Report ENE-0010-
SF-002, HEC Tasmania, Civil and Water Resources Engineering.
Forza, S.J. and Hancock, D.J. (1993) Serpentine dam, Surveillance Report No. 2 (1980 - 1993). Report CDR 602,
HEC Tasmania, Consulting Business Unit.
Forza, S.J. and Hancock, D.J. (1995) Cethana dam, Surveillance Report No. 2 (1982 - 1995). Report ENE-0010-SF-
004, HEC Tasmania, Civil and Water Resources Engineering Group.
Foster, M.A., Fell, R. and Spannagle, M. (1998) Analysis of embankment dam incidents. Uniciv Report R-374, The
University of New South Wales, School of Civil and Environmental Engineering.
Galloway, J.D. (1939) The design of rock-fill dams. Transactions, ASCE, Vol. 104, pp. 1-24.
Gerke, D., Forza, S.J. and Hancock, D.J. (1995) Murchison dam, Surveillance Report No. 2 (1985 - 1995). Report
ENE-0010-SF-005, HEC Tasmania, Civil and Water Resources Engineering Group.
Gerke, D., Quinlan, P. and Hancock, D.J. (1993) Scotts Peak dam, Surveillance Report No. 2 (1985 - 1993). Report
CDR 570, HEC Tasmania, Consulting Business Unit.
Giudici, S., Herweynen, R. and Quinlan, P. (2000) HEC experience in concrete faced rockfill dams - past, present and
future. Proceedings of the International Symposium on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp. 29-
46.
Gonzales, F.V. and Mena, E.S. (1997) Aguamilpa dam behaviour. Non-Soil Water Barriers for Embankment Dams,
17th Annual USCOLD Lecture Series, San Diego, California, USCOLD. pp. 133-147.
Good, R.J. (1976) Kangaroo Creek dam. Use of a weak schist as rockfill for a concrete faced rockfill dam.
Proceedings of the 12th International Congress on Large Dams, Mexico, ICOLD. pp. 645-665 (Q.44 R.33).
Good, R.J. (1981) Behaviour of concrete faced rolled rockfill dams in South Australia. ANCOLD Bulletin, Vol. 59,
pp. 45-56.
Good, R.J., Bain, D.L.W. and Parsons, A.M. (1985) Weak rock in two rockfill dams. Proceedings of the Symposium
on Concrete Face Rockfill Dams - Design, Construction and Performance, (Cooke and Sherard ed.) Detroit,
Michigan, ASCE New York. pp. 40-72.
Gosschalk, E.M. and Kulasinghe, A.N.S. (1985) Kotmale dam and observations on CFRD. Proceedings of the
Gosschalk, E.M. and Kulasinghe, A.N.S. (1987) Closure: Kotmale dam and observations on CFRD. A.S.C.E., Journal
of Geotechnical Engineering, Vol. 113 (10), pp. 1202-1208.
Hacelas, J.E., Ramirez, C.A. and Regalado, G. (1985) Construction and performance of Salvajina dam. Proceedings
of the Symposium on Concrete Face Rockfill Dams - Design, Construction and Performance, (Cooke and
Sherard ed.) Detroit, Michigan, ASCE New York. pp. 286-315.
He, G. (2000) Technical study on crest overflow of concrete face rockfill dams. Proceedings of the International
Symposium on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp. 283-291.
Heinrichs, P.W. (1996) Mangrove Creek dam embankment behaviour and surveillance. ANCOLD Bulletin, No. 102
(April), pp. 22-34.
Howson, G.W. (1939) Discussion on paper by Galloway: The design of rock-fill dams. Transactions, ASCE, Vol.
104, pp. 42-45.
Huang, Y., Peng, Z., Si, H. and Li, G. (1993) Monitoring and performance of Xibeikou CFRD. Proceedings of the
International Symposium on High Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing, Vol. 1, pp. 475-482.
Hunter, G. and Fell, R. (2002, in press) Post failure deformation of embankment dams and cuts in highly plastic
clays. Uniciv Report. The University of New South Wales, School of Civil and Environmental Engineering.
Hydro Electric Commission (1991a) Dams Surveillance Report, Bastyan Dam, 1986 - 1991 (No. 2). Report CDR
560, HEC Tasmania, Safety of Dams Unit.
Hydro Electric Commission (1991b) Dams Surveillance Report, Mackintosh Dam, 1985 - 1991 (No. 2). Report CDR
559, HEC Tasmania, Safety of Dams Unit.
Hydro Electric Commission Tasmania (1987) Anthony power development, specification for excavation and rockfill
for White Spur dam. Specification C.E. 2023.
ICOLD (1989) Rockfill dams with concrete facing. State of the art. Bulletin 70, ICOLD.
Jiyuan, S., Zhu, B. and Liang, C. (2000) Characteristic and experience of the design, construction and performance of
TSQ-1 concrete faced rockfill dam. Proceedings of the International Symposium on Concrete Faced Rockfill
Dams, Beijing, ICOLD. pp. 97-105.
Justo, J.L. (1991) Collapse: Its importance, fundamentals and modelling. In Advances in Rockfill Structures
(Maranha das Neves ed.), Kluwer Academic Publishers, Netherlands, pp. 97-152.
Knoop, B.P. (1982a) Performance design report on Murchison dam. Report CDR 433, HEC Tasmania, Civil Design
Section.
Knoop, B.P. (1982b) Report on the performance of Mackintosh Dam. Report CDR 421, HEC Tasmania.
Knoop, B.P. and Lack, L.J. (1985) Instrumentation and performance of concrete faced rockfill dams in the Pieman
River power development. Proceedings of the 15th International Congress on Large Dams, Lausanne, ICOLD.
pp. 1103-1120 (Q.56 R.58).
- 56 -
Kovacevic, N. (1994) Numerical analyses of rockfill dams, cut slopes and road embankments. Ph.D. Thesis, Imperial
College of Science, Technology and Medicine, London, Faculty of Engineering.
Kulasinghe, A.N.S. and Tandon, G., N, (1993) Technical and behavioural aspects of Kotmale dam. Proceedings of
the International Symposium on High Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing, pp. 233-244.
Lambe, T.W. (1973) Predictions in soil engineering. Geotechnique, Vol. 23 (2), pp. 149-202.
Li, S. (1991) Pieman power development, Reece dam, civil design report (record). Report CDR 551, HEC Tasmania,
Civil Design Section.
Li, S.S.Y., Giudici, S. and Tindall, W.F. (1993) Design of Crotty dam and spillway. Proceedings of the International
Symposium on High Earth-Rockfill Dams, Vol. 1, pp. 263-271.
Li, S.S.Y., Paterson, S.J. and Kinstler, F.L. (1991) A concrete faced rockfill dam constructed on a deeply weathered
foundation. Proceedings of the 17th International Congress on Large Dams, Vienna, ICOLD. pp. 1601-1612
(Q.66 R.85).
Liggins, T. (1971) Mersey-Forth power development, Cethana dam, design report. Report CDR 218, HEC
Tasmania, Civil Design Division.
Macedo, G.G. (1999) Concrete face behaviour of Aguamilpa dam. Proceedings of the Second Symposium on
Concrete Face Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams. pp. 211-222.
Macedo, G.G., Castro, A.J. and Montanez, C.L. (2000) Behaviour of Aguamilpa dam. In Concrete Face Rockfill
Dams, J. Barry Cooke Volume (Mori, Sobrinho, Dijkstra, Guocheng and Borgatti ed.), Beijing, pp. 117-151.
Mackenzie, P.R. and McDonald, L.A. (1985) Mangrove Creek dam: Use of soft rock for rockfill. Proceedings of the
Mahasandana, T. and Mahatraradol, B. (1985) Monitoring systems of Khao Laem dam. Proceedings of the 15th
International Congress on Large Dams, Lausanne, ICOLD. pp. 1- (Q.56 R.1).
Marachi, N.D., Chan, C.K., Seed, H.B. and Duncan, J.M. (1969) Strength and deformation characteristics of rockfill
materials. Report TE-69-5, University of California, Department of Civil Engineering.
Marsal, R.J. and Ramirez de Arellano, L. (1967) Performance of El Infiernillo dam, 1963-1966. A.S.C.E., Journal of
the Soil Mechanics and Foundations Division, Vol. 93 (SM4), pp. 265-298.
Marsal, R.L. (1973) Mechanical properties of rockfill. In Embankment Dam Engineering (Casagrande Volume)
(Hirschfeld and Poulos ed.), John Wiley and Sons, pp. 109-200.
Marulanda, A. and Pinto, N.L. (2000) Recent experience on design, construction, and performance of CFRD dams. In
Concrete Face Rockfill Dams, J. Barry Cooke Volume (Mori, Sobrinho, Dijkstra, Guocheng and Borgatti ed.),
Beijing, pp. 279-315.
Materon, B. (1985a) Alto Anchicaya dam - ten years performance. Proceedings of the Symposium on Concrete Face
Rockfill Dams - Design, Construction and Performance, (Cooke and Sherard ed.) Detroit, Michigan, ASCE New
York. pp. 73-87.
Materon, B. (1985b) Construction of Foz do Areia dam. Proceedings of the Symposium on Concrete Face Rockfill
Dams - Design, Construction and Performance, (Cooke and Sherard ed.) Detroit, Michigan, ASCE New York.
pp. 192-207.
Melbourne and Metropolitan Board of Works (1975) Sugarloaf reservoir project, design report. Water Supply
Division.
Melbourne and Metropolitan Board of Works (1981) Notes on rock materials used at Winneke dam. ANCOLD
Bulletin, Vol. 59, pp. 57-61.
Melbourne and Metropolitan Board of Works (1995) Sugarloaf reservoir, monitoring report (7 June 1995). Water
Assets Section.
Montanez, L.E., Hacelas, J.E. and Castro, A.J. (1993) Design of Aguamilpa dam. Proceedings of the International
Symposium on of High Earth Rockfill Dams, Beijing, ICOLD. pp. 337-363.
Mori, R.T. (1999) Deformation and cracks in concrete face rockfill dams. Proceedings of the Second Symposium on
Concrete Face Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams. pp. 49-61.
Mori, R.T. and Pinto, N.L.S. (1988) Analysis of deformations in concrete face rockfill dams to improve face
movement prediction. Proceedings of the 16th International Congress on Large Dams, San Francisco, ICOLD.
pp. 27-34 (Q.61 R.2).
Morris, S.B. (1939) Discussion on Galloway paper: The design of rock-fill dams. Transactions, ASCE, Vol. 104, pp.
35-38.
Morse, A.R. (1995) White Spur dam, Surveillance Report No. 1 (1986 - 1995). Report ENE-0010-SF-013, HEC
Tasmania, Civil and Water Resources Engineering Group.
Morse, A.R. and Ward, M.J. (1989) Anthony power development, White Spur Dam, Civil Design Report (record).
Report CDR 523, HEC Tasmania, Civil Design Section.
Naylor, D.J., Maranha, J.R., Maranha das Neves, E. and Veiga Pinto, A.A. (1997) A back-analysis of Beliche dam.
Geotechnique, Vol. 47 (2), pp. 221-233.
Naylor, D.J., Tong, S.L. and Shahkarami, A.A. (1989) Numerical modelling of saturation shrinkage. Proc. Numerical
models in Geomechanics NUMOG III, (Pietruszczak & Pande ed.) Amsterdam, Elsevier. pp. 636-648.
- 57 -
Nobari, E.S. and Duncan, J.M. (1972) Effect of reservoir filling on stresses and movements in earth and rockfill
dams. Report TE-72-1, University of California, Department of Civil Engineering.
Parkin, A.K. (1971) Application of rate analysis to settlement problems involving creep. Proceedings of the First
Australia New Zealand Conference on Geomechanics, Melbourne, Vol. 1, pp. 138-143.
Parkin, A.K. (1977) The compression of rockfill. Australian Geomechanics Journal, Vol. G7, pp. 33-39.
Partyka, G. and Bowling, A.J. (1984) P.R.P.D., Lower Pieman dam, Rockfill and filter material control testing.
Report 4413-9, HEC Tasmania, Civil Engineering Laboratories.
Peng, Z. (2000) Analysis of the deformation of Xibeikou CFRD in eight years of operation. Proceedings of the
International Symposium on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp. 555-564.
Peterson, O.W. (1939) Discussion on Galloway paper: The design of rock-fill dams. Transactions, ASCE, Vol. 104,
pp. 40-41.
Pinto, N.L.S., Blinder, S. and Toniatti, N.B. (1993) Foz do Areia and Segredo CFRD dams - 12 years evolution.
Proceedings of the International Symposium on High Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing,
Vol. 1, pp. 381-398.
Pinto, N.L.S. and Marques Filho, P.L. (1998) Estimating the maximum face slab deflection in CFRDs. Hydropower
& Dams, No. 6.
Pinto, N.L.S., Marques Filho, P.L. and Maurer, E. (1985a) Foz do Areia dam - design, construction and behaviour.
Proceedings of the Symposium on Concrete Face Rockfill Dams - Design, Construction and Performance,
(Cooke and Sherard ed.) Detroit, Michigan, ASCE New York. pp. 173-191.
Pinto, N.L.S., Marques Filho, P.L. and Maurer, E. (1985b) Segredo dam - basic design aspects. Proceedings of the
Pinto, N.L.S., Materon, B. and Marques Filho, P.L. (1982) Design and performance of Foz do Areia concrete
membrane as related to basalt properties. Proceedings of the 14th International Congress on Large dams, Rio de
Janiero, ICOLD. pp. 873-906 (Q.55 R.51).
Pinto, N.L.S. and Filho, P.L.M. (1985) Discussion: Post-construction deformation of rockfill dams. A.S.C.E., Journal
of Geotechnical Engineering, Vol. 111 (12), pp. 1472-1475.
Poulos, H.G., Booker, J.R. and Ring, G.J. (1972) Simplified calculations of embankment deformations. Soils and
Foundations, Tokyo, Vol 12 (4), pp. 1-17.
Poulos, H.G. and Davis, E.A. (1974) Elastic Solutions for Soil and Rock Mechanics. Wiley, New York.
Public Works and Services Department NSW (1998) Mangrove Creek dam - Post construction embankment
behaviour. Report DC98072, DPWS NSW, Dams and Civil Section.
Public Works Department N.S.W. (1990) Gosford Wyong water supply, Mangrove Creek dam monitoring survey
(unpublished).
Quinlan, P. (1993) King River power development, Crotty Dam, Performance Report. Report CDR 610, HEC
Tasmania, Consulting Business Unit.
Regalado, G., Materon, B., Ortega, J.W. and Vargas, J. (1982) Alto Anchicaya concrete face rockfill dam behaviour
of the concrete face membrane. Proceedings of the 14th International Congress on Large Dams, Rio de Janiero,
ICOLD. Vol. 4, pp. 517-535 (Q.55 R.30).
Regan, P.J. (1997) Performance of concrete-faced rockfill dams of the Pacific Gas & Electric Company. Non-Soil
Water Barriers for Embankment Dams, Proceedings of the 17th Annual USCOLD Lecture Series, San Diego,
California, USCOLD. pp. 149-162.
Regan, W.M. (1980) Winneke Reservoir, main dam geological report on construction. Engineering Geology Section
Report 81/489, Melbourne and Metropolitan Board of Works.
Robinson, P. (1979) Murchison dam - Control testing of rockfill placing. Report 4352-1, HEC Tasmania, Civil
Saboya, F., Barbosa, R. and Vasconcelos, A. (2000) The influence of the left abutment geometry on the behaviour of
Xingo rockfill dam. Proceedings of the International Symposium on Concrete Faced Rockfill Dams, Beijing,
ICOLD. pp. 565-572.
Saboya, F.J. and Byrne, P.M. (1993) Parameters for stress and deformation analysis of rockfill dams. Canadian
Geotechnical Journal, Vol. 30, pp. 690-701.
Schmidt, L.A.J. (1958) Rockfill dams: Performance and maintenance of Dix River dam. A.S.C.E., Journal of the
Power Division, Vol. 84 (PO3), pp. 1683-1 to 29.
Sherard, J.L. and Cooke, J.B. (1987) Concrete-face rockfill dam: I. Assessment. Journal of Geotechnical
Engineering, ASCE, Vol 113 (10), pp. 1096-1112.
Sierra, J.M., Ramirez, C.A. and Hacelas, J.E. (1985) Design features of Salvajina dam. Proceedings of the
Silveira, J.F. and Sardinha, A.E. (1999) Behaviour of Ita CFRD at the end of the construction period. Proceedings of
the Second Symposium on Concrete Face Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams.
pp. 37-48.
- 58 -
Sinclair Knight Merz (1995) Kangaroo Creek dam - Report on 1st stage of safety evaluation. For Engineering and
Water Supply Department of South Australia.
Singh, A. and Mitchell, J.K. (1968) General stress-strain-time function for soil. Journal of the Soil Mechanics and
Foundations Division, ASCE, Vol 94 (SM1), pp. 21-46.
Sobrinho, J.A., Sadinha, A.E., Albertoni, S.C. and Dijkstra, H.H. (2000) Development aspects of CFRD in Brazil. In
Concrete Face Rockfill Dams, J. Barry Cooke Volume (Mori, Sobrinho, Dijkstra, Guocheng and Borgatti ed.),
Beijing, pp. 153-175.
Sobrinho, J.A., Sardinha, A.E. and Fernandes, A.M. (1999) Ita dam - design and construction. Proceedings of the
Second Symposium on Concrete Face Rockfill Dams, Florianopolis, Brazil, Brazilian Committee on Dams. pp.
377-390.
South Australian Water Corporation (1995) Dam Safety Committee inspection of Kangaroo Creek dam.
Souza, R.J.B., Cavalcanti, A.J.C.T., Silva, S.A. and Silveira, J.F. (1999) Xingo concrete face rockfill dam behaviour
of the dam on the left abutment. Proceedings of the Second Symposium on Concrete Face Rockfill Dams,
Florianopolis, Brazil, Brazilian Committee on Dams. pp. 143-157.
Sowers, G.F., Williams, R.C. and Wallace, T.S. (1965) Compressibility of broken rock and the settlement of rockfills.
Proceedings of the 6th International Conference on Soil Mechanics and Foundation Engineering, Montreal,
Vol. 2, pp. 561-565.
Soydemir, C. and Kjaernsli, B. (1979) deformations of membrane-faced rockfill dams. Proceedings, 7th European
Conference on Soil Mechanics and Foundation Engineering, Brighton, England, Vol. 3, pp. 281-284.
Steele, I.C. and Cooke, J.B. (1958) Rockfill dams: Salt Springs and Lower Bear River concrete face dams. A.S.C.E.,
Journal of the Power Division, Vol. 84 (PO4), pp. 1737-1 to 43.
Steele, I.C. and Dreyer, W. (1939) Discussion on Galloway paper: The design of rock-fill dams. Transactions, ASCE,
Vol. 104, pp. 53-63.
Terzaghi, K. (1960) Discussion on 1958 paper by Steele & Cooke: Rockfill dams: Salt springs and Lower Bear River
concrete face dams. ASCE, Transactions, Vol. 125, (Part II), pp. 139-148.
Vasconcelos, A.A. and Eigenheer, L.P. (1985) The Xingo rockfill dam. Proceedings of the Symposium on Concrete
Face Rockfill Dams - Design, Construction and Performance, (Cooke and Sherard ed.) Detroit, Michigan,
ASCE New York. pp. 559-565.
Wang, P., Shen, Y. and Wang, Y. (1993) Design and construction of Xibeikou concrete-faced rockfill dam.
Proceedings of the International Symposium on High Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing,
Vol. 1, pp. 418-429.
Wang, Y., Chen, J. and Shen, Y. (1988) Xibeikou concrete face rockfill dam. Proceedings of the 16th International
Congress on Large Dams, San Francisco, ICOLD. pp. 1075-1090 (Q.61 R.57).
Watakeekul, S., Roberts, G.J. and Coles, A.J. (1985) Khao Laem - a concrete face rockfill dam on karst. Proceedings
of the Symposium on Concrete Face Rockfill Dams - Design, Construction and Performance, (Cooke and
Sherard ed.) Detroit, Michigan, ASCE New York. pp. 336-361.
Wilkins, J.K., Mitchell, W.R., Fitzpatrick, M.D. and Liggins, T. (1973) The design of Cethana concrete face rockfill
dam. Proceedings of the 11th International Congress on Large Dams, Rio de Janiero, ICOLD. pp. 25-43 (Q.42
R.3).
Woodward Clyde (1998) Lake Eppalock main embankment remedial works. Unpublished report for Goulburn
Murray Water.
Wu, G., Freitas, M.S., Araya, J.A.M. and Huang, Z.Y. (2000a) Planning and construction of Tianshengqiao 1 CFRD
(China). Proceedings of the International Symposium on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp.
481-496.
Wu, G., Freitas, M.S., Araya, J.A.M., Huang, Z.Y. and Mori, R.T. (2000b) Tianshengqiao-1 CFRD - monitoring and
performance - lessons and new trends for future CFRDs (China). Proceedings of the International Symposium
on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp. 573-586.
Wu, H., Wujie, Wang, S., Wu, Q. and Cao, K. (2000c) Ten years surveillance of Chengbing concrete face rockfill
dam. Proceedings of the International Symposium on Concrete Faced Rockfill Dams, Beijing, ICOLD. pp. 595-
605.
Wu, Q. and Cao, K. (1993) Concrete faced rockfill dam of Chengbing project. Proceedings of the International
Symposium on High Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing, Vol. 1, pp. 440-447.
Yang, S. (1993) Design of TSQ-1 concrete face rockfill dam. Proceedings of the International Symposium on High
Earth-Rockfill Dams, (Jiang, Zhang & Qin ed.) Beijing, Vol. 1, pp. 448-543.
UNICIV Report No. R405 – Deformation Behaviour of Rockfill January 2002
APPENDIX A
CFRD Case Study Information

-i-
TABLE OF CONTENTS
1.0 CASE STUDY INFORMATION ON CFRD..............................................................................1
LIST OF TABLES
Table A1.1: Case studies information on concrete face rockfill dams ....................................................................3
-1-
1.0 CASE STUDY INFORMATION ON CFRD

Information on the individual case studies used for the analysis of rockfill deformation behaviour in
embankment dams is presented in Table A1.1. The data set comprises thirty-five concrete faced rockfill dams
and one central core earth and rockfill dam (El Infiernillo dam).
An explanation of some of the terminology used in Table A1.1 is as follows:

• General terms or symbols used and their meaning are:
− - or na = not required, not appropriate or not known
− ( ) in strength class, Cu, dry density and void ratio columns indicates the value has been estimated.
− H to V refers to horizontal to vertical slope angle; i.e. 2H to 1V.
− SMP = surface monitoring point.
− HSG = hydro-static settlement gauge
− IN = inclinometer
− EOC = end of construction
− ff = first filling
• The system for classification of the various rockfill zones used for the main rockfill is in accordance with
Section 1.2.3. The zoning classification used by the author/s of the referenced paper is given separately.
• Under the section construction;

− Height, H = maximum embankment height
− Length, L = crest length
− Slab face area = the area of the upstream face in the plane measured normal to the upstream face.
• In the design section;

− Main rockfill refers to the rockfill forming the main body of the embankment (i.e. Zone 3 for dumped
rockfill, Zones 3A and 3B for compacted rockfill).
− Facing details refers to the embankment zone between the facing slab and the main rockfill (i.e. Zones
2D and 2E in Figure 1.4).
• In the section on rockfill material parameters and properties:

− Terms FR and SW, used in rockfill source column, refer to the rock weathering classification (fresh,
slightly weathered respectively) used by the referenced author/s.
− AS 1726-1993 is used for classification of the unconfined compressive strength (UCS) of the intact rock
(Table 1.1).
− Particle size distribution parameters have been determined from the average grading curve for the
rockfill where available or as reported by the referenced author/s.
− Cu = Grading uniformity coefficient of the rockfill particle size distribution (= d60/d10).
2
− Cc = Grading curvature coefficient of the rockfill particle size distribution (= (d 30 ) (d 60 ⋅ d 10 ) ).
− d10, d30, etc., refer to the particle size (from the average grading curve) related to 10, 30, etc. percent
passing.
-2-
− The dry density, void ratio and porosity are average values reported by the referenced author/s or
calculated from the available data.
− The classification terms used to describe the level of compaction are defined in Section 1.2.1. Good
refers to well-compacted, poor refers to dumped and well sluiced or poorly compacted, and very poor
refers to dry dumped or dumped and poorly sluiced rockfill.
− 4p 10t SDVR = 4 passes of a 10 tonne (dead-weight) smooth drum vibrating roller.
• Calculation of the rockfill secant moduli during construction ( E rc ) and the modulus on first filling ( E rf )
are defined in Section 1.2.2. For E rc the modulus has generally been determined from deformation records
in the lower 60% and central half of the embankment.
− The average secant modulus during construction over the construction period has been determined by
averaging of modulus estimates throughout the construction period. It will include values estimated
when the embankment was in the early stages of construction right through to the end of construction.
− E rc at the end of construction only considers those secant modulus estimates at the end of the main
rockfill construction. A representative value has been obtained by averaging where more than estimate
has been made (i.e. from or between different HSGs). The applied vertical stress in the calculation of
each modulus value (before averaging) has been corrected for valley shape effects according to Figure
3.5.
• Leakage. The time period of the long-term leakage rate is generally given in years after the end of first
filling. For several cases, such as Dix River, it is given in terms of years after the end of main rockfill
construction because the period of first filling was not known. For several embankments where first filling
was not completed in the reported period and where a significant reduction in leakage rate occurred, such as
due to repair, a long-term leakage rate is given (e.g. Khao Laem).
• Deformation monitoring:
− The settlements during construction are generally based on the monitoring records from hydrostatic
settlement gauges (HSG) installed as construction proceeded.
− Post construction crest settlements are from after the end of main rockfill construction. 0.5 to 5 years
indicates the period of monitoring after the end of main rockfill construction.
− The “vertical strains” reported are crest settlements as a percentage of the embankment height.
− For the deformation normal to the face slab the references to % of dam height or distance below the
crest (bc = vertical distance below the crest) refer to the location on the face slab representative of the
deformation given.
− The long-term deformation rates are in units of crest settlement (as a percentage of the embankment
height) per log cycle of time, where time is measured in years after the end of main rockfill
construction.
− ff = first filling.
-3-
UNICIV Report No. R405 – Appendix A: CFRD Case Study Information January 2002
Table A1.1: Case studies information on concrete face rockfill dams

GENERAL DETAILS CONSTRUCTION DESIGN
Construction Timing Dimensions Facing Details (Support Zone for Face Slab)
Name Location /
Geology
Foundation
Main Rockfill (Comments) Thickness References
Owner Year Time Height, Length, Slab Face Upstream Downstream (Soil/Rock) Material Type /
L/H (normal to face Comments
Completed (years) H (m) L (m) Area (m2) Slope Slope Compaction
slab), m
Aguamilpa (Zone 3A) Macedo et al (2000), Macedo (1999)
Rock - gravels in Zone 3A upstream half. Zone 3T dowmstream central River gravels (sandy
Tertiary volcanics - Montanez et al (1993), Mori (1999)
Aguamilpa (Zone 3T) CFE, Mexico 1993 3? 185.5 475 2.6 130000 1.5 to 1 1.4 to 1 part of river 1/3. Zone 3B - downstream outer 2/3. Zone 3B coaser - gravel), rounded, well -
ignimbrites Gonzales & Mena (1997)
section than 3T. compacted.
Aguamilpa (Zone 3B) Marulanda & Pinto (2000)
metamorphic sediments, Gravels in river Zone 3A upstream half of main rockfill, Zone 3B finer well graded rockfill, Materon (1985a), Regalado et al (1982)
Alto Anchicaya Columbia 1974 1.5 140 260 1.9 31000 1.4 to 1 1.4 to 1 1.75 to 7 -
schists & chert section downstream half of main rockfill. well compacted Amaya & Marulanda (2000)
Zone 3A - 90% of main rockfill, Zone 3B in lower part of

finer well graded rockfill, higher densities than HEC (1991a), Bowling (1981-82)
Bastyan Hydro Tasmania Rhyolite 1983 1.5 75 430 5.7 19000 1.3 to 1 1.3 to 1 Rock outer downstream shoulder. Slight influence from 3.7
well compacted main rockfill Knoop and Lack (1985)
arching in lower gully.
Forza & Hancock (1995a), Liggins (1971),
Quartzite & quartzite Zone 3A - 75% of main rockfill, Zone 3B in outer finer well graded rockfill, Fitzpatrick et al (1973), Wilkins et al (1973),
Cethana Hydro Tasmania 1971 2.5 110 213 1.9 30000 1.3 to 1 1.3 to 1 Rock 3.7 -
conglomerate downstream shoulder. Likely arching in lower gully. well compacted Fitzpatrick et al (1982), Pinto et al (1982),
Giudici et al (2000)
Zones 3B and 3C (Zone 3A) similar in rockfill type and crushed rockfill, finer than Higher densities than Wu & Cao (1993)
Chengbing China Tuff lava 1989 2 to 2.5 74.6 325 4.4 15800 1.3 to 1 1.3 to 1 Rock 2.4
compaction 3A and well compacted main rockfill Wu et al (2000c)
Laminated gunite facing damaged early in embankment Lean concrete facing

Cogswell LACFCD Granitic Gneiss ? 1935 3.3 85.3 176.8 2.1 - 1.3 to 1 1.5 to 1 Rock life. Replaced in 1935 with timber facing and in 1947 1.8 to 5.5 derrick placed rockfill placed on derrick Baumann (1939, 1958, 1964)
with reinforced concrete facing. rockfill
Pacific Gas &

derrick placed rockfill, voids Cooke (1958)
Courtright Electricity Co. granite 1959 2 97 274 2.8 20440 1.15 to 1 1.4 to 1 Rock - 2.4 to 3.5 -
chinked Regan (1997)
San Francisco
tunnel spoil, rockfill Davies (1993), Quinlan (1993)
sandstone, phyllite, Facing rockfill (Zones 2A and 2B) likely to have a lower rockfill used, well
Crotty Hydro Tasmania 1991 2 83 240 2.9 14500 1.3 to 1 1.5 to 1 Rock 7.3 from sedimentary Li et al (1993), Cribben (1990)
conglomerate modulus than the gravels used in the main fill (Zone 3A). compacted
geology units. Giudici et al (2000)
Kentucky derrick placed rockfill, 2.7 to
Suspect rockfill likely to be relatively coarse. Some Schmidt (1958)
Dix River Utilities limestone & shale 1925 2 84 311 3.7 - 1.1 to 1 1.4 to 1 Rock 1.4 to 4.1 9 tonne rocks used, voids -
rockfill placed directly from fired shots in abutments. Howson (1939)
Company chinked
El-Infiernillo (Zone 3B) Central core earth and rockfill embankment. Included to
CFE, Mexico silicified conglomerate 1963 1.3 148 344 2.3 - 1.75 to 1 1.75 to 1 Soil/Rock give an estimate of moduli during construction for dry na na na Marsal & Ramirez de Arellano (1967)
El-Infiernillo (Zone 3C) dumped rockfill.
Foz Do Areia (Zone 3A) Complex design. Zone 3A on roughly upstream 1/3, finer graded sound basalt Pinto et al (1985a), Materon (1985b)
Basalt & Basaltic
Brazil 1979 2.5 160 828 5.2 139000 1.4 to 1 1.4 to 1 Rock Zones 3B (1C and 1D) in downstream 2/3 of main 2.3 to 7.6 (very high strength), well Pinto et al (1982), Pinto et al (1993)
Breccia
Foz Do Areia (Zone 3B) rockfill. compacted Sobrinho et al (2000), Cooke (1999)
processed dirty gravels,

interbedded shale and Clean gravels placed as chimney filter and in central slightly higher density Amaya & Marulanda (1985)
Golillas Columbia 1978 1.75 125 107 0.9 17000 1.6 to 1 1.6 to 1 Rock 4.0 finer graded, well
sandstone section of main fill. than main rockfill Amaya & Marulanda (2000)
compacted
Ita (Zone 3A) Zone 3A - upstream 1/3; Zone 3B - downstream 2/3.

Sobrinho et al (2000)
Dumped rockfill placed in river section. Zone 3A crushed dense basalt, fine high densities than
Brazil Basalt 1999 3 125 880 7.0 110000 1.3 to 1 1.3 to 1 Rock 3.5 to 5 Sobrinho et al (1999)
incorporates dense basalt zone closest to upstream face sized, well compacted main rockfill
Ita (Zone 3B) and zone of lesser quality behind this.
Silveira & Sardinha (1999)
Finer graded H to VH Good et al (1985), Good (1976, 1981)

SA Water, South Rock, some Zone 3A is main body of rockfill. Zone 3B is a basal Zone 2 placed in 915
Kangaroo Creek schist and gneiss 1969 1.75 60 178 3.0 - 1.3 to 1 1.4 to 1 8.0 strength granitic gniess, E&WSD (1981), SKM (1995)
Australia gravels layer of very high strength rockfill for drainage purposes. mm layers.
well compacted SA Water (1995)
Khao Laem (Zone 3A) Watakeekul et al (1985)

limestone, shale, Zone 3A - 75% of main rockfill in section, Zone 3B in finer graded rockfill, 0.5 m
Thailand 1984 2 130 1000 7.7 - 1.4 to 1 1.55 to 1 Rock, karstic 4.1 D50 = 24 to 40 mm Mahasandana & Mahatharadol (1985)
sandstone & siltstone outer downstream shoulder (25% of main rockfill). layers, well compacted
Khao Laem (Zone 3B) Aphaiphuminart et al (1988)
Gosschalk & Kulasinghe (1985, 1987)

Limited information on rockfill properties. HSGs in higher finer graded rockfill, well
Kotmale Sri Lanka foliated gneiss 1984 3 90 560 6.2 - 1.4 to 1 1.45 to 1 Rock 3.5 - Casinader (1987)
elevations indicate higher modulus. compacted
Kulasinghe & Tandon (1993)
High strength (UCS = 36 to Better quality rockfill
SA Water, South interbedded quartzite, No water available to add to rockfill for construction to 24 Good et al (1985), Good (1981)
Little Para 1977 2.5 53 225 4.2 - 1.3 to 1 1.4 to 1 Rock 8.5 45 MPa) quartzite rockfill, used for facing
Australia slaty dolomite & slate m height. Boucaut & Beal (1979)
well compacted layers.
Lower Bear No. 1 Pacific Gas & granite 1952 1.5 75 293 3.9 17650 1.3 to 1 1.35 to 1 Rock Coarse rockfill 3 to 6.4 derrick placed rockfill, 1.15
Electricity Co. to 2.7 m3 volume, voids - Steele & Cooke (1958)
Lower Bear No. 2 San Francisco granite 1952 <1.5 46 264 5.7 8400 1 to 1 1.3 to 1 Rock Coarse rockfill 3 to 5.2 chinked
-4-
Table A1.1: Case studies information on concrete face rockfill dams (Sheet 2 of 8)
Material Parameters / Properties of Rockfill Material Parameters / Properties of Rockfill
Rockfill Zone Intact Rock Strength Particle Size Distribution Parameters (from average grading curve) Particle Size Distribution Parameters (from average grading curve)
Name Rockfill Particle
Wet/Dry % finer % finer
n Authors Source Shape Strength Absorption D10 D30 D50 D60 D80 D90 Dmax % finer Clean / D10 D30 D50 D60 D80 D90 Dmax % finer Clean /
Class UCS (MPa) Strength Cu Cc 75 Cu Cc 75
Classn Classification (%) (mm) (mm) (mm) (mm) (mm) (mm) (mm) 19 mm Dirty (mm) (mm) (mm) (mm) (mm) (mm) (mm) 19 mm Dirty
(%) micron micron
clean to clean to
Aguamilpa (Zone 3A) 3A 3B dredged alluvium rounded (Very High) - - - 85 5.8 0.75 16.4 45 62 124 230 600 2 34 85 5.8 0.75 16.4 45 62 124 230 600 2 34
dirty dirty
Aguamilpa (Zone 3T) 3T T ignimbrite angular Very High 180 - - 30 2.4 2.9 24.5 61 88 204 290 500 2 28 clean 30 2.4 2.9 24.5 61 88 204 290 500 2 28 clean
Aguamilpa (Zone 3B) 3B 3C ignimbrite angular Very High 180 - - 22 1.4 9 - 150 - - - 700 - - clean 22 1.4 9 - 150 - - - 700 - - clean
Alto Anchicaya 3A D hornfels angular (Very High) - - - 18 2.5 5.5 31 71 100 205 280 600 <2 22 clean 18 2.5 5.5 31 71 100 205 280 600 <2 22 clean
Rhyolite (SW to
Bastyan 3A 3A angular (Very High) - - - 42 3.3 2.4 28 69 100 190 300 600 3.3 25 clean (?) 42 3.3 2.4 28 69 100 190 300 600 3.3 25 clean (?)
FR)
Cethana 3A 3A Quartzite, quarried angular (Very High) - - - 23 2.0 4.8 32 75 108 250 440 900 2 21 clean 23 2.0 4.8 32 75 108 250 440 900 2 21 clean
Chengbing 3A 3B/3C Tuff lava, quarried angular Very High 80 - - 10.4 - - - - - - - 1000 - - clean (?) 10.4 - - - - - - - 1000 - - clean (?)
Cogswell 3A 3A Granitic Gneiss angular High 45 - - 7 2.3 100 - 650 - - - 1300 3 5 clean 7 2.3 100 - 650 - - - 1300 3 5 clean
550 to 550 to
Courtright 3A 3A granite, blasted angular (Very High) - - - (<7) - - - - - - 1750 - - clean ? (<7) - - - - - - 1750 - - clean ?
1500 1500
Gravels
Crotty 3A 3A rounded (Very High) - - - 70 5.6 0.34 7.3 20 28 60 86 200 3.5 48 dirty 70 5.6 0.34 7.3 20 28 60 86 200 3.5 48 dirty
(Pleitocene)
Limestone & Shale

Dix River 3A 3A angular ? (Very High) - - - low - - - - - - - - - - - low - - - - - - - - - - -
(?), spillway excav
diorite & silicified

El-Infiernillo (Zone 3B) 3B 3B angular Very High 125 - - 13 1.6 6.5 - 62 - - - 600 1 22 clean ? 13 1.6 6.5 - 62 - - - 600 1 22 clean ?
conglomerate
diorite & silicified
El-Infiernillo (Zone 3C) 3C 3C angular Very High Diorite = 125 - - <13 1.6 ? >7 - - - - - >600 - - clean ? <13 1.6 ? >7 - - - - - >600 - - clean ?
conglomerate
basalt (max 25% High to Very Basalt = 235
Foz Do Areia (Zone 3A) 3A 1B angular - 1.0 6 2.1 20 60 105 145 300 440 600 <1 10 clean 6 2.1 20 60 105 145 300 440 600 <1 10 clean
basaltic breccia) High Breccia = 37
mix basalt & High to Very Basalt = 235
Foz Do Areia (Zone 3B) 3B 1C & 1D angular - 1.0 14.2 2.2 - - - - - - - - - clean ? 14.2 2.2 - - - - - - - - - clean ?
basaltic breccia High Breccia = 37
gravels,
Golillas 3A 2 rounded (Very High) - - - 125 2.5 0.65 11.5 31 80 185 225 350 6 40 dirty 125 2.5 0.65 11.5 31 80 185 225 350 6 40 dirty
unprocessed
Ita (Zone 3A) 3A E1/E3' basalt (dense) angular (Very High) - - - 11 1.6 13.4 57 113 152 275 425 700 1 12 clean 11 1.6 13.4 57 113 152 275 425 700 1 12 clean
Basalt - VH
Ita (Zone 3B) 3B E3 Breccia and Basalt angular - - - 13.3 1.6 10 46 100 133 250 385 750 1 15 clean 13.3 1.6 10 46 100 133 250 385 750 1 15 clean
Breccia - H (?)
angular to
Kangaroo Creek 3A 3 schist Medium to High 25 63 3.0 310 6.4 0.14 6.2 26 44 105 180 600 8.5 44 dirty 310 6.4 0.14 6.2 26 44 105 180 600 8.5 44 dirty
subangular
Khao Laem (Zone 3A) 3A 3A limestone, quarried angular Very High < 190 100 low - - - - - - - - 900 - - - - - - - - - - - 900 - - -
Khao Laem (Zone 3B) 3B 3B limestone, quarried angular Very High < 190 100 low - - - - - - - - 1500 - - - - - - - - - - - 1500 - - -
charnockitic / (High to Very

Kotmale 3A 3A angular - - - - - >2 - - - - - 700 - - - - - >2 - - - - - 700 - - -
gneissic High)
angular, close to dirty (soft dirty (soft

Little Para 3A 3A dolomitic siltstone Medium 8 to 14 - 120 4.0 0.6 13 40 70 255 390 1000 5 35 120 4.0 0.6 13 40 70 255 390 1000 5 35
elongated ? 100% rock) rock)
Lower Bear No. 1 3A 3A granodiorite angular Very High 100 to 140 - -

700 to 2000 to 700 to 2000 to
8 to 10 - >100 - - - - low low clean 8 to 10 - >100 - - - - low low clean
800 3000 800 3000
Lower Bear No. 2 3A 3A granodiorite angular Very High 100 to 140 - -
-5-
SECANT MODULI VALUES HYDROGEOLOGY LEAKAGE
Modulus During, Erc (MPa) Modulus on First Filling, Erf Time for First Filling Rainfall,
Maximum
Name (from end of main average Long-term
Average at EOC Reservoir Level / Comments Rate Comment
Average over rockfill annual (l/sec)
(adjusted for Erf (MPa) Comments (l/sec)
construction period construction) (mm)
valley shape)
Aguamilpa (Zone 3A) 305 (250 to 330) 305
Reduced modulus where Zone 3C has an influence on 200, at first 150 to 170 (4 Large increase in leakage rate when reservoir raised above
Aguamilpa (Zone 3T) 104 104 770 0.48 to 1.83 years - -
the deformation of the face slab. filling years post ff) certain level (due to cracking in upper part of face slab)
Aguamilpa (Zone 3B) 36 (25 to 45) 36
375 (@ 30 to High rate of leakage on first filling, through the perimeter joint on
Alto Anchicaya 138 (100 to 170) 110 - 0.18 to 0.2 (Oct 74) years. - - 1800 -
40% height) the right abutment and open joints on left abutment. Repaired.
5 (8 years post Approximately 10 l/sec on first filling. Gradual reduction with time
Bastyan 130 (120 to 140) 102 290 Maximum face deflection at about mid height. 0.86 to 1.05 years. Steady 2100 10
ff) to base flow of 5 l/sec.
160 (120 to 210) in lower 25%

5 to 10 (4 years
Cethana 105 (85 to 120) in mid to lower 103 300 - 0.26 to 0.48 years (1971). Steady 2030 65 to 70 -
post ff)
25%
First filling started before embankment

-0.5 to 4 years (12/88 to Max leakage during construction when reservoir level rose above
Chengbing 43 43 110 Approx 2.5 times E during completed. Fluctuating, drawdowns up to - 75 -
1993). level of face slab. After completion leakage = 60 l/sec.
50 m.
Started filling 2.6 years
Cogswell - - 44 From measured face slab deformation. after EOC. Filled in 71 - - 3600 - Maximum in 1938 on first filling. Large crack in floater slab.
hours in March 1938.
Lower perimetric joints moved soon after completion affecting

Courtright - - - - 0 to 9 years (from 1959). Fluctuating - 1275 - waterstops. Joints on abutments opened (~ 1968) and upper
perimetric joint opened tearing waterstops in 1989.
Relatively low modulus (compared with modulus during

30 to 35 (2 years Peak leakage rate of 45 l/sec. Thought to be seepage through
Crotty 375 (113 to 636) 360 470 construction) may be associated with lower modulus of 0.87 to 2.0 years. Steady 2900 45
post ff) the foundation on the abutments where the grouting failed.
facing rockfill.
2700 (1937, Leakage near intake tower, through cracks and opened joints in
2000 (30 years
Dix River - - - - Started in 1925 - - 12 yrs after slabs and through the foundation. Various methods to try and
post EOC)
EOC) treat problem.
El-Infiernillo (Zone 3B) 39 (27 to 48) 35.5

- - 0.33 yrs (June to Oct 1964) Fluctuating - - - -
El-Infiernillo (Zone 3C) 22 (17 to 27) 20
Foz Do Areia (Zone 3A) 47 (38 to 56) 47 Relatively large deformations in toe region compared with
0.5 to 0.9 years (Apr to 70 (5 years post
80 (65 to 92) other CFRDs. Therefore lower moduli toward toe region Relatively steady, typical 5 m fluctuation - 240 Maximum leakage on first filling. Gradual reduction with time.
Aug 80). ff)
Foz Do Areia (Zone 3B) 32 (29 to 38) 32 (15% height above toe)
155 (145 to 165)

Estimated by hydrostatic settlement cells for section 4 to 5.17 years (June 82 to Fluctuating. First filling interupted due to 270 to 500 (15 Leakage through the foundation during first filling and some
Golillas excludes those points thought to 107 250 - 1080
below reservoir level. Aug 83). excessive seepage. years post EOC) segregation of Zone 1 noted. Repaired and leakage reduced.
be affected by arching
Ita (Zone 3A) 48 48 Reached 1700 l/sec 4 months after first filling, due to cracking in
Relatively low because incorporates low modulus
87 (83 to 91) 0.7 to 0.9 years. - - 1700 - slab. Reduced to 380 l/sec by dumping sand and silty sand on
dumped rockfill in river gully (E during = 17 MPa).
Ita (Zone 3B) 24 (14 to 46) 24 face.
0 to 1.95 yrs (Sept 69 to Fluctuating, seasonal drawdown of up to 42 On initial filling seepage heard through within the embankment.
Kangaroo Creek - - 140 Reported (no data to confirm) 600 to 800 11 2.5
Aug 71). m Possible seepage along horizontal layers in weak rockfill.
Khao Laem (Zone 3A) 59 (43 to 79) 43 High reading possibly affected by arching in deepest 1.1 (after repair Feb 1986 (3 months post first filling) leakage increased from 9 to
0.6 to 1.9 years (June 84 to
130 to 240 section of embankment, low reading possibly affected by Steady (?) - 53 of face slab 53 l/sec as a result of cracks in the face slab. Cracks repaired
Nov 85).
Khao Laem (Zone 3B) 30 30 "plum pudding" foundation. cracks) and leakage reduced.
145 (135 to 0.54 to 1.04 years (Nov 84

Kotmale 61 (47 to 87) 52 Estimated from HSGs near to upstream face. Fluctuating, seasonal drawdown up to 60 m - < 10 - At close to end of first filling.
155) to Aug 85).
Maximum leakage measured just prior to first overflow in August

0.58 to 2.83 years (Aug 77
Little Para 21.5 (19.5 to 23.5) 20.5 - No data for calculation Seasonal fluctuation of about 5 m. - 19.2 - 1981. Wet spots and seepage on downstream slope on right
to Nov 79).
abutment.
At end of 2 yrs and 42 m below crest (39% dam height). 0.07 to 0.58 years (Dec 52 Fluctuating, seasonal drawdown typically 43 (4 years post
Lower Bear No. 1 - - 21 - 113 Maximum at FSL on first filling. Gradual reduction thereafter.
Likely under-estimate to June 53). 35 m ff)
0.07 to 0.58 years (Dec 52 Fluctuating, seasonal drawdown typically Suspect relatively small in comparison to Lower Bear No. 1 and
Lower Bear No. 2 - - 40 At end of 1 year and 29 m below crest (39% dam height). - no leakage -
to June 53). 35 m other CFRDs constructed in the 1950s.
-6-
MONITORING
Settlement Post Construction Crest Deformation Long Term Crest
Deformation Normal
Name During
Settlement Settlement Lateral to Face Slab
Settlement Rate
Comments
Other Comments
Construction Settlement, total Settlement, (% per log time cycle)
on FF on FF Displacement (mm)
(mm) (mm) total (%) (t o at EOC)
(mm) (%) (mm)
Aguamilpa (Zone 3A)
Decreasing velocity of deformation with time. Face slab deformation greatest at
Aguamilpa (Zone 3T) - 307 (0.4 to 5.7 yrs) 0.185 222 0.133 - 320 from ff to 6 years 0.090 crest (due to much lower modulus of Zone 3C rockfill). Limited time for estimation Moduli during construction from literature.
of long term creep rate.
Aguamilpa (Zone 3B)
153 mm (0.11 to 10.3 160 mm to 5/77 (130 mm Decreasing rate of deformation with time. At 10 yrs < 2 mm/yr crest settlement.
Alto Anchicaya - 0.109 15 0.011 - 0.037 Moduli during construction from literature.
yrs) ff) Crest points on top of concrete plinth.
50 mm (0.2 to 7.5 16 mm (SC3), 7 68 mm @ 34 m below

Bastyan > 219 0.066 15 0.02 0.028 Decreasing rate of deformation post first filling (also decreasing on log scale).
years, SC3) mm ff crest
137 mm (0 to 28.6 83 mm to 7/92 170 mm to 10/93 (114 ff Decreasing rate of deformation post first filling (linear on log scale). Upper 12 m of
Cethana > 560 0.124 46 0.043 0.042
yrs) (27 ff) 37m bc) rockfill placed 1 year after main body had been completed.
154 mm after 10 years In 1989, during construction, the impounded water level exceeded the top level of the
Chengbing - - - - - - - Insufficient information on deformation.
(?) constructed face slab resulting in leakage through the embankment above the slab.
In late 1933 a very hevy rain storm resulted in crest settlements of to a maximum
Estimated at 271 mm (0 to 3 403 mm vertical settlement
Cogswell 0.317 < 271 < 0.317 - - of 3.41 m (due to collapse settlement). Artificial sluicing resulted in an additional
15.7 m. years, to 4/38) to 4/38 (first filling)
1.71 m.
1237 mm (0 to 38 0.70 (from 5 year to 20+ Relatively steady rate of crest settlement on log scale, but only limited number of
Courtright - 1.282 844 0.875 - - No data to calculate modulus during construction or on first filling
yrs) years) data points.
55 mm (0 to 8.5 Gradual reduction in rate of deformation after first filling, relatively steady on log
Crotty > 179 0.066 16 0.019 9 mm (SC3 - ff) 46 mm @ 49 m bc 0.045
years, SC4) scale.
1281 mm (Stn 16.9, 970 mm to 1957 0.58 to 1.44 (increasing Decreasing rate of deformation with time. At 30 yrs approximately 20 to 25
Dix River - 1.525 - - - No data to calculate modulus during construction or on first filling
to 1957, 32 yrs) (Stn 16.9) with time) mm/year settlement, and 15 to 17 mm/yr displacement.
El-Infiernillo (Zone 3B) 1095 (1.9%) in

IN3, and 2425 Central core earth and rockfill dam. Data used for analysis of rockfill deformation during
- - - - - - na -
mm (2.2%) in construction.
El-Infiernillo (Zone 3C) IND1.
Foz Do Areia (Zone 3A) 248mm to 1984 Log scale deformation plot shows sharp reduction in rate of deformation after 3
328 mm (0 to 11 780 mm @ 49% dam
- 0.205 73 0.046 (180 ff), d/stream 0.099 years and then acceleration in rate beyond 7 years (to that similar from 1 to 3
years) height (620 mm ff)
Foz Do Areia (Zone 3B) slope years).
160 mm @ 46% dam Acceleration in crest settlement and face deflection toward end of first filling.
52 mm (0.46 to 6.4 7 mm to 10/84 (2
Golillas - 0.042 20 0.016 height by 3/84 (first 0.033 Thereafter, significant reduction in rate of deformation, but limited data. Crest
years, to 10/84) yrs)
filling) SMPs on parapet. Upstream face slab deformation estimated by HSGs and SMPs.
Ita (Zone 3A)

Relatively complex cross section. Face slab cracked resulting in relatively high leakage
- - - - - - 461 mm (during ff) - Limited data.
rates.
Ita (Zone 3B)
Not sure when monitoring first started, possibly shortly after start of first filling or
50 mm to 1979 Initial monitoring possibly missed first filling to EL 220 (26 m below crest, 22 m below
Kangaroo Creek - 116 mm (0 to 26 yrs) 0.193 26 0.043 - 0.126 could have missed most of first filling entirely. Decreasing rate of movement with
(8yrs), 32 mm ff FSL), but cannot be sure. Dam raised approx. 3.4 m in 1983 for additional flood storage.
time post first filling. Crest SMPs on rockfill and face slab.
Khao Laem (Zone 3A)

> 550 (of Calculation of moduli during construction takes into consideration foundation settlement.
- - - - - - - Significant settlement of foundation during construction.
embankment) Erf adjusted for filling to full supply level.
Khao Laem (Zone 3B)
255 mm (0 to 2.46 62 mm to 10/86 98 mm to ff (estimated from 0.258 (limited data, possibly Initial reading very close to time of completion of rockfill. Last available readings
Kotmale > 1020 0.283 96 0.107 Modulus on first filling determined from deformation of HSGs adjacent to face slab.
yrs) (0 to 2 yrs) HSG) hasn't steadied as yet) approx 12 months after first filling, rate has not settled down as yet.
152 mm (0 to 22.6 Monitoring appears to have started about June 1978, most likely when first filling Wet spots and seepage observed on downstream slope (on right abutment). Indicative of
Little Para - 0.288 22 0.042 - - 0.21
yrs) started. high permeability variation between horizontal and vertical directions.
375 mm (0.07 to 4.05 305 mm to 11/56 (4 625 mm to 10/54 (2 yrs) @ Rapid deformation on first filling with significant reduction in rate thereafter. Crest
Lower Bear No. 1 - 0.56 < 335 < 0.50 0.103
years, to 11/56) yrs), ff 270 mm 42 m bc SMPs on upstream edge of crest.
116 mm (0.07 to 4.07 116 mm to 11/56 (4 165 mm to 11/56 (4 yrs) @ Rapid deformation on first filling with significant reduction in rate thereafter. Crest
Lower Bear No. 2 - 0.271 73 0.171 0.128
years, to 11/56) yrs), ff 88 mm 15 m bc, ff 137 mm SMPs on upstream edge of crest.
-7-
GENERAL DETAILS CONSTRUCTION DESIGN
Construction Timing Dimensions Facing Details (Support Zone for Face Slab)
Name Location /
Geology
Foundation
Main Rockfill (Comments) Thickness References
Owner Year Time Height, Length, Slab Face Upstream Downstream (Soil/Rock) Material Type /
L/H 2 (normal to face Comments
Completed (years) H (m) L (m) Area (m ) Slope Slope Compaction
slab), m
high densities Knoop & Lack (1985), Knoop (1982b)
Greywacke, slate & finer graded greywacke
Mackintosh Hydro Tasmania 1981 2.75 75 465 6.2 27500 1.3 to 1 1.3 to 1 Rock Zone 3B not used. 3.7 achieved (higher than HEC (1991b), Bowling (1978)
phyllite rockfill, well compacted
main rockfill) Giudici et al (2000)
Mangrove Creek (Zone
3A) Interbedded Zone 3A - upstream facing (20 to 30%). Random fill Mackenzie & McDonald (1985)
Sydney Water, finer graded siltstone
sandstones, siltstones 1981 2 80 380 4.8 - 1.5 to 1 1.6 to 1 Weathered Rock (Zone 3B) makes up most of rockfill. Drainage layer 1.5 to 2 - PWSD NSW (1998)
Mangrove Creek (Zone NSW rockfill, well compacted
and claystones below random fill. Heinrichs (1996)
3B)
finer crushed rockfill, Knoop & Lack (1985), Knoop (1982a)
Zone 3A - 70% of main rockfill, Zone 3B in outer Vey high strength
Murchison Hydro Tasmania Rhoylite 1982 2.25 94 200 2.1 17000 1.3 to 1 1.3 to 1 Rock 3.7 watered and well Gerke et al (1995), Robinson (1979)
downstream shoulder. rhyolite
compacted Giudici et al (2000)
Knoop & Lack (1985), Li (1991)
Zone 3A - 75% of main rockfill, Zone 3B in outer
laminated quartzite & Gravels in river crushed dolerite, finer Li et al (1991), Davies et al (1995)
Reece Hydro Tasmania 1986 3 122 374 3.1 37800 1.3 to 1 1.4 to 1 downstream shoulder. River gravels left in place (except 3.7 -
amphibolite section grading, well compacted Partyka & Bowling (1984)
near upstream face).
Giudici et al (2000)
some rock pieces
Pacific Gas & Derrick placed rockfill with Steele & Dreyer (1939)
Very coarse rockfill, "lot of large rocks with few fines". were observed to
Salt Springs Electricity Co. granite 1931 3 100 396 4.0 35300 1.3 to 1 1.4 to 1 Rock 4.6 flat face in the plane of the Steele & Cooke (1958), Regan (1997)
Likely low Cu. crush during
San Francisco face slab.
construction.
Cooke & Strassburger (1988)
Salvajina (Zone 3A) Rock and soil, Alluvials in gully were in dense state (moduli approx 320
Tertiary sedimentary - gravels, thin layers, well Amaya & Marulanda (2000)
Columbia 1984 1.5 to 2 148 362 2.4 57500 1.5 to 1 1.4 to 1 gravels in river MPa). 2.8
siltstone and sandstone compacted. Hacelas et al (1985), Sierra et al (1985)
Salvajina (Zone 3B) section Zone 3A - upstream 2/3, Zone 3B - downstream 1/3.
No Zone 3B. Bituminous concrete facing. Zone 2B (well-

interbedded argillite & sound crushed rockfill, Gerke et al (1993), Cole (1974)
Scotts Peak Hydro Tasmania 1972 1 43 1067 24.8 - 1.7 to 1 1.3 to 1 Rock compacted gravels) placed at upstream toe to 13 m 0.3 -
sandstone dolomite, well compacted. Baker (1972), Cole & Fone (1979)
above foundation.
Segredo (Zone 3A) Zone 3A - upstream 1/3. Zone 3B (1C and 1D)
Basalt & Basaltic fine graded sound crushed high densities, higher Pinto et al (1985b), Pinto et al (1993)
COPEL, Brazil 1992 2.5 145 720 5.0 86000 1.3 to 1 1.4 to 1 Rock downstream 2/3. Dumped basalt rockfill in river section 3.1
Breccia basalt, well compacted than main rockfill Sobrinho et al (2000), Blinder et al (1992)
in downstream 2/3.
Segredo (Zone 3B)
Rock, gravels in No Zone 3B used. Grading of 3A indicates significant Forza & Hancock (1993)
Serpentine Hydro Tasmania quartzite and schists 1971 2 38 134 3.5 8000 1.5 to 1 1.5 to 1 0.3 150 mm minus rockfill -
river section breakdown on compaction. Giudici et al (2000)
Similar rockfill for entire embankment. Localised finer graded rockfill, well heavily compacted in
Shiroro Nigeria Granite 1983 - 125 560 4.5 - 1.3 to 1 1.4 to 1 Rock 6.1 Bodtman & Wyatt (1985)
enlargement of Zone 2A below EL 290 (bottom 30 m). compacted thin layers.
Tianshengqiao - 1
Sedimentary - crushed limestone (SW to Wu et al (2000a, 2000b)
(Zone 3A) Zone 3A - upstream half. Zone 3B (mudstone) high densities,
China limestone, sandstone 1999 3.3 178 1168 6.6 173000 1.4 to 1 1.3 to 1 Rock 4.6 FR), relatively fine, well Yang (1993)
surrounded by coarse limestone rockfill (Zone 3D). wetted.
Tianshengqiao - 1 and mudstone compacted. Jiyuan et al (2000)
(Zone 3B)
Knoop & Lack (1985)
finer crushed greywacke high densities
Tullabardine Hydro Tasmania Greywacke & slate 1979 0.33 25 214 8.6 5500 1.3 to 1 1.3 to 1 Rock No Zone 3B used. 3.7 Forza & Hancock (1995b)
rockfill, well compacted achieved
Bowling (1979)
1.5 to 2 fine crushed rock, thin Tuff and HEC (1987), Morse (1995)
White Spur Hydro Tasmania Volcanics - Tuff 1989 43 146 3.4 4300 1.3 to 1 1.3 to 1 Rock No Zone 3B used. 3.7
(?) layers, well compacted conglomerate. Morse & Ward (1989)
finer crushed rockfill, Regan (1980)
Melbourne siltstone with Zone 3A - upstream 1/3. Rockfill drainage layer below High densities in
Winneke 1978 1.4 85 1050 12.4 - 1.5 to 1 2.5 to 1 Rock 2.8 watered and well Casinader & Watt (1985)
Water interbedded sandstone random fill (Zone 3B) in downstream 2/3. Zone 2B
compacted MMBW (1975, 1981, 1995)
Pacific Gas &
derrick placed rockfill, 0.7 to Irregular surface Cooke (1958)
Wishon Electricity Co. Glaciated granite 1958 1.75 90 1015 11.3 60400 1.15 to 1 1.4 to 1 Rock Thinner rockfill lifts placed adjacent to upstream face. 2.4 to 3.5
2 m size, voids chinked finish Regan (1997)
San Francisco
Fresh crushed limestone, Peng (2000), Wang et al (1988)
Rock, gravels in high densities
Xibeikou China Sedimentary - limestone 1989 3 95 222 2.3 23000 1.4 to 1 1.4 to 1 Zone 3A - upstream half, Zone 3B - downstream half. 4.1 finer graded than Zone 1, Huang et al (1993)
lower river bed. achieved.
well compacted Wang et al (1993)
Xingo (Zone 3A) Sobrinho et al (2000), Souza et al (1999)

fine crushed rockfill placed
Metamorphic - granitic Rock, gravels in Zone 3A - upstream 1/2; Zone 3B - downstream 1/2 high densities and Eigenheer & Souza (1999)
CHESF, Brazil 1993 6 140 850 6.1 135000 1.4 to 1 1.3 to 1 4.5 to 5.8 in thin layers, well
gneiss, Precambrian lower river bed. Dumped rockfill at downstream toe. low void ratios. Saboya et al (2000)
Xingo (Zone 3B) compacted.
Vasconcelas & Eigenheer (1985)
-8-
Material Parameters / Properties of Rockfill
Rockfill Zone Intact Rock Strength Particle Size Distribution Parameters (from average grading curve) Density Placement Methods
Name Rockfill Particle
Wet/Dry % finer Dry Void Layer
n Authors Source Shape Strength Absorption D10 D30 D50 D60 D80 D90 Dmax % finer Clean / Porosity, Level of
Class n UCS (MPa) Strength Cu Cc 75 Density Ratio, Thickness Water Added Plant
Class Classification (%) (mm) (mm) (mm) (mm) (mm) (mm) (mm) 19 mm Dirty 3 n (%) Compaction
(%) micron (t/m ) e (m)
Greywacke, some angular,
Mackintosh 3A 3A Medium to High 45 - - 52 3.3 1 13 34 52 130 230 1000 3.5 38 dirty 2.20 0.24 19.4 1.0 10% by volume 8p 10t SDVR Good
slate elongated
Mangrove Creek (Zone fresh siltstone & angular to

3A 1B High 45 to 64 40 - 60% 3.0 - 6.7% 310 9.9 0.45 25 88 140 225 280 400 4 27 dirty ? 2.24 0.18 15.5 0.45 to 0.6 5 to 7.5% by volume 4p 10t SDVR Good
3A) sandstone subangular ?
weathered to fresh
Mangrove Creek (Zone angular to dirty (soft 2.06 Placed slightly dry of
3B 4 siltstone & High 26 to 64 40 - 60% 3.0 - 6.7% 330 7.5 0.3 15 60 100 200 275 450 3 32 0.26 21 0.45 4p 10t SDVR Good
3B) subangular ? rock) t/m3 optimum
sandstone
Rhyolite (SW to
Murchison 3A 3A angular Very High 148 - - 19 2.1 5.3 33 75 100 180 290 600 1.5 22 clean (?) 2.27 0.234 19.6 1.0 20% by volume 8p 10t SDVR Good
FR)
171 (80 to
Reece 3A 3A Dolerite angular Very High - - 10 1.6 16 63 120 160 330 475 1000 1 11 clean 2.287 0.29 22.7 1.0 5 to 10% by volume 4p 10t SDVR Good
370)
low 1000 2000 to sluiced, but poor dumped and Poor to Very
Salt Springs 3A 3A granite, Mesozoic angular, blocky Very High 100 to 130 - - - - - - - - low low clean (1.88) 0.41 29 5 to 52
(<10) (approx) 3000 quality poorly sluiced Poor
Salvajina (Zone 3A) 3A 2 natural gravels rounded (Very High) - - - 9.2 1.2 5.2 17 34 48 80 110 400 0 to 13 32 dirty 2.24 0.25 20 0.6 Yes, water added 4p 10t SDVR Good
weak sandstone
Salvajina (Zone 3B) 3B 4 angular ? Medium (?) - - - 45 3.4 1.2 15 41 58 115 190 600 0 to 16 32 dirty 2.26 0.21 17.4 0.9 Yes, water added 6p 10t SDVR Good
and siltstone
22 dry, 13
Scotts Peak 3A 3A argillite - quarried angular ? Medium 60% - 380 6.6 0.21 10.5 44 80 205 370 914 7 38 dirty 2.095 0.266 21 0.915 no water added 4-6p 10t SDVR Good
wet
basalt (<5% Basalt = 235

Segredo (Zone 3A) 3A 1B angular Very High - - 7.4 1.4 - - 177 - - - - - - clean 2.13 0.37 27 0.8 25% by volume 6p 9t SDVR Good
basaltic breccia) Breccia = 37
basalt (<5% High to Very Basalt = 235 2.01 1D - 0.8

Segredo (Zone 3B) 3B 1C & 1D angular - - 10.2 1.4 - - - - - - - - - clean 0.43 31 no water added 4p 9t SDVR Reasonable
basaltic breccia) High Breccia = 37 t/m3 1C - 1.6
Ripped quartz angular to (Medium to Not sure, suspect

Serpentine 3A 3A - - - 210 0.11 0.025 0.12 1.5 5.3 43 76 152 24 69 dirty 2.10 0.262 21 0.6 to 0.9 4p 9t SDVR Good
schist subangular High) likely
Shiroro 3A 2 granite angular (Very High) - - - 32 2.1 4 33 95 128 260 380 500 - 22 clean 2.226 0.20 17 1.0 15% by volume 6p 15t SDVR Good
Limestone, SW to
Tianshengqiao - 1 15 to 90 to
3A 3B FR - spillway angular Very High 70 to 90 (wet) - - - - - - - - 800 - clean 2.19 0.23 19 0.8 20% by volume 6p 16t SDVR Good
(Zone 3A) 20 120
excav.
Tianshengqiao - 1
3B 3C Mudstone angular Medium 16 to 20 - - 40 - - - - - - - 600 - 20 to 35 dirty 2.23 0.21 17.5 0.8 20% by volume 6p 16t SDVR Good
(Zone 3B)
Greywacke, some angular,
Tullabardine 3A 3A High 45 - - 28 1.7 2.8 19 50 78 155 240 400 2.5 30.5 dirty 2.22 0.23 19 0.9 to 1.0 > 10% by volume 4p 10t SDVR Good
slate elongated
Quarried Tuff - SW 37 to 0.18 to SDVR (4p, 10t

White Spur 3A 3A angular (Very High) - - - - - - - - - - 1000 - - - 2.30 15 to 20 1.0 > 10% by volume Good
to FR 200 0.25 ?)
SW to FR suspect 4 - 6p 10t
Winneke 3A 3 angular High 66 - 1.5 - 2.3% 33 1.7 2.8 21 58 91 200 310 800 4 28 2.07 0.302 23 0.9 15% by volume Good
Siltstone, Quarried dirty SDVR
Sluiced with high
550 to 1500 to 8 to 52 dumped and
Wishon 3A 3A granite, blasted angular Very High - - - (<7) - - - - - - - - clean ? 1.80 0.47 32 pressure jets, 300% Poor
750 2000 (variable) sluiced
by volume
Xibeikou 3A 1 Limestone - Fresh angular Very High 240 28 - - - - - 90 - - - 600 - - clean ? 2.18 0.284 22.1 0.8 25 to 50% by volume 8p 12t SDVR Good
High to Very clean to

Xingo (Zone 3A) 3A 3 granite gneiss angular - - - 18 1.6 10 54 150 180 400 500 650 <3 4 to 33 2.15 0.28 21.8 1.0 15% by volume 4p 10t SDVR Good
High (?) dirty
Sound and
Medium to Very 0.3 to
Xingo (Zone 3B) 3B 4 weathered granite angular - - - 80 - - 190 - - - 750 2 to 7 15 to 60 dirty 2.1 0.31 23.6 2 no water added 6p 10t SDVR Reasonable
High (?) 1
gneiss
-9-
SECANT MODULI VALUES HYDROGEOLOGY LEAKAGE
Modulus During, Erc (MPa) Modulus on First Filling, Erf Time for First Filling Rainfall,
Name Maximum
(from end of main average Long-term
Average at EOC Reservoir Level / Comments Rate Comment
Average over rockfill annual (l/sec)
(adjusted for Erf (MPa) Comments (l/sec)
construction period construction) (mm)
valley shape)
45 Face slab profile shows peak deformation is 24% above 8 to 10 (10 years
Mackintosh 39 63 1.92 to 2.93 years. Fluctuating (FSL to -10 m) 2135 21 Max. at end of first filling. Gradual reduction with time.
(35 to 60) toe, this is quite low. post ff)
55 to 60 57.5
3A) 2.5 (15 years
0.6 to > 15 years (had not
- Not reached FSL. First Filling - Oct 81 to ….. - 5.6 Max. leakage when storage level was at its highest (1991)
Mangrove Creek (Zone reached FSL by 1996). post EOC)
46 (36 to 56) 46
3B)
190 560 (485 to Large fluctuations on daily/weekly basis, up 2 (1995/96, 13

Murchison 154 - 1.04 to 1.09 yrs. - 3.5 3.5 l/sec on first filling.
(170 to 205) 640) to 20 to 22 m. years after ff)
190 (175 to 1.32 to 1.46 yrs (Apr to Fluctuating on daily/weekly basis, up to 8 0.5 to 1.5 (8 to 9
Reece 86 (57 to 115) 72 - - 8 to 12 Initial leakage virtually all from right abutment.
205) June 1986). m. years post ff)
Leakage mostly through the upper part of the face slab (upper 35
Calculated from face slab displacements measured after
Salt Springs - - 20 0 to 1.48 years (1931/32). Fluctuating 1145 565 - m) through cracks in face slab, honeycombe pockets in concrete
2 years
and open joints.
205 (175 to 260)

Salvajina (Zone 3A) 130
(likely affected by arching) By He (2000). GJH estimate is 615 MPa when reservoir Approx. 14 l/sec through abutments and 60 l/sec through face
500 0.33 to >0.75 years (1985). no details after first filling - 74 -
at 14 m below FSL. slab.
Salvajina (Zone 3B) 62 (likely affected by arching) 53
Estimated from HSGs near to upstream face.

59 (Zone 3A) 0.42 to 2.58 years (June 72 2 to 4 (1994, 20 At close to FSL on first filling leakage increased from 5 to 100
Scotts Peak 20.5 (18.5 to 23.5) 20.5 Significantly higher modulus for gravels in lower section Steady 2130 100
420 (Zone 2B) to Aug 74). years post ff) l/sec due to cracking in face slab. Gravel blanket placed.
compared with argillite rockfill.
55 (excludes points thought to

Segredo (Zone 3A) 55
be affected by arching) 45 (1.5 to 2 Max leakage on first filling when at FSL. Leakage reduced by
175 - Approx. 0.6 years. - - 390
years post ff) dumping fill on lower portion of face slab.
Segredo (Zone 3B) 28 (25 to 33) 28
Assumption that the river gravels have a significantly

0.24 to 2.94 years (Dec 71 No estimates possible due to inundation of the downstream toe of
Serpentine 92 (46 to 142) 92 97 (94 to 100) greater modulus (than rockfill) and have negligible Steady - - -
to Aug 74). the embankment
contribution to deformation on first filling.
No deformation quoted for upstream face slab on first 1.44 to >1.8 years (from 100 (0.5 years When close to FSL. Cracking in face slab near junction with
Shiroro 66 (61 to 71) 58 - By Oct 1984 not a FSL (10 m short) - 1800
filling. 5/84). after max.) plinth.
Tianshengqiao - 1
49 (40 to 57) 49
(Zone 3A) Not yet filled to FSL. Stored water during construction First filling started 1.3 years before end of Max at end Dec 1999 when reservoir at highest level. March
- -1.5 to >0.8 years. - 53 -
to within 40 m of FSL. construction. 2000 ~ 30 l/sec. Repairs to cracks in face slab.
Tianshengqiao - 1
37 (32 to 42) 37
(Zone 3B)
2.05 to 2.35 years (5/81 to 0.5 to 1 (10 -12
Tullabardine 74 74 170 - Fluctuating (FSL to -10 m) - 1.5 to 2 Current base flow less than 1 l/sec.
8/81). yrs post ff)
180 On virtually date of reaching FSL. 1 year later E = 200 to 0.18 to 0.24 years (June to 2 (6 years post Limited leakage. &l/sec shortly after first filling. Gradual
White Spur 139 340 Fluctuating 3150 7
(160 to 200) 240 MPa. July 1989). ff) reduction to approx. 2 l/sec.
1.62 to 5.04 years (6/80 to Fluctuating - seasonal (no catchment, off 13 (10 to 12 Max. leakage when reservoir at FSL on first filling. Gradual
Winneke 55 (50 to 59) 55 104 Determined from HSGs close to upstream face. 700 to 1200 58
11/83). stream storage) years post ff) decrease in leakage rate since.
Maximum within 2 months of first filling. Due to cracking in face

0.4 to 0.45 years (May
Wishon - - - - - - 3120 850 slab, mainly at the perimeter joint. Series of repais to face slab
1958).
over the years.
Impounded water during construction.
Query on deflected shape of face slab after September -1.5 to 6 years (June
Xibeikou 80 (60 to 100) 80 260 Overtopped in 1987. Other problems 1150 - - -
1993 1995).
during construction.
Xingo (Zone 3A) 34 (30 to 39) 34

Steady. Springs observed on downstream
1.02 to 1.46 (mid to late 140 (4.5 years Leakage increased post ff from 100 to 200 l/sec. Due to cracks in
76 (73 to 80) Calculated for measurements taken on the left abutment. slope of left abutment (associated with - 200
1994) years. post ff) slab. Dumping soils on face reduced leakage to 140 l/sec.
Xingo (Zone 3B) 13 (12 to 14) 13 seepage through the face slab)
- 10 -
MONITORING
Settlement Post Construction Crest Deformation Long Term Crest
Deformation Normal
Name During
Settlement Settlement Lateral to Face Slab
Settlement Rate
Comments
Other Comments
Construction Settlement, total Settlement, (% per log time cycle)
on FF on FF Displacement (mm)
(mm) (mm) total (%) (t o at EOC)
(mm) (%) (mm)
333 mm (0 to 20.6 Decreasing rates of deformation with time, approximately 5 mm/yr settlement and 5
Mackintosh > 780 0.444 99 0.132 130 mm (75 ff) 228 mm (173 ff) 0.184
years) to 6 mm/yr displacement at 10 yrs.

3A) 0.251 (after 5 years, but still First monitoring of crest SMPs 0.67 years after end of rockfill construction.
287 mm (0.67 to 15 196 mm (from 12/81 Moduli during construction calculated from HSG deformation during construction and also
> 610 0.359 > 287 > 0.359 - not reached full supply Settlement rates with time still high to 1990 (highest water level) then reduced from
Mangrove Creek (Zone yrs, 12/81 to 5/96) to 5/96) taken from published values.
level) 1990 to 1996.
3B)
104 mm (0.08 to 17.6 77 mm to 4/93 (28 ff 67m

Murchison > 255 0.111 9 0.01 22 mm to 4/93 (8 ff) 0.056 Decreasing rates of movement with time, approximately 3mm/yr at 10 yrs
yrs, to 1999) bc)
221 mm (0.12 to 15 68 mm (5/86 to 264 mm to 8/94 (215 ff Decreasing rates of movement with time, approximately 4 mm/yr crest settlement
Reece > 820 0.181 85 0.07 0.063
yrs, to end 1999) 11/93), all post ff 66m bc (45%)) at 5 yrs
Rapid deformation on first filling with significant reduction in rate thereafter.

1276 mm to 1996 550 mm to 1958 (27 1755 mm to 1958 (67 m
Salt Springs - 1.276 380 0.38 0.29 to 0.77 Acceleration at about 20 years when the full supply level was raised by 3 m. Crest
(0.26 to 65 yrs) yrs), ff 230 mm bc), ff 1317 mm
SMPs on upstream edge of crest.
Salvajina (Zone 3A)

90 mm (0.33 to 0.75 55 mm to 14 m short of Constructed of dirty gravels with rockfill used in the downstream 1/3. High stiffness as
- 0.061 > 90 > 0.061 - - Limited data.
years) FSL indicated by high moduli. Limited deformation data in published literature.
Salvajina (Zone 3B)
445 mm (0 to 18 190 mm to 12/89 178 mm to 12/89, 78 mm High rate of deformation post first filling due to leakage through dam and wetting up Cracking occurred as a result of tensile stresses in face slab due to very large differential
Scotts Peak - 1.036 203 0.472 0.178 (after 5 years)
years, to 12/89) (17 yrs), ff 96 mm ff of argillite rockfill. Long term rates approximately 2.5 mm/yr. stiffness in rockfill.
Segredo (Zone 3A)

229 mm (to 0.4 yrs 340 mm (to 4 months Limited information on deformation. Estimation of crest settlement from HSG
- 0.158 Approx. 200 0.138 - -
post ff, 1992) post ff) device below crest.
Segredo (Zone 3B)
77 mm (0.24 to 25.5 38 mm (12/71 to

Serpentine - 0.203 35 0.092 - 0.109 -
yrs, to 1/97) 1/92), ff 23 mm
Modulus during construction taken from Bodtman and Wyatt (1985), however they may
166 mm (0 to 1.8 27 mm to 12/84 (1 90 mm (when 32 m Limited deformation information. Could find nothing on deformation behaviour after
Shiroro - 0.133 > 66 > 0.053 - not have allowed for reduction in the vertical stress due to embankment shape.
years) to 12/84 yr) below FSL) 1984.
Therefore, the values are likely to over-estimate the modulus.
Tianshengqiao - 1 During the early stages of construction the embankment (~40 m high) was overtopped 4
670 mm to Dec
(Zone 3A) 926 mm (0.05 to 0.8 times during floods. Cracks noted in slab cushion zone during construction. Face slab
- 0.52 > 926 > 0.52 1999 (0.75 years), ff - - First filling not completed.
years, to Dec 1999) cracking on first filling. Separation between face slab and cushion zone was noted during
Tianshengqiao - 1 not completed.
staged construction of the face slab, up to 100 mm gap.
(Zone 3B)
19 mm (0.2 to 12.8
Tullabardine - 0.076 2 0.01 - - 0.023 Negligible post construction settlement
years)
58 mm (0.04 to 5.9 38 mm (@ 6 years), 15 Rate of settlement reduced after completion of wave wall in January 1990. At 5yrs
White Spur > 65 0.135 7 0.016 - 0.08
yrs, 4/89 to 2/95) mm ff crest settlement rate approx. 1.25 mm/yr.
207 mm (0.17 to 16.2
M7 - 160 mm (3/80 to 2/94), Monitoring started shortly after end of rockfill construction (SMP 13 on upstream
Winneke - yrs, Jan 79 to Jan 0.244 105 0.124 - 0.107
145 mm ff (all settlements) face used until crest SMP installed). Deceleration of settlement after first filling.
95)
Monitoring started shortly after end of rockfill construction. Rapid movement on
Wishon - 954 mm (0 to 38 yrs) 1.136 189 0.233 - - 0.25 to 0.33 No data to calculate modulus during construction or on first filling
first filling with significant reduction in rate thereafter.
Limited data available. Monitoring shows discrepancies in face slab deformation, and
Xibeikou - - - - - - 75 mm (6.5 years), all ff - Query on results. Discrepancies between SMPs and HSGs in published literature.
between surface monitoring points and hydrostatic settlement gauges.
Xingo (Zone 3A)

SMPs base survey approx. 1 year after the end of rockfill construction. Cracking of Wetting of rockfill in left abutment resulted in acceleration in the rate of deformation (in
320 mm (1.0 to 6.2 510 mm (@ 6 years),
- 526 (1.0 to 6.2 yrs) 0.376 302 0.216 0.25 face slab on left abutment and subsequent leakage resulted in acceleration of this area). Springs observed high above river level on downstream slope of embankment
yrs), 210 mm ff 290 mm on ff.
Xingo (Zone 3B) deformation of face slab and crest (i.e. wetting of rockfill). (on the left abutment) indicative of wetting of Zone 3B rockfill.

The Deformation Behaviour of Rockfill CFRD

Uploaded by

Copyright:

Available Formats

The Deformation Behaviour of Rockfill CFRD

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Deformation Behaviour of Rockfill CFRD

Uploaded by

Copyright:

Available Formats

ISSN 0077-880X

THE DEFORMATION BEHAVIOUR OF ROCKFILL

G. HUNTER and R. FELL

UNICIV REPORT No. R-405 JANUARY 2002

ISBN : 85841 372 8

BIBLIOGRAPHIC DATA SHEET ISSN 0077-880X

4. Title THE DEFORMATION BEHAVIOUR OF ROCKFILL

6. Keywords Concrete face rockfill dam, rockfill, deformation

The deformation behaviour of rockfill during construction, on first filling and

2.0 LITERATURE REVIEW OF ROCKFILL DEFORMATION................................................6

3.0 ANALYSIS OF DEFORMATION BEHAVIOUR OF CFRD................................................18

4.0 DISCUSSION AND RECOMMENDED METHODS FOR PREDICTION .........................48

5.0 CONCLUSIONS .........................................................................................................................52

Appendix A: CFRD Case Study Information

Table 1.1: Classification of unconfined compressive strength or rock (AS 1726-1993).........................................1

Figure 1.1: Determination of rockfill moduli E rc and E rf (Fitzpatrick et al 1985) ..............................................3

1.1 USE OF ROCKFILL, AND OBJECTIVES OF THIS STUDY

1.2.1 Properties of Rockfill

1.2.2 Rockfill Moduli

Figure 1.1: Determination of rockfill moduli E rc and E rf (Fitzpatrick et al 1985)

1.2.3 Zoning of Main Rockfill in Concrete Face Rockfill Dams

2.0 LITERATURE REVIEW OF ROCKFILL DEFORMATION

2.1 HISTORICAL SUMMARY OF ROCKFILL USAGE IN EMBANKMENT DESIGN

2.1.1 Deformation Properties of Rockfill

2.2 PREDICTIVE METHODS FOR ROCKFILL DEFORMATION

2.2.1 Finite Element Analyses

2.2.2 Empirical Predictive Methods of Deformation Behaviour During Construction

2.2.3 Predictive Methods for Deformation Behaviour Post Construction

∆H = α (log t 2 − log t1 ) Equation 2

strength rockfill 2 8. Foz do Areia

3.0 ANALYSIS OF DEFORMATION BEHAVIOUR OF CFRD

3.1 CASE STUDY DATABASE

UCS = 25 angular to 4p 10t SDVR, 100% 0 to 1.95 yrs (Sept 69 to Aug

UCS = 8 - 14 angular, 4p 9t SDVR, no water 0.58 to 2.83 years (Aug 77

UCS = 45 angular, 8p 10t SDVR, 10% 45

Khao Laem UCS < 190 4p 10t SDVR, 15%

El-Infiernillo diorite & silicified UCS = 125 Dumped and spread,

3.2 DEFORMATION DURING CONSTRUCTION

3.2.1 Base Plots of Data

3.2.2 Effect of Compaction/Placement Method

Figure 3.1: Stress-strain relationship for rockfills observed during construction

3.2.3 Effect of Confining Stress

3.2.4 Effect of Valley Shape

Abutment Slope Angle (degrees) .

Abutment Slope Angle (degrees) .

The results of the analysis indicate:

3.2.5 Effect of Particle Size Distribution

E rc = 153 ⋅ e (−0.0045⋅D90 ) for medium to high strength well-compacted rockfill Equation 6

3.2.6 Moduli During Construction of Dumped Rockfill

3.3 DEFORMATION OF THE FACE SLAB OF CFRD ON FIRST FILLING

3.3.1 Simplified Methods of Estimation of Maximum Face Slab Deformation

3.5 Estimation of trendline for upstream

Figure 3.8: E rf E rc ratio versus embankment height

Other factors considered to affect the E rf E rc ratio are:

Points normal to face slab at

Points normal to face slab

3.3.2 Shape of the Face Slab Displacement

Figure 3.14: Scotts Peak dam (courtesy of HEC Tasmania)

3.3.3 Prediction of Face Slab Deflection on First Filling

3.4 POST CONSTRUCTION CREST SETTLEMENT