Testing of Cyclone Sand Tailings at High
Stresses
Matthias Busslinger, PEng, MASc; Howard Plewes, PEng, MSc; Graham Parkinson, PGeo, BSc.
Klohn Crippen Berger Ltd., Vancouver BC, Canada, mbusslinger@klohn.com
ABSTRACT
Over the past four decades, tailings dams have steadily increased in height with new proposed designs reaching close to
300 m. This paper describes research carried out for the pre-feasibility design of a centerline cyclone sand tailings dam
for a Gold-Copper porphyry mine. Cyclone sand dams built by the centerline method require (a) sufficient shear strength
and (b) free drainage in the cyclone sand. Shear strength is required for shell stability, whereas free drainage reduces
liquefaction potential by lowering the phreatic surface. Behaviour of cyclone sands under high dam loads (up to
5,250 kPa) was examined in triaxial and oedometer tests to characterize material deformation, particle crushing effects,
shear strength, and permeability. High shear strengths (effective stress friction angle 38º) were found despite observing
net contractive behavior at higher stresses. The cyclone sand was found to crush (increase in fines) under elevated
stresses, especially under shear deformations.
RÉSUMÉ
Durant les quarante dernières années, la hauteur des barrages de retenue des résidus miniers a progressivement
augmenté pour atteindre environ 300 m dans les nouveaux projets. Cet article décrit la recherche entreprise pour l’étude
de préfaisabilité d’une barrage à base de sable cyclone dans le cadre d’une mine de porphyre cuivre-or. Les barrages
construites à base de sable cyclone avec la méthode de construction longitudinale nécessitent (a) suffisamment de
résistance au cisaillement et (b) le drainage libre dans le sable cyclone; la résistance au cisaillement est requise pour la
stabilité de l’enveloppe de la barrage, et le drainage libre permet de réduire le risque de liquéfaction en abaissant le
nappe phréatique. Le comportement du sable cyclone sous haute pression (jusqu’à 5,250 kPa) a été étudié avec des
essais triaxiaux et oedométriques pour caractériser sa déformation, l’effet du broyage des particules, sa résistance au
cisaillement et sa perméabilité. Les résultats montrent que la résistance au cisaillement est élevée (angle de frottement
effectif de 38°), malgré que le matériel soit sujet à contraction. A pression élevée, le sable cyclone a tendance à être
broyé en fines particules, en particulier lorsque la déformation est cisaillant.
1
INTRODUCTION
This paper assesses strength and permeability of cyclone
sand for construction of the downstream shell of a 240 m
high tailings dam. Owing to the high stresses occurring
under this dam height, laboratory tests with stresses up to
5,250 kPa were carried out. The dam design requires
sufficient shear strength of the cyclone sand shell, as well
as free draining material properties keeping the phreatic
surface in or near the drains at the base of the dam.
Hence, this paper particularly emphasizes shear strength,
deformation behavior, permeability and effects of particle
crushing under elevated confining stresses.
Flotation tailings for this project will have a gradation
with d80 of 110 μm, and 56% fines passing the #200 sieve
(i.e. < 74 µm). Cyclone sand is produced by feeding
tailings to hydrocyclones, to classify the fines and produce
sand suitable for dam fill. The cyclone sand has target
fines content of 17% facilitating ease of compaction and
providing a permeability of about kv = 5x10-6 m/s (initial
compacted state).
2
TAILINGS DAM DESIGN
Klohn Crippen Berger Ltd. (KCB) is currently designing a
tailings management facility to store 2.3 billion tonnes of
tailings for a mining project in British Columbia, Canada.
Four tailings dams are proposed to provide staged
storage over the mine life of which the tallest dam (240 m)
is described here (Fig. 1).
The starter dams, ranging up to 100 m high, will be
earthfill embankments with shells of random fill supporting
a central till core. The dams will be subsequently raised
by the centerline method with a downstream shell of
cyclone sand supporting the vertical till core along the
centerline. No upstream shell will be built, as tailings are
gradually raised with the dam, providing lateral
confinement of the core. The ultimate dam crest will be
240 m above existing ground with a downstream slope of
3H:1V.
Fig. 1 Cross Section Trough Tailings Dam.
The sand shell will be constructed by hydraulic
sluicing of the sand into cells oriented parallel to the dam
crest. “Dykes” of sand pushed up by dozers confine the
perimeter of the cells. The cells will be nominally 30 m to
50 m wide and 150 m to 300 m long. Cyclone underflow
sand will be sluiced into the cells and compacted by dozer
trafficking. For each year of construction, sand placement
will start at the downstream toe and raise up to required
crest elevation.
Due to climatic constraints at the site, downstream
cyclone sand will be placed during restricted periods:

During the “summer” months of April to October,
tailings will be routed to primary cyclones.
Underflow from the primary cyclones will be
pumped to secondary cyclone stations located
along the dam crest. Secondary cyclone
underflow sands will be conveyed in slurry
pipelines to hydraulic fill cells on the downstream
shell. Cyclone overflow, containing the fines, will
discharge onto the upstream beach.

During the “winter” months of November to
March, flotation tailing will not be cycloned, but
directly discharged into the impoundment via
spigots located every 200 m along the crest of
the dams. This will develop the beach and push
the tailing pond towards the center of the cells.
The deposition will also build up the tailing beach
against the dam to support the construction of
the central till core during the following summer.
Because the final crest elevation will not be achieved
until October at the end of each construction season,
each year’s dam raise will provide the required storage
needed until October of the following year. This ensures
that adequate dam freeboard, tailing and flood storage
capacity is available at all times.
A system of finger drains will be installed at the base
of the cyclone sand shell to keep the water level
depressed within the dam. Main drains in the center of the
valley floor will collect and convey seepage to the toe of
the dam. Smaller secondary drains will convey water
laterally into the main drains. Drains will comprise an
inner core of highly pervious drain rock, surrounded by
sand and gravel filter, to prevent piping of the cyclone
sand or native foundation soils into the drain core.
According to the Canadian Dam Safety Guidelines
(CDA 2007) the dam falls under the “extreme”
consequence classification. Therefore the dam is
designed to withstand an earthquake with a 10,000-year
return period, corresponding to design earthquake ground
acceleration of 0.14 g. The dam provides enough storage
to contain the operating pond and the probable maximum
flood (i.e. 30 day flood with 100-year snowmelt) without
discharge. Three stability scenarios were considered in
design with the following factors of safety (FS) criteria:
FSstatic > 1.5; FSpseudo-static > 1.0; and, FSpost-earthquake > 1.2
assuming all uncompacted tailings in the impoundment
are liquefied during an extreme earthquake and with
allowances for cyclic pore pressures induced in the
cyclone sand and alluvium.
3
TESTING
Metallurgical pilot plant tests were carried out during the
ore processing plant feasibility study. Milling that
simulated high pressure grinding rolling (HPGR) was used
to crush drill core material for pilot plant tests. Tailings
derived from these pilot plant tests were wet sieved to
prepare samples matching the anticipated particle size
distribution of cyclone sand. Laboratory tests were carried
out characterizing the cyclone sand; including USCS soil
classification, particle size distribution (PSD), relative
density, and compaction. Mineralogy was assessed with
quantitative x-ray diffraction (QXRD). Particle shape and
angularity was visually characterized using scanning
electron microscope (SEM).
Uniaxial consolidation behaviour was tested with
oedometers. Four consolidated, undrained triaxial test
were carried out to measure shear strength and material
behaviour. Results from PSD and SEM prior and after
testing were compared to analyze particle breakage under
static and shear stresses.
3.1
Sample Preparation
To replicate the particle size distribution of cyclone sand
produced from the tailings, samples of rougher flotation
tailings were divided into separate parts by dry sieving
(149 μm and 74 μm sieves) and washing. The cyclone
sand sample was reconstituted by adding the separate
components proportionally and gradation was tested
again. Target fines content was 17% with an error margin
of ±2%. The cyclone sand PSD is shown as dashed line
in Fig. 6 (i.e. Cyclone Sand).
Table 1. Cyclone Sand Properties.
Property
USCS Symbol
Specific Gravity Gsa)
Particle Size Distributionb)
% Sand
% Fines
d85
Density
Min. Dry Density ρmin,d
Max. Dry Standard Proctor Density
Max. Dry Density ρmax,d
a)
b)
Description
SM Silty Sand
2.75
83%
17%
0.208 mm
1280 kg/m3
1580 kg/m3
1640 kg/m3
whole tailings, prior cycloning
prior testing
3.1
High Pressure Triaxial Shear Strength and
Permeability Testing
Samples for triaxial shear tests were prepared to an initial
water content of 17% and compacted in a 63.5 mm
diameter split mold with a height to diameter ratio of 2:1
after ASTM D4767-11.6.3. Samples were compacted in 7
lifts to a dry density of about 1630 kg/m3 which equates to
103% Standard Proctor density. The compaction
procedure was carefully conducted to ensure soil density
and permeability was consistent throughout the sample.
De-aired water was used to saturate the samples.
Four isotropically consolidated, undrained triaxial
shear tests were conducted. During the sample
consolidation, permeability tests were conducted at
several confining stresses. Details of the triaxial testing
program are given in Table 3.
Table 3. Triaxial Testing Program.
Relative Density
Relative density of cyclone sand was tested according to
BS1377-4:1990. In contrast to relative density testing
following ASTM D-2049-69, this procedure does not
require a shaking table. A saturated sample is compacted
using a vibratory impact hammer into a confining cell to
obtain maximum dry density ρmax,d. Minimum dry density
ρmin,d is measured in graduated cylinders. Results are
listed in Table 1.
3.2
3.3
Sample Mineralogy
Quantitative X-Ray Diffraction (QXRD) analysis of cyclone
sand material was conducted to quantify the relative
proportions of the stronger quartz and K-spar minerals,
compared with weaker micas and non-phyllosilicates.
Three samples of cyclone sand were reduced to the
optimum grain-size range (i.e. <10 μm) for QXRD
analysis. Continuous-scan X-ray powder diffraction data
were collected and analyzed using International Centre
for Diffraction Database PDF-4. Results from QXRD
analysis are listed in Table 2 and represent relative
amounts of crystalline phases by weight, averaged for the
three samples.
Test
Triax1
Triax2
Triax3
Triax4
p0' (kPa)
1000
2000
3000
4000
Permeability testing at p0' (kPa)
500, 1000
500, 1000, 2000
500, 1000, 2000, 3000
500, 1000, 2000, 3000, 4000
p0' = confining pressure
Samples were sheared at 0.1%/min to axial strain εa of
31% to 34%. Fig. 2 shows deviator stress q and excess
pore water pressure ue, versus axial strain εa. Net
contractive pore pressure response was observed for all
samples. All samples experienced strain softening after
20% strain, but this is partly attributed to sample distortion
due to high strain.
Table 2. Relative Percentage by Weight of Minerals in
Cyclone Sand from Quantitative X-Ray Diffraction
(QXRD).
Mineral
Calcite
Clinochlore (chlorite)
Illite-Muscovite
Plagioclase
K-Feldspar
Quartz
Pyrite
Dolomite-Ankerite
a)
Ideal Formula
CaCO3
(Mg,Fe2+)5Al(Si3Al)O10(OH)8
KAl2AlSi3O10(OH)2
NaAlSi3O8 – CaAl2Si2O8
KAlSi3O8
SiO2
FeS2
Ca(Mg,Fe2+, Mn)(CO3)2
%a)
5
5
32
6
4
47
0
1
average percentage by weight from 3 tests.
Fig. 2. Deviator Stress and Excess Pore Pressure vs.
Axial Strain for Triaxial Tests 1 to 4.
Fig. 3. Permeability Measured During Consolidation
Stages for Triaxial Tests 1 to 4.
Fig. 3 illustrates vertical permeability kv as a function
of effective confining stresses. Permeability reduces with
increasing consolidation stress. Hydraulic conductivities
measured during consolidation stages of triaxial tests
ranged from 5·10-6 m/s to 5·10-7 m/s and decreased with
increasing stress.
3.2
High Pressure Oedometer Tests
Two oedometer tests were carried out to assess potential
for particle breakage under static loads, by measuring
particle size distribution prior and after testing.
Samples for oedometer tests were prepared similar to
triaxial test samples with moisture content of 17%. Test
Oed 5 was carried out in a 63.5 mm diameter cell, while a
50.8 mm diameter cell was used for test Oed 6 to reach
higher consolidation pressures. A height to diameter ratio
of 1:4 was taken to mitigate side friction effects along
oedometer wall. Samples were compacted in 2 lifts to a
dry density of about 1510 kg/m3.
Oed 5 was step-wise consolidated by doubling the
load between each step from an initial vertical effective
stress of 15 kPa to a vertical effective stress of 525 kPa,
followed by unloading. Oed 6 was consolidated in the
same manner as Oed 5. Subsequently, it was reloaded,
by adding 1/3 of the previous load in each step to ultimate
vertical effective stress of 5250 kPa, followed by
unloading to 15 kPa. Results are presented in Fig. 4. The
compression index Cc is 0.22 and the swelling index Cs
for both unloading phases is approximately 0.02.
4
4.1
MATERIAL BEHAVIOUR
STRESSES
UNDER
ELEVATED
Stress Paths
Fig. 5 plots the stress paths in terms of q = (σ1' – σ3')/2
versus p' = (σ1' + σ3')/2, where σ1' and σ3' are the effective
major and minor principal stresses, respectively. Effective
peak friction angle φ’peak is nominally 38º (i.e. sin1(0.6126)).
Fig. 4. Void Ratio vs. Vertical Effective Stress for
Oedometer Tests 5 and 6.
Fig. 2 shows the failure deviator stress q f generally
increased with the effective confining stress applied prior
to shearing p0', with Triax 1 (p0' = 1000 kPa) being the
exception, where qf was greater than in Triax 2 (p0' =
2000 kPa). Undrained shear strength ratio, s u/p0' (where
su is undrained shear strength) are listed in Table 4.
Since, initial densities for all triaxial tests were very
similar, ranging between 1620 kg/m3 to 1630 kg/m3, the
difference in su/p0' is attributed to the higher dilatancy of
the sand at lower confining stresses.
Fig. 5. Stress Paths in p’ vs. q Coordinates for Triaxial
Tests 1 to 4.
Table 4. Undrained Shear Strength Ratio from Triaxial
Testing.
p0' (kPa)
1000
2000
3000
4000
p0' = initial effective confining pressure
su = undrained shear strength
su/p0'
1.1
0.47
0.45
0.49
4.2
Stress Dependency of Permeability
Fig. 3 shows decreasing permeability with increasing
confinement for all samples (solid symbols in Fig. 3).
Overall consolidation was observed to reduce
permeability. Some variation exists and with values
ranging from 5·10-6 m/s to 5·10-7 m/s, but overall the
observed permeability decrease is up to one order of
magnitude.
4.3
Particle Crushing
Particle size distributions were measured prior and after
triaxial and oedometer testing (Fig. 6). Particle size
distributions measured after testing lie above the
distribution measured prior testing (i.e. Cyclone Sand).
Table 5 lists fines content measured with the #200 sieve.
Triaxial samples show appreciable increases in fines
content of 5% to 9% after testing, with the degree of fines
generation increasing with stress level. Changes in fines
content for one-dimensional consolidation tests was
negligible at 525 kPa and 2.5% at 5250 kPa. Collectively,
the results show that shearing plays a larger role in
particle crushing than that contributed by confining stress.
Table 5. Fines Content Measured Before and After
Testing.
Sampled
Sample ID
Before Test
After Test
After Test
After Test
After Test
After Test
After Test
Cyclone Sand
Triax 1
Triax 2
Triax 3
Triax 4
Oed 5
Oed 6
% Fines
>74 µm
17.2
22.5
23.3
23.7
25.8
17.1
19.7
Δ % Fines
>74 µm
N/A
5.3
6.1
6.5
8.6
-0.1
2.5
Δ % Fines = (% Fines After Testing) – (% Fines Before Testing).
Particle crushing due to high stresses on grain
angularity were visually examined using trimmings from
Oed 6 and Triax 4 samples. Prior to SEM analysis, the
entire Oed 6 sample was wet sieved to allow for better
assessment of separate grain size fractions, while only an
approximately 1 mm thick layer of material was scraped
from the shear plane of Triax 4 sample, this sample was
not sieved.
SEM micrographs of cyclone sand sampled before
testing generally show varying degrees of angularity for
fine sand size particles, typically with individual particles
possessing many concavities (Figs. 7a and 7b). Material
below sand size tends to be more elongated with very
sharp asperities in a matrix of rock flour (Fig. 7c). This is
consistent with the products of the milling process,
whereby the larger particles tend to comprise the more
crush-resistant minerals, whereas for example less
resistant muscovites and calcites are expected to be
subject to relatively high breakage. The asperities and
relatively low numbers of inter-particle contacts in silt and
clay size particles (Fig. 6c) cause high point loads, further
contributing to particle crushing of these small particles.
The larger sand particles have more inter-particle
Fig. 6. Particle Size Distributions Before Testing (Cyclone
Sand) and After Testing.
contacts and experience less particle crushing.
SEM micrographs of cyclone sand after loading up to
5250 kPa in the oedometer (Oed 6) are shown in Figs. 8a
and 8b. Fig. 9 is an SEM micrograph for material sampled
from the shear plane after consolidating the sample to
4000kPa and subsequent shearing (Triax 4). Post testing
images show apparently ‘fresh’ particle breakage
surfaces, however based on SEM analysis it cannot be
determined whether these surfaces developed while
testing, or already existed before.
a) Sieved Particles > 149 μm (Micrograph 1c-2c);
b) Passing 149 μm Sieve, but Retained on 74 μm Sieve
(Micrograph 1a-1d);
c) Sieved Particles < 74 μm (Micrograph 2a-2a);
Fig. 7 Cyclone Sand Prior Testing: Scanning Electron
Microscope Micrographs of Particles Sieved Into Three
Fractions.
a) Sieved Particles > 149 μm (Micrograph 1d-1c);
b) Passing 149 μm Sieve, but Retained on 74 μm Sieve
(Micrograph 1b-1e1);
Fig. 8 Oed 6 Material After Testing: Scanning Electron
Microscope Micrographs of Wet Sieved Particles.
Fig. 9 Triax 4 Material from Shear Plane After Testing:
Scanning Electron Microscope Micrograph (3-1c).
5
CONCLUSIONS
Results from geotechnical laboratory tests to assess
suitability of cyclone sand for construction of a 240 m high
cyclone sand dam included:
1. Consolidated undrained triaxial tests. The
effective stress friction angle of the cyclone sand
is 38º. The sand displayed increasing contractive
behavior with higher stress levels. The undrained
strength ranged from a su/p0' of 1.10 at 500 kPa
confining stress to 0.49 at 4000 kPa.
2. Hydraulic conductivities measured during
consolidation stages of triaxial tests ranged from
5·10-6 m/s to 5·10-7 m/s and decreased with
increasing stress.
3. Crushing of the sand under the confining
stresses up to 5250 kPa in the oedometer tests
was measured to contribute up to 2.5% increase
in fines content. Particle crushing under shearing
in the triaxial tests (confining pressure up to
4000 kPa) produced increased fines contents of
5% to 9%, and is consistent with the contractive
response of the sand under shearing.
6
ACKNOWLEDGEMENT
The authors thank Dr. Myles Lawler for his contribution to
the coordination and interpretation of the testing program,
Ms. Kate Lunney for preparing figures and tables, and Dr.
Matthieu Sturzenegger for translating the abstract.
REFERENCES
Canadian Dam Association [2007]. “CDA Dam Safety
Guidelines”. http://www.cda.ca.