Professional Documents
Culture Documents
Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng
a r t i c l e
i n f o
Article history:
Received 1 March 2016
Revised 14 June 2016
Accepted 21 July 2016
Keywords:
Tailings disposal
Rheology
Yield stress
Structure
a b s t r a c t
Mine tailings may be sufficiently dewatered prior to deposition such that they exhibit a yield stress and
therefore they will form gently sloped deposits, which result in a number of advantages from an
engineering perspective. Predicting the slope and the shape of these deposits at the field scale remains
challenging, and is probably the key technical unknown holding back more general adoption of thickened
tailings disposal in the mining industry. Methods for estimating the slope are very sensitive to rheological
data, in particular the yield stress. This paper presents data from rheometry on a gold tailings that
presents yield stress values in the range of 18125 Pa. A rheometer with a vane fixture was used in a
number of techniques, including controlled increments in strain rate to generate a flow curve, stress
relaxation, stress growth and creep techniques. A controlled stress technique was used to simulate the
stress history that the material would experience in the field as the tailings slow to a rest. The measured
yield stress varied substantially (18125 Pa) depending on the measurement method. This result is
explained using a structure based viscosity model, modified from work created by others working on
clays. Ageing and shear rate appear to be significant factors that influence the rheology, though the
mechanism for ageing may be partly due to gravity driven particle settling, as opposed to or in addition
to the buildup of a network structure.
2016 Elsevier Ltd. All rights reserved.
1. Introduction
Dewatering tailings using operations such as tank thickening,
in-line thickening, or filtration, is an increasingly common practice,
which is done to realize several possible benefits, including greater
water recovery and reuse, minimization of waste volume, and minimizing reliance on containment structures (dams or dyke). The
last advantage can be realized if the tailings are sufficiently dewatered to the point where they become a non-Newtonian fluid and
exhibit a yield stress, which allows them to be deposited in a
gently sloped stack. As the stack does not require confinement by
dams, this deposition method avoids the type of catastrophic failures (Recently in Minas Geiras Brazil, October 2015, Mount Polley
in British Columbia, Canada July 2014) that unfortunately occur
with some regularity (ICOLD, 2001). Increasing density comes with
increased cost, and whether a given density is economically viable
depends on site-specific requirements such as transport distance,
topography, rate of mining, and cost and availability of land and
water.
Corresponding author.
E-mail address: shabnammizani@cmail.carleton.ca (S. Mizani).
http://dx.doi.org/10.1016/j.mineng.2016.07.011
0892-6875/ 2016 Elsevier Ltd. All rights reserved.
of the particle size effect (Gawu and Fourie, 2004; Liddell and
Boger, 1996). Moreover, sample disturbance during vane intrusion
is minimized. For instance, Yoon and Mohtar (2013) showed that
large strains (sample disturbance) occurring in bentonite suspensions during cone placement resulted in underestimation of yield
stress values compared to measurements using a vane geometry.
A common issue encountered in yield stress fluids is variability
in the results when different techniques are employed. Variation in
the results are often associated with the time scale of the technique, definition of yield stress and the principle inherited in each
method (Cheng, 1986; Nguyen and Boger, 1992; Nguyen et al.,
2006). Nguyen et al. (2006) compared the yield stress values
obtained on TiO2 suspensions, from 6 different laboratories, each
using several techniques, and observed that while there exists a
variation between different techniques, larger variations were
observed among different laboratories. Considering the timedependency and shear dependency behavior of the tested suspension, Nguyen et al. (2006) proposed that the variations in yield
stress was likely due to differences in sample preparation.
Another approach to estimate this is to simulate deposition in
the laboratory and back-calculate yield stress analytically from
experiments on bench scale flows. Early studies focused on relating
the slope in a flume test to the materials rheology (Sofra and
Boger, 2001; Kwak et al., 2005); however these ignored the fact
that the slope of even bench scale deposits varies with the size of
the deposit. More recently, data from laboratory flume have been
fitted by equations derived from lubrication theory, which implicitly handles scale dependency. Several researchers have developed
equations for the profiles of static deposits of a Bingham fluid using
lubrication theory to reduce the Navier-Stokes equation (e.g., Yuhi
and Mei, 2004). These equations have been successfully applied to
bench scale deposits for hard rock tailings and finer-grained tailings from overburden (Simms, 2007; Henriquez and Simms,
2009; Mizani et al., 2013). Mizani et al. (2013), however, showed
that the yield stress back-calculated from such methods may be
dependent on deposition time, as the tailings begin to settle even
as they flow. This settling behavior leads to consideration that
the apparent yield stress of the material is dependent on shear history and ageing, as explored by Coussot et al. (2002a,b) for clays.
This paper presents measurements of the yield stress of a high
density gold tailings using a number of different techniques. As
will be shown, the variation in rheology measured by the different
techniques can be explained by shearing and ageing phenomena,
and can be modelled using a structure based viscosity model, similar to the model proposed by Coussot et al. (2002a).
41
42
other time-independent materials such as mineral tailings suspensions (Clayton et al., 2003; Pashias et al., 1996; Boger, 2009).
In these tests, the samples were mixed (250 rpm for 1 min in a
static mixer) then placed In the cylindrical slump containers
(Height and diameter of 0.30 m), which were immediately inverted
and then raised.
As per Pashias et al. (1996), the yield stress was calculated from
the slump by the following equation:
sy
1 1 p0
s qgH
2 2
3. Results
3.1. Slump tests
Yield stress values for a range of densities estimated by the
slump test are shown in Fig. 3. The values are well-constrained
for wet samples (Cs < 74%). The variability in yield stress at the
drier end is consistent with results reported by other authors for
tailings (e.g. Gawu and Fourie, 2004). For 72% solids the yield stress
is 30 Pa.
2.2.2. Rheometry
All rheometry data were obtained using an Anton Paar Physica
MCR Rheometer and employing an air mounted vane fixture. The
vane consists of four thin blades arranged around a central shaft,
of height 40 mm and diameter of 22 mm. Immediately after mixing
of the sample to the target solids concentration (using a static
mixer at a setting of 250 rpm for one minute), the sample was
poured into a cylindrical sample holder (part number CC27) with
a diameter of 28.92 mm. The vane was always submerged by a
minimum of 2 cm depth of tailings throughout all tests. The small
gap size between cup and vane will reduce the potential for non
Newtonian shear rate effects (Klein et al., 1995). However, the
gap size is several times larger than the largest particle which
prevents larger particles bridging the gap and jamming the
apparatus.
The techniques employed to measure the yield stress were:
0.45
0.5
0.4
0.35
0.3
0.25
0.2
50
100
150
200
250
300
180
160
140
120
100
7.5 sec
80
60 sec
60
40
20
0
50
100
150
200
250
Shear Rate (1/s)
300
350
400
450
Fig. 4. Flow curves measured over 7.5 s and 60 s for tailings at 72% solids.
43
140
120
100
First Ramp up (5 1/s)
80
60
40
20
0
0
10
120
shear rate=15/s
shear rate=5/s
100
shear rate=10/s
80
60
40
off
off
off
20
0
0
500
1000
1500
Time (s)
Fig. 6. Stress relaxation test for tailings at Cs = 72%.
2000
2500
44
10000
150 Pa
130 Pa
120 Pa
115 Pa
110 Pa
100 Pa
50 Pa
20 Pa
Viscosity (Pa.s)
1000
100
10
1
0
10
20
30
40
50
60
70
Time (s)
Fig. 7. Viscosity-time profile for sample prepared at 72% solids.
140
120
Measurd 0.1 / s
100
Measured 10 / s
Measured 100 / s
80
60
40
20
0
0.01
0.1
10
100
Time (s)
Fig. 8. Stress growth tests for samples at 72% solids.
1000
Viscosity (Pa.s)
Measured 100(1/s)
100
Measured 10(1/s)
Measured 0.1(1/s)
10
20
40
60
80
100
Time (s)
Fig. 9. Viscosity plots for samples at 72% solids for different rotational speed.
120
45
7000
Strain (%)
6000
5000
4000
3000
150 (Pa)
125(Pa)
100(Pa)
75(Pa)
50(Pa)
40(Pa)
30(Pa)
20(Pa)
10
2000
1000
0
0
10
20
30
40
50
60
70
80
90
Time (s)
Fig. 10. Constant decreasing stresses applied to samples prepared at 72% solids. Each stress was applied for 10 s.
1,000
Viscosity (Pa.s)
100
10
75 Pa (10s)
100 Pa (10s)
125 Pa (10s)
150 Pa (200s)
125 Pa (200s)
100 Pa (200s)
150 Pa (30min)
125 Pa (30min)
0
1
10
100
Time (s)
1000
10000
Fig. 11. Viscosity versus time in three controlled decreasing stress tests on samples at 72% solids, initially sheared at 150 Pa and subsequently sheared at values decreasing by
25 kPa. The duration of each shear step was 10 s, 200 s, and 30 min in the three different tests.
the vane was recorded when the stress reached 30 Pa (strain rate of
0 (s1)). This test was also repeated at different shearing times,
where stress was held for 200 s and 30 min for each step. The viscosity profile for all three tests is shown in Fig. 11, which shows
jumps in viscosity at 75, 100, and 125 Pa.
4. Comparison between different yield stress measurements
Table 1 summarizes the results obtained using different rheometry techniques and slump test.
While there is some consistency for those tests that impose
increasing value of shear and strain (stress growth, controlled
shear in increasing steps, ascending flow curve) for a value of
90125 Pa, there is clearly a large discrepancy with the results of
tests that decrease shear and strain (slump, stress relaxation, controlled decreasing shear). For these latter tests, the duration of the
test affects the measurement, with tests with shorter duration
having the lowest yield stress value: The stress relaxation test is
almost instantaneous, the slump lasts a few seconds.
5. Discussion
To explain the results, two viscosity models were empirically fit
to the rheological data. Both are variants on the model developed
by Coussot et al. (2002a). In this model, transition from solid-like
to liquid-like behavior is described by viscosity bifurcation rather
than by a yield stress. Viscosity is related to a structure term,
which is itself governed by ageing (rate of restruction) and shear
rate (rate of destruction), as per the following equations:
g g0 1 kn
dk 1
ac_ k
dt h
46
Table 1
Yield stress estimates from different rheometry techniques for gold tailings at
Cs = 72%.
Technique
Yield stress
(Pa)
125
4552
1822
115
90
75, 100,
125a
30
a
For stress steps of 10 s, 200 s and 1800 s respectively and times before viscosity
jump of 30 s, 400 s, and 1800 s.
levels, from 20 Pa to 150 Pa. Three distinct regions may be identified. For stress levels below 100 Pa, viscosity increases to the point
where flows come to an apparent stoppage. For stress levels above
150 Pa, the viscosity decreases monotonically over time. Between
these two values, the viscosity alternates between increasing and
decreasing states. It is apparent from Fig. 12 that the Coussot
model is not capable of capturing the material behavior over this
intermediate range, despite capturing the bifurcation of viscosity.
This maybe due to the assumption of a single characteristic time
for the ageing term (Coussot et al., 2002b). It is noted that the bentonite suspensions used by Coussot et al. (2002b) showed an opposite behavior to the above gold tailings in the intermediate range.
That is, for that material, the viscosity decreased initially but dramatic ageing was seen over longer times of shearing over a longer
period of shearing to a point where shear rate approaches zero.
One potential source of error for application of the Coussot
model, and indeed, a potential explanation for some of the variance
in yield stress reported by the different techniques, is the possibility of coarse particles settling out of the suspension. As reported by
Klein et al. (1995), settling in concentrated suspensions may affect
rheometry measurements if grain size segregation occurs. In order
to investigate particle separation in the gold tailings, two samples
were collected at the top and bottom of the cup after 5 min of
shearing at a constant shear stress of 150 Pa. These samples were
analyzed for grain size distribution using sieve and hydrometer
methods (ASTM D 422-63, 2002), and the results are shown in
Fig. 13. No significant segregation was observed despite a settlement of 12 mm that occurred during the shearing of the tailings.
20 Pa
50 Pa
10000
100 Pa
115 Pa
Viscosity (Pa.s)
120 Pa
1000
130 Pa
150 Pa
Coussot 50 Pa
100
Coussot 100Pa
Coussot 150Pa
Coussot 200Pa
10
0.1
10
100
Time (s)
Fig. 12. Viscosity versus time as predicted by the Coussot model and as measured by the rheometer, for h = 0.5 and g0 = 1.2.
Fig. 13. Particle size distribution (PSD) of particles collected at the top and bottom after shearing the sample at 150 Pa for 5 min.
47
dk kmax k
ac_ k
dt
h
This ageing term causes the rate of structure buildup to be faster at low values of structure and to decrease as the structure term
increases, thereby reflecting the effect of density changes due to
settling. The value for g0 was assumed to be equal to the final viscosity reached after substantial time at the high shear rate of
150 Pa. The value of the ageing term was found from the stress
relaxation test (when the vane is stopped and shear rate is zero).
Values for kmax, and a were then obtained through calibration so
as to arrive at best fits for all the creep tests (Fig. 13) and stress
growth tests (Fig. 14). When compared to the original model, the
fits in the 100130 Pa range are improved. The modification does
two important things: first, all viscosity values will eventually tend
to a different final viscosity value, depending on the constant shear
stress value, second, this limits the maximum viscosity reached for
the low shear stresses (100 Pa and lower).
In Fig. 15 the modified model is applied to the stress growth
data presented in Fig. 8. A very good fit can be obtained to the data
from the two lower shear rates (0.1 and 10 s1), and somewhat of
an overestimate (15%) of peak stress for the highest shear rate
160
140
120
Shear Stress (Pa)
100
80
60
Measurd 0.1 / s
Predicted 0.1 / s
Measured 10 / s
Predicted 10 / s
Measured 100 / s
Predicted 100 / s
40
20
0
0
10
20
30
Time (s)
40
(100 s1), using the same parameters used to model the creep data,
excepting a lower structure-less viscosity value (0.45 Pa s instead
of 0.5 Pa s). Using a value of 0.5 Pa s still results in a relatively good
fit for the 0.1 s1 data and 10 s1 data (overestimate of peak stress
by 8 and 10 Pa respectively, less than 10%), but a poorer fit for the
100 s1 data (overestimate of peak stress by 30 Pa, or 24%).
Though both the original Coussot model and the modified
model are not perfect, they reasonably explain the ageing and
shearing processes in the tailings. While the use of Coussot model
is empirical, viscosity bifurcation appears to explain the discrepancy between measurements of yield stress by the different
rheometry techniques.
6. Practical implications
Most commonly used methods to predict tailings geometry
assume constant rheology, and often a constant yield stress and
viscosity. However, this study has shown that the apparent yield
stress and viscosity of gold and most likely other hard rock tailings
20 pa
50 Pa
10000
100 Pa
115 Pa
Viscosity (Pa s)
120 Pa
1000
130 Pa
150 Pa
100
Mode 50 Pa
Model 100 Pa
Model 120 Pa
10
Model 130 Pa
Model 150 Pa
1
1
60
Fig. 15. Modelled stress growth tests, using the same parameters for Fig. 10 except
g0 = 0.45 Pa s.
100000
0.1
50
10
100
Time (s)
Fig. 14. Creep tests modelled using modified model (g0 = 0.5 Pa s, n = 1, kmax = 104, h = 100, a = 0.43).
1000
48
Cheng, D.C.-H., 1986. Yield stress: a time-dependent property and how to measure
it. Rheol. Acta 25, 542554.
Clayton, S., Grice, T.G., Boger, D.V., 2003. Analysis of the slump test for on-site yield
stress measurement of mineral suspensions. Int. J. Miner. Process. 70 (3), 321.
Coussot, P., Nguyen, Q.D., Huynh, H.T., Bonn, D., 2002a. Avalanche behaviour in yield
stress fluids. Phys. Rev. Lett. 88 (17).
Coussot, P., Nguyen, Q.D., Huynh, H.T., Bonn, D., 2002b. Viscosity bifurcation in
thixotropic, yielding fluid. J. Rheol. 46 (3), 573589.
DAvino, G., Maffettone, P.L., 2015. Particle dynamics in viscoelastic liquids. J.
Nonnewton. Fluid Mech. 215, 80104.
Dzuy, N.Q., Boger, D.V., 1983. Yield stress measurement for concentrated
suspensions. J. Rheol. 27 (4), 321349.
Fitton, T.G., Bhattacharya, S.N., Chryss, A.G., 2008. Three-dimensional modelling of
tailings beach shape. Comput.-Aided Civ. Infrastruct. Eng. 23, 3144.
Gawu, S.K.Y., Fourie, A.B., 2004. Assessment of the modified slump test as a measure
of the yield stress of high-density thickened tailings. Can. Geotech. J. 41, 3947.
Henriquez, J., Simms, P., 2009. Dynamic imaging and modeling of multilayer
deposition of gold paste tailings. Miner. Eng. 22 (2), 128139.
Hewitt, D.R., Balmforth, N.J., 2013. Thixotropic gravity currents. J. Fluid Mech. 727,
5682.
ICOLD, 2001. Tailings dams, risk of dangerous occurrences. Bulletin 121: Lessons
learnt from practical experiences Paris, p. 14.
Keentok, M., 1982. The measurement of the yield stress of liquids. Rheol. Acta, 325
332.
Klein, B., Laskowski, J.S., Partridge, S.J., 1995. A new viscometer for rheological
measurements on settling suspensions. J. Rheol. 39, 827.
Kwak, M., James, D.F., Klein, K.A., 2005. Flow behaviour of tailing paste for surface
disposal. J. Miner. Process. 77 (3), 139153.
Liddell, P.V., Boger, D.V., 1996. Yield stress measurement with the vane. J.
Nonnewton. Fluid Mech. 63 (23), 235261.
Magnin, A., Piau, J.M., 1987. Shear rheometry of fluids with a yield stress. J.
Nonnewton. Fluid Mech. 23, 91106.
McPhail, G., 2008. Prediction of the beach profile of high density thickened tailings
from rheological and small scale trial deposition data. In: Proceedings 11th
International Seminar on Paste and Thickened Tailings (Paste08), Perth,
Australia.
Mizani, S., He, X., Simms, P., 2013. Application of lubrication theory to modeling
stack geometry of high density mine tailings. J. Nonnewton. Fluid Mech. 198,
5970.
Nguyen, Q.D., Boger, D.V., 1992. Measuring the flow properties of yield stress fluids.
Annu. Rev. Fluid Mech. 24, 4788.
Nguyen, Q.D., Akroyd, T., De Kee, D.C., Zhu, L., 2006. Yield stress measurements in
suspensions: an inter-laboratory study. Korea-Australia Rheol. J. 18 (1), 1524.
Pashias, N., Boger, D.V., Summrs, J., Glenister, D.J., 1996. A fifty cent rheometer for
yield stress measurement. J. Rheol. 40 (6), 11791189.
Roussell, N., Coussot, P., 2005. Fifty-cent rheometer for yield stress measurements:
from slump to spreading flow. J. Rheol. 49 (5), 705718.
Simms, P., 2007. On the relation between laboratory flume tests and deposition
angles of high density tailings. In: Proceedings of the 10th International
Seminar on Paste and Thickened Tailings. Australian Centre for Geomechanics,
Perth, Australia.
Simms, P., William, M.P.A., Fitton, T.G., Mcphail, G., 2011. Beaching angles and
evolution of stack geometry for thickened tailingsa review. In: Proceedings of
the 14th International Seminar on Paste and Thickened Tailings (Paste 2011).
Australian Centre for Geomechanics, Perth, Australia.
Sofra, F., Boger, D.V., 2001. Slope prediction for thickened tailings and pastes. In:
Proceeding of the Eighth International Conference on Tailings and Mine Waste,
Fort Collins, Balkema, Rotterdam.
Yoon, J., Mohtar, C., 2013. Disturbance effect on time-dependent yield stress
measurement of bentonite suspension. Geotech. Test. J. 36 (1), 110.
Yuhi, M., Mei, C.C., 2004. Slow spreading of fluid mud over a conical surface. J. Fluid
Mech. 519, 337358.