Professional Documents
Culture Documents
DOI 10.1007/s10706-006-9103-6
ORIGINAL PAPER
Received: 11 October 2005 / Accepted: 18 August 2006 / Published online: 25 October 2006
Springer Science+Business Media B.V. 2006
123
192 Geotech Geol Eng (2007) 25:191202
Atterberg limits (Gupta and Larson 1979; Therefore, research was focussed on the devel-
Aubertin et al. 1998; Fredlund et al. 1997; Vana- opment of an experimental technique using large
palli and Lobbezoo 2002). tailings specimen in the laboratory to determine
Many of the prediction procedures available in the flow behaviour. A permeameter was specially
the literature that use the saturated coefficient of designed and an experimental program was
permeability, ksat, and the SWCC provide undertaken. The features incorporated in the new
reasonably good comparisons between the mea- device in comparison to a conventional perme-
sured and predicted values of the coefficient of ameter include the use of a large specimen and
permeability, kunsat of unsaturated soils. These monitoring specimen consolidation while collect-
comparisons are usually made between the ing the flow characteristics data using non-
predicted results and the laboratory test data destructive testing techniques. Design details of
determined using small size specimens (approxi- the new permeameter are described in this Paper.
mately 5063.5 mm diameter and 20 mm height).
Small size specimens are used in laboratory testing
for various reasons that include the availability of 2 Background
testing equipment, testing time and fewer logistic
requirements. Several factors such as the density, Several experimental methods are available to
soil structure, compaction water content, stress determine the coefficient of permeability of
state, mineralogy, and hysteresis influence the unsaturated soils (Klute 1972; Elzeftawy and
SWCC behaviour (Vanapalli et al. 1999; Zapata Mansell 1975; Raimbault 1986; Abu-Hejleh et al.
et al. 2000). The parameters that influence the 1993; Amraoui et al. 1998; Meerdink et al. 1996;
SWCC behaviour also influence the flow behav- Benson and Gribb 1997). Benson and Gribb 1997,
iour in unsaturated soils. have summarized 14 different techniques that
A number of studies have shown significant include both laboratory and field methods for
differences between the predicted unsaturated determining the coefficient of permeability of
flow behaviour using the SWCC and the field unsaturated soils. The laboratory methods are
measured unsaturated flow behaviour (Meerdink based either on steady state or transient state flow
et al. 1996; Amraoui et al. 1998). The differences conditions. The steady state methods require a
between the predicted and measured flow series of tests to obtain water content, w, and soil
behaviour can be attributed to 3-D seepage pat- suction, w, data to interpret the unsaturated flow
tern and more representative nature of the soil behaviour (Klute 1972; Benson and Gribb 1997).
mass being tested. The limitation of non-repre- The main drawback associated with these methods
sentative size of the specimen used in the labo- is the long time period required to attain
ratory to estimate the flow behaviour in the field, equilibrium conditions to properly assess the flow
in both saturated and unsaturated soils has now behaviour characteristics.
been well recognised (Daniel 1984; Garga 1988; The centrifuge method is not feasible for rou-
Elsbury et al. 1990; Benson et al. 1994). For this tine geotechnical testing as it requires a com-
reason, many researchers recommend reliance on mercial centrifuge and elaborate testing facilities
field measurements or the use of large size spec- (Nimmo et al. 1987). This method also applies
imens rather than laboratory measurements on large stresses to the specimen through centrifugal
small size specimens (Fredlund and Rahardjo forces during testing and is suitable only for soils
1993; Benson et al. 1994; Houston and Houston whose flow behaviour is not sensitive to applied
1995; Benson and Gribb 1997). stress state.
The studies summarized in this paper was ini- Some of the transient state methods, namely
tiated when a preliminary search for permeability the BruceKlute Method (Bruce and Klute 1956),
functions (i.e., relationship between the unsatu- and the different outflow methods (Kool et al.
rated coefficient of permeability, kunsat versus 1985) are based on diffusivity relationships that
suction) for nickel mine tailings yielded limited require the estimation or determination of a large
data (Gonzalez and Adams 1980; Vick 1983). number of parameters. In addition, these methods
123
Geotech Geol Eng (2007) 25:191202 193
123
194 Geotech Geol Eng (2007) 25:191202
Table 1 Physical properties of tailings materials content. The volumetric water content, h, in the
Properties Overflow Underflow compacted soils can be reliably determined using
fixed TDR probes in compacted specimens since
Clay (%) 0 3.5 there is no concern related to the possibility of
Silt (%) 100 27.4
Sand (%) 0 69.1
formation of water pockets near the TDR parallel
D10 (mm) 0.0161 0.0261 probes.
D30 (mm) 0.0180 0.0727 The permeameters used for determining the
D60 (mm) 0.0206 0.1356 coefficient of permeability of compacted soils
Coefficient of curvature, Cc 1.0 1.5
Coefficient of uniformity, Cu 1.3 5.2
cannot be directly used to determine the flow
Specific gravity, Gs 2.91 2.99 characteristics of tailings under unsaturated con-
Liquid limit, wL (%) 31.1 8.4 ditions. Tailings are an industry by-product and
Plastic limit, wP(%) N/A N/A have characteristics that are different from natu-
Saturated coefficient 1 106 2.4 103
of permeability,
rally occurring soils or compacted soils. It is dif-
ksat (cm/s) ficult to duplicate the depositional conditions of
Volumetric water 0.60 0.47 natural tailings and its drainage conditions that
content (m3/m3) are typically found in tailings dams. If fixed TDR
Gravimetric water 50.5 35.0
content (%)
probes are used in the permeameter with tailings
that are typically in a state of slurry, then during
the initial stages of the measurement of flow
undisturbed specimens in this paper. The dis- behaviour, it is likely that water pockets will be
turbed tailings, or re-slurried tailings, refer to formed around the TDR probes. The water
the tailings specimens that were stirred to attain pocket formation can also occur during consoli-
their original slurry consistency. Unless otherwise dation of the sample while the sample is desatu-
specified, the term tailings is used in this Paper rating. Design modifications are therefore
refers to the re-slurried tailings. necessary to measure the volumetric water con-
The nickel tailings under dry conditions are tent of the sample during the entire duration of
grey in colour. The same tailings under wet con- the test.
ditions turn black in color. They also release a
particular odour, associated with the presence of
sulphur in the host rock. Physical properties of 5 The modified permeameter
these tailings along with the saturated coefficient
of permeability are summarized in Table 1. The modifications include using a large size
sample; the ability to accommodate mobile sen-
sors to alleviate the problems associated with the
4 Permeameter and the instantaneous profile formation of water pockets and to use the same
method test specimen and determine the SWCC and
unsaturated coefficient of permeability of the
Several investigators have determined the unsat- tailings over a large range of saturation.
urated coefficient of permeability of compacted Figure 1 shows the schematic diagram with the
specimens in the laboratory following the wetting details of the instrumentation. This permeameter
front using the IPM (Hamilton et al. 1979; can accommodate a large sample of tailings,
Merdink et al. 1996). The IPM requires mea- 200 200 400 mm high. The matric suction
surements of water content and matric suction to in the soil specimen was proposed to be deter-
determine the unsaturated coefficient of perme- mined using thermal conductivity sensors and
ability. In this procedure, soil specimen has to be tensiometers. Attempts were made to measure
dismantled at different suction values to deter- the volumetric moisture content using the TDR
mine the water content or a technique such as technique as described below. However, this
Time Deflection Reflectometry (TDR) probes technique showed a significant scatter in results
has to be used to measure the volumetric water that was attributed to the mineral content in the
123
Geotech Geol Eng (2007) 25:191202 195
tailings sample. Hence a specially designed sam- values determined and the measurements using
pling tube was used for this purpose to determine TDR. In an attempt to explain the discrepancies
the water content of soil sample at different val- between the direct laboratory results and the
ues of suction. TDR results of volumetric water contents, the
tailings were tested for their electrical conduc-
tivity and magnetic field properties. The nickel
5.1 Time domain reflectometry (TDR)
tailings proved to be highly conductive both with
respect to electrical and magnetic fields (most
The TDR sensors were accommodated such that
likely due to the presence of iron found in the
they were mobile and not fixed in the modified
pyrite and pyrrhotite). Robinson et al. 1994,
permeameter. The provision of mobile TDR
studies show that iron minerals significantly affect
sensors was included in the design to avoid the
the determination of water content using TDR.
formation of voids around the probes rods during
Due to this reason, a second system was required to
the settlement of tailings. A counterweight system
validate the results of water content measurements
was used to accomplish the mobility of the sen-
sors with very little friction during the testing
period. The TDR probes were essentially placed
100
in a state of buoyant condition in the sample with
90
the provision of counter weight balancing system.
80
Water Content (%)
Measured (TDR)
Only a small vertical pressure was required for
70 Theoretical
the movement of TDR sensors under buoyant Gravimetric
60
conditions. The tailings sample consolidating un-
50
der its own weight provided the necessary vertical
40
pressure such that the TDR sensors could slide
30
without causing voids. More details about the
20
counterweight system are detailed in a later
10
section.
0
Preliminary calibration of the TDR equipment 1 2 3 4 5 6 7 8
in tailings material provided erratic results. Test No.
Figure 2 shows the scatter in results between the Fig. 2 Comparison of theoretical and experimental volu-
directly measured volumetric water content (h) metric water contents
123
196 Geotech Geol Eng (2007) 25:191202
obtained using TDR. This was achieved by grease applied at the probes sliding arms, (shown
collecting small samples at different depths using as 4 in Fig. 3). This grease does not creep when
a sampling tool from the modified permeameter applied on a vertical surface.
using minimally intrusive method for direct
determination of the water content. More details 5.3 Sampling tool
of the sampling tool are provided in a later
section. The sampling tool, shown in Fig. 4, is a device
that provides access to collect a small specimen
5.2 Counterweight system and measures from the large size tailings specimen and allows
for arresting leakage easy, accurate and direct determination of the
gravimetric water content at different elevations
Figure 3 shows a schematic of the balancing at different time intervals of testing. The sampling
counterweights (shown as 1 and 2) that maintain tool also allows the test to be continuous by not
the TDR probes in a free-floating condition. requiring the test to be halted for dismantling
Leaks were anticipated at the interface between purposes. The advantage of such a device lies in
the sliding arms, holding the TDR probes, and the its ability to allow tailings to consolidate in and
permeameter walls. Combinations of Teflon,
various foam tapes, and numerous types of grea-
ses were evaluated for eliminating leakage. The
water-tightness and mobility of the sensors was
best achieved with a bead of graphite based
123
Geotech Geol Eng (2007) 25:191202 197
around it, even though the sampling tool is fixed was left bare. A high air-entry porous ceramic
to the permeameter wall. Hence, there is very disk of 1 bar was used in the modified perme-
little disturbance in the tailings specimen. This ameter in conjunction with the hanging column of
feature eliminates the concern of void formation water to create a negative pressure head on the
around intrusions in the specimen during testing. sample above the porous ceramic disk.
Figure 4 (label 1) indicates the exterior wall of Figure 5 shows the design details of the base of
the permeameter, through which a support (5) is the modified permeameter. The porous ceramic
introduced to hold the sampling tool in a hori- disk was fitted into an aluminum collar (Arrow 1)
zontal position while preventing the flow of air in such a way that the ceramic plate or stone
and water from and out of the system by means of could be disassembled from the permeameter.
O-ring seals (6). The sampling tool is composed of Such a provision helps to first saturate the cera-
two cylinders. The outer cylinder acts as a sleeve mic plate externally and then allows its placement
for the inner cylinder. An opening is introduced in the permeameter. The ceramic disk was held in
at the same location along both cylinders, which the aluminum collar using all weather silicon
allows the slurry sample to flow through the grease (2).
sampling tool during testing. In order to collect a
small sample for determining the water content,
5.4 Tensiometers
the knob connected to the inner cylinder is
rotated by 90 degrees to trap a small sample of
Thermal conductivity matric suction sensors (TC
the soil in the opening of the interior cylinder.
sensors) and tensiometers were originally selected
The O-rings located just besides the opening (4)
to measure the matric suction in the tailings
seal the sample in the outer sleeve preventing
specimen. However, erroneous voltage readings
changes in the water content.
from the matric suction sensors were observed
In the present study, three sampling tubes were
due to very minor fluctuations from the power
installed on different sides of the permeameter at
source. Hence, matric suction measurements in
different heights. Provisions were made to
this research study were measured using only
determine the soil suction using the tensiometers
tensiometers.
at approximately same heights. The fourth side
Commercial tensiometers (model 2100 F from
Soilmoisture Equipment Corporation) were used.
These tensiometers can be used in the 085 kPa
range of suction. The tensiometers were capable
of measuring suction almost instantaneously; in
comparison, the TC sensors required a heating
time of 50 s prior to providing a measurement.
123
198 Geotech Geol Eng (2007) 25:191202
6 Trial tests
123
Geotech Geol Eng (2007) 25:191202 199
123
200 Geotech Geol Eng (2007) 25:191202
Overflow
50
GravimetricWater Content (%)
80
Feed Experimental
40
60 Prediction
Underflow
30
40
20
20
10
0
0.001 0.01 0.1 1
0
Grain Size (mm) 0.1 1 10 100
Soil Suction (kPa)
Fig. 10 Grain size analysis of overflow, feed and under-
flow tailings samples Fig. 11 Underflow SWCC laboratory results
123
Geotech Geol Eng (2007) 25:191202 201
10 1.E-12
0 1.E-13
0.1 1 10 100
Soil Suction (kPa) 1.E-14
1 10 100 1000
Fig. 12 Overflow SWCC laboratory results Suction (kPa)
Underflow
1.E-05 (Brooks &
1.E-06 Corey) A modified permeameter has been designed and
1.E-07 developed to determine the coefficient of per-
1.E-08 meability of unsaturated nickel tailings using
1.E-09 large size samples. The method to calculate the
Overflow
1.E-10 coefficient of permeability is based on the
1.E-11 instantaneous Profile Method. The permeameter
1.E-12 provides values of unsaturated permeability with
1.E-13
remarkably little scatter over low suction range.
1.E-14
1 10 100 1000 Further experimental work is required to extend
Suction (kPa) the results to higher range of suction values. This
Fig. 13 Predicted conductivity curve for underflow and promising apparatus can also be used for simul-
overflow of tailings using Brooks and Corey (1964) model taneous determination of the SWCC.
123
202 Geotech Geol Eng (2007) 25:191202
123