Compressive Stress Dewaterability Limit in Fluid Fine Tailings

Abstract

Reclamation of fluid fine tailing (FFT) storage facilities to their pre-disturbance equivalent landforms is hampered because micrometer size fines, whose surface-area-to-volume ratio is remarkably high, are occupied with siloxane and hydroxy groups, which bind water strongly. The purpose of this study is to differentiate the forms of water physically distributions in FFT and determine their propensities for dewaterability under compressive stresses. Two thermal and two mechanical methods were used to analyze water distributions in FFT. Dynamic and isothermal thermogravimetric analyses of FFT gave a transition from predominately bulk water to coevolution with water of higher enthalpy of vaporization at 81% (w/w) solids. Differential scanning calorimeter studies were used to determine the non-freezable water amount, with the premise that water that does not freeze by −30 °C is also unlikely to be removable by compressive stresses encountered in tailing treatment processes. The solid weight percent of FFTs with the non-freezable water was 79.6%. A 1D finite-strain model simulation using the constitutive relations of void ratio and effective stress, void ratio, and hydraulic conductivity show that deep-pits filled with such clayey-silt FFT will consolidate to a maximum solids content of 74% (w/w). For separation by centrifugation, the solids content plateaued to a mean of 74% (w/w) for total centrifugal force of ≥30 mega Newtons. These solid contents represent upper thresholds and demonstrate dewatering limit property of an FFT under compressive stresses.

1. Introduction

The extraction of resources through mining operations often starts with crushing the lumps of ore followed by mixing in boxes, stirred tanks or cyclo-feeders. The conditioned slurry is then introduced to hydro transport pipelines or to tumblers, where lumps of the ore are farther subjected to mechanical shearing so as to release the valuable resource from the gangue. These operations crush, splay and delaminate the clay minerals creating non-settling fluid fine tailings (FFT) waste stream which require storage in containment tailing dams.

FFT poses challenges because the solids are dominated by hydrophilic clay minerals, which do not dewater easily. The lateral dimension of the fine solids in tailings range from 20 to 200 nm, and their surface-to-thickness aspect ratios are very large. As a result, the particles have highly specific surface areas ranging up to 800 m²/g, as has been reported by Johnson [1]. In the aqueous media, the clay surface charge is counter-balanced by cations and polar water molecules as described by the double layer theory. According to the theory, the clay surface is in direct contact with a rigid layer of water molecules that, depending on the charge density and specifically adsorbed cations, involve several layers of water molecules designated as the outer Helmholtz layer. Beyond the outer Helmholtz layer is a diffuse space charge distribution that shields the excess charge of the inner layers. The water in the diffuse layer is ordered and of varying activity as it transitions to the bulk water (free water) [2]. Chemical treatments with large-molecular-weight polymers known as flocculants are a necessary upstream step that modifies the double layer structure around the particles so that the solids can gather closer to each other, i.e., form flocs and make the FFT more amenable for dewatering. Unlike thermal desiccation, not all of the moisture in the FFT can be removed by compressive stress (a mechanical dewatering action), even after treatment, and this dewaterability limit is dictated by the forms the water distributes as described at a macroscale below.

In the tailings treatment field, the distribution of the water is commonly divided in to four segments, namely free water, interstitial water, vicinal water, and water of hydration [3,4,5]. The free water, also named bulk water, is water that freely moves and is not affected by the solids. This free water is the easiest to remove by mechanical means. The interstitial water is water trapped between flocs and aggregates, and could become free water. It is likely that most of the differences in separation performances between various processes is attributable to their effectiveness in releasing the interstitial water. The interstitial water designation in tailings is different from its use in clay science, in which it refers to the very small quantity of tightly bound water molecules found between lamellae of clay particles. Vicinal (or surface) water is closely attached to the solids surface without free movement [5,6]. Vicinal water is treated as a solid skeleton, and forms part of the solid particle. Vicinal water is the reason that mineral solids show lower specific gravities than their dry crystal forms [7,8]. The water of hydration is chemically bound to the solids, and can be removed only by high temperature thermal drying. Thermal drying that occurs at different ranges of temperature can show the water distribution in the tailings [3]. Bulk water and surface water can be removed by heating at approximately 60 °C and 105 °C, respectively [3,9,10]. Clays can farther lose hydration water at three temperature steps of up to 150, 500 and 800 °C. Each temperature range corresponds to differences in the level of binding energy of water molecules to the solids. Water that is driven off at these high temperatures is outside the effect of mechanical dewatering, and thus not the subject of this study [3,9].

The concept of the compressive stress limit is related to the binding energies of water to the solids. The understanding of bound water varies from field to field. Bound water in food science is different from that of cryobiology, which is again different from that of soil science. The differences arise because there are no standard methods for direct determination of bound water. The methods currently available are empirically based, in which a whole system is subjected to a set of conditions that removes or changes a portion of water properties in distinction from that of free water, which is then designated as bound water [11].

The purpose of this paper is to examine the forms of water distributions in FFT using thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), benchtop centrifuge separation studies, and large-strain consolidation modelling. The most important form is the portion of water in FFT that is associated with the solids, with such strength that it cannot be separated by conventional industrial mechanical separation processes. It is evident that the dewaterability will depend on the mechanical or physical separation method employed, as well as the composition of the FFT itself. However, a practical or functional moisture content dewatering limit can be established by consensus of the relevant industry or society analogous to the classification of mineral solids based on particle size distribution (PSD), as an example. As an illustration of professional association established standards, within the oil sands industry, the minus 44 μm fraction is known as fines, while in the wider unified soil classification system used in engineering and geology fines are defined as smaller than 75 μm particles. Similarly, 2 μm and smaller particles in soil are taken to be clay minerals, all based on the preponderance of observations, although some clay particles are bigger than this threshold size. Knowledge of the compressive stress dewaterability limit for a given FFT is an essential input in designing the geotechnical stability of their deep deposits.

2. Materials and Methodology

2.1. PSD of FFT

An FFT sample was obtained directly from one of the oil sands companies’ tailing storage facilities in Northern Alberta. Particle size distributions (PSD) of the sample was measured using a Mastersizer™ 2000 (Malvern Instruments, Orsay, UK) coupled with a Hydro 2000MU wet sample dispersion unit, which enables continuous stirring and circulation of the suspension to pass through the measurement cell of the instrument. The sample was also sonicated in an ultrasound bath to maintain the particles in suspension. The PSD was determined based on the scattering of blue and red laser lights, as shown in Figure 1.

The same batch’s tailing sample PSD obtained by Sedigraph, an X-ray attenuation method, is found in bibliographic listing [12]. The difference between the two PSD results is a result of their limits of detection. The laser diffraction analysis’ minimum measurement size is 20 nanometers, while that of the Sedigraph is 300 nanometers. For this reason, the Sedigraph gives higher mass% in the lower size ranges than the laser diffraction method. The PSDs by both methods are comparable in the relatively large size zones.

2.2. Solids Minerology

Methylene blue titration of the as-received FFT slurry revealed that clay minerals constituted 58% (w/w) of the solids and particles of sizes less than 2 μm made up 16% of the solids in the FFT. All of the solid particles were <75 μm in size, as is shown in Figure 1.

The crystalline solid phases in the FFT were quantitatively determined by X-ray diffraction (XRD) using the internal standard alumina addition method. A Bruker D8™ diffractometer instrument and the database of the International Centre for Diffraction Eva 4.2 were used for peak identification. More details on the XRD characterization and quantification are available in this Special Issue [13]. The results of the mass normalized quantitative phase minerals analysis are given in

Table 1. Rietveld method quantitative phase analysis (wt.%) of FFT.

Mineral		Ideal Formula	Weight %
Carbonates	Ankerite—Dolomite	Ca(Fe2+,Mg,Mn)(CO3)2—CaMg(CO3)2	0.2
	Siderite	Fe2+CO3	4.0
	Calcite	CaCO3	0.2
Clays	Illite—Muscovite	~K0.65Al2.0(Al0.65Si3.35O10)(OH)2–KAl2(AlSi3O10)(OH)2	10.1
	Illite–Smectite (mixed layer)	~K0.65Al2.0(Al0.65Si3.35O10)(OH)2–(Na,Ca)0.3(Al,Mg)2Si4O10(OH)2∙nH2O	15.3
	Kaolinite	Al2Si2O5(OH)4	27.6
Fieldspars	K-feldspar (microcline)	KAlSi3O8	2.1
Pyrite	Pyrite	FeS2	0.8
Quartz	Quartz	SiO2	25.0
Titanium oxides	Anatase	TiO2	1.0
	Rutile	TiO2	0.5
Amorphous			13.2
Total %Clay			53

.

2.3. Thermogravimetric Moisture Analysis of FFT

A TA-Q500 (TA Instruments, New Castle, DE, USA) was used to perform thermogravimetric analysis (TGA). The FFT as a slurry is convenient to spread thinly and uniformly minimizes the effect of temperature gradients within the sample. Thermal studies were conducted on untreated FFT. The analyses were carried out with nitrogen carrier gas at 60 mL/min flow rate. TA Universal analyzer™ software was used to acquire the data.

2.4. Calorimetric Water Distribution Analysis

Differential scanning calorimeter (DSC) measurements were made employing a TA-Q2000 DSC instrument (TA Instruments, New Castle, DE) equipped with a refrigeration system to a −90 °C temperature. The TA-Q2000 DSC device has a built-in platinum resistance thermometer, having an uncertainty of 0.01 K. The conventional DSC mode of the apparatus was used to measure the heat flux by comparison with an empty measuring cup. A standard sapphire sample supplied by the manufacturer was used to calibrate the instrument by the Tzero calibration method. Typically, less than 15 mg of a sample was placed in a hermetically sealed aluminum cup. Crystallization and fusion phase changes of water in the FFT were conducted by cooling–heating cycles between 10 °C and −15 °C under nitrogen carrier gas flow of 50 mL/min. The TA Universal analyzer™ software (version 2000) was used to acquire the data.

2.5. Centrifuge Solid–Liquid Separation of FFT

The benchtop centrifuge FFT dewaterability tests were conducted using a Sorvall WX Ultra 80 ultracentrifuge with a Beckman Type 45Ti rotor at a constant temperature of 20 °C. The tests were conducted using about 50 g of FFT placed in polycarbonate tubes. Each separation test was conducted with a triplicate number of samples. The centrifugation run time and angular acceleration reported in this paper are those that resulted in two distinct layers of overflow water at the top and sediment (cake) at the bottom. The moisture in the sediment and the solids content in the overflow water of the samples were determined by the gravimetric drying method. A high-molecular-weight, partially charged, anionic polyacrylamide commercial macromolecule, A3338™ (SNF, Riceboro, GA, USA), was used as flocculant in the bench top centrifugation studies.

3. Results and Discussion

Due principally to the intrinsic properties of the FFT particles, the industrial mechanical dewatering processes leaves much more moisture with the fines than is desired for creating geotechnically stable landscapes. Thermal methods are used in this study to examine the distribution of water in FFT, so that the portion that could not be removed by mechanical separation can be evaluated.

TGA is a mature method to study thermal behavior of materials accompanied by mass changes [14,15]. The mass change with temperature provided by a TGA device can be used to calculate the vaporization enthalpy. When coupled to other techniques such as Fourier transform infrared spectroscopy (FTIR), mass spectroscopy (MS), or FTIR-MS, TGA provides advanced characterization of thermal effect on samples. In this study, TGA is used to allocate the various water fractions by their ease of vaporization. The ease of evaporation is related to the degree of association (binding) of the water to the solid particles. Consequently, the mass loss in the initial stages of temperature controlled drying tests is predominantly due to the evaporation of free water. Isothermal mass loss data are usually easier to analyze and interpret. The second-order derivatives (DDTGAs) of the mass loss with time at three isothermal drying temperatures of FFT are shown in Figure 2.

The isothermal temperatures are well below temperatures (<105 °C) that are used to drive off surface water from soil samples. The evaporation rate at low temperature isothermal TGA are potentially more controlled, and could enable differentiation of overlapping forms of water distributions. Steps in gravimetric mass loss or peaks of first- or second-order differential mass changes with respect to time or temperature, which occur prior to the sample reaching its dry solids content are indicative of transitions between distinct evaporative forms. The DDTGA of Figure 2 suggests three transition steps. The initial transition occurred within the first 4 min, and when the solids content of the FFTs were on the average about 54% (w/w) solids. The instrument takes a few minutes to heat up and stabilize to the isothermal temperatures. The first peaks coincide with the temperature rise period, and are inevitable artifacts as instruments do not reach the test isothermal temperatures instantaneously. Comparable peaks did not occur in temperature ramp DDTGA measurements, proving that the first peak in the isothermal measurements was method-dependent and not related to the evolution of vapor from FFT. The higher the isothermal heating temperature, the sooner the second peak occurred in correlation to the increased mass loss rate and the diminishing moisture content of the FFTs. The results show that TGA can differentiate the most evaporative form (bulk water) from the follow up forms that have higher heats of evaporation. Temperature ramp TGA tests are more sensitive than isothermal tests, and this method was utilized to determine the solids content at the point of transition to the slower evaporative or higher heats of vaporization water forms [14,15,16].

Figure 3 shows a representative moisture loss of the FFT at 2 °C/min heating rate to a maximum temperature of 105 °C. Mass loss was practically complete before the maximum drying temperature was reached. The peak of the DTGA corresponds to the inflection of overlapping evaporation rates. It should be noted that the mass loss rate decreased while the temperature was increasing. For a thin uniformly distributed sample, the decrease correlates with the diminution of readily evaporating water form. The moisture content per unit dry mass of the sample at which it transitions to a falling rate of evolution indicates the depletion of the easy to evaporate water form and the continued evolution of the more tightly bound water [17]. The mass loss up to the peak can thus be considered as potentially the mechanical dewatering limit of FFT. The solids content of the sample at the peak was 81.1% (w/w) (or water content of 23.3%). Repeated TGA measurements at varying heating rates and sample masses all occurred in approximately 82% of solids (w/w) or less.

As many as three calorimetric forms of frozen water have been observed, due to the way water interacts with clays [18]. One of the forms is freezable water, but with crystallization and melting temperatures different from bulk water. Another form is a rigid layer of water-adjacent clay particles that only freezes at far lower temperatures of −30 °C and lower. The non-freezable water is so strongly bound that it is unlikely to be removed by mechanical means. The phase transitions between the different forms are accompanied by evolution or absorption of latent heats. It is, in principle, possible to quantify these three forms of water from measurement of the latent heats using DSC measurements, as explained below. Thermal energy exchanged during a phase change is directly proportional to mass [19].

$$\frac{dh}{dt}=m\times C_P\times \nu$$

(1)

where dh/dt is the heat flow rate (J/min), m is mass of sample (g), C_P is the specific heat capacity (J/(g.°C), and ν is the temperature scanning rate (°C/min). As the DSC detects differences in temperatures between the sample and reference, it provides additional power that appear as peaks in the calorimeter signal (thermogram). The peak area provides the heat (or heat of phase transition) [18].

$$∆h={\int }_{T_i}^{T_f}dh\left(T\right)dT$$

(2)

where h(T) is the power function, T_i and T_f are, respectively, the initial and final temperatures of the transition peak. With the assumption that tightly bound water would not freeze up to the threshold cooling temperature of −30 °C, the heat absorbed in the melting stage from initial temperatures >−30 °C is equal to the latent heat required for freezable water. The bound water content in this study was determined by the subtraction of the freezable water amount from the total moisture content of the sample obtained by gravimetric drying method.

Only one solidification and one melting peak were observed with FFT when its temperature in the DSC was cycled between −30 °C and 25 °C. The melting curves of successive cycles whether cooled slowly or rapidly were superimposable. The free water was determined from the melting curves of FFTs that were quench cooled and thermally equilibrated at −15 °C for 40 min and scanned to 10 °C at 1 °C /min, as shown in Figure 4 [14,15,19].

Division of the heat obtained from the integration of the fusion peak by the latent heat gives the amount of freezable water in the FFT. The water melted from −1.2 °C, and the peak melting temperature was at 0.63 °C. The latent heat of fusion of ice is known to decrease with decreasing temperature, and the following empirically obtained formula has been used for its calculation [20].

L(T) = 7.3 × T + 334

(3)

where L(T) is the latent heat of fusion at temperature T in J/g and T is the melting temperature (°C). The non-freezable water content per unit dry mass by the DSC method was 26%. The solids content of the FFT excluding the freezable water mass provides indications of the highest possible mechanical dewatering limit, and it was found to be 79.6% (w/w), pointing to a mechanical dewatering limit under 80% solids (w/w). It should be noted that there can be bound water that freezes at slightly different temperatures than bulk (free) water. The DSC measurements did not resolve bound but freezable water from free bulk water.

It is not plausible that water forms zones of sharply differing binding energy structures around the fine solids. Instead, the water forms layers with binding energies that continuously and gradually transition from the highest for vicinal water to that of null for bulk water. TGA drying mass loss, therefore, does not occur in discrete steps, but is a continuum. For this reason, differing forms of water evaporated concurrently in varying proportions. This means that there remain unseen water forms that are being lumped with bulk water, and the mechanical dewatering limits reported are upper thresholds of what could be obtained in practice.

Centrifugal separation, in distinction to the thermal methods covered above, is a mechanical separation method and directly relatable to FFT mechanical separation dewatering limit. The unit weight of soil can be increased by increasing the gravitational acceleration, and it is on this basis that large centrifugal equipment is used to model self-weight consolidation of deep-pit deposits [21,22]. The centrifugal force, F_c, of the centrifuge is expressed by Equation (4).

F_c = m × r × ω²

(4)

where m is the mass of the FFT in the test tube (kg), r is the radius (m) which is the centroid of the sample, and ω is the angular velocity (rad/s). The obscured property of F_C is that it is force exerted per second of the centrifuge runtime. The liquid–solid separation and moisture content of sediment are dependent on both F_c and centrifuge runtime. For this reason, the centrifuge sediment results are reported in terms of the product of centrifuge runtime and centrifugal force, as shown in Figure 5 [21,23,24].

Centrifugation conditions, which did not give easily decantable overflow water layer at the top and sediment (cake) at the bottom of the tubes, are not included in the results, as the objective was to determine mechanical dewatering limit. Figure 5 shows the dewatering plateaued with total centrifugal force levelling from about 30 MN. The mean solids content of the settlement at the plateau was 74% (w/w).

Consolidation under self-weight or overburden stress is another form of compressive dewatering and consolidation test results can be used for the prediction of mechanical dewatering limit. The geotechnical properties and constitutive equations of this FFT have been reported [12]. The constitutive relations are used to calculate the settlement (dewatering) using a finite-strain modelling software; FSConsol, (GWP Geo Software Inc., Edmonton, AB) [25] (Pollock 1988). The application provides the sought output of solids profile and settlement of a deep pit filled with thickened FFT. The deep pit dimensions and filling rate are not unique, but practical parameters were selected for the modelling. The model output for the FFT is given in Figure 6. The results show that the solids content of the FFT deposit converges both with years of consolidation and depth (i.e., compressive stress). The computation clearly shows that the deposit’s solids content converges to a maximum solid content close to 74% (w/w). This agrees with the maximum solids content of the sediment obtained at the plateau of the total centrifugal force separation test above.

Yuan et al. [26] have investigated the consolidation of thickened FFT from the same mining operations as in this work, using long beam centrifuge measurements. The long beam centrifuge test conditions were set to simulate the settlement over a 150-year period of a 50 m deep-pit deposit. The beam centrifuge simulation demonstrated that flocculated FFT and centrifuged FFT cake with SFR 0.05 will consolidate to just 65% and 75% solids content, respectively. According to Yuan’s et al. work the practical dewatering limit of FFT is under 75% (w/w) solids and close to the values reported in this paper [26]. The implication of the compressive dewatering limit and the inordinately long consolidation time taken by FFT is that depositing merely thickened FFT will not result in reclamation-ready landforms, and co-disposal (or co-mingling) with coarse tailings and mine waste rock have been proposed as realistic routes [13,26,27,28,29].

Figure 7 shows that blending freshly produced cake with coarse tailings instantly reaches the solids content that a cake only deep-pit deposit will take hundreds of years to consolidate to. The solids content of a freshly produced cake blended at a 1.5 SFR will achieve solids content higher than the mechanical dewatering limit of this FFT. Suppose an 80% (w/w) solids content is desired (orange box); a freshly produced cake blended with coarse at an SFR of 2 can achieve the condition saving time, avoiding risk, and eliminating the cost of thickened FFT only deep-pit perpetual maintenance.

Operationally, blending (or co-mingling) an already-thickened FFT to the coarse tailings avoids interference of the coarse tailings in the dewatering treatment process and conserves material handling and tear and wear of equipment by delaying it to the final stage of discharge. Conducting the thickening and blending steps sequentially is more reliable in the co-disposal scheme, as it simply combines commonly used operations. Another alternative to co-disposal, but whose viability has not yet been extensively examined by multiple agencies, is thermal drying of the centrifuge separated cake using stack flue gas heat and exhausted steam from the other mining operations [30,31]. The attraction of the method is that the waste can then be disposed directly as a reclamation-ready landfill. Future works to advance the mechanical dewatering limit concept of fines we have planned are application of the double-layer theory and measurement of the different water forms using low-field time-domain nuclear magnetic spectroscopy.

4. Summary and Conclusions

Reclamation of fluid fine tailing (FFT) dams to their pre-disturbance equivalent landforms is hampered because micrometer size fines whose surface-area-to-volume ratio is very high bind water strongly. The water in the FFT is bound to the fine solids with varying strength and, therefore, the moisture content of the dewatered fines is dependent on both the treatment technology and the fines composition too. The vast experiences of the oil sands fine tailings management proves that a significant amount of moisture on dry fines mass basis remains with the fines, even after dewatering, indicating a practical dewatering limit. Knowing the FFT dewatering limit of the mechanical separation methods provides input for the implementation of measures to create reclamation ready landform.

Two thermal methods and two mechanical methods were used to evaluate the dewatering limit. Isothermal and dynamic temperature ramp TGA test results gave two forms of water besides the clay surface water. The transition from the initial evolving water to the less evaporative water forms occurred at a solids content of 81% (w/w). Since evaporation is not a discrete process, the solids content of the FFT with just only the evolution of less tightly bound water is likely lower than the 81% (w/w) reported. DSC was used to determine the mechanical dewatering limit, on the premise that water that does not freeze up to −30 °C is also not likely to be removable by mechanical means. The solids content of the FFT excluding the freezable water was 79.6% (w/w). It was not possible to differentiate the freezable but bound water from the bulk water that freezes at normal temperature by DSC analyses.

The constitutive constants of the void ratio–effective stress–hydraulic conductivity interrelationships for this FFT in the literature were used as model inputs to predict tailings settlement using finite-strain 1D model software. Numerical calculations indicated that a deep pit starting with a 58% (w/w) solids filling will only consolidate to a maximum solids content convergent to 74% (w/w). For separation by benchtop centrifugation, the solids content plateaued both with increasing time and g-force. A mean solids content plateau of 74% (w/w) was obtained for total centrifugal force of ≥30 MN. These results point out the highest possible solids content of FFT that can be produced or achieved by mechanical dewatering excluding desiccation is around 74% (w/w). This final dewatering level is indicative of the long-term settlement of deep pits filled with thickened FFT.