Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Effect of Particle Size on the Rheology of

Athabasca Clay Suspensions


Olusola B. Adeyinka, Sepideh Samiei, Zhenghe Xu and Jacob H. Masliyah*
Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB, Canada T6G 2V4

The success of the separation process conventionally used in Alberta for oil sands extraction is highly influenced by the rheology of the oil sands
slurry. In the gravity separation vessel, high slurry viscosities can hinder the rise of aerated bitumen and reduce the efficiency of the recovery
process. In this study, the effect of particle size on the viscosity of oil sands slurries is investigated. Solids from mature fine tails (MFT) obtained
from tailings pond were fractionated into three fractions of different particle size distributions and their rheological properties were studied.
The solids in each fraction were characterized by XRD analysis which showed the presence of different types of clays in each fraction. For the
rheological measurements, dispersions of the three fractions were prepared in the supernatant water decanted from centrifuged MFT to maintain
the solution chemistry of the solids. Suspensions of each fraction showed a non-Newtonian shear thinning behaviour as well as yield stress that is
characteristic of structure formation within the suspensions. For all solids fractions, increasing solids concentration led to higher viscosities and
higher yield stress values. Viscoelastic properties of the suspensions showed stronger solid-like behaviour at higher particle concentrations. Among
the three fractions numbered from 1 to 3, solids in fraction 3 were coated with organic matters, exhibiting the highest suspension viscosities.
Also for fraction 3, higher gelling potency was observed at much lower weight fractions of solids as compared to the other fractions.

La réussite du processus de séparation normalement utilisé en Alberta pour l’extraction du pétrole à partir de sables bitumineux est fortement
influencée par la rhéologie de la boue des sables bitumineux. Dans le récipient de séparation par gravité, les viscosités élevées de la boue peuvent
empêcher la remontée du bitume cellulaire et réduire l’efficacité du procédé d’extraction. Dans le cadre de cette étude, l’effet de la dimension
des particules sur la viscosité des boues de sables bitumineux est analysé de près. Les solides provenant des résidus fins mûrs (RFM) obtenus des
bassins de résidus ont été fractionnés en trois fractions de granulométrie des particules différentes et leurs propriétés rhéologiques ont été étudiées.
Les solides de chacune des fractions ont été caractérisés au moyen d’une analyse diffractométrique qui a révélé la présence de différents types
d’argiles dans chacune des fractions. Pour ce qui est des mesures rhéologiques, des dispersions des trois fractions ont été préparées dans l’eau
surnageante décantée à partir des RFM centrifugés afin de maintenir la chimométrie des solides. Les suspensions de chaque fraction ont montré
un comportement de fluidisation par cisaillement non newtonien, de même qu’une contrainte d’écoulement qui est typique de la formation de
structure à l’intérieur des suspensions. Pour toutes les fractions de solides, l’augmentation de la concentration de solides a entraîné des viscosités et
des valeurs de contrainte d’écoulement plus élevées. Les propriétés viscoélastiques des suspensions ont montré un comportement plus rigoureux
semblable à celui d’un solide à des concentrations de particules plus élevées. Parmi les trois fractions numérotées de 1 à 3, les solides de la fraction
3 ont été imprégnés de matières organiques, affichant les viscosités des suspensions les plus élevées. On a également observé, pour ce qui est
de la fraction 3, une activité de gélification plus élevée à des charges des fractions de solides beaucoup plus basses, comparativement aux autres
fractions.

Keywords: oil sands, ultra-fines, rheology, gelling

INTRODUCTION environment facilitates the separation of aerated bitumen and


solids based on their density difference. The aerated bitumen of

S
eparation of bitumen from Alberta oil sands is convention-
ally carried out through a water based extraction process.
The oil sands ore is slurried with warm water and trans-
∗ Author to whom correspondence may be addressed.
ported to the extraction plant via hydrotransport pipelines.
E-mail address: jacob.masliyah@ualberta.ca
Flowing through the pipeline, bitumen is liberated from the sand
Can. J. Chem. Eng. 87:422–434, 2009
grains and attaches to air bubbles. The first stage of separa-
tion takes place in a gravity separation vessel where a quiescent
© 2009 Canadian Society for Chemical Engineering
DOI 10.1002/cjce.20168

| 422 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
overall density less than the density of its suspending medium useful insight into the gelation problem with 2 H NMR and jar
floats to the top of the vessel and forms a bitumen froth layer. The tests on ultra-fines separated from oil sands and waste process-
coarse solids separate fairly effectively to the bottom of the sep- ing unit samples. Their study further established that, given the
aration vessel, forming the tailings stream. From the middle of right water chemistry, sludging conditions will be reached when
the gravity separation vessel, an intermediate stream normally ultra-fines in the slurry exceed a certain concentration. They also
referred to as the middlings stream comprising water contain- found that at 460 ppm of sodium ions in water, only 1.5 wt% of
ing non-buoyant bitumen and mainly fine solids is withdrawn particles smaller than 0.3 ␮m is sufficient to cause gel formation
for further processing. The middling region of the gravity sepa- as compared to 16 wt% for 10 ␮m particles. Previous work by
ration vessel is normally a dilute suspension typically at a level Sparks et al. (2003) has also identified a number of solids com-
of 10 wt% solids content (Schramm, 1989), but could be as high ponents of the fines fraction, particularly clays and organic rich
as 21 wt% (Syncrude Analytical Methods, 1979) under normal solids that under certain circumstances have adverse effects on
conditions. bitumen recovery. These organic-rich solids which are either bi-
The tailings stream is typically directed to tailings ponds wettable or hydrophilic ultra-fines can interact strongly to form
where the coarse fraction of solids rapidly segregates from the a gel and cause the gravity separation vessel to sludge. Chow
fine fraction and settles. The fine solids, on the other hand, et al. (2006) used inter-bedded clays and rheology to map out
are of slow settling rates. After 2–3 years of settling, these fine conditions of particle interaction, solids concentration and shear
solids reach a particle concentration of about 30–40 wt% and that influence recovery and gelation propensity. In their study, the
form a mature fine tails (MFT) with virtually no further con- particle interaction effects for a primarily quartz (55%) and kaoli-
solidation. It has been estimated that complete settling of MFT nite (25%) suspension were incorporated at different sodium and
solids would take more than a century to occur (Eckert et al., calcium ion concentrations.
1996). The results from the afore-mentioned studies suggest that gela-
Occasionally, a high build up of solids within the gravity sepa- tion can be due to a combination of the following factors: high
ration vessel in the middlings zone causes sludge formation, that solids content, high ultra-fines content, high electrolyte content,
is, the formation of a thick, highly viscous suspension. The rheo- and the presence of interstratified clays. The ultra-fines modify
logical properties of this thick suspension depend on the applied the rheological properties of the slurry and middlings due to
shear. The rise of aerated bitumen and settling of solid particles colloidal inter-particle forces. At high concentrations, smaller dis-
through this region is thus hindered, leading to a low bitumen tances between the particles lead to stronger interactions. For the
recovery and non-segregated settling of the solids. This opera- purpose of this study, rheological measurements were utilized to
tional condition is referred to as thickening or sludging of the determine the effect of particle size on the flow properties of oil
separation vessel. A common industrial practice to alleviate sludg- sands slurries. Because gelled middlings are not readily available,
ing is to use dispersing agents, decrease throughput and/or dilute mature fine tails (sludge from a tailings pond) were chosen as the
the feed slurry with water. These approaches however are not suc- source of particles. Originating from the gravity separation ves-
cessful for some problematic ores, in which case other remedial sel middlings, MFT is conveniently accessible. In one occasion
actions are required. where a small amount of gelled middlings was provided, compar-
Research on the cause and methods to avoid sludging is on- isons were made between the mineral composition of MFT and
going. Generally, two rheological conditions have been attributed gelled middlings. The results of this comparison are discussed in
to this reduced performance in the separation vessel. Sludging the following sections.
could results from an abnormal increase in slurry viscosity and, Three fractions of solids with different particle size distribu-
in some cases, form gel. It could also be the result of struc- tions were separated from MFT and their rheological properties
ture formation, causing the suspension to show very high yield were monitored at different particle concentrations. Former stud-
stresses (Schramm, 1989; Tu et al., 2005; Chow et al., 2006). ies (Abend and Lagaly, 2000; Amorós et al., 2002; Mpofu et al.,
Although both conditions may occur simultaneously, especially 2003; Rasteiro and Salgueiros, 2005; McFarlane et al., 2006; Faers
in localized regions of the gravity separation vessel, an increase et al., 2006) showed the effect of solution chemistry on the rhe-
in viscosity is sufficient to cause sludging. While ore classifi- ological behaviour of solid dispersions. With reference to the oil
cation based on fines (i.e., particles having diameters less than sands extraction process, Wallace et al. (2004) showed the impact
44 ␮m) and bitumen content have traditionally provided some of aqueous phase composition on bitumen recovery in the pres-
insight into separation behaviour (Sanford, 1983; Chong et al., ence of fine solids. All samples in this study were prepared in
2003), it is now apparent that they are insufficient for prediction supernatant water separated from centrifuged MFT. While the
of sludging or thickening in a gravity separation vessel (Tu et al., effect of particle size on rheological characteristics of solid/liquid
2005). Therefore, recent research efforts have been focused on the suspensions has been well established in other areas (Rasteiro and
interaction between clay and/or ultra-fine components of the oil Salgueiros, 2005), there have been few studies focused on the role
sand fines fractions as possible cause for sludging. The approach, of particle size in changing viscosities of slurries encountered in
materials, methodology and important parameters investigated in the oil sands industry. The objective of this study is to provide
a number of previous studies on sludging are briefly discussed insights of the types of particle interactions that are directionally
below. consistent with changes in rheological properties that might be
Schramm (1989) used an in situ approach to measure the found in middlings of a primary separation vessel. With empha-
viscosity of middlings in a commercial separation vessel. A vibra- sis on the effect of particle size, the approach in this study is to
tional viscometer was placed in the middling zone and viscosities measure suspension rheology of ultra-fines obtained from MFT.
were measured at different pH and positions in the separation ves- However, the exact viscosity and yield stress values reported here
sel. Using the Stokes equation, Schramm found that high bitumen should not be used for any engineering or design purposes. In
rise velocities and recoveries are realized for suspensions with low addition, this study investigates roles of ultra-fines of various size
fines content over a certain pH range and locations in the gravity fractions in determining rheologic properties of fine tailings as
separation vessel. More recently, Tu et al. (2005) provided more they settle down and consolidate.

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 423 |
Figure 1. Solids fractionation procedure.

EXPERIMENTAL
Materials Used
MFT suspensions obtained from Syncrude Canada Ltd. (Fort
McMurray, AB, Canada) were used as the source of solids. Three
fractions of fines were separated from MFT following the proce-
dure described in the next section. Gelled and ungelled middlings
samples were provided by Shell Canada (Fort McMurray, AB,
Canada) and Syncrude Canada Ltd, respectively. Rheological mea-
surements (not shown here) by Shell Canada show that sludge
samples were gelled.
The bitumen, solids and water content of the middling samples
were determined by the Dean Stark analysis. Mineral composition
of MFT and gelled middlings was determined by X-ray diffraction
analysis. X-Ray diffraction patterns were collected with a Rigaku
D Max B rotating anode (Cu K˛ and Co K˛). Calcium-saturated Figure 2. Particle size distribution of the three solids fractions.
and glycolated slides of solids were prepared for mixed layer clay
identification and quantification as described in Omotoso et al.
(2002). The procedure developed by Moore and Reynolds (1997) are 0.14 and 0.16 ␮m, respectively. Figure 2 shows the particle
was used for slide preparation. size distributions of the three solids fractions.
After separating each fraction, particles were reintroduced to
the original water chemistry by being washed twice in the super-
Solids Fractionation natant water decanted from centrifuged MFT. The wash removed
Figure 1 schematically illustrates the procedure for MFT fractiona- the deionized water used during the separation process. Stock
tion into three fractions. Initially, to break the ultra-fine structures samples of solids were prepared by centrifuging and concentrating
and release of fine solids, MFT was diluted by adding four volumes the washed solids.
of de-ionized water per volume of suspension. After dilution and Mineral components of the solids fractions were determined by
settling for about 1 h, the coarser fraction of solids settled to the X-ray diffractionation analysis identical to the procedure applied
bottom. The suspension of fine solids was then decanted from to the mineralogy analysis of MFT and gelled middlings. Diffrac-
the top and passed through 75, 45, and 20 ␮m sieves. Settling the tion patterns were collected on a Bruker D8 Advance with an
sieved suspension for about 48 h formed a sediment with parti- incident beam parabolic mirror (Co K˛), a 25 mm (or 35 mm
cle sizes ranging between 2 and 20 ␮m. The samples obtained as for oriented slides) sample diameter, and a VANTEC-1TM linear
such were in this study referred to as fraction 1 solids. detector. An exit slit of 0.2 mm was used to minimize intensity
After removing the sediments from the sieved suspension, the aberrations above 2◦ 2.
remaining suspension was further diluted with deionized water
at a four to one volume ratio and left to settle for about 24 h. Rheology
Centrifuging the diluted suspension at about 40,000g for 20 min Rheological measurements were conducted using a TA Instru-
after settling led to the appearance of two layers with different ments AR 2000 rheometer. The geometry used was cone and
shades of grey. After carefully separating the lighter colour sedi- plate (40 mm 2◦ steel cone) suitable for non-Newtonian fluids.
ments from the darker ones, the particle size distributions were This system allowed measurements at low shear rates which were
measured with Malvern Mastersizer 2000. The mean particle size comparable to the shear environment in gravity separation vessels
of the lighter fraction (fraction 2) and the darker one (fraction 3) (Schramm, 1989; Chow et al., 2006).

| 424 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
Table 1. Rheological experiments set up Table 2. Dean stark analysis of stable middle layers separated from
gelled and ungelled middlings samples
Test Test type Parameters Results
No. Middlings sample Bitumen (wt%) Solids (wt%) Water (wt%)

1 Conditioning ˙ = 1, 500 s−1 Ungelled 0.65 8.9 90.4


t = 10 s
2 Equilibration t = 180 s Gelled 4.7 29.5 65.8
3 Flow, steady ˙ = 2 → 2, 000 s−1 Stress (), yield stress
state ( y ), relative viscosity
␮ () ˙ tained some of the ultra-fines and other solids originally entrapped
4 Conditioning ˙ = 1, 500 s−1 in the middlings. Due to these sampling difficulties, an exhaus-
t = 10 s tive characterization of the middlings samples was not possible.
5 Equilibration t = 180 s Nevertheless, it can be envisioned that the ultra-fines remaining
6 Oscillatory, ˙ = 1 × 10−3 → 0.1 s−1 Storage modulus (G ), in the middle layers contribute to the separation characteristics of
strain sweep (angular loss modulus (G ), the original middlings suspensions. Therefore, to reach a sense of
frequency = 6.283 rad/s) phase angle (ı), the difference between the gelled and the ungelled middling sam-
oscillatory stress ples, the stable middle layer was analyzed further. Table 2 shows
7 Conditioning ˙ = 1, 500 s−1
the results of Dean Stark analysis for the gelled and ungelled
t = 10 s
8 Equilibration t = 180 s
middlings samples. As anticipated, the stable middle layer of the
9 Oscillatory, Oscillatory G , G gelled middlings was characterized by higher residual bitumen
frequency stress = constant and solids content.
sweep (measured in test 6) It is instructive to point out that extreme care should be taken
angular while determining the percentage of ultra-fines present in the
frequency = 1 → 100 rad/s gelled middlings sample as the resolution of the laser diffrac-
All tests were performed at 25◦ C. tion equipment can produce measurement artifacts. An instance
of such artifacts occurred in PSD determination of our gelled
middlings sample. After a diligent cleaning of samples from the
Flow and oscillatory tests were performed on the samples with-
stable middle layer for gelled middlings, the PSD measurements
out unloading the samples from the instrument. In order to
showed a mean particle diameter range of 35–50 ␮m while the
prevent drying for the duration of the tests, a solvent trap was
used to cover the sample maintaining it at the vapour pressure of
the suspending medium. Experimental procedure used in rheol-
a 9
ogy measurements is summarized in Table 1. Prior to each test
pre-shearing was performed, followed by an equilibration step to
avoid the effects of shear history which may be present in the
Volume (%)

suspensions as a result of particle aggregation and flocculation. 6


The tests were conducted on suspensions of a fraction or
mixtures containing different weight percentages of the three
fractions. Desired solid concentrations were obtained by adding
3
supernatant water from centrifuged MFT to the stock samples
prepared as described earlier.

Thermogravimetry 0
STA 409 PC Luxx from NETZSCH was used for thermal analy- 0.01 0.1 1 10 100 1000
sis of dried particles. Suspensions of the fractionated particles Particle Size (µm)
were dried in vaccum oven at −10 kPa and 60◦ C for half an hour.
The solid agglomerates formed as a result of heating were then b 12
crushed by mortar and pestle. The dried particles were heated
from room temperature to 600◦ C in a nitrogen environment. Mass
loss of the samples was monitored by the integrated software of 9
Volume (%)

the equipment.
6
RESULTS AND DISCUSSION
Middlings samples received from the field were partially segre- 3
gated. This partial segregation indicates the presence of coarse
sands and bitumen in the sampled middlings. In order to achieve
consistent characterization results, both gelled and ungelled mid- 0
dlings samples were gently re-mixed and allowed to stand for a 0.01 0.1 1 10 100 1000
few minutes. Both samples, however, were observed to be sepa- Particle Size (µm)
rated into three distinct layers comprising of a bitumen layer that
floated to the top, coarse sand sediment containing some bitumen, Figure 3. Particle size distribution of (a) non-separated gelled middling
and a middle layer of muddy suspension. The bitumen layer con- (b) separated gelled middling supernatant.

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 425 |
Table 3. Mineral composition of MFT and gelled middlings determined by XRD

Clay minerals (wt%) ± 15% relative uncertainty


Sample
Illite-smectite (77:23) Illite Kaolinite-smectite (90:10) Kaolinite Quartz Total surface area (m2 /g)*

MFT 44 20 10 26 0.1 186

Gelled middling 31 32 5 31 0.6 137



Derived from XRD domain size in the c* crystallographic direction.

particles less than 1 ␮m were not visualized. A sample of these


Table 5. Water chemistry of fractions 1 to 3 and MFT supernatant
measurements is shown in Figure 3a. Upon fractionation using
(pH 8–8.5) measured by atomic absorption spectroscopy
the procedure outlined in Solids Fractionation Section, PSD of the
final suspension obtained was similar to that obtained for solids Ionic concentration (mg/L)
fractions 2 and 3 from MFT (see Fig. 3b). While an attempt was
made to develop a proper means of determining the actual amount HCO−
3 CO2−
3 Na K Mg Ca
of ultra-fines in the gelled middlings, our results were inconclu-
sive due to limited supply of gelled samples and are not reported MFT supernatant 568.0 116.1 866.3 14.6 7.4 7.7
in this communication. Nevertheless, Figure 3b clearly shows that
ultra-fine components are present in the gelled middlings. Fraction 1 661.5 62.8 811.8 15.4 8.7 15.8
The mineral composition of these ultra-fine fractions of the Fraction 2 601.2 59.9 786.5 19.6 12.7 21.2
gelled middlings and MFT were determined using XRD analy-
sis. The X-ray diffraction patterns of calcium saturated glycolated Fraction 3 548.5 30.3 669.2 12.7 10.1 17.6
slides of solids from MFT and gelled middlings are shown in Fig-
ure 4a. Figures 4b and c respectively show the diffraction patterns
of calcium saturated oriented slides of gelled middlings and MFT The diffraction patterns of MFT and gelled middlings show
in 54% relative humidity and an ethylene glycol environment. similar peaks. Both samples contain discrete kaolinite and illite
The presence of illite-smectite mixed layers was speculated as a in addition to illite-smectite and kaolinite-smectite mixed layers.
result of the broadening of the peak by around 1 nm in the X-ray Tables 3 and 4 show that the type and the amount of minerals in
diffraction pattern when the clays were Ca-saturated and exposed MFT are similar to that found in the gelled middlings.
to 54% relative humidity. At 54% relative humidity, a swelling
pure smectite layer having 2 water molecules causes an increase in Solids Fractions
d-spacing from 1 to 1.5 nm. With ethylene glycol, each molecule Table 5 shows the ion concentration of the water in which
is about 0.35 nm, leading to an expansion from 1 to 1.7 nm. When fractions 1, 2, and 3 were suspended as stock samples. The
smectite occurs as a mixed layer with illite as commonly observed ion concentrations were determined by atomic absorption spec-
in oil sands, there will be a 0.2 nm shift going from 54% relative troscopy. The results in Table 5 show that the concentration of
humidity to ethylene glycol vapour. However, the actual position these ions is similar for all fractions. In addition, all fractions
of the 1 nm peak depends on the extent of mixed layering. Usu- have a similar ionic content as compared to MFT supernatant.
ally, discrete illite and kaolinite do not respond to ethylene glycol Mineral composition of the three fractions was determined
solvation. Upon treatment with organics such as ethylene glycol using XRD analysis. Diffraction patterns of calcium saturated ori-
and glycerol, the d-spacings became more precise and the illite ented slides at 54% relative humidity of the three fractions are
peak became more well-defined, an event that is typical of a clay shown in Figure 5. Mineral composition was identified through
with swelling properties. The presence of randomly interstratified the same procedure as described for the MFT and gelled mid-
kaolinite-smectite is evident by the observed asymmetry of the low dlings. The diffraction data were analyzed using Rietveld analysis
angle peak of the kaolinite at the 0.72 nm peak upon hydration software, AUTOQUANTM from GE Technologies (2005) and the
or glycolation. Quantification (Table 3) was carried out through results are summarized in Tables 6 and 7.
NEWMOD modeling of the diffraction patterns (Reynolds, 1995). The three fractions contain discrete kaolinite and illite in addi-
The surface area calculation (Table 4) was based on the double tion to illite-smectite and kaolinite-smectite mixed layers. Based
Voigt method (Balzar, 2001) implemented in TOPAS (Bruker-AXS, on the results in Table 6, fraction 1 consists mainly of quartz,
2003). The quantification procedure as applied to oil sands has kaolinite and a small amount of illite. The main components of
been reported previously (Omotoso et al., 2002, 2005; Omotoso fraction 2 solids are kaolinite and illite both in the form of discrete
and Mikula, 2004). and mixed-layers forms as well as a detectable amount of quartz.

Table 4. Microstructural parameters derived from XRD domain size in the c* crystallographic direction

Specific surface area (m2 /g) [mean fundamental crystallite thickness, nm]
Sample
Illite-smectite (77:23) Illite Kaolinite-smectite (90:10) Kaolinite

MFT 293 [2.6] 68 [11] 273 [3.3] 61 [15.3]

Gelled middling 273 [2.8] 68 [11] 255 [3.5] 54 [17.5]

| 426 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
Figure 4. Diffraction patterns of gelled middlings and MFT. a)
calcium-saturated solids from gelled middlings and MFT, measured in
ethylene glycol vapor (EGV); b) calcium saturated solids from gelled
middling in EGV and 54% relative humidity (RH); c) calcium saturated
solids from MFT in EGV and 54% relative humidity (RH). [Colour figure
can be viewed in the online issue, which is available at
http://www.interscience.wiley.com.]

In fraction 3, there is no quartz and the amount of discrete illite


and its mixed form with smectite is higher than in the other two
fractions.
Based on the total surface area of each fraction and the weight
percent of the mineral components shown in Table 6, it is evident
that clay minerals are the most significant contributor to the sur-
face area. Table 7 shows the contributions of these components
to the surface area of each fraction. The last row in Table 6 shows
that the surface area of fraction 3 is considerably higher than that Figure 5. Diffraction patterns of calcium saturated oriented slides of the
three solids fractions in 54% relative humidity.
of the other two fractions. The particles of fraction 3 are therefore
expected to have stronger surface activities as these properties are
vastly influenced by the extent of the available surfaces interacting
with each other. Typically, a number of methods are employed to study the rhe-
ological properties of materials. Steady state flow tests monitor
the shear stress variations with changing shear rate. Viscosity,
Rheology the coefficient relating shear stress () and shear rate (), ˙ is
Among the critical parameters of interest in rheology are viscous determined from the results of flow tests. Multiple equations in
properties such as shear viscosity ␮, elastic properties such as literature correlate shear stress and shear rate through a variety of
yield stress  y , and viscoelastic properties such as storage modu- coefficients. In such equations, yield stress is a common param-
lus G and loss modulus G . Changes detected in these properties eter arising for materials having elastic properties. As stated by
under shear flows are indicative of structures that might exist in Macosko (1994) when the transition of the flow from Newtonian
suspensions being investigated. to viscoelastic region occurs gradually, the Casson model having

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 427 |
Table 6. Mineral composition of clay fractions (wt%)

Mineral group Mineral type Fraction 1 (10-␮m) Fraction 2 (0.3-␮m) light Fraction 3 (0.3-␮m) dark

Quartz 27.9 ± 0.5 8.5 ± 0.6

Carbonates Ankerite 0.6 ± 0.2 0.3 ± 0.2

Calcite 1.8 ± 0.6 2.1 ± 1.1

Siderite 5.2 ± 0.3 1.1 ± 0.4

Feldspars K-spar 1.3 ± 0.4 1.9 ± 0.5

Plagioclase 0.8 ± 0.3 0.7 ± 0.4

Pyrite 0.4 ± 0.1 0.2 ± 0.1

Anatase 1.0 ± 0.2 1.3 ± 0.2

Rutile 0.6 ± 0.2 0.8 ± 0.3

Clay minerals Chlorite 1.8 ± 0.5 2.2 ± 0.6

Kaolinite (90)–smectite BDL 7.5 ± 0.7 14.6 ± 0.8

Kaolinite 41 ± 0.7 41.6 ± 0.7 19.7 ± 1.0

Illite (77)–smectite 3.9 ± 0.7 16.7 ± 0.7 33.5 ± 1.7

Illite 14.0 ± 0.7 15.8 ± 0.7 32.4 ± 1.7

Estimated total surface area (m2 /g) 22 ± 1 86 ± 4 174 ± 9


BDL, below detection limit.

Table 7. Surface area derived from XRD domain size in the c* crystallographic direction

Sample Specific surface area (m2 /g)

Kaolinite (90)-smectite Kaolinite Illite (77)-smectite Illite

Fraction 3 (0.3-␮m) dark 260 52 293 87

Fraction 2 (0.3-␮m) light 258 35 280 38

Fraction 1 (10-␮m) 26 228 17


Clay minerals are the most significant contributors to the total surface area.
Contribution of chlorite is negligible.

the following format provides a good fit of the data: oscillatory deformation can be assumed. In rheology measure-
ment with low amplitude oscillatory shearing, the response of a
 1/2 = y1/2 + ␮1/2
c ˙ 1/2 (1) viscoelastic material is solid-like at high angular frequencies or
short time intervals and liquid-like at lower angular frequencies
In Equation (1), ␮c represents Casson viscosity. Yield stress val- or longer time intervals (Larson, 1999).
ues can be calculated through Casson model provided that the For determination of the strength of the solid-like or liquid-like
square roots of shear stress and shear rate correlate well through state of a concentrated suspension, oscillatory frequency sweep
a straight line. tests can be conducted. For a viscoelastic material subjected to
For viscoelastic materials with properties intermediate between a sinusoidal deformation () with amplitude of  0 and angular
purely elastic or solid-like and purely viscous or liquid-like, a com- frequency of ω,
mon measurement technique is to apply oscillatory shear while
monitoring stress. An effective approach to investigate the struc-  = 0 sin(ωt) (2)
ture or state of a material is to impose oscillatory shear with
amplitudes sufficiently small to ensure that the fluid microstruc-
ture is not deformed significantly during the measurement. In this or
regime, also known as the linear viscoelastic regime, a direct pro-
portionality between the oscillating shear stress and the applied ˙ = ˙ 0 cos(ωt) (3)

| 428 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
The shear stress will also oscillate with the same frequency except
with a phase lag (ı) which depends on the viscoelastic character-
istics of the material:

 = 0 sin(ωt + ı) (4)

The shear stress and strain can then be correlated using a complex
shear modulus, G* :

(t) = G∗ (t) (5)

G* can be written in terms of its real and imaginary components


as:

G∗ = G + iG (6)
Figure 6. Relative viscosity of fraction 1 (open symbols) and 2 (solid
symbols) with changing shear rate at different particle concentrations
leading to: (wt%).

 = G 0 sin(ωt) + G 0 cos(ωt) (7)

G , the real component of the viscoelastic modulus G* , represents


the elastic properties of the material.
Equation (7) shows that the elastic component of the mate-
rial response is out of phase with the strain rate, . ˙ Here, G is
referred to as the storage modulus as it represents the capacity of
the material to resist deformation and store an applied stress. G ,
the imaginary component of G* representing the viscous proper-
ties of the material is called the loss modulus as it is a measure of
the energy dissipated per cycle of deformation. The G response
in Equation (6) is in phase with the rate of deformation as shown
in Equation (2) (Macosko, 1994; Barnes et al., 1989).
In a frequency sweep test, when G is higher than G the mate-
rial shows less energy dissipation due to the applied deformation
and has stronger elastic (solid-like) properties. This can be inter- Figure 7. Relative viscosity of a suspension made of different ratios of
preted as gel formation, the extent of which can be estimated by fractions 1 and 2 (10 wt% total concentration).
the difference between G and G .
To characterize the rheological properties of the oil sand solids trend with variations in particle concentrations, the viscosity
fractions both flow and oscillatory tests were performed. Viscos- transition to higher values appeared to be more significant for
ity values reported here with the symbol ␮r , are relative to the fraction 2 having particles of smaller sizes. In all cases, at a given
viscosity of water under the test conditions. solids concentration and shear rate higher viscosity values were
Figure 6 shows the viscosities of fractions 1 and 2 at three differ- obtained for fraction 2 than for fraction 1.
ent solids concentrations. The suspensions, as is expected for clay Figure 7 shows the rheological behaviour of a mixture of
suspensions, show non-Newtonian, shear-thinning behaviour. fractions 1 and 2. While keeping the total solids content at 10
Higher viscosity at lower shear rates indicates the presence of wt%, the percentage of fraction 2 was increased. The suspension
a structure in the samples, which breaks down as a result of viscosity was monitored for simple shear tests following the same
increased shear. In Figure 6, the break down of the structure in procedure as for the single fractions. The results in Figure 7 show
the sample is manifested as a gradual decrease of viscosity until that although the total concentration is kept at 10 wt%, adding
a high shear rate viscosity plateau is reached. Depending on the particles of fraction 2 to the suspensions of fraction 1 increased its
fraction of ultra-fines studied and the particle concentration of the viscosity. This trend continued till the suspension was completely
sample, the high shear rate plateau is observed at different values composed of fraction 2 solids where the maximum viscosity was
of shear rate. Figure 6 shows that the plateau is reached at lower reached. This monotonic increase in viscosity with increasing
shear rates for lower solids concentrations and for lower concen- solids fraction of fines was different from observations reported
trations of fraction 1 solids. As shown in Figure 6, the increase by Chong et al. (1971), where a minimum in the viscosity was
in the viscosity with decreasing shear rates at low shear regime predicted for a concentrated bimodal suspension of spherical par-
is more pronounced for fraction 2. At 10 wt% solids, for exam- ticles. The monotonic increase in viscosity with increasing the
ple, the viscosity of fraction 1 is not a strong function of shear content of fines fractions as observed in our system appears to
rate while for fraction 2 there is a sharp rise in the viscosity with suggest minimal interactions between coarse and fine particles.
decreasing shear rate. Figure 8 shows the viscoelastic moduli for fractions 1 and 2.
Figure 6 also shows that for fractions 1 and 2, increasing the At 40 wt% solids content, Figure 8a shows that the value of
particle concentration led to an increased viscosity. This increase G exceeds G indicating the presence of a weak structure. At
in viscosity would be expected as at high concentrations, particles 20 wt% solids concentrations and higher frequencies the G and
are closer to each other leading to more effective particle–particle G lines cross each other and G becomes higher than G . With the
interactions. However, although both fractions show a similar loss modulus, G , higher than the storage modulus, G , for most

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 429 |
Figure 8. Viscoelastic moduli of fractions 1 and 2 at different solids
concentrations (wt%).

Figure 9. The square root variations of shear stress with shear rate for
frequencies tested, fraction 1 could not resist deformation at con- fractions 1 and 2 at different solids concentrations (wt%).
centrations lower than 30 wt%. For this fraction, gelling appears
when the solids concentration reaches values around 40 wt%.
On the other hand, the results in Figure 8b for fraction 2 show
a different trend. At 10 wt% solids concentration the value of G is
higher than G for most test frequencies. Also, at 20 wt% a strong
gel is formed as G at this concentration is significantly higher than
G . Noting the different scales of Figure 8a and b and comparing
the viscoelastic behaviour of fractions 1 and 2 show that frac-
tion 2 has stronger particle interactions thereby leading to more
structured suspensions with higher gelation propensity. It is also
interesting to note that for both fractions 1 and 2, increasing solids
concentration transformed the suspension from an ungelled sys-
tem to a gelled structure. This transformation, however, occurred
more readily for solids of fraction 2.
The results of simple shear flow tests were used to calculate the
Figure 10. Yield stress values of fraction 1 and fraction 2 with changing
yield stress values of the suspensions tested. Figure 9 shows the weight fraction.
square root variations of shear stress with shear rate for fractions 1
and 2. Interestingly,  1/2 vs. ˙ 1/2 variations gave a straight line
which suggests that for fractions 1 and 2, Casson model would be The other two regimes were designated as weak and no gel. This
a suitable correlation for yield stress calculation. graph shows that fraction 2 mainly falls in the gel regime even
Figure 10 summarizes the values of yield stress calculated for at relatively low solids content of ca. 13 wt%. It also shows that
fractions 1 and 2 using the Casson model. Incorporating the results fraction 1 did not form a strong gel until its suspension contained
of the frequency sweep tests discussed earlier into this figure, the high concentrations of solids.
status of gelation was mapped for suspensions of fractions 1 and Yield stress was also calculated for some mixtures of fractions 1
2. At 20 wt%, fraction 2 started to show a gelled structure. Simi- and 2. Figure 11 shows that for these mixtures there is a linear
lar structured behaviour was not achieved for fraction 1 until 40 relationship between the square roots of shear stress and shear
wt% solids concentration. Based on these observations, a yield rate. Thus, the Casson model can be used to calculate yield stress.
stress of about 0.1 Pa in Figure 10 is chosen to mark the starting Figure 12 shows the yield stress values calculated for mixtures of
point of gelation as it corresponds to gel forming concentrations. fractions 1 and 2 with 10 wt% total solids. The results show that

| 430 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
Figure 11. The square root variations of shear stress with shear rate for Figure 14. Thermogravimetric analysis on solids of fractions 2 and 3.
mixtures of fractions 1 and 2.

Figure 2 showed nearly the same particle size distributions for


fractions 2 and 3, they were distinguishable by their colour. It was
speculated that this colour difference was due to high amounts of
adsorbed organic matter on the surface of the particles in fraction
3. The higher content of organic matter on the surface of fraction
3 particles was confirmed by a great mass loss in thermogravimet-
ric analysis (Figure 14). A relatively small mass loss of fraction 2
in Figure 14 is due to the loss of interlayer water of the clays.
To visualize the structures formed in fractions 2 and 3, scan-
ning electron micrographs of the two solid fractions were taken.
A much less well-defined morphology of particles in fraction 3
than in fraction 2 is evident. Particles in fraction 2 appear more
platy like with sharper edges as compared with particles in frac-
tion 3. The SEM micrographs shown in Figure 15 also illustrate
Figure 12. Yield stress values for bimodal systems of fraction 1 and 2 at a notable difference in the particle sizes of these two fractions.
different ratios of each fraction in 10 wt% total solids concentration. Contrary to the light scattering particle size measurements in Fig-
ure 2, solids in fraction 3 appears to have smaller particle size as
addition of fraction 2 to a suspension of fraction 1 solids increased compared to the solids in fraction 2. The overestimation of par-
the solid-like behaviour of the system. The increase of yield ticle sizes of fraction 3 with Malvern Mastersizer can be due to
stress in Figure 12 is non-linear with increasing the percent of the formation of tactoids as the solids are dispersed in an aqueous
fraction 2. system during the measurements.
Figure 13 shows the results of simple shear flow tests on frac- The higher viscosity of fraction 3 could be attributed to either
tions 2 and 3 at a 1:1 mass ratio in a suspension of 20 wt% total the organic coating of the solids in the fraction or their smaller par-
solids concentration. The combination of fractions 2 and 3 showed ticle sizes of less well-defined morphology. Further simple shear
a higher viscosity than fraction 2 alone. Figure 13 thus shows tests were performed on a 23 wt% suspension of fraction 3 after
that at a given total concentration, the more fraction 3 solids removing of the organic matter on the particle surfaces in order to
are present in the suspension, the higher is the viscosity. While study the effect of organic coating. Hydrogen peroxide was used
to remove the organic coatings. The cleaned solids were then
washed and redispersed in MFT supernatant. Figure 16 shows
the SEM images of the solids in fraction 3 in the absence of the
organic matter on the particle surfaces. The solution chemistry of
these solids following the wash in MFT supernatant is shown in
Table 8. According to the SEM micrograph and the water chem-
istry of the solids, removing the organic matter with hydrogen
peroxide did not cause a significant alteration in particle size and
solution chemistry of fraction 3. However, the results of simple
shear flow tests in Figure 17 show a drastic decrease in relative
viscosity at a given shear rate when fraction 3 solids are treated
with hydrogen peroxide (H2 O2 ) to remove organic coatings on the
solids, even at a slightly higher suspension solids content (23 wt%
for H2 O2 treated solids vs. 20 wt% for untreated solids). Clearly,
surface wettability controlled by the presence of organic matters
on particle surfaces plays a critical role in determining rheological
Figure 13. Relative viscosity of a 20 wt% solids suspension formed of behaviour of fine solids in oil sands processing streams. With a
different combinations of fraction 2 and 3. different approach, Angle et al. (1993) also observed higher elastic

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 431 |
Figure 16. Scanning electron micrograph of fraction 3 particles after
wash with hydrogen peroxide.

Figure 15. Scanning electron micrographs of solids from (a) fraction 2 Figure 17. Relative viscosity of fraction 3 suspensions with and without
and (b) fraction 3. the organic coating of the particles.

Table 8. Solution chemistry of the solids in fraction 3 redispersed in CONCLUSIONS


MFT supernatant after washing in hydrogen peroxide In oil sands extraction, recovery is very much dependent on the
viscosity of the slurry in the primary separation vessel. At high
Ionic concentration (mg/L)
slurry viscosities, the rise of aerated bitumen in the vessel is
HCO− CO2− Na K Mg Ca
hindered and the bitumen separation efficiency is reduced. In
3 3
this study, rheological measurements were utilized to investigate
624.2 68.7 855.2 15.8 8.16 11.8 the role of particle size on the viscosity of oil sand suspensions.
Mature fine tails (MFT) were chosen as the source of solids. XRD
analysis confirmed the similarity of these solid particles to those
properties for organic rich solids separated from mature fine tails obtained from the extraction vessel.
and related the structures of the sludge to bound organic matter Three fractions of solids with three different particle size distri-
on the surface of the clays. butions were separated from MFT. XRD analysis showed that each
The DLVO theory suggests that particle interactions are influ- fraction is composed of a variety of minerals at different weight
enced both by particle size as well as the electric properties of fractions. Samples of these fractions at certain weight percentages
the solid surfaces (i.e., mineralogy). While the focus of this paper were prepared by addition of supernatant water separated from
is to investigate the effect of particle size on suspension rheol- centrifuged MFT to maintain solution chemistry.
ogy, it is possible that more active clay particles are increasingly According to the simple shear flow tests, all fractions behave
concentrated in finer size fractions, contributing to an increased as non-Newtonian fluids showing shear thinning behaviour. The
rheology. Further work is needed to resolve the contributions of observed yield stress values indicated a structure in the solids
clay mineralogy from the contributions of finer sizes of particles suspensions. For all samples, increasing solids content led to
and organic coatings to rheology. higher viscosities and yield stresses as well as stronger viscoelas-

| 432 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |
tic properties. At a given particle concentration, the oscillatory Fraction, Shear Stress and Deflocculant Content,” Br. Ceramic
tests showed that compared to fraction 1, fraction 2 has higher Trans. 101, 185–193 (2002).
viscoelastic properties where it formed stronger solid-like gels. Angle, C. W., R. Zrobok and H. A. Hamza, “Surface Properties
The result of the oscillatory tests also showed that by increasing and Elasticity of Oil-Sands-Derived Clays Found in a Sludge
solids content, an ungelled suspension can form a gel. Pond,” Appl. Clay Sci. 7, 455–470 (1993).
Among the three fractions, the particles in fraction 3 with a high Balzar, D., “Voigt-Function Model in Diffraction Line-Broadening
content of organic matter on the particle surfaces showed the high- Analysis,” In “Microstructure Analysis from Diffraction,”
est viscosities and gelation potency. More solid-like characteristics IUCR monograph, R. L. Snyder, et al. Eds., Oxford University
of suspensions of fraction 3 were diminished after removal of the Press, New York, NY (2001).
organic matter from the surfaces of particles in fraction 3. Barnes, H. A., J. F. Hutton and K. Walters, “An Introduction to
These results further suggest that organic matter adsorbed on Rheology,” Elsevier, Amsterdam (1989), pp. 46–48.
the mineral surfaces results in stronger particle interactions, caus- Bruker-AXS. “TOPASTM Rietveld Refinement Software,”
ing an increase in suspension viscosity. A thorough understanding Bruker-AXS, Madison, WI (2003).
of the nature of these organics and their presence in the gravity Chong, J. S., E. B. Christiansen and A. D. Baer, “Rheology of
separation vessel is of great importance and should be investigated Concentrated Suspensions,” J. Appl. Polym. Sci. 15,
further. 2007–2021 (1971).
Chong, J., S. Ng, K. H. Chung, B. D. Sparks and L. S. Kotlyar,
“Impact of Fines Content on a Warm Slurry Extraction
END NOTES Process Using Model Oil Sands,” Fuel 82, 425–438 (2003).
1 It is for this reason that thickening or sludging in the gravity separation Chow, R., J. Zhou and D. Wallace, “Rheology of Oil Sands
vessel is also referred to as gelation by some operators.
Slurries,” Oil sands Conference, University of Alberta (2006).
Eckert, W. F., J. H. Masliyah, M. R. Gray and P. M. Fedorak,
ACKNOWLEDGEMENTS “Prediction of Sedimentation and Consolidation of Fine
Tails,” Am. Inst. Chem. Eng. 42, 960–972 (1996).
The authors gratefully acknowledge the financial support from
Faers, M. A., T. H. Choudhury, B. Lau, K. McAllister and P. F.
the Natural Sciences and Engineering Research Council of Canada
Luckham, “Syneresis and Rheology of Weak Colloidal Particle
through the NSERC Industrial Research Chair in Oil Sands Engi-
Gels,” Colloids Surf A Physicochem Eng Aspects 288,
neering and a fellowship to O.B.A. Special thanks to Mr Brad
170–179 (2006).
Komishke for valuable discussions which contributed immensely
GE Technologies. “AUTOQUANTM Rietveld Refinement
to the progress of this work. The help of Dr. Dipo Omotoso and
Software,” GE Technologies, Lewistown, PA (2005).
Heather Kaminsky with the XRD analysis is also acknowledged.
Larson, R. G., “The Structure and Rheology of Complex Fluids,”
We acknowledge the work carried out by Dean Wallace, Ross
Oxford University Press, New York, NY (1999).
Chow and Joe Zhou under a project sponsored by the Canadian Oil
Macosko, C. W., “Rheology: Principles, Measurements and
Sands Network for Research and Development (CONRAD), from
Applications,” Wiley-VCH, New York, NY (1994), pp. 95,
which some test protocols in this paper were adapted.
121–123.
McFarlane, A., K. Bremmell and J. Addai-Mensah, “Improved
NOMENCLATURE Dewatering Behavior of Clay Minerals Dispersions Via
Interfacial Chemistry and Particle Interactions Optimization,”
d50 mean particle diameter (␮m)
J. Colloid Interface Sci. 293, 116–127 (2006).
G* complex shear modulus (Pa)
Moore, D. M. and R. C. Reynolds Jr., “X-Ray Diffraction and the
G storage modulus (Pa)
Identification and Analysis of Clay Minerals,” Oxford
G loss modulus (Pa)
University Press, New York, NY (1997).
Mpofu, P., J. Addai-Mensah and J. Ralston, “Investigation of the
Greek Symbols Effect of Polymer Structure Type on Flocculation, Rheology
ı phase lag (rad) and Dewatering Behaviour of Kaolinite Dispersions,” Int. J.
 deformation Miner. Process 71, 247–268 (2003).
0 zero frequency deformation Omotoso, O. and R. Mikula, “High Surface Area Caused by
˙ shear rate (s−1 ) Smectitic Interstratification of Kaolinite and Illite in
␮ viscosity (Pa s) Athabasca Oil Sands,” Appl. Clay Sci. 25, 37–47 (2004).
␮c Casson viscosity (Pa1/2 s1/2 ) Omotoso, O., R. Mikula and P. W. Stephens, “Surface Area of
␮r relative viscosity Interstratified Phyllosilicates in Athabasca Oil Sands From
 Shear stress (Pa) Synchrotron XRD,” Adv. X-Ray Microanal. 45, 391–396
0 zero frequency stress (Pa) (2002).
y yield stress (Pa) Omotoso, O., R. J. Mikula, S. Urquhart, H. Sulimma and P. W.
ω angular frequency (rad/s) Stephens, “Characterization of Clays From Poorly Processing
Oil Sands Using Synchrotron Techniques,” Clay Sci. 12, 88–93
(2005).
REFERENCES Rasteiro, M. G. and I. Salgueiros “Rheology of Particulate
Abend, S. and G. Lagaly, “Sol-Gel Transitions of Sodium Suspensions in Ceramic Industry,” Particul. Sci. Technol. 23,
Montmorillonite Dispersions,” Appl. Clay Sci. 16, 201–227 145–157 (2005).
(2000). Reynolds, R. J., NEWMODTM Clay modeling software (1995).
Amorós, J. L., V. Sanz, A. Gozalbo and V. Beltrán, “Viscosity of Sanford, E. C., “Processability of Athabasca Oil Sand:
Concentrated Clay Suspensions: Effect of Solids Volume Inter-Relationship Between Oil Sand Fine Solids, Process

| VOLUME 87, JUNE 2009 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 433 |
Aids, Mechanical Energy and Oil Sand Age After Ining,” Can.
J. Chem. Eng. 61, 554–567 (1983).
Sanford, E. C., “Processability of Athabasca Oil Sand:
Inter-Relationship Between Oil Sand Fine Solids, Process
Aids, Mechanical Energy and Oil Sand Age After Ining,” Can.
J. Chem. Eng. 33, 103–107 (1998).
Schramm, L. L., “The Influence of Suspension Viscosity on
Bitumen Rise Velocity and Potential Recovery in the Hot
Water Flotation Process for Oil Sands,” J. Can. Petrol Technol.
28, 73–80 (1989).
Sparks, B. D., L. S. Kotlyar, J. B. O’Carroll and K. H. Chung,
“Athabasca Oil Sands: Effect of Organic Coated Solids on
Bitumen Recovery and Quality,” J. Petrol Sci. Eng. 39,
417–430 (2003).
Syncrude Analytical Methods, “Syncrude Analytical Methods of
Oil Sand and Bitumen Processing,” Syncrude Canada Ltd,
Edmonton Alberta (1979).
Tu, Y., J. B. O’Carroll, L. S. Kotlyar, B. D. Sparks, S. Ng, K. H.
Chung and G. Cuddy, “Recovery of Bitumen From Oil Sands:
Gelation of Ultra-Fine Clay in the Primary Separation Vessel,”
Fuel 84, 653–660 (2005).
Wallace, D., R. Tipman, B. Komishke, V. Wallwork and E.
Perkins, “Fines/Water Interactions and Consequences of the
Presence of Degraded Illite on Oil Sands Extractability,” Can.
J. Chem. Eng. 82, 667–677 (2004).

Manuscript received July 5, 2007; revised manuscript


received November 20, 2007; accepted for publication July 3,
2008.

| 434 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 87, JUNE 2009 |

You might also like