Sei sulla pagina 1di 10

International Journal of Mineral Processing 87 (2008) 90–99

Contents lists available at ScienceDirect

International Journal of Mineral Processing


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o

Using turbulent pipe flow to study the factors affecting polymer-bridging flocculation
of mineral systems
A.T. Owen, P.D. Fawell ⁎, J.D. Swift, D.M. Labbett 1, F.A. Benn, J.B. Farrow
Parker Cooperative Research Centre for Integrated Hydrometallurgy Solutions (CSIRO Minerals) PO Box 7229, Karawara, WA 6152, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Batch flocculation tests in cylinders (jars) or stirred vessels suffer from poor control over the reaction time and
Received 28 June 2007 involve broad shear rate distributions. The use of a linear pipe reactor provides continuous turbulent mixing of
Received in revised form 4 February 2008 flocculant and slurry at well-defined mean shear rates. Combined with in-line, real-time monitoring of the
Accepted 16 February 2008
aggregation state with a focused beam reflectance measurement (FBRM) probe, reaction times can be
Available online 29 February 2008
determined to within a fraction of a second. The resultant reaction profiles of aggregate growth and breakage
obtained while varying different conditions (dosage, solids concentration, shear rate, split dosing) provide
Keywords:
new insights into flocculation kinetics that are more relevant to mineral processing applications. In this study,
Flocculation
Kinetics the reaction profiles are examined in terms of the effects of flocculant adsorption, applied shear and solids
FBRM dilution on the aggregate structures formed. Also presented is the first definitive evidence of “post-
Flocculant aggregation”, where surface adsorbed flocculant remains active at short reaction times (up to the peak in
Aggregate aggregate size in the measured profiles), allowing additional aggregate growth under subsequent reduced
Settling rate shear. At reaction times beyond the peak, little or no additional growth is observed. This behaviour explains
Turbulence the observed discrepancy between size and settling rate reaction profiles, and highlights the importance of
optimising hydrodynamics in industrial flocculation applications to maximise settling flux.
Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.

1. Introduction whereas flocculation is normally defined as involving polymer-


bridging mechanisms. While coagulation does still have relevance in
The flocculation of mineral slurries by high molecular weight water hydrometallurgical applications, this is only as a precursor to floc-
soluble polymers has been extensively studied over the last 50 years. This culation for the treatment of feeds with sub-micron fines.
has led to a vast body of published literature which has been the subject of • Flocculation is frequently studied as a batch process, either in
numerous reviews (e.g. Henderson and Wheatley, 1987; Gregory, 1989; cylinder (jar) tests or in stirred vessels. Such batch tests offer poor
Hocking et al.,1999; Fellows and Doherty, 2006; Bolto and Gregory, 2007). control over the reaction time and the applied hydrodynamics.
Despite this effort, the actual link between fundamental knowl-
As part of an ongoing research focus on the design and op-
edge and practical application in hydrometallurgy has not been well
timisation of thickener feedwells, we have developed detailed pipe
developed, and indeed, a substantial fraction of these literature
flocculation procedures to characterise the kinetics of aggregate
studies have only limited relevance to real systems. There are several
growth and breakage. Flocculation in pipe flow offers the advantages
factors contributing to this:
of relatively homogeneous turbulence, a mean shear rate that is
• Many fundamental studies have been carried out on sub-micron readily estimated and excellent control of the reaction time under
sized particles, typically low density standard polymer spheres. the applied shear. Pipe flow has been applied to aggregation studies
While such polymers offer narrow particle size distributions and the (e.g. Delichatsios and Probstein, 1975; Klute and Amirtharajah, 1991;
resultant aggregates are readily characterised, the nature of the Byrne and Fitzpatrick, 2002; Carissimi and Rubio, 2005, Carissimi
flocculants plus the flocculation conditions and hence the actual et al., 2007; Zumaeta et al., 2006), although generally with off-line
mechanisms are usually very different from those encountered with determination of aggregate size distributions. Off-line size measure-
comparatively coarse mineral particles. ments are questionable for any aggregation process, and are certainly
• As a consequence, many studies that purport to be of flocculation not practical for the fragile aggregates formed by polymer-bridging
actually deal with coagulation through surface charge effects, flocculation.
The in-line determination of chord length distributions by focused
beam reflectance measurement (FBRM) makes pipe flocculation
⁎ Corresponding author. Fax: +61 8 9334 8001.
E-mail address: phillip.fawell@csiro.au (P.D. Fawell).
studies a viable option. This was first applied by Peng and Williams
1
Current address: Food Science Australia, 671 Sneydes Rd, Werribee, VIC 3030 (1994), who examined the effect on aggregate breakage of flow
Australia. through a fixed pipe length. We expanded upon this approach to

0301-7516/$ – see front matter. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.minpro.2008.02.004
A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99 91

examine both aggregate growth and breakage, varying the distance Slurry feed to the pipe reactor was pumped (between 10 and 30 L min− 1) from the
feed tank using a positive displacement “mono-pump” fitted with a variable speed
between flocculant addition and FBRM detection to generate profiles
drive. Dilution liquor (water) was pumped (between 10 and 20 L min− 1) into the system
of aggregate size as a function of reaction time (Swift et al., 2004). by an equivalent pump. Electromagnetic flow meters were used to monitor the slurry
Subsequent studies of calcite flocculation (Heath et al., 2006a) were and dilution flow rates, which were individually adjusted to give a desired solids
used to develop the first population balance (PB) model for concentration, with care taken to ensure the total flow rate (and therefore mean shear
flocculation that includes irreversible aggregate breakage, a critical rate) was always constant across runs. Solids concentrations at the outflow were
checked at intervals gravimetrically. A peristaltic pump was used to add the diluted
aspect of the behaviour of polymer-bridged aggregates under shear flocculant solution down the centre of the pipe reactor through a 3.91 mm ID stainless
(Heath et al., 2006b). The experimental approach, termed the linear steel pipe (6.4 mm OD × 5.8 m). Flocculant solution velocities at the entry point varied
pipe reactor, has been used to characterise flocculation kinetics from 0.06 to 1.1 m s− 1. When required, the length of the pipe reactor was changed by
through application of the PB for numerous industrial thickener feed stopping the slurry, dilution and flocculant flows, flushing the system with plant water,
disconnecting the pipe reactor and adding additional pipe lengths.
suspensions. The PB has now been incorporated within a computa-
Fig. 1 shows that after the FBRM probe the slurry enters a Perspex column (25 mm
tional fluid dynamics (CFD) model for thickener feedwells, allowing ID) which can be isolated to allow hindered settling rates to be measured. The distance
3D predictions of aggregate size in addition to predicted flow patterns, from the FBRM probe to the top valve of the settling column is 1.7 m, which is taken into
solids distribution and flocculant adsorption (Nguyen et al., 2006). account when fixing pipe lengths, to ensure that settling rates are measured at reaction
While the main purpose of the linear pipe reactor has been to times with corresponding FBRM data. While wall effects are known to influence
standard cylinder test procedures, Farrow and Swift (1996) have shown that this mostly
provide data for PB modelling (Heath et al., 2006a,b,c), the measured reflects variations in flocculation conditions in different cylinder diameters. When
profiles for aggregate size and settling rate as a function of reaction flocculation is done external to the cylinder (as in this case), wall effects are negligible
time obtained in response to variations in flocculation conditions for cylinder diameters of 25 mm and above.
(dosage, shear rate and solids concentration) have provided valuable The effect of reduced shear was examined by incorporating different lengths of
16.6 mm ID pipe prior to the existing 22.1 mm ID pipe reactor segments. The different
insights into the basics of the polymer-bridging flocculation processes.
size pipes were joined with short tapered (4° angle) sections to permit a gradual
The capacity to tightly control reaction time and flocculation con- expansion of the flow and therefore permit a smooth transition from one flow (shear)
ditions also allows the preparation of aggregated suspensions for condition to the other.
other off-line characterisation methods. This paper describes some of Slurry flow rates of 10, 15, 20 and 30 L min− 1 were examined in the 22.1 mm ID pipe,
these insights gained directly from linear pipe reactor studies on the which correspond to linear flow velocities of 0.43, 0.65, 0.87 and 1.30 m s− 1,
respectively. For the 16.6 mm ID pipe, flow rates of 10 and 20 L min− 1 corresponded
flocculation of standard kaolin and calcite slurries.
to linear flow velocities of 0.77 and 1.54 m s− 1, respectively. Experiments with a
10.8 mm ID pipe were for flocculant adsorption studies only, and are described in
2. Experimental
Section 2.4.
The principle of FBRM has been described by Peng and Williams (1994) and Fawell
2.1. Feed solids
et al. (1997). In all experiments the focal point of the laser was set at the outer surface of
the probe window. Measurements at each pipe length were made over 2 s periods and
Two substrates were used in this study, namely kaolin clay (RF Kaolin, Unimin
every measurement recorded. Five measurements, taken once steady state was
Australia) and calcite (Omyacarb, Omya Australia). Bulk slurries of either substrate (in
reached, were averaged during subsequent data processing.
the range 30 to 300 g L− 1) were prepared by adding the solids to Perth (Western
The primary data collected by the FBRM probe is in the form of the number of
Australia) tap water in a stirred 1 m3 mixing tank. Calcite studies were carried out using
chords measured within a certain chord range (size channel) per unit time. In this study
Omyacarb 10 (d50 11.7 μm, density 2.71 g cm− 3). The calcite slurries were stirred for
90 channels were used, ranging from 0.8 to 1000 µm in a logarithmic progression. The
at least 1 h prior to use to disperse and fully wet the particles. For kaolin (density
statistical weightings that are applied to this data have been described elsewhere (Swift
2.68 g cm− 3), preliminary testwork established that 15 h of continual stirring was
et al., 2004). The unweighted chord length distribution is number-sensitive and
required to breakdown the natural aggregates existing within the solids to a stable and
provides a good indication of the presence of fines. Applying a length square-weighting
reproducible particle size (d50 5.5 μm, compared to ~ 4 μm achieved after sonication).
to the chord length distribution accentuates the volumetric contribution of the
Kaolin slurries were therefore prepared the day prior to experiments and stirring
aggregates within the slurry. Distributions are presented as line graphs for ease of
continued until used.
comparison — in reality they should be histograms over the full 90 chord length
channels. For weighted distributions, the counts in each channel are weighted at the
2.2. Flocculants
channel midpoint.
The mean shear rates (calculated from energy dissipation) and Reynolds Numbers
For the bulk of the pipe reactor studies, a high molecular weight, 35% anionic
for the pipe reactor at each flow rate are given in Table 1. The mean shear rates ranged
flocculant (acrylamide/acrylate copolymer) was used. For adsorption experiments
from 240 to 1660 s− 1, with turbulence expected at all conditions studied. Note that these
within pipe flow, a low molecular weight, 20% anionic flocculant was used. Both are
values are for the use of water in the pipe reactor at 25 °C, and do not correspond exactly
commercially available flocculants well known within the mineral processing industry.
with those of the slurries which have varying viscosity.
Only anionic flocculants were considered, as they are used in by far the majority of
In turbulent flow there is a range spanning from very high shear rates in the viscous
relevant thickening applications (Connelly et al., 1986). However, work within our
sub-layer (that only occupy a very small fraction of the flow volume, typically less than
laboratories has shown the same general trends for flocculation with nonionic
1%) to quite moderate rates in the core, which occupies approximately 75% of the flow
flocculants and anionic flocculants applied after coagulant addition. The same trends
volume. Intermediate shear rates apply in the inertial sub-layer. For the highest flow
should also be expected for cationic flocculants, provided their molecular weight is
rate flow rate studied in the 22.1 mm pipe (mean shear of 1100 s− 1), we estimate peak
sufficiently high to give predominantly a bridging mechanism.
wall shear rates to be 5000 s− 1 and values in the core of the flow at approximately one
Concentrated (0.10 wt.%) flocculant solutions were prepared the day prior to use to
order of magnitude less than the mean. Of course there is good mixing between the core
minimise the effect of flocculant aging. Powder flocculant (4.00 g) was introduced
and inertial sub-layers and most particles experience shear in the intermediate zone
slowly into the vortex of rapidly stirred deionised water (4.0 L), with the impellor speed
multiple times in their passage through the pipe.
maintained at 500 rpm for 10 min. After this time the impellor speed was reduced to
200 rpm for 2 h, and the solution was then left to stand overnight. The following day the
entire volume of stock solution flocculant was then diluted to the final working 2.4. Flocculant concentration determinations
concentration (0.01 wt.% unless otherwise stated) with Perth tap water and stirred for
10 min at ~ 200 rpm. Any unused flocculant was discarded each day. For adsorption Residual flocculant concentrations were determined by size exclusion chromato-
experiments, additional dilutions to 0.0033% and 0.0067% were made as required. graphy, based on the approach described by Beazley (1985) and Hunt et al. (1988). A
All flocculant dosages are expressed in terms of grams per tonne of dry solid substrate Shimadzu LC-10AD Liquid Chromatography pump was used for solvent delivery, with a
(g t− 1). Shimadzu SPD-10AV UV–VIS detector attached (wavelength 208 nm). Separation was
achieved with a Waters Ultrahydrogel™ 250 column, giving a nominal upper exclusion
2.3. Linear pipe reactor (LPR) operation limit of 80 000. The mobile phase was 0.05 M sodium sulphate in 0.2 µm filtered water,
pumped at a flow rate of 0.4 mL min− 1. Samples were injected through a Rheodyne
The LPR consisted of a number of stainless steel tubes (ID 22.14 mm × 5.5 m) that 7125R injector fitted with a 100 µL sample loop. Data acquisition and integration were
could be joined together to provide the required length (reaction time). The slurry feed performed with DAPA Chromatography software (version 1.4×). Polymer recovery was
and flocculant entered at one end, with an FBRM probe (M500, Mettler Toledo, not calculated, but is expected to be N 90% (Beazley, 1985).
Redmond, Washington, USA) located at the other, its window projecting into the flow at Concentrated kaolin slurry prepared as described in Section 2.1 was diluted to 40 g
an angle of 45°. The measurement end of the stainless steel probe was a 25 mm OD L− 1 with deionised water prior to use. The pipe set-up was essentially a scaled-down
cylinder with a 12 mm diameter flat sapphire window set into its base. A schematic of version of the LPR using 10.8 mm internal diameter stainless steel pipes, with a septum
the LPR system is shown in Fig. 1. (Puresep T) instead of the FBRM probe in the flowcell. A syringe needle (1.0 mm ID)
92 A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99

Fig. 1. Representation of the layout and principle of operation for the Linear Pipe Reactor (LPR).

inserted through the septum into the middle of the pipe was used to rapidly withdraw a 0.0033% and 0.0067% dilutions only the highest flocculant flow rate (3.6 L min− 1) was
flocculated sample into a 20 mL syringe that was inverted to very rapidly settle the used, thus maintaining the same dosages of 300 and 600 g t− 1, respectively.
solids and then the liquor passed through a 0.45 µm syringe filter (Gelman Acrodisc) Adsorption isotherm data was obtained from solution depletion tests, in which
prior to analysis. The intention was to achieve separation as quickly as possible, and the substrates (100 g L− 1 at pH 7) were mixed with known concentrations of flocculant (1 to
high flocculant dosages ensured the well flocculated solids were extremely fast settling. 1000 ppm) for 2 h and the liquor separated by centrifugation. Residual flocculant
After withdrawing sample into the syringe (~ 1 s), mixing was considered to be concentrations were then determined as described above.
negligible. The time to obtain a solids free sample was typically 3 s. Any breakdown of
the aggregates during the 1 s under the high shear conditions in the syringe needle 2.5. Floc Density Analyser (FDA)
would have little effect on the amount of unadsorbed flocculant. However, it is
inevitable that the adsorption process will progress further during this sample
The Floc Density Analyser (FDA) is a device developed for measuring “microscopic
treatment, and therefore the reaction times must be considered as nominal values.
properties” such as size, shape, density and settling velocity of individual aggregates (or
Despite this, the measured trends proved to be reproducible. A slurry flow rate of 10 L
“flocs”) within their process liquor (Farrow and Warren, 1993). A similar approach has
min− 1 was maintained, corresponding to a linear flow rate of 1.82 m s− 1. The flow rate of
been applied by numerous groups, most recently reviewed by Bushell et al. (2002).
the 0.01 wt.% flocculant solution through the inner pipe was 1.2, 2.4 and 3.6 L min− 1 for
With the FDA a video camera, coupled to selected optical lenses, is used to record the
300, 600 and 900 g t− 1, respectively (linear rates of 1.66, 3.32 and 4.98 m s− 1). For the
sedimentation characteristics of individual aggregates as they settle under the
influence of gravity within the analysis cell. The maximum horizontal and vertical
Table 1 dimensions, together with the settling rate, are measured for individual aggregates in
Flow velocities and calculated mean shear rates covering the range of flow rates used in the recorded images. The effective density can be calculated from the aggregate size/
the pipe reactor studies described in Section 3 settling velocity data using a modified version of Stokes' Law — this requires that the
aggregates are diluted sufficiently to ensure free settling. A statistically representative
Flow rate (L min− 1) Linear velocity (m s− 1) Mean shear rate (s− 1) Reynolds number number of aggregates (usually two to three hundred) are measured, covering a broad
size range (for more detail, see Farrow and Warren, 1993, and Owen et al., 2002).
16.6 mm ID pipe
Samples collected from the linear pipe reactor require a high degree of dilution
10 0.77 640 12,740
prior to FDA settling rate and density measurements. The sampling and manipulation of
20 1.54 1660 25,500
fragile aggregates will inevitably lead to some degree of rupture and densification — if
the impact of these effects are not minimised, comparisons may prove meaningless.
22.1 mm ID pipe
While the linear pipe reactor represents an ideal method for the preparation of
10 0.43 240 9500
flocculated suspensions under well-controlled conditions, obtaining a representative
15 0.65 423 11,040
sample is far from a trivial matter. To this end, a sampling port was located within the
20 0.87 632 19,200
flowcell directly opposite the FBRM detector (Fig. 1) ensuring that both the reaction
30 1.30 1100 28,700
time under shear and the corresponding chord length information were known.
A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99 93

Samples were carefully withdrawn into a short length of tubing (100 mm, 6.4 mm ID)
over about 2 s using a syringe to supply a slight vacuum. While the sampling process
typically took 2 s, the slurry was under much milder shear than experienced within the
pipe system (~ 27 s− 1 compared to 423 and 642 s− 1 for 15 and 20 L min− 1, respectively).
Each sample was transferred to the FDA dilution cell without delay, with the images of
free settling aggregates then recorded.

3. Results and discussion

3.1. Flocculant adsorption

3.1.1. Optimum flocculant dosage and residual flocculant


The concept of an optimum flocculant dosage for rapid aggregation is frequently
invoked in flocculation studies, invariably referring back to the seminal work of Smellie
and La Mer (1958), which suggested the optimal collision efficiency between particles
occurs for 50% surface coverage by polymer. More advanced modelling has refined this
estimate (see for example De Witt and Van de Ven, 1992; Moncho-Jordá et al., 2003),
but all such work has focused on the aggregation of sub-micron particles, for which
high dosages producing large fractional surface coverages may be expected.
Dosages of this magnitude have almost no relevance to most mineral systems,
where the majority of particles are N1 μm in size. At the dilute solids concentrations
encountered in hydrometallurgical thickeners, aggregate sizes become massive
(N 1 mm) at dosages that represent much less than 20% surface coverage. Table 2
shows the plateau adsorption isotherm values (expressed as dosages) obtained for three
common substrates on extended mixing, which equates to effective complete cover- Fig. 2. Percentage of unadsorbed flocculant remaining following pipe reactor
age of the surfaces. In contrast, the range of dosages required to give settling rates of flocculation of kaolin with a low molecular weight 20% anionic flocculant in 0.2 M
10–20 m h− 1 are at least ten-times lower. NaCl at a slurry flow rate of 10 L min− 1, pipe diameter 10.8 mm. Solid symbols: applied
When flocculant adsorption isotherms are considered, it can be seen that floc- flocculant concentration maintained at 0.01% and the flocculant flow rate increased to
culation takes place at dosages that are expected to represent close to quantitative give required dosage; open symbols: flocculant concentrations reduced to 0.033% and
adsorption under equilibrium conditions. However, total residence times within 0.067% to give dosages of 300 and 600 g t− 1 while maintaining a constant flocculant
feedwells are short, typically less than 30 s, and while flocculant adsorption at these linear flow velocity.
lower dosages should be rapid, the potential for significant levels of residual flocculant
reporting to thickener overflows is often a concern. This is especially so where the linear flow velocity at all dosages (open symbols in Fig. 2). Increasing the flocculant
overflows are feeds to filtration, solvent extraction or electrowinning circuits, for which velocity for the two lower dosages did indeed improve mixing and the adsorption
low levels of water soluble polymers can influence performance. achieved, such that the effect of different flocculant dosages was more in keeping with
expectations.
3.1.2. Adsorption during turbulent pipe flow
Flocculant adsorption kinetics onto a standard kaolin slurry was examined using 3.1.3. Implications for mixing of flocculants in feedwells/feedpipes
samples withdrawn from the pipe reactor. Studies of adsorption on real mineral slurries Poor mixing of flocculant throughout a slurry may also lead to non-homogenous
are limited by the sensitivity of techniques for residual polymer analysis in the presence distribution of the polymer on the solid particles (i.e. higher dosages in terms of g t− 1 on
of other potential interferences. The size exclusion chromatography (SEC) approach some particles than on others). If localised high concentrations persist for extended
described by Beazley (1985) and Hunt et al. (1988) provides robust quantification at reaction times, it becomes more likely that adsorption will progress to high levels on
sub-ppm levels, but this still required the application of dosages well in excess of what the solids contained in those areas. The effectiveness of flocculant distribution may
would normally be used for kaolin (≥ 300 g t− 1) to make analysis viable. therefore be worse than indicated on the basis of residual flocculant concentrations.
Over the reaction time range of interest (1 to 4 s), almost no adsorption was Based on the expected adsorption behaviour, residual flocculant should not report
detected for the anionic flocculant when it was diluted to 0.01% in pure water. This was to the overflow in most thickening applications. The observation of flocculant leaving a
undoubtedly a consequence of the high viscosity of anionic flocculant solutions in the feedwell implies:
absence of salts (Connelly et al., 1986), and is consistent with flow visualisation studies,
• Massive flocculant overdosing, which can occur under poor control in response to
where an Allura Red dye was added to flocculant solutions that were then mixed with
feed variations.
water in a transparent pipe (White et al., 2001). To minimise the impact of the solution
viscosity, all further adsorption studies were conducted with flocculants prepared and
diluted in 0.2 M NaCl. While the presence of salt may also reduce the magnitude of the
kaolin surface charge, it will not change the proportion of surface sites available for
adsorption by the amide functionalities of the flocculants.
Fig. 2 shows the percentage of unadsorbed flocculant remaining in solution as a
function of reaction time for three different dosages of 0.01% flocculant solutions. As
expected, the percentage unadsorbed decreased at longer reaction times, but was
higher for dosages in excess of 300 g t− 1. The most interesting observation was that for a
fixed flocculant concentration of 0.01%, the 900 g t− 1 results were not greatly dissimilar
to the 600 g t− 1. However, this may be in part attributed to the relative linear velocities
of the feed and flocculant streams. The flow visualisation studies in water by White et al.
(2001) identified that the extent of flocculant mixing increased as the ratio of flocculant
to water (feed) linear flow velocities was raised from 1.0 to 4.0 (feed velocity was fixed
at 0.44 m s− 1), reflecting greater turbulence. In the 0.01% flocculant solution results in
Fig. 2, the equivalent ratio at 900 g t− 1 was ~ 2.7, while at 300 g t− 1 it was only 0.9.
To test if this was indeed a factor affecting mixing and subsequent adsorption, the
flocculant concentration was reduced from 0.01% to 0.0033% and 0.0067% for the
dosages of 300 and 600 g t− 1, respectively, thereby maintaining a constant flocculant

Table 2
Comparison of the flocculant dosages required at pH 7 for effective settling with those
that represent plateau adsorption on extended mixing

Substrate d50 (µm) Dosage (g t− 1) required to give

Adsorption isotherm plateau Settling rates of 10–20 m h− 1


Fig. 3. Effect of flow rate (shear) on aggregate size as a function of reaction time for
Calcite 4.1 1260 30–60
kaolin (21 g L− 1) flocculated with a high molecular weight acrylamide/acrylate
Kaolin 6.9 700 50–70
copolymer (64 g t− 1) through a 22.14 mm ID pipe. Reaction times corresponding to
Hematite 4.0 650 25–35
sampling for FDA measurements are indicated.
94 A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99

Fig. 4. Effect of reaction time on FDA settling rate and density versus size relationships for kaolin (22 g L− 1) flocculated with a high molecular weight acrylamide/acrylate copolymer:
(a) and (b) 34 g t− 1, (c) and (d) 64 g t− 1.

• Poor mixing, where the feedwell hydrodynamics are such that feed and flocculant can Establishing that aggregate breakage was the dominant process was important to
exit without proper contact. This can occur in feedwells operating at feed rates below the development of a PB model for the flocculation process (Heath et al., 2006b), as it
their optimum range, or if the flocculant concentration is too high, leading to an meant that a constant fractal dimension can be assumed throughout. It is important to
excessive flocculant solution viscosity. note that this assumption only applies at the moderately low solids concentrations
• Flocculant may have been added at a point very low in the feedwell, leading to a very normally encountered in feedwell flocculation. In addition, low shear applied post-
short flocculant residence time. Such a strategy is used when there is a fear of
excessive aggregate rupture under shear, but should only be considered when the
feedwell hydrodynamics are well understood.

Significant residual flocculant concentration may also be expected in the case of


thickeners used for low solids clarification, i.e. where clarity is the main priority. Such
clarifiers are often operated at solids concentrations significantly below the optimum
for flocculation (see Section 3.4 below), and flocculant dosages are then elevated in an
attempt to compensate. The effect of poor feedwell hydrodynamics on the passage of
flocculant can therefore be heightened in the absence of sufficient particles to provide
sites for adsorption.

3.2. Aggregate densification or breakage?

The profiles produced from the linear pipe reactor typically display a rapid increase
in size up to a peak value, beyond which the size decreases at a slower rate with time.
The relative extent of this decrease will vary depending upon the flow rate (shear rate)
as shown in Fig. 3, as well as the dosage, solids concentration, the substrate mineralogy
and the chosen flocculant. While FBRM detection can show a clear change in size, it
provides no direct evidence as to whether this change is due to aggregate densification
or breakage under the extended shear.
To examine the effect of extended reaction time on aggregate density, samples of
flocculated kaolin were withdrawn from the linear pipe reactor at times that
represented either the peak in aggregate size or a point well beyond the peak (these
times are indicated on the profiles shown in Fig. 3). Fig. 4 shows FDA results for such
samples obtained at two different flow rates (15 and 20 L min− 1), and in both cases
there was no evidence to support aggregate densification. In fact the results suggest
that the longer reaction times led to a slight reduction in aggregate density. Such a trend Fig. 5. Effect of flow rate (shear) on aggregate size and settling rate as a function of
may seem counter-intuitive, but may be consistent with phenomena discussed in reaction time for kaolin (21 g L− 1) flocculated with a high molecular weight acrylamide/
subsequent sections. acrylate copolymer (60 g t− 1) through a 22.14 mm ID pipe.
A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99 95

flocculation at higher solids concentrations (i.e. during raking within a settled bed) is
known to produce aggregate densification (Farrow et al., 2000).

3.3. Short-term activity of flocculant on an aggregate surface

3.3.1. Comparison of size and settling rate reaction profiles


A general premise of most flocculation studies is that a larger aggregate size will
result in a higher settling rate, on the proviso that the solids concentration is not so high
that the effective aggregate volume adversely influences hindered settling behaviour.
The well-established relationships between particle size and settling rate have
frequently been extended to aggregated systems, although in most cases the size
measurements have been conducted off-line or following batch flocculation processes.
Such measurements can provide an important indication of the aggregate size that gave
a particular settling rate, but by failing to properly consider aggregation kinetics, may
not necessarily give much insight into the actual flocculation conditions required to give
that result.
Fig. 5 compares the size and settling profiles as a function of reaction time for kaolin
flocculation at two different flow rates. Both size and settling rate profiles achieve peak
values at short reaction times, but at longer reaction times the settling rates diminish to
a much greater degree than would be anticipated on the basis of the change in
aggregate size. Similar observations have been made for pipe reactor flocculation of
calcite over a wide range of conditions (Heath et al., 2006c).
Hecker et al. (1999) demonstrated reasonable correlations between settling rates
and mean square-weighted chord lengths for kaolin flocculated in a continuous Couette
mixing device. However, the reaction times for solids and flocculant within the Couette Fig. 7. Effect of a reduction in agitation intensity on aggregate size for calcite flocculation
section under the strongest shear generally represented times beyond the peak in (dosage 7 g t− 1, slurry flow rate 10 L min− 1, solids concentration 20.5 g L− 1). Shaded
points represent a change of pipe diameter from 16.6 mm to 22.1 mm after the marked
intervals.

aggregate size for the expected reaction profiles. Indeed, continuous Couette
flocculation at the scale described does not provide access to short reaction times,
and the additional seconds taken from the base of the Couette section to isolation of a
settling column and measuring a settling rate is thought to allow for additional ag-
gregate growth.
This highlights that for any flocculation study, even for the case of continuous
flocculation in the linear pipe reactor with control of reaction times to a fraction of a
second, settling rates are by definition a batch measurement of the aggregation state. A
settling column can be isolated at chosen reaction time of (say) 3 s, but it may take at
least that long again for a mudline to form, while subsequent monitoring of the rate of
mudline fall may take at least an additional 5 to 60 s.
It was therefore proposed that the greater dependence upon reaction time
observed for settling rates relative to aggregate size is a consequence of additional
aggregate growth at short reaction times. This can be attributed to the presence of
active flocculant on the external surface of aggregates while still within the growth
phase, as shown in Fig. 6. Under reduced shear within a settling column, further
aggregation (post-aggregation) to larger sizes can take place. At reaction times beyond
the peak in aggregates size, the majority of this surface flocculant may lose its activity,
either through shear degradation or reconformation of the polymer tails closer to the
particle surface. Little or no post-aggregation will be observed.

Fig. 8. Effect of a reduction in agitation intensity on aggregate size for kaolin flocculation
Fig. 6. Schematic representation of why additional aggregation may be favoured after (dosage 67 g t− 1, slurry flow rate 20 L min− 1, solids concentration 20.5 g L− 1). Shaded
shorter reaction times. points represent a change to a 22.1 mm diameter pipe after the marked intervals.
96 A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99

significantly shorter reaction time but with a smaller peak magnitude, and a much
greater degree of aggregate rupture at longer reaction times.
Increasing the internal pipe diameter after a reaction time of 1.7 s (while still in the
growth region) led to a marginal increase in the maximum aggregate size relative to
that achieved in the narrow pipe alone, but the reaction time required to achieve this
maximum was significantly higher. While the maximum did not approach that
obtained in the 22.1 mm pipe alone, it is important to note that the extent of aggregate
breakage under the milder shear conditions is much reduced compared to the full
profile in the narrow pipe. However, even under this reduced shear, the very large
aggregates formed in the 22.1 mm pipe are still susceptible to breakage. Fig. 7 shows
that both the large pipe profile and that obtained on reducing shear at 1.7 s appear to be
approaching a similar equilibrium size at extended reaction times. In contrast,
increasing the internal pipe diameter at reaction times beyond the peak gave no
evidence of any subsequent aggregation.
The flocculation of a kaolin slurry was examined in a similar manner but under
stronger shear conditions, and in this case the impact of post-aggregation was more
distinct. At a flow rate of 20 L min− 1, a 20.5 g L− 1 kaolin slurry flocculated at a dosage of
67 g t− 1 in the 16.6 and 22.1 mm pipes alone (mean shear rates of 1660 and 632 s− 1)
gave maximum mean aggregate sizes of 140 and 240 μm, respectively (Fig. 8). The
differential between these profiles was much larger than that seen for calcite (Fig. 7);
this was not an indication of any intrinsic difference in the response to shear for the two
substrates, but rather a reflection of a number of competing factors, including the solids
Fig. 9. Effect of solids concentration on aggregate size for kaolin flocculation (flocculant volume fraction, the aggregate fractal dimension and the size of aggregate formed.
dosage 71 g t− 1, slurry flow rate 18 L min− 1). Indeed the flocculation of both substrates can be described over a wide range of
conditions by the same generic population balance model for aggregate growth and
3.3.2. Effect of reduced shear breakage in turbulent pipe flow. For a more detailed discussion of this model, see Heath
In order to obtain more definitive evidence of post-aggregation, a series of pipe et al. (2006b).
reactor experiments was conducted where flocculation was initiated in a region of high In Fig. 8 the internal pipe diameter was increased from 16.6 to 22.1 mm after
shear (narrow pipe) before passing into lower shear (wide pipe), maintaining turbulent reaction times of 1.2 s (growth), 2.5 s (peak) and 7.3 s (breakage dominated). Reducing
conditions throughout. The duration under high shear was varied to represent times on the shear while still in the growth region for kaolin led to a larger aggregate size
the full reaction profiles that occur prior to the peak in aggregate size (i.e. while still (~ 155 μm, compared to 140 μm for the narrow pipe alone) and a much lower degree of
within the growth phase), near the peak and well after the peak. breakage at longer reaction times. Reducing shear at the peak gave a slightly smaller
Fig. 7 shows data for calcite flocculation at a flow rate of 10 L min− 1, with full enhancement in size and more breakage beyond the peak, although still not
profiles presented for flocculation in either a 16.6 or 22.1 mm ID pipe alone (mean shear approaching that seen for the 16.6 mm pipe. In both cases the enhancement in size
rates of 640 or 240 s− 1, respectively), together with profiles obtained after increasing under reduced shear was still well below the maximum aggregate size of 240 μm
from the narrow to the wide pipe at four different reaction times. The full profiles for achieved for flocculation in the 22.1 mm pipe alone.
each pipe diameter display quite distinct behaviour — the higher shear in the smaller When the shear was reduced at a reaction time well into the breakage dominated
pipe gives a faster initial rate of aggregate growth, achieving a peak aggregate size at a region, almost no increase in aggregate size was observed (Fig. 8). This was consistent

Fig. 10. Effect of reaction time on (a), (c) unweighted and (b), (d) length square-weighted chord length distributions for kaolin flocculation at two different solids concentrations
(flocculant dosage 71 g t− 1, slurry flow rate 18 L min− 1).
A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99 97

with expectations that at such times, flocculant on the external surfaces of aggregates is
effectively deactivated and unable to participate in bridge-forming collisions.
These results highlight the critical distinction between pure coagulation and
bridging flocculation processes. For coagulation, where aggregation is in effect induced
by charge neutralisation reactions, a reduction in shear will usually lead to aggregate
growth, with any aggregate fragments initially formed under the higher shear readily
able to recombine and establish a new equilibrium.
In contrast, aggregation by polymer-bridging does not lead to an equilibrium over
the timescales experienced within a feedwell or most feedpipes — breakage proceeds
through the rupture of the bridging polymers, and in the absence of additional floc-
culant, breakage to smaller aggregate fragments is irreversible. Fig. 8 demonstrates that
applying strong shear for even a very short period can significantly reduce the extent of
aggregation achieved, which cannot be recovered by then applying a lower level of
shear.
In practical terms, thickener throughput can be substantially reduced by excessive
residence times within a feedwell under higher shear. The incorporation of a PB for
flocculation within a CFD feedwell model allows design options or flocculant addition
points to be identified that produce the optimal residence time and flocculation
response (Nguyen et al., 2006). Indeed, judicious feedwell optimisation may favour
post-aggregation, providing flocculation conditions where additional aggregate growth
can take place under the milder shear experienced immediately below the feedwell
exit.

3.4. Solids concentration

Solids concentration has long been known as a crucial factor within aggregation
processes (e.g. La Mer and Healy, 1963; Moss, 1978). For coagulation at low solids
concentrations, an increase in concentration means a greater probability of effective
collisions, and therefore an expectation of larger sizes (e.g. Lach-Hab et al., 1996).
However, most polymer-bridging flocculation processes take place at higher concen-
trations, where viscosity effects can become significant, limiting and in some cases
greatly reducing the size of aggregates that are formed. In developing a PB model from
linear pipe reactor kinetic results for calcite flocculation, Heath et al. (2006b) were able
to predict an optimum solids concentration (i.e. giving the largest aggregate size), but
did not have experimental results at sufficiently low concentrations to confirm this
prediction.
The effect of solids concentration on kaolin flocculation at a fixed flocculant dosage
and pipe flow rate is shown in Fig. 9. In this case, the experimental range of solids
concentrations does show a clear optimum, with the largest mean square-weighted
chord length (300 µm) attained at 9.2 g L− 1. Above this concentration the measured size
diminished rapidly, and was less that 150 µm at 36.8 g L− 1. This sharp change reflects the
very porous nature of the aggregates formed with the plate-like kaolin particles, leading
to a very high effective volume fraction at quite low concentrations.
A greater insight into this effect can be obtained by examining the full FBRM chord
length distributions at selected reaction times for two different solids concentrations.
At 9.2 g L− 1 the unweighted chord length distributions dropped to very low counts
after only 4.3 s, with incremental decreases to the plateau value after 10 s (Fig. 10a). At
27.5 g L− 1 the unweighted distributions (Fig. 10c) were shifted to the left for the shorter
reaction times (i.e. 4.3 and 5.6 s), indicating a substantial proportion of primary par-
ticles remaining within the slurry. Reasonably efficient flocculation was only achieved
after 10 s, and even then the counts remained high. The length square-weighted
distributions (Fig. 10b and d) clearly demonstrate the formation of larger aggregates at
the lower solids concentration for the same flocculant dosage (71 g t− 1).
The expectation from flocculation modelling is that a higher solids concentration
should provide a higher frequency of collisions, therefore increasing the aggregation
rate (Heath et al., 2006a). However, this assumes that the flocculant is well mixed
throughout the slurry, giving a uniform distribution of adsorbed polymer chains across
all particles, which is rarely the case. As shown in Section 3.1, flocculant adsorption is
very rapid, and with the added flocculant volumes invariably being small relative to that
of the bulk slurry, a proportion of the solids is unlikely to make direct contact with
unadsorbed polymer chains. This proportion will rise for higher concentration slurries,
and indeed the large number of residual counts in the unweighted chord length
distributions at long reaction times for flocculation at 27.5 g L− 1 (Fig. 10c) are consistent
with poor flocculant/slurry mixing.

3.5. Multi-stage addition of flocculant

The effect of flocculant dilution on aggregation is well known, with benefit ob-
tained through more uniform dispersion throughout the slurry (Moss, 1978). Sub-
stantial enhancement of aggregation may also be achieved from multi-stage (split)
addition of the same flocculant, although the benefit results from more than just
improved dispersion. Moss (1978) observed that splitting the dose into two or more
streams and applying them at different points gave maximum fines capture and larger
aggregates. Peng and Williams (1993) used FBRM in a stirred beaker to clearly dem-
onstrate the enhancement of silica aggregate size from adding the flocculant in five Fig. 11. Effect of split flocculant dosing on mean square-weighted chord length and
portions at regular intervals rather than a single dose, while Glasgow and Liu (1995) hindered settling rate for kaolin flocculation, with total dose split at ratios of (a) 30:70,
used light obscuration in a similar study of kaolin flocculation. (b) 50:50 and (c) 70:30 (total dosage 66 g t− 1, slurry flow rate 20 L min− 1, solids
The above examples all deal with flocculation in beakers, and as a consequence the concentration 20.7 g L− 1).
reaction times were very long compared to what might be seen in a feedwell, as was the
duration between staged additions of flocculant (at least 40 s). Much shorter reaction
98 A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99

times were achieved by Suharyono and Hogg (1996), who examined kaolin flocculation The decline in the corresponding settling rates at longer reaction
through a series of static mixer elements. They observed a positive effect on settling rate
times was much greater than would be expected on the basis of the
from staged addition of flocculant, although their experimental set-up almost certainly
involved a significant delay between flocculation and cylinder settling tests. size measurements alone. Experiments in which turbulent floccula-
The pipe reactor is ideally suited to looking at multi-stage addition, as it provides tion was initiated in a region of high shear (narrow pipe) before then
excellent control over the reaction time and the possibility of looking at much shorter passing into lower shear (wide pipe) strongly indicated that surface
durations between additions. Fig. 11 compares the reaction profile obtained under adsorbed flocculant remained active at short reaction times, allowing
standard conditions (20.7 g L− 1 kaolin, 20 L min− 1) for flocculant added in a single dose
(giving 66 g t− 1) to profiles obtained for different splits of applied flocculant, with the
additional aggregate growth under subsequent reduced shear. This
second dose added either at 3.3 s (i.e. still in the growth phase), 7.5 s (close to the peak surface flocculant becomes deactivated at longer reaction times under
in aggregate size) and 14.8 s (well beyond the peak). It can be seen that while the the higher shear (either by reconformation or shear degradation),
addition of the second dose early in the aggregate growth process gave a significant thereby preventing or greatly reducing additional growth. Optimising
enhancement of aggregate size, addition at times corresponding to either the peak in
the shear conditions and residence times during flocculation within
size or well after were both more effective.
The benefit of split dosing is a consequence of the fact that the initial dose forms thickener feedwells therefore has considerable potential to improve
aggregates that are small but still consist of many particles, and hence the second dose settling performance.
is only exposed to the external surface area of these aggregates, with particles inside the
aggregates not accessible to the polymer. This effectively produces a higher flocculant Acknowledgements
adsorption density per unit area, and significantly reduces the dosage required to
achieve rapidly settling aggregates. The concept of excluded surface area has been used
to explain the reduction in plateau adsorption densities with increasing solids Chad Loan and Alton Grabsch are acknowledged for the measure-
concentrations in aggregating systems (Chaplain et al., 1995) and the contribution of ment of adsorption isotherms on kaolin and hematite, respectively.
aggregation to effective surface coverage has been theoretically modelled (Hsu and Lin, This work was conducted as part of the AMIRA project P266D:
1998; Kislenko, 2000).
“Improving Thickener Technology”. The authors acknowledge the
The critical observation from this work was that the enhancement in aggregate size
rose as the duration between doses was lengthened. The smaller increase seen in Fig. 11 financial support given to this research from the following project
from adding the second dose while still in the initial growth phase (at 3.3 s in this case) sponsors: Albian Sands Energy, Alcan International, Alcoa World
was consistent with results obtained at other dosages and with other mineral sub- Alumina, Anglo Gold, Anglo Platinum, BHP Billiton, Cable Sands, Ciba
strates. This suggested that the aggregates may not have taken a well-defined form at an Specialty Chemicals, Cytec Australia Holdings, De Beers Consolidated
early stage in the flocculation process, such that the second dose is not greatly restricted
Mines, EIMCO Process Equipment, GL&V/Dorr Oliver, Glencore AG,
from penetrating to the internal surfaces. Alternatively, it may simply reflect the smaller
aggregate size (and hence larger available external surface area) at the time of addition Iluka Resources Limited, Kumba Resources, Metso Minerals, Mt Isa
for the second dose. Mines, ONDEO Nalco, Pasminco, Pechiney Aluminium, Queensland
Varying the volumetric ratios of the split doses from 30:70 to 50:50 gave very Alumina Limited, Queensland Nickel (QNI), Rio Tinto, Tiwest, True
similar aggregate size results (Fig. 11a and b, respectively). For a 70:30 split (Fig. 11c) the
North Energy, WMC Resources Ltd and Worsley Alumina Pty Ltd.
benefit from sequential dosing was clearly reduced, although still remained significant.
These results suggest that split dosing is most effective when the greater proportion Support was also received under the Australian Government's Co-
of the flocculant is applied at the second dosing point. While this leads to a smaller operative Research Centre Program, through the Parker Centre for
aggregate size after the initial dose, provided that fines capture is largely complete (i.e. Integrated Hydrometallurgy Solutions. This support is gratefully
that the applied dosage itself is sufficiently high), then the effective surface area will acknowledged.
have already been reduced substantially. Having more flocculant to apply to this lower
effective surface area, thereby achieving a greater fractional coverage, may more than
compensate for the smaller initial extent of aggregation. References
Fig. 11 shows a clear, large benefit from split dosing of the flocculant, which is
generally consistent with plant experience. However, instances are known where such Beazley, P.M., 1985. Quantitative determination of partially hydrolysed polyacrylamide
an approach has been less than convincing, with little or no performance enhancement polymers in oil field production water. Anal. Chem. 57, 2098–2101.
observed. This highlights the importance of understanding the prevailing hydrody- Bolto, B., Gregory, J., 2007. Organic polyelectrolytes in water treatment. Wat. Res.,
namic conditions during flocculation. In Fig. 11 (100% added at 0 s) the relatively mild Organic polyelectrolytes in water treatment, vol. 41, pp. 2301–2324.
mixing allowed the formation of aggregates over 200 μm in size, with a minor degree of Bushell, G.C., Yan, Y.D., Woodfield, D., Raper, J., Amal, R., 2002. On techniques for the
rupture at longer reaction times. Compare this to the situation in Fig. 8 for a similar measurement of the mass fractal dimension of aggregates. Adv. Colloid Interface
flocculant dosage but stronger shear conditions, for which the aggregate size reaches Sci. 95, 1–50.
only 140 μm and then undergoes substantial rupture. In this situation, split dosing may Byrne, E., Fitzpatrick, J.J., 2002. Investigation of how agitation during precipitation
still increase the aggregate size, but not necessarily to a level approaching that achieved and subsequent processing affects the particle size distribution and separation of
α-lactalbumin enriched whey protein precipitates. Biochem. Eng. J. 10, 17–25.
under milder shear.
Carissimi, E., Rubio, J., 2005. The flocs generator reactor — FGR: a new basis for floc-
The solids concentration during flocculation will similarly influence the potential
culation and solid–liquid separation. Int. J. Miner. Process. 75, 237–247.
extent of performance enhancement. The results in Fig. 11 were obtained at 20.7 g L− 1, Carissimi, E., Miller, J.D., Rubio, J., 2007. Characterization of the high kinetic energy
which is a moderate kaolin concentration that still allows the formation of very large dissipation of the Flocs Generator Reactor (FGR). Int. J. Miner. Process. 85, 41–49.
aggregates. At concentrations above 30 g L− 1, the effective solids volume fraction will Chaplain, V., Janex, M.L., Lafuma, F., Graillat, C., Audebert, R., 1995. Coupling between
limit the size of aggregates (and the hindered settling rates) that can be achieved. Under polymer adsorption and colloid particle aggregation. Colloid Polym. Sci. 273, 984–993.
such conditions, the benefit from split dosing may be limited, possibly even eliminated. Connelly, L.J., Owen, D.O., Richardson, P.F., 1986. Synthetic flocculant technology in the
Bayer process. Light Metals, pp. 61–68.
Delichatsios, M.A., Probstein, R.F., 1975. Coagulation in turbulent flow: theory and
4. Conclusions
experiment. J. Colloid Interface Sci. 51, 394–405.
De Witt, J.A., Van de Ven, T.G.M., 1992. The effect of neutral polymers and electrolyte on
Turbulent flow in a linear pipe reactor that utilises continuous the stability of aqueous polystyrene latex. Adv. Colloid Interface Sci. 42, 41–64.
FBRM monitoring of the resultant aggregation state has been shown to Farrow, J.B., Johnston, R.R.M., Simic, K., Swift, J.D., 2000. Consolidation and aggregate
densification during gravity thickening. Chem. Eng. J. 80, 141–148.
be a powerful tool for the study of polymer-bridging flocculation Farrow, J.B., Swift, J.D., 1996. A new procedure for assessing the performance of
in mineral systems. This arrangement enables far better control over flocculants. Int. J. Miner. Process. 46, 263–275.
the applied shear rate than can be achieved from the majority of Farrow, J.B., Warren, L.J., 1993. Measurement of the size of aggregates in suspension. In:
Dobiás, B. (Ed.), Coagulation and Flocculation — Theory and Applications. Marcel
laboratory-based approaches for inducing flocculation. Critically, Dekker, New York, pp. 391–426.
reaction times for slurry and flocculant under applied shear during Fawell, P., Richmond, W., Jones, L., Collisson, M., 1997. Focused beam reflectance mea-
continuous flocculation in a linear pipe reactor are known to within a surement in the study of mineral suspensions. Chem. Aust. 64 (2), 4–6.
Fellows, C.M., Doherty, W.O.S., 2006. Insights into bridging flocculation. Macromol.
fraction of a second, providing the ability to examine in detail aspects
Symp. 231, 1–10.
of the aggregate growth and breakage processes that are not Glasgow, L.A., Liu, S.X., 1995. Effects of dosing regimen and agitation profile upon floc
accessible through batch flocculation. characteristics. Chem. Eng. Commun. 132, 223–237.
Gregory, J., 1989. Fundamentals of flocculation. CRC Crit. Rev. Environ Control 19, 185–230.
Over the mean shear rate range studied (240–1660 s− 1), reaction
Heath, A.R., Bahri, P.A., Fawell, P.D., Farrow, J.B., 2006a. Aggregation of calcite by
times of only 2 to 10 s were required to achieve peak aggregate sizes, polymeric flocculant — experimental observations in turbulent pipe flow. AIChE J.
with longer times favouring breakage over aggregate densification. 52, 1284–1293.
A.T. Owen et al. / International Journal of Mineral Processing 87 (2008) 90–99 99

Heath, A.R., Bahri, P.A., Fawell, P.D., Farrow, J.B., 2006b. Aggregation of calcite by Moncho-Jordá, A., Odriozola, G., Tirado-Miranda, M., Schmitt, A., Hidalgo-Alvarez, R.,
polymeric flocculant — population balance model. AIChE J. 52, 1641–1653. 2003. Modelling the aggregation of partially covered particles: theory and
Heath, A.R., Bahri, P.A., Fawell, P.D., Farrow, J.B., 2006c. Aggregation of calcite by simulation. Phys. Rev. E. 68, 0114404.
polymeric flocculant — relating the aggregate size to the settling rate. AIChE J. 52, Moss, N., 1978. Theory of flocculation. Mine Quarry J. 7 (5), 57–61.
1987–1994. Nguyen, T., Heath, A., Witt, P., 2006. Population balance — CFD modelling of fluid flows,
Hecker, R., Kirwan, L., Fawell, P.D., Farrow, J.B., Swift, J.D., Jefferson, A., 1999. Focused solids distribution and flocculation in thickener feedwells. In: Witt, P.J., Schwarz,
beam reflectance measurement for the continuous assessment of flocculant M.P. (Eds.), Proceedings of the Fifth International Conference on CFD in the Process
performance. In: Laskowski, J.S. (Ed.), Polymers in Mineral Processing. Canad. Industries. Melbourne, Australia, 13–15 December.
Inst. Mining, Metall. Petrol. Montreal, Quebec, pp. 91–105. Owen, A.T., Fawell, P.D., Swift, J.D., Farrow, J.B., 2002. The impact of polyacrylamide
Henderson, J.M., Wheatley, A.D., 1987. Factors affecting the efficient flocculation of flocculant solution age on flocculation performance. Int. J. Miner. Process. 67, 123–144.
tailings by polyacrylamides. Coal Prep. 4, 1–49. Peng, S.J., Williams, R.A., 1993. Control and optimisation of mineral flocculation and
Hocking, M.B., Klimchuk, K.A., Lowen, S., 1999. Polymeric flocculants and flocculation. transport using on-line particle size analysis. Miner. Eng. 6, 133–153.
J. Macromol. Sci. Rev. Macromol. Chem. Phys. C39, 177–203. Peng, S.J., Williams, R.A., 1994. Direct measurement of floc breakage in flowing
Hsu, J.-P., Lin, D.-P., 1998. Transient behaviour of polymer-induced flocculation: effect of suspensions. J. Colloid Interface Sci. 166, 321–332.
dosing mechanism. J. Chin. Inst. Chem. Eng. 29 (3), 219–228. Smellie, R.H., La Mer, V.K., 1958. Flocculation, subsidence and filtration of phosphate
Hunt, J.A., Young, T.S., Green, D.W., Willhite, G.P., 1988. Size-exclusion chromatography slimes VI. A quantitative theory of filtration of flocculated suspensions. J. Colloid Sci.
in the measurement of concentration and molecular weight of some EOR polymers. 23, 589–599.
SPE Reservoir Eng. 3, 835–841. Suharyono, H., Hogg, R., 1996. Flocculation in flow through pipes and in-line mixers.
Kislenko, V.N., 2000. Mathematical model of polymer adsorption accompanied by Miner. Metall. Process. 13, 93–97.
flocculation. J. Colloid Interface Sci. 226, 246–251. Swift, J.D., Simic, K., Johnston, R.R.M., Fawell, P.D., Farrow, J.B., 2004. A study of the
Klute, R., Amirtharajah, A., 1991. Particle destabilization and flocculation reactions polymer flocculation reaction in a linear pipe with a focused beam reflectance
in turbulent pipe flow. In: Amirtharajah, A., et al. (Ed.), Mixing in Coagulation measurement probe. Int. J. Miner. Process. 73, 103–118.
and Flocculation. American Water Works Association Research Foundation, USA, White, R.B., Simic, K., Strode, P.R., 2001. The combined use of flow visualisation,
pp. 217–255. electrical resistance tomography and computational fluid dynamics modelling to
Lach-Hab, M., Gonzalez, A.E., Blaisten-Barojas, E., 1996. Concentration dependence of study mixing in a pipe. 2nd World Congress on Industrial Process Tomography,
structural and dynamic quantities in colloidal aggregation: computer simulations. Hannover, Germany, 29–31 Aug.
Phys. Rev. E. 54, 5456–5462. Zumaeta, N., Byrne, E., Fitzpatrick, J.J., 2006. Predicting precipitate particle breakage in a
La Mer, V.K., Healy, T.W., 1963. Adsorption–flocculation reactions of macromolecules at pipeline: effect of agitation intensity during precipitate formation. Chem. Eng. Sci.
the solid–liquid interface. Rev. Pure Appl. Chem. 13, 112–133. 61, 7991–8003.

Potrebbero piacerti anche