J.R. MACDONALD
Century Autoflote Pty. Ltd. (formerly of Catoleum Pty. Ltd.)
ABSTRACT
A quantitative semi-empirical
model of sedimentation was presented at the Second Australian Coal Preparation Conference.
This model has been used as the basis for a complete on-line automatic control system
for thickener circuits called AUTOFLOC.
The AUTOFLOC system has been successfully trialed
at three Hunter Valley Coal Preparation Plants using Hi-Rate thickeners.
The
AUTOFLOC system provides for control of flocculant dose rates, underflow
pumping, thickener rake torque, thickener bed level and cationic polymer dose
rates in those instances where poor thickener overflow clarity requires the use
of cationics. These parameters are controlled in the following way:
i)
Flocculant dose rates are varied on a feed forward
basis so as to achieve a constant settling rate, in accordance with the
sedimentation model; provision exists for feedback adjustment in response to
the measurement of bed level.
ii) Underflow pumping
is varied on a feed forward basis so as to ensure that solids are removed from
the thickener at the same rate at which they enter, thereby preserving
thickener mass balance; provision exists for feedback adjustment in response to
the measurement of rake torque and/or bed level.
iii) Thickener rake torque
and bed level are normally maintained within acceptable limits by ensuring that
both constant settling rates and thickener mass balance are preserved. However
when this is not the case, provision exists for feedback adjustment to the
underflow pumping rate so as to control both rake torque and bed level; in
addition, provision exists for feedback adjustment to flocculant dose rates so
as to control bed level.
iv)
Cationic dose rates, when required, are
varied on a feedback basis so as to control thickener overflow clarity.
The mathematical techniques used in developing the sedimentation
model have been modified and extended to provide the basis for a complete on-line
flotation control system which is currently bring trialled.
INTRODUCTION
Because of the importance of particulate
suspensions in industrial processes, the properties of sedimentary slurries
have been widely investigated. The more significant references are listed here (1-9).
However, there does not appear to be any
reference in the literature to the quantitative effect of variations in the
particle size distribution or in flocculant dose rate. Consequently no model existed
which could be used as a sound theoretical basis for a feed forward flocculant control
system.
The author of the present paper conceived a
quantitative semi-empirical model of sedimentation which described the settling
velocity (V) of a slurry in terms of only three variables, namely flocculant
dose rate (D), slurry solids concentration (C) and a particle size distribution
parameter (S). The model was developed in terms of these parameters because it
was believed that such a formulation would provide the basis for a simple but
reliable, site-specific, automatic feed forward flocculant control system. This
model was presented at the Second Australian Coal Preparation Conference (10).
This paper reports that a simple, reliable
automatic feed forward flocculation control system has been developed from the
above model. This control system is currently being marketed by Catoleum under
the name AUTOFLOC. The flocculant control system has been successfully tried and
proven at three Hunter Valley Coal Preparation Plants using Hi-Rate thickeners
- Warkworth, Hunter Valley and Wambo.
The AUTOFLOC control system has been
extended to achieve complete control of thickener operation -by including
control of thickener underflow pumping, thickener rake torque and bed level.
This extended control system has been
proven under actual operating conditions at Hunter Valley Coal Preparation Plant.
The control system has been extended
further to provide feedback control of cationic polymer dose rates in those
instances where poor thickener overflow clarity requires the use of cationics.
This system has not yet been tested under plant conditions, but difficulties
are not anticipated.
The present paper describes certain aspects
of the total control system, together with other results of interest that have
been developed from the paper presented at the Second Australian Coal Preparation
Conference (10).
OUTLINE OF THE CONTROL SYSTEM
This section outlines the philosophy, the
features and the instrumentation requirements of the AUTOFLOC control system.
Other sections deal with the mathematical
basis for certain aspects of the control system, practical aspects of the
control system and other areas of interest.
Control System Philosophy
Because of the low residence times and fast
response, Hi-Rate thickeners are much more difficult to control than
conventional thickeners. Consequently the following comments are restricted to Hi-Rate
thickeners on the grounds that the corresponding problems in conventional thickeners
are much easier to solve.
The
basic assumption of the control system is that by
(i) Achieving constant
settling rates at all times (which means that the variation in floc formation has
been minimised)
(ii) Removing solids from the
thickener at the same rate that they enter
Conditions within the thickener will be kept close to steady-state,
thereby minimising or totally eliminating problems with rake torque and/or bed level.
Actual
performance at Hunter Valley Coal Preparation Plant indicated that
(i) Problems with rake
torque were completely eliminated.
(ii) Bed level fluctuations were
minimised but not eliminated.
It should be noted, however, that the gap
to the inner ring (where bed level fluctuations were greatest) was larger than
design specifications. Appropriate modifications have been carried out subsequently.
This may eliminate bed level fluctuations. It is important to note that
satisfactory performance of the thickener circuit was achieved which enabled
production rates above the design figure to be routinely achieved.
Features of the Control System
Flocculant Control: Flocculant dose
rates are controlled automatically on a feed forward basis, using the
semi-empiric sedimentation model. Provision is made for feedback adjustment to the
calculated dose rates in the event that bed level starts to get out of control,
although the primary means of controlling bed level is by adjusting the rate of
underflow pumping.
Thickener Control; The rate of underflow pumping can be altered so as to provide
control of
i) Mass balance in
the thickener; that is, remove solids via the underflow at the same rate at
which they enter via the feed
ii) Thickener rake torque
iii) Thickener bed level.
Mass
balance can be controlled automatically on a feed forward basis provided that
the following information is available
i) the flow rate
and solids content of the thickener feed,
ii) the density of the
solids in the underflow and the density of the underflow slurry,
iii) the underflow pump characteristics;
that is, flowered versus pump speed at various slurry densities.
In most instances, maintaining both mass
balance and const settling rates ensures that rake torque and bed level and
control within acceptable limits. However, where required, provision exists for
feedback adjustment to the calculated underflow pump speed order to control rake
torque and/or bed level.
Control of Overflow Clarity:
In some instances, particularly in Hunter Valley region, adequate overflow clarity
cannot be achieved with the use of flocculants alone. In such cases cationic polymers
or inorganic coagulants are necessary to improve overflow clarity
The control; software has been extended to
provide automatic control of thickener overflow clarity by adjusting cationic
dosing rates on a feedback basis. Although this has not yet been tested under plant
conditions, problems-are not anticipated.
Microcomputer Facilities:
Microcomputer flexibility enables the 'mathematical sophistication of feed
forward calculations to be carried out. It also allows feedback control algorithms
to be undertaken which are more effective than the usual PID feedback methods.
Plant Performance
The AUTOFLOC flocculant control system has been
successfully proven at Warkworth, Hunter Valley and Wambo Coal Preparation Plant.
The thickener control system has been successfully
proven at Hunter Valley Coal Preparation Plant. Wambo intends to use the thickener
control system when the instrumentation pre-requisites have been completed.
The author is unaware of any case at any of
the three sites where flocculant or thickener problems have been due to the inability
of the control system to control.
It should be noted that at Hunter Valley
Coal Preparation Plant rake torque was, at all times, maintained within
acceptable limits by ensuring that the underflow pumps maintained mass balance.
That is as expected, feedback adjustment was never required. On the other hand,
feedback adjustment for bed level was required at times. However it was noted
subsequently that the gap to the inner ring was greater than the design
specifications. It is possible that appropriate modifications will make feedback
adjustment unnecessary.
Instrumentation Requirements
Flocculant/Cationic Control: Both systems require
i) Light transmission
probes which generate 4-20 mA input signals
ii) Variable speed pumps
which can be controlled via 4-20 mA output signals.
Thickener Control: It is preferable
that variable speed underflow pumps be used, particularly with Hi-Rate thickeners,
but fixed speed pumps can be controlled by varying the length of time that the
pumps operate.
Control
of thickener mass balance requires
i) Variable
speed underflow pumps which can be controlled via 4-20 mA output signals, or
ii) Fixed speed
underflow pumps which can be controlled via digital output signals.
Control of rake torque requires a
measurement of rake torque which can provide a 4-20 mA input signal.
Control
of bed level requires a measurement of bed level which can provide a 4-20 mA
input signal. For example, this may be provided by
i) a Gem sensor
ii) a mechanically
movable light sensor
iii) taps at the side of the
thickener, together with a light sensor.
Benefits of the Control System
Since there are no moving parts, maintenance
problems and costs are low,
The software package communicates in
user-friendly language, and can be easily extended as newer advances are made.
In
addition, the control system ensures that
i) flocculant/cationic
usage is optimised
ii) Thickener
operating objectives are constantly ach eved e.g. rake torque, bed level and
overflow clarity
iii) Plant downtime due to
rake torque problems is eliminated
iv) Satisfactory overflow
clarity is achieved, which optimises the performance of other process units e.g.
magnetite recovery and, particularly, flotation performance
v) Operator
involvement is minimised, which provides time for operators to optimise the
performance of other unit processes.
The reduction in plant downtime can have
significant cost benefits. For example, increased plant availability of 10
hours per annum for a plant producing 1000 t/h at a value of $50/tonne results in
increased production of $500,000 per annum.
3. THE MATHEMATICAL BASIS OF THE
CONTROL SYSTEM
There
are four major components of the total control system
i) Flocculant
control
ii) Cationic/overflow
clarity control
iii) Thickener mass balance
control
iv) Rake torque, bed level
control.
As outlined in the previous section,
flocculant dosing is varied on a feed forward basis so as to achieve a constant
settling rate (or other suitable operating objective) while underflow pumping
is varied on a feed forward basis so as to achieve thickener mass balance. In most
instances, ensuring a constant settling rate and thickener mass balance will
control rake torque and bed level within acceptable limits. On those occasions
when this is not so, provision for feedback adjustment to the rate of underflow
pumping exists. Cationic dosing is varied on a feedback basis so as to ensure
that the desired thickener overflow clarity is maintained.
For reasons of space, this section will
deal mainly with the flocculant control system, but will deal briefly with the
cationic control system also.
The Basis for the AUTOFLOC
Feedforward
Flocculant Control System
It
is generally agreed that the settling velocity of particulate slurries is a function
of
i) The properties
of the material being settled
ii) The properties of
the process water
ill) the properties of the flocculant
used
iv) The fluid dynamics of
the slurry in the specific piece of equipment in which settling occurs.
However, in any given plant, it would be
unusual for some of the above variables to vary significantly over the period
of a day. For example, a given flocculant would be used, the process water chemistry
should be approximately constant (if not, the problem should be examined
thoroughly) and, to a lesser extent, the process hydrodynamics and, to an even
lesser extent, the surface chemistry of the slurry particles should be reasonably
constant.
This suggests that an adequate
site-dependent description of settling can be given in terms of the slurry
solids content, C the particle size distribution parameter, 5, and the
flocculant dose rate in g/t, D.
It has been shown (10) that a
semi-empirical quantitative model relating these variables has the form
(1)
where a0, a1, a2, a3
and n are empirically determined, site-specific constants.
For control purposes we wish to express the
flocculant dose rate in terms of the other parameters.
(2)
At any particular plant, the settling rate objective,
V, will be a constant. Thus we have
Dv = exp (a3 S) [a4 Cn - a5]
(3)
where Dv
= dose rate in g/t required to achieve the desired
settling
velocity, V
a4 = (V+a0)/a2
a5 = a1/a2
Under actual plant conditions, it is easier
to work with dose rates expressed in units of ppm rather than g/t. This is
because plant dose rate in ppm can be calculated from the flocculant dose rate
in l/m, whereas a dose rate in g/t requires that, in addition the slurry solids
content be known.
Since D
= d/C
where d =
flocculant dose rate in ppm
C
= fractional solids content (not % solids)
we have
dv = exp(a3 S)[a4 Cn+1 - a5C]
(4)
where dv =
dose rate in ppm required to achieve the desired
settling velocity, V.
The basis for many flocculant control systems
is the belief the quantity of flocculant which must be added to achieve a desired
settling rate is proportional to the quantity of solids present the slurry. In other
words, it is believed that adequate control can be achieved by varying the flocculant
do se rate in accordance with
dv = k C (5a)
or Dv = k (5b)
It can easily be seen that there is a major
difference between the two models.
The proportional or ratio model makes no
allowance for variation in the slurry solids particle size distribution, while
the semi-empiric model predicts that the dose rate required will increase exponentially
with increases in the particle size distribution parameter. A detailed
examination of A.I.S. raw coal thickener feed has shown conclusively that
variation in the particle size distribution does have a dramatic effect on the
flocculant dose rates required to achieve a given settling rate (10). This can
be seen from Fig 1, where approx 25 g/t is required to achieve a settling rate
of 10 m/h with slurries containing 35-38% of -53 microns material (with solids
content approximately 12$) whereas less than 10 g/t is required to achieve the
same settling rate when only 26-28% of the solids present are -53 microns (with
solids content also 12%).
Furthermore, the proportional model
predicts that flocculant lose rate, D, is independent of C, whereas the
semi-empiric model predicts that the flocculant dose rate depends on C, From
Fig 1, noting that dose rates are given in g/t, it can be seen that D is not constant,
as is required by the proportional model, but does in fact increase with C, as required
by the semi-empiric model.
It is instructive to compare the magnitude
of the effect of changes in solids content versus changes in particle size distribution.
It can be seen either from equation 4 or Figure 1 that the influence of the
particle size distribution is more important than the solids content. This has
not been generally recognized in the coal industry, or elsewhere, although it
could be argued that it is somewhat obvious. For example, Stokes Law predicts
that a spherical particle of 100 microns diameter will have a settling velocity
25 times as large as that of a 20 microns diameter spherical particle. furthermore,
in interface settling (10), virtually all particles settle at the same rate. On
the other hand it seems difficult to avoid the assumption that the actual
settling rates for interface settling is largely determined by the
"natural", or unflocculated, settling rate of the smallest particles.
Confirmation of the validity of the
semi-empiric model is given from a practical viewpoint as well. For example,
control of Hi-Rate thickeners has been notoriously difficult because of their
very small residance times. This can lead to loss of thickener control in such
a short period of time as 15-30 minutes. However, by using the semi-empirical
model as its basis, three Hunter Valley Hi-Rate thickeners have been
successfully controlled for long periods of time without, to the author's
knowledge, a -single instance of loss of control due to the control system. If
the quantitative predictions of the model were significantly in error, then it
is reasonable to assume that these errors would have manifested themselves in
the rapid loss of control of the Hi-Rate thickeners.
FIGURE 1 - THEORETICAL CURVE AND
EXPERMENTAL CATA
A.I.S THICKENER FEED
The Use of Light Transmission/Scattering
Probes
The theoretical model depends on both solids
content and a particle size distribution parameter. But there is no simple convenient
way of monitoring both on-line.
However the problem can be viewed in another
way.
The surface area of slurry solids is
dependent on the solids content and the particle size distribution. A parameter
that is related to the surface area of slurry solids can be measured by means of
light transmission/scattering methods using radiation of a suitable wavelength.
By examining the mathematical properties of
the relationship describing the correlation between transmitted/scattered light
intensity, T, and the slurry % solids and particle size distribution parameter,
it has been possible to arrive at a correlation between dv and T which is
described by a mathematical form which is an invariant.
We are therefore left with the problem of
determining the value of the "constants" which specify a definite
function within the family of functions which have the given form.. Under
actual operating conditions the value of the "constants" will depend
on variations in particle surface chemistry, process hydrodynamics and other
process variations (such as pump wear etc.) as well as variations in the particle
size distribution.
A simple calibration procedure has been
devised which determines the values of the constants. This enables a single,
specific function to be selected from the infinite number of functions belonging
to that family of curves which describes the plant behavior on that particular
day.
This is indicated schematically in Figure 2.
FIGURE 2 FAMILY OF CURVES FOR dv
VERSUS T
The Cationic Control System
The feedback control algorithm devised for
the addition of cationic polymers is based on simple but well-defined physical principles.
Suitable limits on the overflow clarity
must be established by plant observation. Giver these limits it makes physical
sense to require the dosing pump to increase its output exponentially up to maximum
capacity as the overflow clarity approaches its lower limit and to decrease its
output exponentially to zero (or some small amount) as the overflow clarity
approaches its upper limit. When the overflow clarity is near the set point
then changes in pump output should approach zero exponentially.
If the average feed conditions change, then
the nominal pump operating point is automatically adjusted in an exponential fashion.
The
above changes can be specified mathematically in a unique way. This can be used
to advantage in trouble-shooting situations.
PRACTICAL ASPECTS OF THE AUTOFLOC
FLOCCULANT CONTROL SYSTEM
Daily Calibration Procedures
The initial motivation for the semi-empiric
settling model was to determine the quantitative effect of those variables
which might be expected to vary over the short term. Those variables which were
expected to vary over a longer period were neglected. However a plant must
operate over a long period, so it is necessary that a procedure be developed which
takes account of this variation.
In practice, a simple calibration procedure
has been devised which takes account of any variations in the particle size
distribution particle surface chemistry, concentration of the flocculant
solution characteristics of the flocculant solution pump, process hydrodynamics
and other process variations which have arisen since the previous calibration
procedure.
The calibration procedure can be carried out
daily, or on as-required basis. However since the time-required to calibrate is
only 10 mins approx, daily calibration is recommended.
The
calibration procedure is:
i) Ensure that the
thickener is performing satisfactorily at one instant of tine. For example,
vary the flocculant dose rate so as to achieve the desired settling velocity.
ii) Record the flocculant
pump speed and the transmitted light signal at that instant.
Flocculant Pump Calibration
The control program calculates the dose rate
required in ppm. However the flocculant solution pump is controlled by a 4-20 mA
signal. Consequently it is necessary to calibrate the flocculant pump for l/m
delivered vs % pump speed (or mA).
The daily calibration takes account of any
pump wear or variation in the concentration of the flocculant solution that has
occurred since the previous calibration. Naturally, however, no account can be taken
of any variations which occur between calibration periods. If poor control were
to occur then this would suggest some sudden change due to such (or other) variations.
This could be easily overcome by a recalibration of the control system.
OTHER AREAS OF INTEREST
The fact that the sedimentation model could
be used to make reliable quantitative predictions as to the effect of various parameters
generated confidence to approach practical problems in flocculation on a proper
scientific cause-effect basis, rather than avoid such problems on the basis
that "what works in the laboratory doesn't work on the plant".
Indeed the view was taken that the
differing performance in laboratory and plant could be used to identify those
relevant factors which differ in the plant. This approach has met with success
in many different areas.
For example, a particular area in which
this approach was predicted to provide significant improvement was in the
examination of the efficiency of flocculant/cationic dosing systems (10). At B.H.P.
Newcastle Coal Preparation Plant the use of cationic polymers has been
necessary since the use of salt water as process water has ceased. It was
anticipated that the change to fresh water would result in significant
increases in the total chemical costs. However, 'by implementing plant
modifications suggested by laboratory test work which was carried out by
Catoleum, B.H.P. have been able to reduce the annual consumption of flocculants
and cationics to approximately one half of the previous flocculant usage rates,
while the usage rate of cationics has been reduced to one quarter of the usage rate
achieved with the original dosing system. Similar successes have been achieved
at other plants.
As another example, many problems in
flocculation are strongly influenced by the details of process water chemistry
and water quality. Wambo Coal Preparation Plant has long been renowned for its very
difficult clay problems, which have required exceptionally high chemical costs
to keep under control. Through the combined use of system, recommended by
Catoleum, and the AUTOFLOC flocculant control system, chemical costs have been
reduced to approximately ones-third of previous usage rates. In addition, the
increased production rates now possible following the recent plant upgrading,
have been handled without bottlenecks arising from problems in the thickener
circuit.
In another example, the recovery of coal
from the flotation circuit it at Coal Cliff Coal Preparation Plant was
increased by approximately 9% by means of the combined use of improved dosing
of cationics and improved flotation reagents. These results are the subject of
another Paper at this conference.
Another, more significant, prediction was
made (10) by claiming that, with appropriate generalization, the mathematical
methodology used in developing the sedimentation model could be used to
determine physically significant relationships in any field between those parameters
which have an underlying causal relationship. This approach has been applied to
the problem of predicting how to vary flotation reagent dose rates in
accordance with varying feed conditions.
Although the problem in flotation is more
complex e.g. the number and type of important parameters, this approach has
provided the basis for a total flotation control system which monitors and controls
those parameters which have a crucial significance on flotation performance,
This system is known as AUTOFLOTE and is currently being developed and marketed
by Century Autoflote Pty. Ltd.
These findings will be the subject of a
subsequent publication.
CONCLUSIONS
The quantitative semi-empiric model of
sedimentation that was presented at the Second Australian Coal Preparation
Conference (10) has been used as the basis of a simple, reliable system which provides
complete on-line automatic control of thickener circuits, The system has been
proven in practice in three Hunter Valley Coal Preparation Plants which use Hi-Rate
thickeners.
The following process parameters may be controlled.
Flocculant dose rates:
controlled on a feedforward basis achieve a desired settling velocity (or other)
operating objective with feedback adjustment based on bed-level measurement.
Underflow pumping: controlled on
a feedforward basis so as to ensure that solids are removed at the rate at which
they enter; feedback adjustment can be made on the basis of measured rake torque
and/or bed level.
Rake torque, bed level:
these parameters are normally controlled within acceptable limits by ensuring both
constant settling rates and thickener mass balance. However, when this is not the
case, underflow pumping rate can be adjusted on a feedback basis so as control rake
torque and bed level.
Thickener overflow clarity:
can be controlled automatically adjusting the dose rate of cationic polymers on
a feedback basis.
The sedimentation model also provided the
basis for the follow developments.
Significant reductions in the usage rates
of cationics as B.H.P. Coal Washery have been achieved by means of appropriate
alteration to the dosing system.
Significant reductions in chemical costs
have been achieved Wambo Coal Preparation Plant via the combined effects of
adjust process water chemistry and the AUTOFLOC flocculant control system
Recovery from the flotation circuit at Coal
Cliff Coal Preparation Plant was increased by 9% approximately by improving
water quality through the use of cationic polymers, in conjunction with
improved flotation reagents.
The mathematical techniques used in
developing the automatic thickener/flocculant control system have been adapted
to development automatic flotation control system.
ACKNOWLEDGEMENTS
The author of this Paper would like to thank
Catoleum Pty. for assistance given in developing the AUTOFLOC system and permission
to publish this Paper and particularly the valuable co-operation and assistance
given by the staff of Warkworth, Hunter Valley and
Wambo Coal Preparation Plants who demonstrated that model did work under plant conditions.
Achieved with the original dosing system. Similar
successes have been achieved at other plants.
As another example, many problems in
flocculation are strongly influenced by the details of process water chemistry
and water quality. Wambo Coal Preparation Plant has long been renowned for its very
difficult clay problems, which have required exceptionally high chemical costs
to keep under control. Through the combined use of gypsum recommended by
Catoleum, and the AUTOFLOC flocculant control system, chemical costs have been
reduced to approximately one-third of previous usage rates. In addition, the
increased production rates now possible following the recent plant upgrading,
have been handled without bottlenecks arising from problems in the thickener
circuit.
In another example, the recovery of coal
from the flotation circuit at Coal Cliff Coal Preparation Plant was increased
by approximately 9% by means of the combined use of improved dosing of cationics
and improved flotation reagents. These results are the subject of another Paper
at this conference.
Another, more significant, prediction was
made (10) by claiming that with appropriate generalization, the mathematical
methodology used in developing the sedimentation model could be used to
determine physically significant relationships in any field between those part
meters which have an underlying causal relationship. This approach has been
applied to the problem of predicting how to vary flotation reagent dose rates
in accordance with varying feed conditions.
Although the problem in flotation is more
complex e.g. the number and type of important parameters, this approach has
provided the basis for a total flotation control system which monitors and controls
those parameters which have a crucial significance on flotation performance.
This system is known as AUTOFLOTE and is currently being developed and marketed
by Century Autoflote Pty. Ltd.
These findings will be the subject of a subsequent
publication.
CONCLUSIONS
The quantitative semi-empiric model of
sedimentation that was presented at the Second Australian Coal Preparation
Conference (10) has been used as the basis of a simple, reliable system which provides
complete on-line automatic control of thickener circuits. The system has been
proven in practice in three Hunter Valley Coal Preparation Plants which use Hi-Rate
thickeners.
The following process parameters may be controlled.
REFERENCES
Coe H S and Clevenger G H, "Methods for Determining the Capacities
of Slime-Settling Tanks", Trans. AIME 55 (1916) 356.
Robinson C S, "Some Factors Influencing
Sedimentation", Ind. Eng. Chem. 18 (1926). 869.
Egolf C B and McCabe W B, "Rate of Sedimentation of Flocculant
Particles", Trans, A. Inst. Chem. Eng. 33, (1937). 620.
Work L T and Kohler A S, "The Sedimentation of Suspensions",
Trans. Am. Inst. Chem. Eng. 36 (1940) 701.
Steinour H H, "Rate of Sedimentation", Ind. Eng.
Chem. 36 (1944), 618, 840, 901.
Kynch G J, "A Theory of Sedimentation", Trans.
Faraday Soc, 48 (1952) 166.
Richardson J F and Zaki W M, "Sedimentation and Fluidisation,
Part 1", Trans. Inst. Chem. Eng. 32 (1954), 35.
Talmage W P and Fitch E B, "Determining Thickener Unit Areas",
Ind. Eng. Chem. 47 (1955) 38.
Dell C C and Brown D W, "Testing Polyelectrolyte Flocculants
for Sedimentation", Filtration and Separation Sept/Oct 1973, 507.
Macdonald J R etal Proceedings of the 2nd Australian Coal Preparation
Conference, Rockhampton, 1983 p. 397-420.
