Note: Descriptions are shown in the official language in which they were submitted.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
METHOD OF AND CONTROL SYSTEM FOR CONTROLLING ELECTROSTATIC
SEPARATOR
FIELD OF THE INVENTION
THIS INVENTION relates a method of and a control system for
controlling an electrostatic separator.
BACKGROUND TO THE INVENTION
Electrostatic separators have been in use in the mineral sands industry
for over fifty years and in the recycling industry for many years.
Figure 1 of the accompanying drawings diagrammatically illustrates an
electrostatic separator which includes a metal roll 10 which is electrically
earthed. A
brush 12, which maybe electrically conductive and electrically earthed serves
to
sweep non-conductive particles off the roll. The roll as illustrated in Figure
1, rotates
clockwise.
A high voltage wire 14 extends along the roll close to its outer surface
and there is a coronal discharge between the wire 14 and the roll 10. Mineral
sand,
or other solid particles containing the materials to be separated, is fed, as
shown by
arrow A, as a thin stream onto the roll. As the particles pass the wire 14
they are
charged. The so-called conductor particles lose their charge almost
immediately as
it leaks away to the earthed drum 10 and are thrown off the roll as indicated
by the
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
2
arrow B. The non-conductor particles retain their charge for longer and are
"pinned"
to the drum. They either fall off later than the conductor particles, as
indicated by
arrow C, or are swept off by the brush 12.
Adjustable splitter plates 16 and 18 keep the conductor and
non-conductor particles apart. There is inevitably, between the separated
conductor
and non-conductor particles, an area where what are called "middlings" fall.
Middlings are a mixture of conductor and non-conductor particles. The arrows D
and E indicate the path of the middlings.
An earthed static electrode 20 may be used, charged to the same
electric polarity as the corona wire 14 assists in discharging the conductor
particies,
attracting them by electric induction and thus freeing them from the roll 10.
A conventional electrostatic separator has a number of controls which
the operator uses to obtain maximum yield and acceptable mineral grade.
Specifically the speed of the roll 10, the voltage on the wire 14, the voltage
on static
electrode 20, the positions of the corona wire 14 and the electrode 20, the
positions
of the splitter plates 16 and 18, air humidity, mineral particle temperature
and
mineral feed rate can usually be adjusted.
Analysis of the conductor and non-conductor particles can be used to
determine whether the grade of the products emerging from the separator is
acceptable. The difficulty is that a proper.analysis may take many hours. The
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
3
results, when given to the operator, are not current and in fact are so far
out of date
as to have little value. A quick but error-prone "snap-shot" analysis can be
obtained
within thirty or sixty minutes. The difficulty with this is that the sample
taken at any
specific instant may not be a true reflection of the average yield and purity
being
obtained. The complexity of manually adjusting all the important controls of
an
electrostatic separator to achieve firstly a stable grade and secondly a
maximum
yield, is mostly too much for a human operator. As a result of this, it is the
skill and
experience of the operator that plays the predominate roll in determining the
efficiency of the separation process.
An ideal solution to the manual problem described above would be an
automatic multiple-input, multi-output modern control system that measures
both
grade and yield, and adjusts some or all of the separator controls
automatically to
achieve a stable grade at a pre-determined set point and to achieve a maximum
yield at the same time, as is well known in the art of advanced automatic
control,
e.g. Model Predictive Control. However such an advanced automatic control
system
has practical drawbacks in a plant environment, where the standard
instrumentation
skill levels are more focussed on single-input, single output control systems
for
instance standard PID (Proportional-Integration-Derivative) control systems
and
therefore any maintenance may be expensive. Another practical drawback is that
the technical complexity of such a solution, specifically the implementation
of
combinations of safe operating ranges for all the separator controls, can make
the
commissioning of such a system expensive as well.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
4
One object of the present invention is to provide a method of
controlling an electrostatic separator in a semi-automatic manner thereby to
improve
the efficiency of the separation process, but in a low cost manner, and
technically
simple enough to be maintained in a plant environment.
Another object of the present invention is to provide a control system
for an eiectrostatic separator which improves the efficiency of the separator
process,
but a low cost system, that is technically simple enough to be maintained in a
plant
environment.
BRIEF DESCRIPTION OF THE INVENTION
According to one aspect of the present invention there is provided a
method of controlling an electrostatic separator, said method comprising
selecting a
set point value for the grade of one of the product streams emerging from the
separator, measuring the grade of said one product stream, using the result of
the
measurement to adjust a first of the control variabies of the separator
automatically
by application of a simple, industry-standard controller to maintain the grade
of said
one product stream generally at the set point value, manually adjusting a
second
control variable of the separator, allowing said automatic adjustment of said
first
control variable in response to the measurement result to take place, to
maintain
said output stream grade generally at the set point value, manually monitoring
the
yield and manually adjusting the second control variable or another control
variable
to optimise the yield, while continually and automatically maintaining said
output
stream grade generally at the set point value, with the simple control system.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
The grade of said one product stream may be measured optically, e.g.
it may be measured spectroscopically or by measuring X-ray fluorescence.
The term "grade" means the percentage of said one stream which is
comprised of the material which is preferably required to be in said one
stream.
Thus a grade of 90% means that 90% by weight of the material in the product
stream is of the material which is required in that product stream.
"Yield", also termed "recovery",, is the ratio between the amount of the
required or preferable material in said one stream and the amount of the
preferable
material in the feed stream, expressed as a percentage.
The set point is preferably an optimum value of the grade of the
product stream.
The first control variable is preferably the voltage of the corona
discharge wire or it can be the rate of rotation of the roll, or any one of
the other
control variables of the separator. Further, the second control variable can
be any
one of the control variables of the separator, such as the voltage of the
corona
discharge wire, rotation of the roll, voltage of a static electrode, position
of the
corona discharge wire, position of the static electrode, splitter plate
position, material
feed rate, material temperature and/or air humidity, provided that the second
control
variable is not the same as the first control variable.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
6
According to another aspect of the present invention there is provided
a control system for an electrostatic separator, said control system
comprising
means for determining the yield of at least one product stream emerging from
the
separator and manually operable means for adjusting at least one manually
operable control variable of the separator, wherein said control system
further
includes means for analysing the grade of said one product stream emerging
from
the separator, means for automatically controlling one of the control
variables of the
separator using a simple, industry-standard controller, in response to an
output
signal from the analysing means to adjust the grade of said one product stream
to a
set point value by application of a simple controller
The means for analysing the grade of said one product stream may be
an optical meter such as a spectrometer or a meter for measuring X-ray
fluorescence.
BRIEF DESCRIPTION OF THE DRAWINGS
For a better understanding of the present invention, and to show how
the same may be carried into effect, reference will now be made, by way of non-
limiting example, to the accompanying drawings in which:
Figure 2 is a schematic representation of a test rig;
Figure 3 shows the correlation of input parameters with product grade, in the
test rig of Figure 2;
Figure 4 shows the impact on the test rig of Figure 2, of variation of the
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
7
separation roll drive speed;
Figure 5 shows the impact on the test rig of Figure 2, of variation of the HT
voltage;
Figure 6 is a schematic representation of the basic control concept for the
test
rig of Figure 2;
Figure 7 is a table of coupling factors for the major control parameters;
Figure 8 is a schematic of the control model for the test rig of Figure 2;
Figure 9 shows the effect on the test rig of Figure 2, of roll motor speed
change;
Figure 10 shows the effect on the test rig of Figure 2, of changed roll speed
and reiative humidity during an early morning start-up;
Figure 11 shows the effect on the test fig of Figure 2, of a change in
temperature and relative humidity;
Figure 12 shows the effect on the test fig of Figure 2, of a change in
humidity
and set point; and
Figure 13 shows changes in parameters while recovery in the test rig of
Figure 2 is optimised.
DETAILED DESCRIPTION OF THE DRAWINGS
The invention is described by way of non-limiting example with
reference to its application to electrostatic separators used for separating
particulate
minerals. However, the invention is useful in all the applications of
electrostatic
separators of which the applicant is aware, e.g. the removal of metals from a
feed-stream of plastic for recycling. The invention also finds application in
the
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
8
processing of other conductor and non-conductors, for instance different types
of
recycled materials.
Referring to Figure 2, a pilot-scale test installation was constructed at
the facilities of the Department of Process Engineering at the University of
Stellenbosch to enable continuous tests to be carried out in a controlled
environment. A laboratory electrostatic roll separator 22, similar to the
electrostatic
separator shown in Figure 1, was set up to process sand in the form of
particulate
ore samples in a closed loop. The feed chute 24 was fitted with electrical
heating
elements by which the feed temperature could be adjusted. The speed of both
feed
and main roller drive motors could be controlled as well as the setting of the
voltage
of the high voltage corona wire 14. The position of the corona wire 14, static
electrode 20 and splitter flaps or plates 16,18 on the discharge were all
manually
adjustable. Mass flow of the feed was measured by calibration of a roll feeder
26.
It is to be understood that, while the invention is described with
reference to a research pilot plant setup, it is primarily intended for
application to
electrostatic separators in an industrial production environment.
The splitter plates 16,18 were set to produce two exit product streams
only, i.e. a "conductor" stream that dropped off the front of the separation
roller 10
and a "non-conductor" stream that was discharged at the earthed brush 12 at
the
rear of the separation roller. Setting the splitter plates 16,18 in this way
avoided
discharging particles in a middlings stream and mass balance analysis to the
test
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
9
installation was simplified. Mass-flow of the conductor stream was measured in-
line
with a calibrated slot discharge mass flow meter 28.
The product, i.e. the non-conductor stream, was continuously
spectroscopically analysed with a Blue Cube MQi in-line mineral quantifier 30
and
the product mass flow was calculated by the difference between the feed mass
flow
and the conductor stream mass flow. Both discharge streams were combined and
re-circulated in an endless loop by the combination of a vibration conveyor 32
and a
bucket elevator 34.
The test installation was provided with instrumentation and facilities to
achieve automatic control and logging of the ore feed mass flow, ore feed
temperature, separation roll speed and the voltage of the corona wire. In
addition,
the product mass flow, product grade and relative humidity (inside the
separator
casing) were logged and were used as input to the control program. The actual
values of all these settings were logged at one second intervals. The other
settings
were adjusted and logged manually.
A large sample of typical heavy minerals dry mill feed, with known
composition, was continuously circulated to determine the characteristics of
the
installation and to prove the control system. The installation was used to
separate
non conductive Zircon containing particles from the sample.
Initially, the installation was set up and run continuously, with the
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
corona wire 14 and static electrode 20 (HT plate) switched off, to homogenise
the
ore sample. Thereafter the corona wire 14 and static electrode 20 were
switched on
and separation started. Under manual control, parameters were adjusted until
stable
operation was achieved. The parameters were then adjusted up and down to
determine relative sensitivity. A multi-input, multi-output dynamic model of
the
electrostatic separator was extracted by the application of standard system
identification techniques for automatic control and is shown in Figure 3. The
initial
tests showed that the separator roll speed and corona wire voltage (HT
voltage)
setting had the most significant impacts on the product grade and recovery. A
series
of tests were then conducted to quantify the sensitivity and practical use of
these
two parameters.
The impact of variation of the separation roll drive speed is shown in
Figure 4. The roll drive motor speed was increased from 50 Hz, in 5 Hz steps,
to 70
Hz. As more "in-between" particles were thrown off the roll, the non-
conductors
product stream mass flow decreased and the product grade improved inversely in
steps from 88% to 98% in Zircon content. With the known feed grade of 81 %,
the
calculated recovery, or yield, reduced from 93% to 86%.
Figure 5 shows the effect of the HT voltage being decreased from 21
kV, stepwise to 14 kV. With the lowered HT voltage applied, less of the in-
between
or middlings particles pinned to the roll 10 and the product mass flow reduced
from
63 g/s down to 46 g/s but product grade improved inversely from 88% to 98%
Zircon
content. However, as to be expected, recovery of Zircon dropped from 97% to
80%.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
11
The tests to determine the effects on the installation of varying
separation roll drive speed and HT voltage, were conducted within 20 minutes
and
provided, quantified and consistent results.
Figure 6 shows a schematic of the basic control concept envisaged for
controlling the separator. The initial tests were used to determine the time
constants for the control system for the installation and the sensitivity of
the roll
separator speed to the most important adjustable parameters. The
experimentally
determined coupling factors for the major control parameters are shown in the
table
of Figure 7.
In these experiments, it was observed that plant reaction time to
changes in high voltage and roll speed was about 2 seconds. The flow rate
through
the MQi sensor 30 was estimated at 4 seconds and the MQi data integration time
was set at 9 seconds. The total lag of the MQi output after changes in control
parameters thus added up to 15 seconds. To allow for potential system noise
and to
ensure stability, the overall time delay was set to 25 seconds for modelling
purposes. The time constant of the scale, at 10 seconds, was comfortably
within this
margin.
The control model finally used, is shown in Figure 8. The model was
designed to maintain automatically, a constant grade of output product (in
this case
the non-conductor stream) as the set point and to allow a human operator to
maximise product mass flow automatically, implying maximised recovery, while
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
12
maintaining the desired grade.
The data logging and control program were implemented on a laptop
computer. All interfacing was done on the basis of industry standard 4-20 mA
loop or
0-10 V analogue signals and A/D converters or RS232 data link. The test rig
was
instrumented to log and display the following parameters at a rate of 1 Hz.
Output
was calibrated in the stated units:
Roll speed Drive Motor supply frequency: Hz
Feed rate Drive Motor supply frequency: Hz
High voltage setting kV
Ore temperature c
Relative humidity %
Product flow rate gls
Product grade % Zircon
The parameters of the control model were conservatively set as a
simple proportional control with a time constant of 200 seconds. An integral
function
was used to ensure return to the set point. No derivative function was
necessary. To
test the control function, automatic control was directed at the single input
parameter
of High Voltage (corona wire voltage or HT), directed at maintaining product
grade at
a set point. Other parameters were then adjusted manually for creating
disturbances.
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
13
Figure 9 shows the effect of a sharp roll motor speed change from
65Hz to 45Hz. All other input parameters were held constant. The product grade
dropped sharply from the stable condition of 92% to 88% Zircon. At the same
time
the product mass flow increased. The automatic control action, reducing the HT
voltage, then corrected the grade back to the set point over a period of 300
seconds.
Mass flow adapted to the new operating conditions. The new setting ensured
that
on-grade production was maintained, although at a lower recovery. It is
therefore
clear that the operator has direct and simple control over recovery, while the
control
system continuously keeps the grade constant. The rather slow control action
ensured that no unstable situation arose. The small kink in the product flow
graph
around 1250 s was due to occasional product hang-ups in the discharge chutes.
Figure 10 shows the results of an early morning start-up of the
separator, with high relative humidity. The feed heater thermostat controlled
the feed
temperature within a band of 6 C. The controller quickly stabilised the grade
at the
92% set point and maintained grade, by adjusting the HT, as the relative
humidity
dropped. The operation was stable within 10 minutes after start-up. A step
reduction
in roll speed disrupted the operation but the controller adjusted the HT
setting back
to a new balanced setting and maintained product grade. A small ripple is
visible in
Figure 10, due to the cyclical variation in product temperature.
Figure 11 shows how the controller handled a change in ore
temperature. The relative humidity, measured within the machine housing,
changed
inversely with the ore temperature. The controller could keep the product
grade
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
14
close to the set point by adjusting the HT Voltage. Product mass flow varied
due to
the changing conditions.
Figure 12 shows how the settings parameters changed in response to
a change in humidity and set point. The left hand side of the graph shows the
response due a change in relative humidity. The ripple, due to the cyclic ore
temperature regulation, is also visible. When the set point was reduced from
93% to
87% in one step, the voltage settings followed in small steps until the new
set point
was reached in about 2 minutes. Non-conductor mass flow increased accordingly.
Figure 13 demonstrate the final control model. From the initial working
point, product grade was controlled at 92% Zircon content. To improve
recovery, the
roll speed was increased. The product mass flow showed a small drop and the
grade increased above the set point. As the control system increased the
voltage to
bring the grade back to the set point, product mass flow increased and
recovery
improved. Recovery was calculated and presented in real time to the operator.
The
same pattern repeated itself until the roll speed reached 70Hz, when recovery
dropped slightly although grade was still maintained. From the initial
setting, the
optimum for roll speed and HT was reached with a production gain of more than
10%, while maintaining grade. In this case, the operator was thus able to find
a set
of separator settings leading to better recovery, at the same product grade,
in an
easy, practical way.
The invention described allows automatic control of electrostatic
CA 02616564 2008-01-24
WO 2007/014402 PCT/ZA2006/000090
separators and rapid optimisation of machine settings, with the advantages of
continuously maintaining product output grade, optimum recovery while
maintaining
the grade, improved production without the burden of regular manual tuning,
improved plant yield by optimised flow circuits, higher throughput by less
recycling of
off-grade product, and allowing multi-parameter plant optimisation.
