PROCESS CONTROL IN COMMINUTION
In a beneficiationplant, Process control is an essential component of any comminution system. The
widespread adoption of automation in mineral processing plants began more than
four decades ago, when rudimentary regulatory strategies for regulation of ore,
water, and slurry were successfully deployed on single-loop analogcontrollers.
Virtually all plants built today have a sophisticated digital control system
that enables all basic control functions, providing the human–machine interface (HMI), and acting as
the gateway to plant management information systems, which couple process and
business controls. In addition, most new plants adopt advanced process control
applications to deal with the multivariable nature of process optimization in
real time. Although the journey has not always been smooth, the industry has
increasingly embraced process control as one of the most capital-effective
investments available in the pursuit of lower costs and increased revenues.
The need
for process control is made evident in early stages of the comminution circuit
design process. To illustrate, Figure1 shows a hypothetical distribution of a
hardness index in an ore body.
The need
to adapt to changing feed conditions is readily apparent. If the plant is
designed to produce the desired product size for a hardness of 14 at design
tonnage, the ore will be harder 17% of the time; unless the tonnage is reduced,
the product size will be too coarse and possibly some loss in liberation will
occur. On the other hand, 83% of the time, the ore will be a softer, finer
product, and more liberation will possibly result.
Fig1. Cumulative probability density
function for ore hardness from testing
The key
implicit assumption here is that a process control system will allow the
circuit to achieve steady-state targets and overall operational stability in
the face of feed-ore variations.
These
temporal changes in ore characteristics in the feed to a comminution process
are called “disturbances” because nothing can be done by the operator or
control system to modulate them. The nature (frequency and amplitude) of these
disturbances will dictate the severity of the control problem and the
complexity of the solution. For example, Figure 2 illustrates three possible
variants of circuit feed for the ore characterized in Figure 1. Case A
illustrates a well-blended feed; case Bshows greater short-term variability;
and case C shows longer-term variability. As Murphy’s law would correctly
predict, disturbances, such as those in cases B and C, are more common in
practice. In fact, all of these disturbances coexist.
Fig 2. Temporal variation of ore hardness
Disturbances
will arise in ore hardness (e.g., different ore types and genesis modes), feed
size (e.g., blasting practices and stockpile segregation), and liberation
requirements (e.g., requiring a change in grind). It is also quite common to
see disturbances arising from internal sources; for example, a sump pump that
intermittently delivers a dilute slurry to a cyclone feed pump box, or a mechanical
feeder prone to flow interruptions. Although the comminution process may well
be stable to these fluctuations without intervention, control systems are
normally required to ensure stability and to enhance the overall economic
performance of the process.
To
continue with the illustration above, Figure 3 A shows a hypothetical grind-size recovery curve. If we were able to
maintain a target product size of 70% at 200 mesh, we would expect to see a valuable
metal recovery of 88.6%. However, if the feed tonnage is constant, the grind
will vary based on the nature of the disturbances. Combining the information of
Figures 2 and 3A produces Figure 3B. As we would expect, the greater variability in hardness
for case C produces a wider distribution in grind size to the separation
circuit. Figure 3.45B also shows the overall recovery
expected under each open-loop operating scenario. (Open loop implies no
intervention by the operator or the control system.) Clearly, the more stable
feed of case A provides a recovery much closer to the optimal value shown in
Figure 3A, while case C incurs an -2% recovery
loss.
Fig 3. Illustration of the impact of disturbances in comminution
on downstream separation Processes
In this
hypothetical case, if a control system were to be applied to maintain the grind
at the average or target value, the total tonnage treated would be essentially
the same, but a 2% recovery gain would be seen for case C. The latter number is
more or less typical of the recovery gains associated with supervisory control
applications. Throughput increases frequently lie in the range of 3% to 15%. The
magnitude of these numbers underscores the attractiveness of such investments.
Industrial
process control systems are powerful tools for maintaining process stability
and ensuring optimum economic performance in the face of disturbances. The
complexity of the control strategy depends in part on the complexity of the
process and in part on the nature of the
disturbances. Estimates of the nature of disturbances are increasingly available
at the design stage, opening new avenues for the a priori design of control
strategies. Operationally, efforts to mitigate disturbances upstream (at the
mine or crusher) will simplify the control requirements, although the spatial
variability of ore characteristics often precludes effective blending.
The
combination of the magnitude and frequency of the disturbances will also have
an impact on
control
requirements. Simply put, minor amplitude variations are easier to handle. Similarly,
low-frequency disturbances can often be very effectively rejected by control,
while the process effectively filters very high-frequency disturbances. Those
lying in the intermediate range can usually be rejected to a greater degree
(the shorter the frequency). This frequency range is related to the time constant
of the process. For example, the dynamics of a response to a feed-hardness
change in a grinding circuit are much slower than the change in water flow in a
pipe to a response in supply pressure.
The Control Triad
The
control triad illustrated in Figure 4 provides a useful framework for this
overview. This schematic conveys the message that an effective process control
system will include the proper blend of field instrumentation, hardware, and
control strategies.
Fig 4. The control triad
Figure 5
provides a practical illustration of the control triad in the context of a
simple water flow regulation loop. In this instance, the field instrumentation
consists of the orifice meter and ball valve. The hardware comprises the
input/output (I/O) subassembly, the computer with the basic software, and the
HMI. The strategy employs a simple well-tuned Proportional Integral and
Differential (PID) Actions control law, where the operator determines the set
point or target for the water flow rate. Of course, there are many important
and related subjects that are beyond the scope of this discussion. These
include signal filtering, sampling intervals, loop tuning, and dead-time
compensation.
Fig5. A simple flow control loop
Instrumentation
Because
final control elements are largely restricted to devices that control position
(e.g., valves, knife, or flop gates) or electrical motor speed (e.g., feeder,
pump, and mill), this section will focus on sensors, offering a much broader
range of devices. The first law of process control—“All control starts with
measurement, and the quality of control can be no better than the quality of
measurement,” —or in the vernacular—“Garbage in, garbage out”—helps validate
this choice.
Table 1
lists some of the more common sensors used to monitor comminution circuit
equipment. (Because there are many manufacturers of competing instruments, we
have elected to distinguish instruments on the basis of the technology employed
to make the measurement.) Although the list is not exhaustive, it does show
that there is a good capacity for measurement in such process systems.
It is
evident from this table that the process control system designer often faces a
problem related to choices. In other words, what kind of technology is best
suited for the measurement problem at hand, and, which vendors manufacture
proven products employing this technology? In the lower level stabilizing loops
typically associated with the regulation of ore, slurry, reagent, and water
flows, the preferred sensors are generally well established. For example,
electronic belt scales are the sensor of choice for measuring solids mass flow
on a conveyor belt.
Instrumentation
provides the interface between the process and the control strategies. The
proper selection, installation, and maintenance of these field devices is
essential to ensure that the benefits associated with process control
applications are sustained for the life of the project. Ongoing sensor development
efforts also means that the process control engineer needs to stay abreast of
measurement technology, looking for opportunities to further develop or enhance
the performance of a process control system.
Hardware
Control
hardware most frequently encountered in mineral processing plants are the Distributed
Control Systems (DCS) or the programmable logic controllers (PLC). In many
plants hybrid architectures involving a combination of DCS and PLC technologies
are common. Figure 6 is an illustration of such a hybrid structure and shows
the hardware layout for a typical process control system. This picture is
expected to change in the coming years as smart instruments and equipment
displace the more traditional I/O interfaces. Moreover, as bandwidth increases,
the likelihood of delivering control applications over the Internet increases,
and remote hardware and application maintenance and development support will be
simplified.
Fig6. Components of a distributed control
system
To
provide some notion of scale, the I/O count (i.e., the total number of
discrete, analog, and digital inputs and outputs) will range from about 2,000
for simpler, smaller plants to 6,000 for larger, more complex operations.
评论
发表评论