Automatic Control of Flotation Columns - Advanced Control and Supervision of Mineral Processing Plants - page 271

cross-sectional area. This variable has recently attracted researchers' attention and
many sensors have been developed mostly for measuring local superficial velocity
in an attempt to evaluate ill gas distribution or sparger malfunctioning [45], [47,
48]. They are all based on some sort of tube submerged in the aerated pulp and
a vacuuming system to drive the gas into a measuring device (typically a gas mass
flow meter). Since regular commercial flow meters ( e.g. turbines, Venturi tubes, etc. )
are perfectly adapted for this sort of measurement in flotation columns and they are
entirely adequate for process control purposes, these newly developed sensors will
not be discussed here.
6.4 Automatic Control
6.4.1 Control Hierarchy
In the present context, with globalization and increasingly strong international com-
petition, mineral processing plant managers are constantly seeking to reduce pro-
duction costs while increasing throughput and product quality. One way to reach
this objective is to use better equipment, i.e. , to replace existing parts of the plant
with more efficient ones or to add new equipment. Improving plant operation can
also be achieved through on-line data processing by making judicious on-line use of
process sensors and large plant operation databases. Of course, this approach will
only be successful if equipment is working properly.
On-line data processing consists in using measurements provided by plant sen-
sors and knowledge of dynamic plant behavior (evaluated from measurement databases
recorded during plant operation) to automatically, continuously and adequately ma-
nipulate plant actuators, such as valves and variable speed drives, in order to achieve
a specific objective. The ideal on-line data processing is made up of four hierarchic
layers: instrumentation and regulatory control; process observation; process control;
and RTO. This scheme is depicted in Figure 6.10.
Instrumentation for the process being considered has been thoroughly discussed
in the previous section. Regulatory control, the basic control layer, consists of local
monovariable control loops (usually PID controllers) providing steady operation for
the flotation column inputs (air, wash water, pH and reagents), but refers in general
to all control elements (valves and pumps, including that of the tailings).
Process observation aims at getting useful on-line information about plant behav-
ior, therefore allowing the “visualization” of process states and detection of operat-
ing problems, abnormal performances, etc. Reliable sensors are the basic equipment
required for this. However, using data processing techniques such as fault detection
and isolation (FDI)[57, 58], data reconciliation [59] and observers [60], may greatly
improve the process “picture”. FDI rapidly detects and physically localizes prob-
lems such as sensor biases and leaks, whose rapid correction will reduce detrimental
effects on production. Data reconciliation techniques improve the quality of noisy