Question 7 of 7: Grinding Circuit Process Control and pH Control

Question 7: Grinding Circuit Process Control and pH Control (20%)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

a) Grinding circuit process control

Process control in mineral processing exists to hold a circuit as close as possible to its economic optimum – maximum throughput at target product size and minimum specific energy – in the face of continuous disturbances (ore hardness, feed size, ore-type blend changes) that would otherwise force wide, costly swings around that optimum if left to manual operation. In a grinding circuit the most commonly controlled variables are cyclone overflow product size (P₈₀, or a proxy such as %-passing a reference mesh from a particle-size analyzer or, more commonly, cyclone overflow density/pressure), mill sump level, and mill power draw/bearing pressure (a proxy for volumetric mill charge/load); the corresponding manipulated variables are fresh ore feed rate, mill (dilution) water addition and sump water addition to cyclone feed, and, on a SAG mill, sometimes ball addition rate or mill speed on a variable-speed drive. DCS stands for Distributed Control System – the plant-wide networked hardware/software platform (PLCs, I/O, operator HMI screens, historian) that executes every control loop, alarm and interlock across the plant from a central control room. PID stands for Proportional-Integral-Derivative – the standard three-term feedback controller (output = proportional to the error, plus its time integral to eliminate steady- state offset, plus its derivative to anticipate/dampen fast changes) that implements the vast majority of individual grinding-circuit control loops (level, density, feed-rate) within the DCS.

b-i) pH control block diagram

b-ii) Terminology

Setpoint: the target pH the loop is trying to hold. Error: setpoint minus measured pH, the signal the PID controller acts on. Manipulated variable: here, the lime (alkali) or acid dosing rate. Feedback: a correction computed from the ALREADY-occurred deviation, measured directly on the process (the pH sensor reading). Feedforward: a correction computed from a measured DISTURBANCE (here, ore/feed rate, which drives reagent demand) applied before that disturbance has had time to move the measured pH at all.

b-iii) Feedforward/feedback system types

The dominant configuration is feedback-only PID control on the pH sensor signal, sufficient for slow disturbances. Because pH response to reagent addition is often fast but the pH probe itself (and slurry transport lag to it) introduces significant dead time, many plants add a feedforward trim based on a measured disturbance that is known to drive reagent demand – most commonly ore/feed tonnage, sometimes combined with a measured feed pH or acid-generating mineral content – summed with the feedback PID output to pre-position the dosing valve ahead of the disturbance reaching the sensor. A cascade arrangement (an outer pH loop setting the setpoint of an inner reagent-flow loop) is also common where reagent delivery itself is not perfectly linear or repeatable.

b-iv) Advantages

Feedback alone is simple, robust to unmeasured/unmodelled disturbances (it corrects whatever the sensor actually sees, regardless of cause), and needs no separate disturbance model. Adding feedforward on top of feedback substantially reduces the amplitude and duration of pH excursions following a known disturbance (a feed-rate change), because correction begins before the pH has actually drifted rather than only after – important in a flotation circuit where a pH excursion outside the selective-collector's working window can immediately depress recovery or grade.

b-v) Disadvantages

Feedback alone cannot correct a disturbance until the pH has already moved and the sensor/transport dead time has elapsed, so it is inherently reactive and can allow real, if temporary, off-spec pH excursions. Feedforward requires an accurate model of how the measured disturbance actually affects reagent demand (an inaccurate gain/lag makes the feedforward trim itself a source of upset), needs its own dedicated instrumentation (feed-rate measurement), and can never fully replace feedback because it cannot see or correct for disturbances it was not designed to measure – it is therefore always used in combination with, not instead of, feedback trim.