Gold plant process control

1. Solids feed control

Most gold plants have some potential problems with the control of solids feeds. These can include the adverse effects of ore segregation, non-linear or erratic actuator dynamics (especially vibratory feeders), moisture content, conveyor-belt dynamics and blockages. All of these issues are readily solved by appropriate advanced technologies that produce the best possible results. When there are multiple withdrawal points from an ore silo, blending can be maximized, or sometimes segregation can be exploited to optimize short- to medium-term grinding conditions. Invariably, a fast and noise-free response of the actual solids feed rate to its setpoint is required for good higher-level control, e.g. for the control mill load.

2. Crushing stage control
Crushers have relatively fast dynamics. Screens, too, have quite fast dynamics. As a result, these two processes can be essentially regarded as instantaneously responding units, connected by conveyor belts to other units or solids storage bins. Control can be applied in a strategy that allows particularly for the dynamics of conveyor belts and for constrained control and optimization. Attention should also be given to start-up and shutdown sequences and their control.

3. Mill circuit control
In the early 1980s, successful multivariable control of industrial milling brought recognition to the fact that milling circuits have complex interactive dynamics that can be modelled and controlled very well by advanced methods. For operation away from the influences of process limits, good multivariable control is easily implemented by many of the advanced methods that are now standard. Because the optimum operation of milling circuits is almost invariably near process limits, optimum control requires a multivariable controller that has switching control strategies, for use when various combinations of process inputs and outputs change between being constrained and unconstrained. Model predictive control (MPC) is a good general technology for this, but it is complex, cumbersome and less reliable in its general form, so specialist adaptation should be applied to exploit specific characteristics of milling circuits. A switching control strategy has been developed to allow for easy specification and tuning. It provides for controlling process outputs to setpoints, minima or maxima, or combinations of these. A typical list of setpoints for a simple closed milling circuit would be setpoints for sump level, product size, circulating load density or underflow angle and cyclone feed density or flow. Minima and maxima could be set for mill power, sump level, product size and cyclone feed density. Priorities can be set for the various setpoints and limits for when not all of them can be satisfied. Besides handling the process limits properly, the controller should optimize the operation of a milling circuit while controlling product size to a setpoint. The setpoint used for product size itself might need to be adjusted to give a required long-term solids throughput. In autogenous milling, conditions inside the mill might need to be made such that preferential grinding of either the fine material or the coarse ore takes place.

4. Thickener control
The most common process controllers used in thickening are those for thickener underflow density and flocculant addition. The thickener underflow density is controlled to a setpoint by a variable-speed pump. The setpoint can be manipulated in a plant-wide optimization scheme that minimizes gold losses and prevents any undesirable bottlenecks. The flocculent is usually added in ratio to the flow of solids to the thickener. If unknown, this flow of solids can sometimes be estimated from a smart sensor based on measurements of an upstream milling circuit.

5. Carbon-in-pulp and carbon-in-leach control
Carbon-in-pulp (CIP) and carbon-in-leach (CIL) are widely used for the extraction of gold by the use of cyanide. Pulp levels are easily controlled, usually mechanically, by means of overflow weirs. Carbon transfer is done in batches or, in some cases, continuously. The long time constants in CIP and CIL plants make it possible for carbon-related measurements from hand samples to be used. However, cyanide concentrations are best determined and controlled automatically and with minimum measurement delays, as they reflect faster process variations and disturbances, and these need to be reacted to quickly. The carbon and cyanide inventories should be optimized to give the optimum financial performance of the plant as a whole. The associated index of performance should include the income from gold recovered minus the costs of cyanide, carbon and any other reagents used.

6. Flotation control
A flotation plant should have advanced stabilizing control of its pulp levels if it has three or more flotation banks or columns operating in a cascade. This is because conventional PI control operates poorly on the related third- or higher-order dynamics of flotation plants and simply transmits disturbances between stages instead of eliminating them. Where there are cleaner flotation stages, flow rates and residence times
should also be optimized dynamically, while the levels are being controlled.