Por: Leonardo Almeyda Tejada y Leonardo Bravo Thais, ingenieros de Innovación, Nexa Resources. 

Abstract

Knowing the size and type of ore entering a ball mill, as well as controlling the amount of feed entering, is crucial to increase throughput and reduce the specific energy consumed in the grinding process.

Feed to the mill is still controlled manually in many mineral processing operations, however, it is not very effective, since it depends on the perception of the person and does not respond to the immediate variations of the mining activity.

In this paper we present the case of the Cerro Lindo mining unit, where a fuzzy control strategy was developed and implemented using synchronous information captured by intelligent sensors and virtual sensors to optimize the grinding process.

The fuzzy control strategy was subdivided into two stages: (i) control of mill feed tonnage and (ii) control of water entering the mill. To regulate the amount of ore fed to the ball mill, the following variables were used: (i) feed grain size F80 measured synchronously and (ii) discharge pump amperages. Once the feed tonnage is obtained, the fuzzy logic calculates the amount of water to be dosed. For this calculation, the algorithm uses the Passing -3mm variable, which indicates the percentage of fine fraction present in the sample, and from this, the mineral water ratio is calculated.

For the development of the logic, a series of rules were constructed to represent the operational behavior of the Cerro Lindo mining unit. The most important criterion was the size of the ore fed. If the ore entering the mill had a size above 8mm, it was considered coarse ore and the system suggested decreasing the feed tonnage to increase the residence time of the ore inside the mill, thus ensuring a more stable product P80 size. 

On the contrary, if the ore had a particle size greater than 3mm and less than 8mm, it was considered a medium-fine size ore, and the logic indicated to increase the tonnage processed. The pump amperage measurement served as an indicator of the circulating load in the grinding system, and under some scenarios, was instrumental in indicating the decrease in tonnage processed.

The paper also details the tests carried out in the field, where a 12% increase in the amount of ore processed per day was validated by enabling the control logic.

Finally, a proposal to incorporate variables obtained from virtual sensors in the operation of fuzzy logic is detailed. The incorporation of: (i) a virtual sensor to estimate the P80 size of the product and (ii) a geo-metallurgical sensor of the feed is proposed. Both represent an opportunity to guarantee an increase in tonnage by ensuring an optimal particle release size.