Development and Calibration of an Autogenous/Semi-Autogenous Mill Simulation Model for Process Control Applications

Additonal authors: Bouchard, J.. Book title: Proceedings of the 58th Conference of Metallurgists Hosting Copper 2019. Chapter: . Chapter title:

Proceedings, Vol. Proceedings of the 58th Conference of Metallurgists Hosting Copper 2019, 2019

Pérez-García, E. M.

Process control and optimization have always been crucial aspects in the mineral processing industry, and expectations constantly increase to cope with more stringent security, quality, efficiency, and environmental performances. The main challenges to reduce variability of key process variables, and steer profits to higher values while maintaining sustainability indexes within acceptable ranges come from nonlinear responses and circulating loads. The most commonly implemented control strategies rely on simple PID controllers for practical implementation and maintenance considerations. If arguably good performances are accomplished when these are adequately set up, there is also room for further gains by using more advanced methods such as model-based techniques. There is thus a need for more realistic process and equipment models. Based on advances of the past few decades, this paper introduces an autogenous/semi-autogenous mill (AG/SAG) model to be included in a modular comminution circuit simulation toolbox comprising ball/rod mills, material handling, and classification devices. The proposed mill model incorporates a modified version of the classic comminution equation that makes grinding rates proportional to the energy input. It requires the determination of a selection function (fragmentation rates) and a breaking function (fragments distribution). The former is modelled with an S-shaped curve characteristic of AG/SAG processes and the latter is a size-relative distribution of the produced fragments. The power draw is predicted employing the empirical equation of Austin. The paper also develops a case study to illustrate how the proposed simulation framework can capture the behaviour of an actual plant. Model calibration consists in an optimization problem that search the global minimum of the difference between predicted and measured/estimated size distributions, power draw, solid fractions, etc. Results demonstrate that the mill simulator can mimic an actual process in its operation conditions. INTRODUCTION Comminution is a key stage in the mineral processing industry, obviously because the phases of interest must be liberated to be able to recover them. In general terms, the global performance of a plant depends largely on the efficiency of the grinding circuit and on the material properties such as its grindability, density, and mineral grades. The subject has vastly been studied since as early as the 19th century, when Rittinger (1867) and Kick (1883) developed some of the first empirical models to relate grinding and energy input. The great interest in the subject comes from the fact that this operation is costly and, probably most concerning, highly inefficient. Buckingham et al. (2011) estimated that ore comminution can represent up to 60% of the overall plant power draw. Moreover, only a modest fraction of the power draw effectively translates into rock fragmentation. For grinding, estimates range between 1% and 25%, depending on the reference used, while the rest is mainly lost as heath (Fuerstenau & Abouzeid, 2002; Tromans & Meech, 2002; Bouchard et al., 2017).
Keywords: Copper 2019, COM2019