Review and Upgrade of the Throughput Forecast Model of Minera Los Pelambres
Francisco Abbott1, Liduvina Rodriguez1, Michel Morales1, Walter Valery2, Roberto Valle3, Rodrigo Hayashida4, Benjamin Bonfils2, Kristy-Ann Duffy2 and Claudia Plasencia3
- Minera Los Pelambres, Antofagasta Minerals, Chile
- Hatch, Australia
- Hatch, Peru
- Hatch, Brazil
*This paper was presented at Procemin-Geomet 2022: 18th International Conference on Mineral Processing and Geometallurgy on 5-7 October 2022 Chile.
ABSTRACT
Minera Los Pelambres (MLP) is an open pit mine and processing operation located 200 km north of Santiago in Chile. The processing plant treats 175000 tonnes per day producing both copper and molybdenum concentrates. The comminution circuit consists of two primary crushers followed by three Semi-Autogenous Grinding (SAG) mill - ball mill - pebble crusher (SABC) grinding circuits. A fourth line will be installed in the future.
The MLP drill core database is very comprehensive and detailed for geometallurgical mapping purposes. MLP has developed an empirical throughput forecast model using this database, and the model parameters are ore hardness (JKMRC breakage Axb parameter calculated from SMC test results), SAG mill feed 80 % passing size in mm (F80), and the proportion of hard ore in the blend. The model has been refined over an extended period of time, resulting in mean relative errors of only 3.5 % and 3.2 % on a monthly and annual basis, respectively. Nevertheless, MLP wished to improve the accuracy of the model, particularly over the longer term, thereby further enhancing production forecast and planning for the Life-of-Mine (LOM).
In late 2021, Hatch was engaged to review MLP’s modelling, ore domain definition, ore characterization (rock strength and structure), and blast fragmentation measurements. Current plant operation was also reviewed, the comminution circuit was analyzed, and a new power-based throughput forecast model was developed. To provide a better description of the SAG mill feed size distribution (coarse and fine material in the feed), both F80 and content of fines (% passing 10 mm) were included in the model. The new model resulted in a mean relative error of 3.0 % and 1.4 % on monthly and annual basis, respectively, and is more responsive to variations in ore and operating conditions than the previous model. Other opportunities were also identified that would allow further improvements in the accuracy of the model as well as circuit optimization.