Optimization of SAG Mill Process Control at the Raglan Mine

"The Xstrata Nickel Raglan concentrator, located in Quebec’s Nunavik region, has been in operation since 1997. Originally designed to process 800,000 tonnes per year of high-grade nickel ore, the plant’s current capacity is 1.3 million tonnes per year due to various optimization projects. In 2011 and 2012, the Metallurgical group, with assistance from BBA, updated and optimized the variables affecting SAG throughput in order to optimize process stability while maintaining maximum plant throughput. This project consisted of reviewing the control strategy and completely updating the pre-existing Fuzzy Logic controller. This paper describes the successful update of the controller including a complete revision of control rules, a new process control strategy and the integration of new process variables, most notably mill rotational speed. It will also show how this new approach has provided a simple and robust way of delivering both an increase in throughput while reducing process variability.INTRODUCTIONThe Xstrata Nickel Raglan Mine concentrator is located in Northern Quebec along the 62th parallel. It has been in operation since 1997 and was originally designed to treat 800,000 tonnes of ore annually. The Raglan Agreement held with the Makivik Corporation and the Inuit communities limits the annual tonnage of ore that can be processed at 1.32 million tonnes.The ore is processed in a 24 foot variable speed, semi-autogenous grinding mill (SAG). The conversion from Autogenous Grinding (AG) to SAG was completed in 2006. Before the ore is processed by the SAG, its size is analyzed by a Wipfrag camera that provides a granulometry estimate. A cone crusher reduces the SAG’s discharge screen’s oversize material. A 14 foot diameter ball mill ensures final grinding before the flotation stage.Many control strategies were used at the SAG over the past few years. Fuzzy Logic was first used in 2007 in order to control tonnage feed to the SAG. This first version only measured the value of the variables without taking into account their rate of change. Revisions were made over the course of the last few years to improve controller performance. Other variables were added such as slurry density and crusher Closed Side Setting (CSS). The previous revision of the Fuzzy Logic was done in 2009 with a review of the rules only."