Search

The installation is described of an OCS expert system and three VisioRock online particle size analysis cameras to optimise the performance of the grinding circuit at the Cannington, Queensland, Ag-Pb operation. The main purpose of the system, which incorporates a dynamic SAG/AG mill model, was to stabilise grind size and feed to the flotation circuit. Online vision-based particle size measurements of the AG mill feed and recycle crusher product are integrated into the logic of the control algorithm. Significant stable throughput increases have been achieved, AG mill overload events have been reduced and pebble crusher utilisation has also increased., The installation is described of an OCS expert system and three VisioRock online particle size analysis cameras to optimise the performance of the grinding circuit at the Cannington, Queensland, Ag-Pb operation. The main purpose of the system, which incorporates a dynamic SAG/AG mill model, was to stabilise grind size and feed to the flotation circuit. Online vision-based particle size measurements of the AG mill feed and recycle crusher product are integrated into the logic of the control algorithm. Significant stable throughput increases have been achieved, AG mill overload events have been reduced and pebble crusher utilisation has also increased.

CVRD operates a complex of seven iron ore pelletising plants at Vitoria, Espirito Santo, Brazil, all of which have the same flowsheet, together producing 25 million tons of pellets a year. The automation of these plants is described, comprising instrumentation, control and information systems. The advanced control systems using artificial intelligence, including fuzzy logic, expert systems and dynamic physical models. Implementation of the project, lessons learned and improvements in plant performance and energy consumption achieved are described., CVRD operates a complex of seven iron ore pelletising plants at Vitoria, Espirito Santo, Brazil, all of which have the same flowsheet, together producing 25 million tons of pellets a year. The automation of these plants is described, comprising instrumentation, control and information systems. The advanced control systems using artificial intelligence, including fuzzy logic, expert systems and dynamic physical models. Implementation of the project, lessons learned and improvements in plant performance and energy consumption achieved are described.

A description is given of the advanced control systems installed in two of Anglogold Ashanti's Au ore grinding circuits, with three lines at Mponeng and six lines at Kopanang. The control system is based on the OCS software and combines the flexibility of an expert system with fuzzy logic and the power of an on-line dynamic phenomenological model, used as an advanced soft sensor of the whole process. The history and implementation methodology is described and the resulting increases in throughput and Au recovery are analysed with particular reference to the effects of improved control of particle size., A description is given of the advanced control systems installed in two of Anglogold Ashanti's Au ore grinding circuits, with three lines at Mponeng and six lines at Kopanang. The control system is based on the OCS software and combines the flexibility of an expert system with fuzzy logic and the power of an on-line dynamic phenomenological model, used as an advanced soft sensor of the whole process. The history and implementation methodology is described and the resulting increases in throughput and Au recovery are analysed with particular reference to the effects of improved control of particle size.

The Swedish concentrator, which processes 2 000 t/h of low-grade copper ore, has five grinding lines feeding four bulk flotation banks with a scavenger concentrate regrind mill, and a column cleaner circuit with regrind mill. A model-based, optimising expert system was installed to maximise Cu recovery in the bulk flotation circuit while maintaining an appropriate grade of cleaner feed. The control system uses on-line assays, calculating and implementing set points for cell pulp levels, air flow rates and reagent addition rates. It combines the power of a dynamic population balance model which estimates mineral liberation in each stream, an optimiser which computes appropriate set points and a rule-based module which implements the strategy. There has been a measurable increase in Cu recovery, of around 0.5%, and better control of rougher concentrate grade., The Swedish concentrator, which processes 2 000 t/h of low-grade copper ore, has five grinding lines feeding four bulk flotation banks with a scavenger concentrate regrind mill, and a column cleaner circuit with regrind mill. A model-based, optimising expert system was installed to maximise Cu recovery in the bulk flotation circuit while maintaining an appropriate grade of cleaner feed. The control system uses on-line assays, calculating and implementing set points for cell pulp levels, air flow rates and reagent addition rates. It combines the power of a dynamic population balance model which estimates mineral liberation in each stream, an optimiser which computes appropriate set points and a rule-based module which implements the strategy. There has been a measurable increase in Cu recovery, of around 0.5%, and better control of rougher concentrate grade.

The identification and handling of factors which influence the profitability of an expert system are discussed based on experience at the Troilus concentrator, Quebec. Short- and long-term process control improvements are considered in relation to the steps involved: grinding circuit process and control audit, instrumentation and process control loop, expert system software; expert system integration; performance testing and expert system optimisation and maintenance. Currently the expert system is fully functional for the semi-autogenous grinding circuit and a system for the ball mill circuit is being finalised. Expansion of the system to the flotation circuit is being considered., The identification and handling of factors which influence the profitability of an expert system are discussed based on experience at the Troilus concentrator, Quebec. Short- and long-term process control improvements are considered in relation to the steps involved: grinding circuit process and control audit, instrumentation and process control loop, expert system software; expert system integration; performance testing and expert system optimisation and maintenance. Currently the expert system is fully functional for the semi-autogenous grinding circuit and a system for the ball mill circuit is being finalised. Expansion of the system to the flotation circuit is being considered.