IntelliSense.io Case Study Thickener Circuit Optimisation

Transcription

1 IntelliSense.io Case Study Thickener Circuit Optimisation The IntelliSense.io Thickener Circuit Optimisation application was implemented at one of the largest copper gold mines in Chile. The mine belongs to a London listed mining company and is considered as one of the top copper producers in the world. Project Scope - Thickener Circuit Optimisation Stability Project Target System The Thickener Circuit of High Density and Paste Thickeners

2 Project Hypothesis The existing thickener circuit Expert System didn t enable continuous set point changes based on the type of materials entering the thickener resulting in inability of the operations team to take pre-emptive action to minimise variance at the circuit, with action being taken only after the event. This resulted in reduced underflow % solid and water recovery at the circuit. The instability problem would have been exacerbated in the medium term due to the planned increase in tailings volume, the addition of planned increase in paste thickeners, the environmental commitment provided by the mine to the local authorities and the continuing dynamic nature of the material composition from the mine. Project Scope Challenges at the Thickener Circuit Increased Feed Mineralogy Variability Low Underflow % Solids and Water Recovery High Flocculant Consumption The scope of the project for IntelliSense.io was to model and optimise the thickener circuit in real time, identify the root cause of variability and continuously predict set-point recommendations for optimal predictive process control. The project integrated data from SCADA & other control systems (including upstream data) with advanced statistical data modelling, machine learning algorithms and first principle models to derive a digital model of the thickener circuit that can predict and simulate future performance of the circuit under various feed conditions and deliver continuous optimised control recommendations that result in; Delivery of predicted material composition and mineralogy input to the thickener circuit, Stable underflow % solid Online thickener circuit simulator. Deployed Solution The overall technical solution to be deployed has been described below. The key components of the technical solutions are: Brains WSN - SCADA Integration Module to various data sources (Upstream & Circuit) Brains.app - Real-time performance insights and predictions at equipment and system level Brains VOS (Virtual Optimisation Simulator) Diagnostics and Operator Training

3 Brains WSN (Wireless Sensor Network) In order to develop an understanding of the data relationships between the output thickened tailings and the thickener circuit operations, we collected data across from the areas outlined in the diagram below with the Brains WSN (Integration Module) used to collect, process and transmit data from existing SCADA systems continuously. Brains.app Used as an analytics and decision support tool for real time diagnostics of Thickener circuit performance, predict future performance and simulate impact with daily reporting features. Predictions are driven from a library of equipment and process models using advanced statistical modelling real time (a combination of generic black box and first principle/equipment models; real time learning models and modern bayesean models). Models are updated/upgraded regularly to adapt to specific system and any system changes e.g. site modifications. The app analyses variations in geometallurgy and upstream processes to recommend optimal operational and economical thickener circuit set points delivering 3 control variables: Flocculant flow rate, Flocculant dilution flow rate & Underflow flow rate. With self-service dashboards and reporting allowing different types of users (operations, engineers & management) to source information based on their needs. A on premise version of the application is deployed to deliver optimisation set points continuously to existing expert / control system.

4 Brains VOS (Virtual Optimisation Simulator) A simulation environment was created through the brains.app that can be used to test various configuration changes with their expected outputs. The offline environment that used real data to train and understand "what if" scenarios & undertake diagnostics. Preliminary Findings - System Bottlenecks Preliminary Findings Incorrect Mass Balance Operational Limitations (Torque) Light feed material Homogenous feed particle size Physical system constraints (e.g. Flocculant, dilution flowrate restrictions) Ineffective Flocculant type Vs. Feed Material Type Volumetric Capacity Ineffective Loop Controller Short-circuiting Imbalanced Material Distribution Economic/Performance Balance

5 Sample Shown Below Impact of Findings The key impact from our technology has been identified across 3 areas: Reduce variance in underflow % solids Reduced variance in bed pressure Reduced variance in torque Summary Output Main Benefits: Increase in Water Recovery Decrease Flocculant Consumption Reduce Environmental OPEX based technology with No Capex Payback period of less than 12 months with projected direct savings calculated at $400k in the first year alone.