Antti Remes

Data-based skill evaluation of human operators in process industry

The operation of industrial process plants is typically supervised by human operators, even though the routine control tasks may be mainly performed by the automation system. When analyzing the efficiency of the plant it is important to consider also the performance of the operators in addition to the automatic control algorithms. In this paper a systematic data-based approach for comparing the skill levels and working methods of process operators is presented. Real process data from a mineral processing plant is used to evaluate and compare how five human operator groups control the same process. The data is clustered and a self-organizing map is used to detect and visualize the operator-specific behaviour and the differences in their skills. It is shown that even though all the operators have at least some experience of the process, their performance varies notably. The final objective of the research is to improve the operation of the process by utilizing the obtained results in training and coaching of the operators.

Soft-Sensor Estimation of an Apatite Thickener Operation at the Siilinjärvi Concentrator

Abstract In slurry thickening, even and high slurry density is often a key requirement for successful operation of the subsequent process. In many cases, after a concentrate thickener, the slurry is fed to a filtration section. Therefore, disturbances in thickening can lead to increased filtration costs or even reduced plant throughput. Here, a Kalman filter based soft-sensor for an apatite concentrate thickener at the Siilinjarvi concentrator was set up. The aim is to monitor the thickener operation, especially changes in the solids mass in the thickener, and thereby to be able to reduce the number of process disturbance periods. Also, flow rate step experiments were carried out, indicating good operation of the estimator.

Prediction of Concentrate Grade in Industrial Gravity Separation Plant – Comparison of rPLS and Neural Network

Abstract Control of the concentrate quality is usually one of the main targets in the operation of mineral concentrator processes. Availability of the estimates of end product properties in advance – based on upstream process measurements – offers an opportunity to develop higher level control strategies for the unit processes. Here, the recursive PLS and adaptive neural network models are compared in the prediction of the concentrate grade at a gravity separation plant. The methods are applied in the Outokumpu Tornio Works Kemi Mine plant data. The chromite concentrate grade can be predicted relatively accurately based on the slurry properties measured in the grinding circuit. Accordingly, the predicted chromite grade decreases about 0.2 %-units when the slurry D50 passing size is increased by 10 %. This enables further development of the grinding control, especially the control of the slurry particle size, to meet the concentrate specifications.

INTEGRATED APPROACH TO MONITORING AND CONTROL OF MINERAL GRINDING PROCESSES

Abstract To enhance the operation of mineral grinding processes, a greater number of monitoring services and control schemes are nowadays being offered by the equipment manufacturers. In this paper an integrated approach to grinding process monitoring and control is formulated and the components of the integrated automation for typical grinding processes are proposed. Furthermore, the benefits of the process monitoring services are studied on the basis of a specific case study - the Outokumpu Chrome Kemi concentrator. Finally, the results are discussed and a new control scheme is outlined.

Pilot Plant Simulation as a Tool for More Efficient Mineral Processing

Abstract This paper describes a static flow sheet simulation model developed for the pilot scale mineral beneficiation plant taken recently into use at University of Oulu, Finland. The mini-pilot plant aims to serve as a research platform for students and researchers from universities, mining companies, and other organizations in the field of mining and mineral engineering. Currently, the mini-pilot plant is set up according to the concentrator plant at the Pyhasalmi deposit but can be configured for different types of ores and beneficiation processes as well. The simulator is built using the HSC Sim® simulation software. The simulator will support research and development of the new mini-pilot line. It can also be used in process design and optimisation. This paper presents preliminary results demonstrating the promising potential of the simulation model for replicating the pilot process behaviour. The future aim is to convert the static model into dynamic operating mode and study different control scenarios of mineral beneficiation plants using the simulator.

Using a Virtual Flotation Process to Track a Real Flotation Circuit

Abstract This paper presents a novel way of using a dynamic flotation process simulation environment for concurrent simulation alongside a real flotation process. The simulation environment is actively adapted to a real flotation process. This yields new possibilities on the usability of the simulation model. For example, virtual measurements can be introduced that provide completely new variables or increased fault tolerance through redundancy. Another option is to provide feedback for the process operators in the form of textual reports. These reports contain e.g. shift-specific performance indices that can be compared against reference data. Furthermore, by speeding up the simulation, it is possible to make prognosis on the effects of different control actions before actually making them.

Process Operator Skill Exploration by a Data Analysis Software

Abstract Experienced process operators have a key role in ensuring a smooth and high quality operation of process plants in many fields of industry. Consequently, there is a need to identify the skills and knowhow of the operators which leads to high performance, and educate new operators of this knowledge. To this end, an analysis software for comparing the performance of operators in a mineral processing plant has been developed for the plant management. The software implements methods that have been developed earlier for data-based analysis of the performance, and automatizes and streamlines the analysis procedure. The software, which is installed at the process plant for generating the analyses, enables the plant management to monitor the performance of the operator groups at the plant. The analysis results can serve as a starting point for identifying the most efficient operating principles. In addition, an efficient manner for sharing the operating knowhow between the operators is provided. An example case study of using the software is also given.

Economical effects of on-line elemental analysis performance on flotation control

Abstract The paper studies the control performance of the flotation process as a function of the measurement accuracy and sampling frequency of an on-stream XRF analyzer, using a simple dynamic model. An additional aim is to assess how the economical result of a mineral flotation process depends on the quality of process control based on the online elemental assays. The economical performance of the control was studied by calculating the overall net smelter return with various analyzer parameter combinations. The economical performance of the simulated process declines strongly when the sampling cycle is increased.

Particle Size Analyzer – A Case Project of an Extended Product in Mineral Processing

Abstract Process, measurement and on-line analysis equipment are becoming technically more detailed and sophisticated. In many of the solutions information and computer technology are highly integrated in the product’s operating and maintenance procedures. This means new possibilities for the suppliers to provide services and to utilize the concept of an extended product. An embedded automation solution case is presented, and some service concepts for a particle size analyzer are considered.

Analysis of Industrial Grinding Circuits Using PCA, PLS and Neural Networks

Abstract The use of variable rotation speed control in mineral grinding circuits is continuously increasing. High precision on-line particle size distribution measurement of the grinding product has also recently become available. These tools have increased the potential for the development of monitoring and optimization methods for grinding circuits. This paper presents some experiments to evaluate the effect of mill operating variables on the particle size distribution of the circuit product. In the evaluation the PCA, PLS and neural networks methods were used for analyzing industrial experiment data. The experiments have been carried out at the Kemi chromite concentrator plant of Outokumpu Chrome in northern Finland.