SHORT COURSE - Process Control in Mining and Metallurgy

Topics and Presenters:

This two-day short course covers a broad range of available process control technologies for the mineral and metal processing industries. The course is divided into four sections, with each section focused on specific topics which include process control fundamentals, model predictive control, robust and H∞ control, and multivariate methods for analysis, monitoring and control. The four instructors are internationally recognized experts in research and industrial control applications.

Process Control Fundamentals

The course is intended for senior and junior engineers responsible for, or involved in the operation or design of mineral processing plants. The session is especially well suited to those requiring an introduction or refresher on what is now a cornerstone technology. The objective of the course is to provide an overview of topics germane to conventional or basic controls such as process dynamics, conventional control, sampling rates, signal filtering, aliasing, PID Control, PID loop tuning and loop performance monitoring. The participant should expect to gain a broad understanding of modern industrial controls, and to be better prepared to participate in decisions or programs relating to the design and development of effective process control systems.

Model predictive control

This section of the course is intended for automation, operations, and process engineering staff embarking or contemplating the use of Multivariable Predictive Control (MPC) strategies in mining or metals processing plants. The course focuses on fundamental concepts, as well as practical details, of MPC technologies pertinent to mining and metal processes. Due to the wide variety of process and many possible applications of MPC, this course will not focus MPC implementation for a particular process, but will site some real world examples of MPC in mining industry. The practical and real-world fundamentals of MPC will the focus of this short course, and not the mathematical and theoretical basis of MPC.

Robust Control & H∞

This section of the course presents an overview of advanced controller design strategies for multivariable industrial processes, starting from the familiar PID control structure to the more advanced H-infinity design technique. The theory behind new multivariable control algorithms will be presented briefly, as the emphasis is rather being put on design techniques and tools. Simple yet realistic process examples, including a heated water tank and a room heating system, are used to illustrate the controller design techniques. Finally, current technologies for implementing the controllers are discussed.

Learning from Process Data: Multivariate Methods for the Analysis, Monitoring and Control in the Mineral and Metal Processing Industries

This section of the course gives an overview of multivariate statistical methods for extracting information from large process databases and using that information for process analysis, monitoring and optimisation. Latent variable methods such as Principal Component Analysis (PCA) and Projection to Latent Structures (PLS) are introduced as powerful methods for treating these problems. These multivariate statistical methods make use of data on all the measured variables, but they project the information contained in the data down onto a low dimensional latent variable space where one can easily visualize and interpret the behaviour of the process through the use of simple plots. They are now widely used in many industries as the major means for data-mining and analysis. The lectures will present the methods with minimal use of equations and theory, relying instead on providing a good conceptual understanding of the methods and illustrating their potential through numerous industrial applications. The intent of the lectures is to make the attendees aware of the methods and of the current state of their use in industry rather than on the details of the methods themselves. The following areas will be covered using industrial applications to mineral and metallurgical processes as well as other process industries: (i) the analysis of historical databases and trouble-shooting process problems (applications from the chemical industry and on autoclave leaching, and nickel refining); (ii) process monitoring / multivariate SPC (applications at AccelorMittel/Dofasco); (iii) extraction of product quality information from on-line digital images (flotation monitoring - Agnico-Eagle and surface appearance control - DuPont); (iv) optimization and control of processes (applications from the chemical industry). In each of these problems latent variable models provide an important framework on which solutions are based.