Development of performance functions for economic performance assessment of process control systems

Abstract

Economic performance assessment (EPA) of control systems is receiving increasing attention in both academia and industry. It addresses the estimation of the potential benefits resulting from control upgrade projects and monitoring and improvement of economic performance of the control system. Economic performance of control systems can often be related to crucial controlled variables dynamically and when controlled variables move away from an optimal operating point either more profit will be made or more cost will be incurred. This relation can be modelled by performance functions (PFs). When the multivariate nature of a process’s economic model is not considered, PFs of different controlled variables are referred to as individual performance functions. Otherwise, PFs of dependent controlled variables are referred to as joint performance functions. PFs play an important role in the latest techniques of EPA. There appears, however, to be no systematic method for developing PFs. The lack of such a method restrains further research into EPA, as without well-established PFs an EPA cannot be conducted smoothly and therefore cannot effectively support decision-making for management. The development of PFs is a bottleneck in the further research into EPA. Furthermore, the multivariate nature of processes has not been taken into account sufficiently as far as the relevant literature is concerned, which hampers the accuracy of PFs and accordingly the accuracy of economic assessment results. The contributions of this thesis lie in the following aspects: • A methodology for developing PFs is proposed, based on the PF development for an electric arc furnace, a grinding mill circuit and a stage of a bleach plant. • A comprehensive case study of an EPA of three controllers of a grinding mill circuit is conducted using a newly published framework to show the significance of PFs and how to perform an EPA systematically. • The current practice and guidelines on the control and functional/economic performance assessment of grinding mill circuits are captured using a survey study. The multivariate nature of an electric arc furnace’s economic model is investigated and joint performance functions are built based on individual performance functions. A multivariate economic assessment is conducted that shows how joint performance functions can help to provide a more accurate estimate of the economic performance of a controlled process. A web-based survey study on grinding mill circuits in mineral processing industries is conducted. One of its objectives is to obtain general PFs of grinding circuits. The survey results provide instructive insight into the PFs of grinding circuits. Furthermore, an in-depth literature review is conducted and the relationship between the product’s particle size distribution of grinding mill circuits and mineral recovery in downstream flotation circuits is revealed. The PFs of a grinding mill circuit being considered are formed, based on the survey results and literature study. An investigation into the PF development of a stage of a bleach plant is performed and crucial ideas used for their development are abstracted. A methodology for developing PFs for the EPA of control systems is then proposed by synthesising the methods used in the PF development described above. This methodology mainly includes the following stages: Stage 1: Determine information required for PF development. • Process operation and control understanding. • Process economics understanding. Stage 2: Gain required information on PF development. • PF-related information elicitation using survey research. • PF-related information available in the literature, including textbooks, journal papers, conference papers. • PF-related information from plant tests. Stage 3: Obtain suitable performance measures. Stage 4: Make suitable assumptions. Stage 5: Determine PFs. Stage 6: Develop Joint PFs. An economic assessment of three controllers (a nonlinear model predictive controller, a decentralized controller and three single-loop proportional-integral-derivative controllers) of the considered grinding mill circuit is conducted, using an EPA framework published recently to show the central role of PFs in the EPA and how to perform an EPA systematically. The circuit’s PFs, developed as described above, are used for the assessment. The EPA also shows that the improvement in the economic performance with the nonlinear model predictive controller mainly results from the improvement of the operating point and the controlled variables’ variation reduction only contributes a small part to the overall improvement, due to the characteristic of the PF of the circuit’s product particle size distribution. In addition, a web-based survey study is conducted and the current practice and guidelines on the control and functional/economic performance assessment of grinding mill circuits are captured. The questionnaire used for the study includes five segments. The first part identifies the respondents and the second part is intended to obtain background information on the milling circuits. The third part concerns the choice of key process variables and their economic impact. Part four involves the control of milling circuits and control loop performance and part five covers economic issues.