Advanced Process Control Applications in Complex Industrial Processes

Reitano, Kieran (2021) Advanced Process Control Applications in Complex Industrial Processes. [USQ Project]

Text (Project)
REITANO Kieran dissertation_redacted.pdf

Download (17MB) | Preview


The purpose of this dissertation was to determine the feasibility of Advanced Process Control (APC) techniques on complex industrial processes. This research is primarily motivated by the fact that limited application of APC has penetrated the industry in recent years, with an estimated 90% of control solutions using PID controllers (Copot, et al., 2018). The research objective was to conduct a case-study on a complex sugar industry plant, involving mathematical and logical modelling in appropriate software, and implementation and analysis of conventional and advanced process controllers. It was expected that the case-study would provide evidence on the required expertise for industry APC implementation and provide a foundation for companies to pursue applications were they are required and feasible.

Initial research conducted firstly involved explicitly identifying the definitions and characteristics of complex industrial plants, as well as methods of analysis. A large focus was put on the consequences of dead-time due to its prevalence in the case-study industry. Research was then conducted on the conventional PID algorithm and its applications for dead-time, however, due to the selected scope of methodology, tuning strategies were not reviewed. Finally, APC techniques, including the Smith Predictor, Internal Model Control and Model Predictive Control were overviewed, with the primary focus on the Smith Predictor as it is directly used in the Methodology and Results. Internal Model Control and Model Predictive Control are not covered in detail in this dissertation. Further industry research was finally conducted into the case-study, including process control audits of sugar control systems, modelling techniques for thermodynamic processes and APC implementations on similar processes.

Model development involved the design of a Simulink® based real-time construction of a Multiple Effect Evaporator set used to concentrate cane juice in the sugar industry. The model is based on industry correlated data, primary principles and logic modelling, using the physical plant for validation. The model was found to aptly simulate the most dominant transients and steady-state behaviours observed in the target system, making it appropriate for further use.

The control theory methodology used was to implement a conventional PI controller in the same configuration as the case-studied plant to benchmark performance of non-APC solutions. Optimal results were found using iterative tuning over a large range of parameters and mapping cost function results to identify a minimum. The candidate APC solution, being the Smith Predictor, was then implemented, showing a best case improvement of almost 100% in error performance while maintaining similar levels of stability. Rigorous stress testing then took place to determine the performance of the Smith Predictor when exposed to disturbances, parameter variance and model inaccuracies. The results showed the Smith Predictor still outperformed the PI controller in approximately 93% of cases. Furthermore, the locality of optimum tuning was extremely broad for the APC solution, whereas the PI controller requires tuning to be within a very fine range to result in acceptable performance.

Final investigation of the results showed that the study proved APC’s ability to compensate for deadtime in complex industrial processes, however, further work is necessary in order to allow for research application to be expanded. The current contribution is that PI controllers are not as suitable as APC solutions for complex processes as defined in this report, however, they are much simpler, requiring minimal expertise or capital expenditure.

Statistics for USQ ePrint 51833
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021)
Supervisors: Hills, Catherine
Qualification: Bachelor of Engineering (Honours) (Electrical and Electronic)
Date Deposited: 03 Jan 2023 04:05
Last Modified: 26 Jun 2023 01:50
Uncontrolled Keywords: advanced process control, APC, industrial, control, dead-time, smith predictor, internal model control, model predictive control, Simulink

Actions (login required)

View Item Archive Repository Staff Only