Advanced Control of District Heating Processes in Estonia

Abstract

The current thesis considers the problem of implementation of model based control to real industrial processes. It covers theoretical preliminaries for building model predictive controller, its development as a stand-alone software and successful implementation to heating process in the combined heat power plant located in Tallinn, Estonia. Before considering model based control an unmodeled advanced process control was considered to show that MPC is not able to control processes with unmodeled dynamics, but there are also other techniques that can handle such cases. Fuzzy controller implementation to the boiler house located in Rapla town, Estonia was considered as an example of such solution.

Overall the thesis covers all stages of MPC implementation starting with process study, MPC design, testing in simulation environment (specially designed for our case in real DCS software), implementation to the real process and further analysis of results. Implementation was successful showing significant improvement in process control comparing to previously used solution based on cascade PI controller. MPC controller was used and tested in the power plant during heating season of years 2017/2018. Process owner is satisfied with achieved results and is going to use the controller in the future as well.

Related publications

• B. Godoy, V. Vansovits, A. Tepljakov, E. Petlenkov, and K. Vassiljeva, “Control-oriented identification using maximum likelihood for a district heating plant,” in 2017 11th Asian Control Conference (ASCC). IEEE, 2017, pp. 2720–2725. [DOI] [Bib]
• V. Vansovits, B. Godoy, A. Tepljakov, K. Vassiljeva, and E. Petlenkov, “Model-based control design for a district heating plant,” in 2017 IEEE 15th International Conference on Industrial Informatics (INDIN). IEEE, 2017, pp. 615–620. [DOI] [Bib]