Dynamic plant-wide optimization of wastewater treatment plant performance using neuro-fuzzy models and predictive control
- 1Faculty of Electrical Engineering, University of Ljubljana, Tržaška cesta 25, Ljubljana, 1000, Slovenia; Corresponding author.
- 2Jožef Stefan Institute, Jamova cesta 39, Ljubljana, 1000, Slovenia
- 3Faculty of Electrical Engineering, University of Ljubljana, Tržaška cesta 25, Ljubljana, 1000, Slovenia
Abstract
Wastewater treatment plants (WWTPs) are complex facilities that must be operated efficiently from both environmental and economic perspectives. They need to balance a range of often conflicting performance objectives, necessitating a plant-wide approach and multi-objective optimization. This paper presents an approach for dynamic plant-wide optimization of the effluent quality index ( E Q I ) and operational cost index ( O C I ) by employing reduced neuro-fuzzy models to represent performance indices, combined with simplified predictive control. The approach is validated through a one-year simulation using the plant-wide benchmark simulation model no. 2 (BSM2PSFe) and by dynamically optimizing the dissolved oxygen (DO) concentration set-point in an aerobic reactor. The Pareto front of optimal solutions is analyzed, and the individual indices are integrated into a single objective function by adjusting the weights of the criteria. The results show that the proposed optimization approach provides a better trade-off between the E Q I and O C I indices than constant DO set-point values. Compared to the DO set-point of 1.5 g/m 3 , the presented optimization case improves E Q I for 2.64% and O C I for 0.87%. The advantage of the approach, apart from plant-wide consideration, is its reliance on measurements available in practice, making it suitable for implementation in real WWTPs.
Related
- Functional Redundancy Enables Stable Treatment Performance Across Taxonomically Diverse Dairy Wastewater Microbiomes.
- Green biosynthesis of multifunctional copper oxide nanoparticles using a leaf extract for potential applications: process optimization, wastewater treatment, phytotoxicity assessment, and antioxidant activity
- Sequential Fenton–ozonation treatment of textile wastewater: treatment performance and operational trade-offs
- Electrode performance in hybrid electrocoagulation and biochar adsorption systems for olive mill wastewater treatment
- Optimization of UV-C irradiation time for biologically treated saline wastewater using artificial neural networks and a hybrid multi-criteria decision-making framework
BibTeX
@article{karer2026dynamic,
title = {{Dynamic plant-wide optimization of wastewater treatment plant performance using neuro-fuzzy models and predictive control}},
author = {Gorazd Karer and Darko Vrečko and Nadja Hvala and Igor Škrjanc},
journal = {Desalination and Water Treatment},
year = {2026},
month = oct,
doi = {10.1016/j.dwt.2026.102008},
url = {https://doi.org/10.1016/j.dwt.2026.102008},
note = {CC BY},
}
RIS
TY - JOUR
TI - Dynamic plant-wide optimization of wastewater treatment plant performance using neuro-fuzzy models and predictive control
AU - Gorazd Karer
AU - Darko Vrečko
AU - Nadja Hvala
AU - Igor Škrjanc
PY - 2026
DA - 2026/10/01
JO - Desalination and Water Treatment
DO - 10.1016/j.dwt.2026.102008
UR - https://doi.org/10.1016/j.dwt.2026.102008
AB - Wastewater treatment plants (WWTPs) are complex facilities that must be operated efficiently from both environmental and economic perspectives. They need to balance a range of often conflicting performance objectives, necessitating a plant-wide approach and multi-objective optimization. This paper presents an approach for dynamic plant-wide optimization of the effluent quality index ( E Q I ) and operational cost index ( O C I ) by employing reduced neuro-fuzzy models to represent performance indices, combined with simplified predictive control. The approach is validated through a one-year simulation using the plant-wide benchmark simulation model no. 2 (BSM2PSFe) and by dynamically optimizing the dissolved oxygen (DO) concentration set-point in an aerobic reactor. The Pareto front of optimal solutions is analyzed, and the individual indices are integrated into a single objective function by adjusting the weights of the criteria. The results show that the proposed optimization approach provides a better trade-off between the E Q I and O C I indices than constant DO set-point values. Compared to the DO set-point of 1.5 g/m 3 , the presented optimization case improves E Q I for 2.64% and O C I for 0.87%. The advantage of the approach, apart from plant-wide consideration, is its reliance on measurements available in practice, making it suitable for implementation in real WWTPs.
SN - 1944-3986
SN - 1944-3994
ER -