Abstract
Full scale wastewater treatment plants (WWTPs) generate extensive sensor and laboratory datasets that remain challenging to translate into real-time operational insight, while increasing hydraulic variability, energy constraints, tightening effluent regulations, and events such as sensor faults, hydraulic shocks, and effluent-quality excursions require predictive decision-support tools beyond conventional mechanistic modelling. This study presents a full-scale, industry-deployed artificial intelligence (AI) framework integrating machine learning (ML), simulation, and operational boundary definition across a nitrogen-removal WWTP in Australia. Up to nine years of high-frequency online instrumentation data (≤419,000 records per target; 5–30 min resolution) and ~3000 laboratory samples were used to train and evaluate 19 ML algorithms across 13 hydraulic, energy, pressure, and effluent-quality targets across full-scale WWTP. Extra Trees (ExT) and deep neural networks (NN) demonstrated superior performance. Deterministic operational variables, including plant inflow, aeration flow, blower energy, effluent pump power and speed, and ocean discharge pressure, were predicted with R2 ~ 0.96–0.99 on validation and unseen datasets. Biological and laboratory-derived parameters achieved R2 ~ 0.86–0.92, reflecting intrinsic process variability. NN models enabled extrapolative simulation beyond historical operating ranges. The models generated and evaluated millions of feature combinations to define safe operational boundary conditions associated with screen bypass, overpressure risk, energy overuse, and effluent non-compliance. The framework enabled dynamic optimisation of aeration and pumping configurations, improved energy efficiency, and provided robust ML-based soft sensors during instrumentation faults. This work establishes a scalable pathway for embedding AI into real-world WWTP engineering infrastructure, measurable improvements in predictive accuracy, operational resilience, and risk-aware optimisation.
Keywords
artificial intelligence, energy optimisation, machine learning, predictive modelling and simulation, soft sensors, wastewater treatment
Document Type
Journal Article
Date of Publication
7-1-2026
Article Number
110361
Volume
89
Publication Title
Journal of Water Process Engineering
Publisher
Elsevier
School
School of Engineering
RAS ID
99975
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Hassnain, M., Veal, C., Lee, S. M., & Azhar, M. R. (2026). Artificial intelligence based predictive simulation and decision-support framework for full scale wastewater treatment systems. Journal of Water Process Engineering, 89, 110361. https://doi.org/10.1016/j.jwpe.2026.110361