Economic Data-Driven Framework for Quality Monitoring and Aeration Energy Management in Municipal Wastewater Treatment Plants
Economic Data-Driven Framework for Quality Monitoring and Aeration Energy Management in Municipal Wastewater Treatment Plants
Nadia Mehrabi1 Ameneh Taleei2 Ali Shabani3 Jaseb Zangene4 Hamidreza Ebrahimi5
1) 1. PhD Candidate, Department of Mathematics, Shiraz University of Technology,
2) 2. Assistant Professor, Department of Mathematics, Shiraz University of Technology,
3) Chairman and Managing Director, Fars Water and Wastewater Company (ABFA),
4) Deputy for Wastewater Operation and Development, ABFA,
5) Manager of Energy and Control Systems, ABFA,
IOI Code
:
AAHF-ABHFDE
Publication :
National Conference on Architecture and Art Technologies, Smart City, Soft and Creative Industries and Technological Culture(icact.ir)
Abstract :
This study presents a low-cost, data science-based decision-support framework for energy management and quality monitoring in wastewater treatment plants. The framework relies solely on data already recorded by the plant, equipment electricity consumption, and routine quality tests, and requires no new instrumentation. Using a gradient-boosting regression model with a day-blocked validation scheme to prevent time-series data leakage, two complementary tools were developed on 158 days of operational data from the Kazerun municipal wastewater treatment plant. First, a virtual (soft) COD sensor that continuously estimates effluent COD from electrical data alone enabled quality monitoring with an effective lag of less than one hour, eliminating the need to wait for laboratory results, and achieved an R² of 0.788. Second, an energy-performance analysis comparing actual COD removal (79%) with the minimum removal required under the Iranian Department of Environment s discharge standard for agricultural reuse (ST-04/00, COD limit = 200 mg/L, equivalent to 35.7% removal) identified a 65% aeration energy saving potential with no risk of non-compliance. Furthermore, short-term aeration energy forecasting demonstrated that, due to the plant s highly regular daily pattern, a simple seasonal forecasting method (sMAPE = 3.51%) outperformed deep recurrent models such as LSTM and GRU. These tools, presented as a lightweight digital twin, show how a data-science approach can equip managers of treatment plants lacking advanced sensor networks with smart decision-making capabilities, in line with smart-city goals, and without requiring new hardware. This research exemplifies the application of data science to smart urban resource management: by leveraging existing operational data, it enables quality monitoring and energy reduction in wastewater treatment plants, constituting a practical step toward the sustainable development of critical urban infrastructure.
Keywords : smart city machine learning virtual sensor wastewater treatment aeration energy efficiency data driven governance