(124e) Improving Operations of Wastewater Treatment Plants Using Economic MPC

Greenhouse gas (GHG) emissions from wastewater treatment plants (WWTPs) have emerged as a critical issue in sustainable environmental management, necessitating advanced modeling and control strategies. Conventional WWTP operation often focuses on meeting effluent quality standards without explicitly addressing the growing concerns of N₂O, CO₂, and other GHG outputs. In response, this study proposes an integrated economic model predictive control (eMPC) framework aimed at mitigating GHG emissions while maintaining or improving treatment performance. We utilize Julia’s ModelingToolkit.jl [1] to develop a modular, equation-oriented flowsheet that accommodates multiple activated sludge models, including ASM1 [2], ASM3 [3], and the GHG-explicit ASMN_G [4]. Because ASMN_G explicitly incorporates GHG pathways, this approach enables direct simulation of N₂O and CO₂ emissions, while maintaining the flexibility to switch to simpler models for other comparative studies.

The modular WWTP flowsheet is carefully designed to capture the key biological, chemical, and physical processes within a typical biological nutrient removal (BNR) unit, clarifier, and associated recirculation loop. By adopting an equation-oriented/acausal modeling paradigm, each unit operation and reaction network is represented through symbolic equations, enabling flexible system composition and reconfiguration. This approach is particularly advantageous when switching among different activated sludge models, as it streamlines testing and analysis of their predictive accuracy under diverse loading conditions and control scenarios. The ASMN_G [4] configuration is employed to represent key aspects of the University of Connecticut’s WWTP, where we calibrate the model with site-specific experimental data to ensure realistic simulations of organic matter degradation, nutrient removal, and GHG emissions such as N₂O. Consequently, the flowsheet serves as a robust and flexible testbed for evaluating advanced control strategies that target both environmental impact and operational efficiency.

A tailored eMPC controller is integrated with the flowsheet to achieve optimization, prioritizing GHG minimization. The controller focuses on regulating the aeration rate, which significantly impacts both energy consumption and nitrogen removal performance. Using the dynamic system model built in ModelingToolkit [1], the eMPC forecasts plant behavior over a finite time horizon and computes optimal aeration setpoints accordingly. While the current implementation focuses on controlling two aerators separately within a single-input single-output architecture, one major advantage of using eMPC is its ability to simultaneously optimize multiple control variables in a multiple-input multiple-output architecture. The modular nature of the framework further allows for future expansion to additional actuators beyond aeration, providing a flexible platform for plant-wide real-time optimization. This setup provides a flexible basis for investigating the interaction between energy use, effluent quality, and greenhouse gas emissions in advanced process control strategies.

The results of this study demonstrate that integrating an eMPC controller with a GHG-explicit process model offers a practical pathway for optimizing WWTP operations toward both environmental and economic goals. By controlling aeration based on predicted plant dynamics, the system achieves meaningful reductions in GHG emissions while maintaining effluent quality and operational stability. The modular structure of the flowsheet also supports reconfiguration for different models, operational settings, and control objectives, making it suitable for broader scenario testing. This flexibility is especially valuable for WWTP operators seeking actionable strategies to improve sustainability within budget constraints. In future work, the system could be extended to incorporate real-time sensor feedback and more adaptive control strategies, enabling better response to varying influent conditions and plant disturbances.

References

[1] Ma, Yingbo, Shashi Gowda, Ranjan Anantharaman, Chris Laughman, Viral Shah, and Chris Rackauckas. "ModelingToolkit: A composable graph transformation system for equation-based modeling." arXiv preprint arXiv:2103.05244 (2021).

[2] Henze, Mogens, CP Leslie Grady Jr, W. Gujer, G. V. R. Marais, and T. Matsuo. "A general model for single-sludge wastewater treatment systems." Water research 21, no. 5 (1987): 505-515.

[3] Gujer, Willi, Mogens Henze, Takahashi Mino, and Mark Van Loosdrecht. "Activated sludge model No. 3." Water science and technology 39, no. 1 (1999): 183-193.

[4] Kim, Dongwook, James D. Bowen, and Ertunga C. Ozelkan. "Optimization of wastewater treatment plant operation for greenhouse gas mitigation." Journal of Environmental Management 163 (2015): 39-48.