Feedforward-feedback Mechanistic Ammonium-based Aeration Control optimization at Blue Plains advanced wastewater treatment plant

1.Introduction Ammonia-Based Aeration Control (ABAC) is widely used to reduce aeration energy demand in biological nutrient removal processes1,2. Conventional ABAC systems rely primarily on downstream ammonium feedback to trim dissolved oxygen (DO) setpoints, which results in delayed responses due to long hydraulic retention times and limits performance under dynamic loading conditions3. To address these challenges, black-box feedforward–feedback (FF–FB) ABAC systems have been implemented at Blue Plains since 2015, incorporating influent flow information to improve response to diurnal variability. However, these empirical approaches remain sensitive to wet-weather events and do not explicitly represent nitrification kinetics or the nonlinear relationship between DO and microbial activity4,5. To overcome these limitations, a mechanistic FF–FB ABAC strategy was developed based on linearization of the Monod DO term through a nitrification factor (N.F = DO/(K_DO + DO))6. This approach enables a more proportional and robust linkage between control actions and biological nitrification capacity. This study compares the two ABAC strategies and documents full-scale implementation, tuning, and performance outcomes. 2.Methodology This study evaluated two ABAC implementations in a full-scale nitrification system from April 1, 2024, to November 1, 2025. Black-box ABAC was operated from April 2024 to June 2025, followed by mechanistic ABAC from June to November 2025. System performance was assessed using Average Absolute Error (AAE), Average Percent Absolute Error, and the percentage of time within allowable thresholds for dissolved oxygen, airflow, and ammonium concentrations7. 2.1 Black Box feedforward-feedback ABAC: data-driven system that modulates the DO setpoint to achieve the target ammonia concentration in a feedforward action. The feedforward component utilizes a model to predict the required DO setpoint based on influent flow rates and empirical DO, flow – NH4 relationships. A nonlinear feedback loop adjusts the DO setpoint based on the error between the online ammonium concentration and the desired setpoint, Fig1A. 2.2 Mechanistic feedforward-feedback ABAC: model-based control strategy designed to linearize the relationship between the output and the nitrification factor (N.F), which represents the DO Monod terms within nitrification kinetics: DO/ (KDO + DO)6. The feedforward model calculates the required N.F using influent ammonia loads, nitrification kinetics, and an estimated mass of active nitrifiers. The feedback action corrects this N.F. fraction to meet NH4 setpoint and then the N.F, fraction is converted to DO setpoint to further control airflow rates and valve openings, Fig1B. 3.Results 3.1 Process optimization of the new ABAC Air-Valve Response Optimization: Prior to optimization, air-valve response times exhibited a 15–20 min delay between airflow setpoint changes and valve actuation, Fig2A. This mechanical lag prevented timely achievement of DO setpoints during rapid load transitions, resulting in ammonium deviations and oscillatory control behavior. By removing deadbands and delays in the airflow control loops, response times were reduced to 3–5 min, Fig2B, enabling rapid tracking of airflow setpoints, Fig2C and substantially improving DO controllability. Optimization of N.F Bounds: Initial mechanistic control constrained the N.F to a narrow range 0.55–0.88, corresponding to DO values of 0.48–2.93 mg O₂/L. Under low-load conditions, this limited flexibility resulted in excess aeration and suppressed ammonium concentrations. Reducing the lower N.F bound to 0.30 expanded the controllable DO range and mitigated over-aeration, Fig3. This adjustment reduced saturation at the upper N.F limit, enabled smoother transitions across diurnal loading conditions, and improved overall control stability. Integration of SRT adjustments: Early mechanistic runs revealed discrepancies between modeled and actual nitrifier inventories, leading to under-aeration during high-load morning periods. Process data indicated overestimation of nitrifier mass, prompting an update of the assumed aerobic SRT from 15 to 11 days, Fig3. This change improved alignment between modeled and real nitrification capacity, resulting in more accurate feedforward DO predictions and fewer ammonium excursions. Feedback Tuning: Initial PID feedback tuning resulted in conservative responses to ammonium deviations, limiting the benefits of the mechanistic feedforward layer, Fig4A. To enhance responsiveness, the lambda parameter was reduced from 3.0 to 2.5, proportional gain increased (0.79 to 1.06), and integral time decreased (2314 to 1543). Post-tuning performance showed faster correction of NH₄ deviations without inducing instability or collapse of the N.F term during load spikes, Fig4B. 3.2 Performance comparsions Table 1 shows that mechanistic ABAC outperformed the black-box approach, achieving lower AAE and maintaining ammonium within target limits for a greater fraction of time (84%). These improvements are attributed to enhanced feedforward predictions that explicitly incorporate ammonia loading and estimated nitrifier mass, allowing the controller to better anticipate oxygen demand under operating conditions. Conclusion: Transitioning from black-box to mechanistic ABAC improved ammonia control stability by explicitly incorporating nitrification kinetics, while indicating the need for further refinement.