Keeping It Simple: A Flexible Low-Code Digital Twin for a Capacity-Constrained Wastewater Facility

[b]Motivation and Background[/b] Wastewater utilities are increasingly embracing digitization as a path to enhance real-time decision making through accelerated access to and analysis of operational data. While this holds the promise of producing operationally relevant visuals and recommendations, there are challenges to adoption. Complex or customized workflows require substantial development and can still break with process changes or data issues. A practical approach to common process control issues which leverages available software would be valuable to the industry. In this project, Trinity River Authority (TRA) and Black & Veatch are piloting a platform centric approach with a low-code emphasis to streamline implementation of process control digital twins (PCDTs). This work began with a digital master plan to determine how to expand on ongoing work to centralize data and develop operational dashboards. Involvement from TRA's IT and operational staff was critical to understanding their digital assets and data management / governance practices. TRA leverages a Water Information System Data Repository to aggregate operational data, along with Process Modeling Simulator software and data visualization tools for dashboarding. Concurrently, plans were underway to implement cloud-based warehousing solutions (now in place) to consolidate all organizational data and enhance availability. The decision was made to pilot a commercially available platform for model deployment and integrate the model derived data into the MS environment for further analysis and visualization to achieve the desired PCDT functionality. Selected model derived data is then fed to a Fabric hosted data lake, which in turn feeds the PowerBI dashboard; the primary interface for operators. Fabric's AutoML features enables low-code/no-code machine learning (ML) operationalization to compliment mechanistic model outputs.

[b]Denton Creek Piloting[/b] TRA's Denton Creek Regional Wastewater System (DCRWS) facility was selected for piloting for several reasons: with capacity limitations and upcoming process changes being the major drivers. The facility is nearing design capacity but expansion projects, including the addition of primary treatment and expansion of secondary basins, won't be completed until 2028. As such the primary aim of the PCDT pilot is to help operators target SRTs and develop peak flow management practices to balance nitrification stability and capacity risks.
• The nitrification safety factor (NSF) is the operating aerobic solids retention time (SRT) over the theoretical washout aerobic SRT. DCRWS must often operate at relatively low NSFs while still maintaining complete nitrification, Calculating the NSF in real time and comparing past NSFs with effluent ammonia over time helps in selecting an appropriate NSF. In this way, adequate but not excessive SRT targets which account for seasonal temperatures changes can be selected.
• The online SUMO model helps to assess data quality associated with parameters needed for the SRT calculation and to troubleshoot associated SRT control issues. Online calculation of a dynamic SRT which factors in daily organic loadings enables proactive SRT control, while maintaining the NSF closer to the target.
• Even at reduced SRT operation, the clarifier solids loading rates (SLRs) are high enough that, depending on sludge volume index (SVI), the sludge blanket (SB) height can approach risky levels. Mechanistic models do not reliably capture clarifier over-loading risks, but an analysis of historical DCRWS data has shown that simple statistical models factoring in settling characteristic, loading, and operation can reasonably predict sludge blanket height at DCRWS. As such, given an SVI a maximum SLR can be set and in turn inform wet weather management decisions or to determine if lower SRT operation is worth the risk of operation at a lower NSF. These data driven models are continually verified and updated using Fabric's AutoML.
• Future flow predictions are made using weather forecasts and ML models. Upcoming wet weather conditions are then evaluated through the SUMO model by embedding standard high flow operating procedures in the model's structure and controls. Ultimately, the goal is to inform operators when equalization basins may overflow and what the impact of step feeding may be on nitrification stability.
• Dissolved oxygen (DO) control is also important to stable nitrification. The aerobic zones are plug flow but only have one DO sensor per train at the end of the zone. The online mechanistic model can provide a soft sensor for DO profile to inform aeration control. For example, how do the DO setpoints and manual airflow control valve positions impact the risk of under-aeration at the head of the aerobic zone.

Digital twin architecture diagrams and example dashboard outputs are attached, though these will evolve over the coming year as new sensors are installed to provide real-time nitrogen and organic loadings and process changes are implemented. Further in early 2026 the facility will be installing Membrane Aerated Biofilm Reactors to manage capacity, and the Digital Twin will guide startup and operation of the system to maximize capacity of the secondary treatment system. Considering these expected changes, the presentation will also cover how this approach is adaptable to process changes and new data availability.