Practical Application Of A Model-Based Digital Twin For Monitoring and Scenario Analysis Of A Water Resource Recovery Facility

1. INTRODUCTION
The effective management of modern water resource recovery facilities (WRRF) requires a holistic approach that considers multiple objectives, such as process efficiency, energy efficiency, resource efficiency and carbon footprint. Recent developments in the field of digitalization have proven instrumental to achieving these goals. Online sensors enable a detailed monitoring of the systems in real time, while mathematical models of the processes enable to create a virtual replica of the WRRFs which can be utilized to evaluate a variety of operational strategies, prior to their actual implementation [1]. Gruppo CAP manages the integrated water service of the Metropolitan City of Milan (Italy) and has historically been an early adopter of digital solutions to manage operations of its water- and wastewater facilities. This study presents the development and practical use of a Digital Twin (DT) Decision Support System (DSS) for the Bresso-Niguarda WRRF.
2. MATERIALS & METHODS
2.1. Description of the WRRF The Bresso-Niguarda WRRF (Figure 1) treats an average 2,550 m3/h flow from two combined sewers in the northern area of Milan, has a design capacity of 250,000 PE and presently treats a daily load of approximately 210,000 PE. The plant includes the following: mechanical preliminary treatment, primary sedimentation, conventional activated sludge process, tertiary treatment, anaerobic digestion and dewatering of sewage sludge. Typical indicators for operational performance are shown in Table 1. Bresso-Niguarda is one of Gruppo CAP's flagship WRRFs where advanced control strategies are implemented to regulate e.g., biomethane separation, sulfur recovery, low-OPEX impact control logics. The plant is also equipped with nearly 100 sensors and energy meters, which provide a detailed insight in its operation in real-time. Bresso-Niguarda was therefore an ideal candidate for the implementation of a DT that would enable the evaluation of operational alternatives to optimize the process efficiency and energy footprint.
2.2 Digital Twin implementation The DT includes three key components (Figure 2): 1) Data layer: a MIKE OPERATION (DHI A/S) workbench was set up to gather data from the SCADA (present on site); generate input files for the WEST (DHI A/S) model; and to retrieve output data from the WEST model to display on the dashboard. Through the SCADA, data are retrieved in real time, quality checked and pre-processed (e.g., anomaly detection and gap filling), from over 60 sensors, including 18 energy meters which monitor the electricity consumption in key sections of the plant.
2) Process model: a comprehensive model of the plant was implemented in WEST (DHI A/S), including the water line, the sludge line and the control layer. The model was pre-configured with input and output files as well as custom key performance indicators (KPIs) as the main interface handles to the data layer and to the dashboard. The modelling approach is described in detail in section 2.3.
3) Dashboard and performance benchmarking: a dedicated, user-friendly dashboard enables operators to modify the boundary conditions for the underlying process model and to assess a variety of scenarios by comparing process indicators and energy consumption time series throughout the plant, such as: treatment efficiency, energy and chemical consumption for water and sludge treatment, biogas production and valorization and sulfur recovery.
2.3 WEST model The model includes the water, the sludge line and the biogas handling line (gasholder, flare, boiler and engine, sulfur recovery). Specific focus was given to (i) the mechanistic modelling of automatic controllers for chemical dosing, aeration (termed 'Liquicontrol'), sludge recirculation and digesters loading, based on Gruppo CAP's specifications; and to (ii) the accurate simulation of the energy balance over the WRRF. The process model chosen for the water line was ASM2dISS, an extension of the Activated Sludge Model no. 2d (ASM2d, [2]) with extra components and processes to handle inorganic suspended solids (ISS) and sulfur. An extension of the Anaerobic Digestion Model no. 1 was used for the sludge line and included biological sulfur reduction (inspired by [3]) and chemical H2S precipitation.
3. RESULTS
3.1. Preliminary model calibration Prior to its implementation as a DT component, the WEST model was calibrated against historical data (both derived from online sensor and laboratory measurements) for different target variables pertaining to solids balances, nutrient removal, biogas production and composition, and energy balance. Figure 3 presents selected results for solids concentration in process tanks, ammonium concentration in the denitrification effluent and final effluent and biogas flow rate produced in the two parallel digesters. Good agreement is shown between simulation results and measurements, which was the precondition to the incorporation of the model into the DT.
3.2. Application of Digital Twin in daily operation
Currently, the DSS is installed on a working station of Niguarda-Bresso WRRF. Operators, process engineers, as well as the energy management team can run simulations and create scenario analyses with 24-h forecast horizon, to assess the process performance under current or modified boundary conditions (e.g., inlet flow rate, pollutant loads, temperature) and operational settings (e.g., controller set-points). There are 7 key-handles available to the user: the oxygen and ammonia set points of the aeration controller; the mixed liquor recirculation set-point; the reagent dosing set-points; the option to vary flow and enabling/disabling certain sections (e.g., one of the primary settlers, dewatering of secondary sludge) in order to reproduce maintenance and future revamping; the sludge loading schedule to the anaerobic digesters; the biogas utilization (biomethane production, thermal valorization). Examples of scenario analyses are shown in Figures 4 and 5. A six hour-rainfall event, resulting in a constant flow of 5,000 m3/h, while only one mixed liquor pump was operated (Q = 1500 m3/h); and a 50% increase in the sludge loading to digester over 6 hours. The scenario analysis shows a temporary reduction in COD, N and P removal and an increase in biogas production and methane content, as compared to the baseline operation.
4. CONCLUSIONS
A Digital Twin solution was implemented on the Niguarda-Bresso WRRF for the model-based evaluation of the activated sludge aeration system, chemical dosage, mixed liquor recirculation, and sludge treatment. The KPIs focus on reducing energy and chemical consumption, increasing biogas recovery, while ensuring a high process efficiency. Following a successful validation period, this first-generation DT will be upgraded to a full-scale tool for monitoring, decision support and early warning using both scenario analysis and optimization. Furthermore, it is envisaged that the future DT may allow for automated transfer of control actions, e.g. optimal set-points derived from model-based evaluations, back to the plant.
ACKNOWLEDGEMENTS
This project is the first example of Digital Twin applied to wastewater sector in Italy and was financed by 'Industry 4.0' Italian Fund. The system has been proposed for the upcoming release of the white paper in SWAN FORUM-The Smart Water Networks Forum (https://www.swan-forum.com/).