Informatics solution for deploying digital twins and MPC in WRRFs

Introduction Mathematical models play a central role in the monitoring and control of water resource recovery facilities (WRRFs), and recent advances in hybrid methods are expanding their capabilities. We present a scalable informatics solution for deploying model-based digital twins, together with a general module for model predictive control (MPC).Methods The architecture is modular, container-based, and designed for real-plant implementation through APIs and interactive dashboards for operator assistance. To demonstrate its applicability, we consider the digital twin of a biological reactor based on the Activated Sludge Model No. 1 (ASM1), along with an MPC algorithm aimed at minimising energy consumption costs.Results The MPC was first tested offline on a generic plant layout based on the Benchmark Simulation Model No. 1 (BSM1) using a reduced-order model, and then online on the digital twin of a real WRRF. In both cases, the controller maintained effluent quality, respected process constraints, and optimised energy-related objectives. However, further tuning of the cost function is needed to reduce excessive variation in the manipulated variables.Conclusion The platform demonstrates a practical framework for integrating real-time data, mechanistic models, and predictive control in WRRFs, combining solid theoretical foundations with an agile and scalable implementation.