DeSw Project Results on Development of Soft Sensors in Water Distribution Networks

Abstract

Reliable drinking-water operation requires more than periodic compliance sampling; it requires timely knowledge of hydraulic conditions, residual disinfectant, and contaminant-related quality variables throughout the distribution network. This article presents the results of the research project “Development of Soft Sensors in Water Distribution Networks” for real-time estimation and Discrete Event Systems based decision support. The approach combines a coupled hydraulic–water-quality model with a bank of linear state observers that operate as soft sensors. Each observer is designed for a local operating region, while a switching mechanism selects the observer most suitable for the current state of the network. The methodology targets variables whose direct measurement is difficult or spatially sparse. The soft-sensing layer is integrated with supervisory control and fault diagnosis units. The resulting architecture links field instrumentation, edge computing, and cloud-based reasoning through standard SCADA systems. The project demonstrates a scalable, interpretable, and computationally light alternative to dense deployment of physical sensors, while preserving compatibility with regulatory monitoring and human operator oversight.