Opti-SENSE: Optimal Placement of Sensors in Stormwater and Wastewater Networks

The project optimised the placement of sensors in stormwater and wastewater networks, with a focus on improving the monitoring of water flows.

Initially, a literature review was conducted to examine existing sensor placement techniques. Based on this review, a methodology was developed for placing level sensors within NSVA’s network. The pilot project evaluated the performance of the methodology with the aim of establishing a proof of concept ahead of the deployment of 1,000 sensors between 2024 and 2026. By optimising sensor placement, water flows could be monitored effectively using fewer sensors.

Level Sensors and IoT

Traditional flow measurement is expensive and resource-intensive. By using lower-cost level sensors and Internet of Things (IoT) technology, water flows can be monitored more cost-effectively. Through the strategic placement of sensors, we gained a better understanding of the system’s hydraulics, leading to reduced maintenance costs and improved operational performance. Early in the project, NSVA observed promising results from initial trials, where level sensors successfully identified issues such as infiltration and inflow.

With the ongoing digitalisation of the water sector and increasing requirements for monitoring, this project was particularly important. By using level sensors connected via IoT, we were able to continuously collect and analyse data to improve operations and planning. The outcomes of the project are beneficial for water and wastewater utilities across the country.

Key outcomes of Opti-SENSE

1. Mathematical models of water levels in stormwater manholes formulated in state-space form provide highly valuable information for the implementation of digital twins. To date, these formulations have primarily been available in open-source software, such as the Pipedream Solver package developed in the Python programming language.
2. We have developed a framework for optimising the placement of a specified number of level sensors within a stormwater network, enabling us to estimate water levels in manholes without sensors with a high degree of accuracy.
3. By allowing the number of level sensors in the pipe network to be selected when applying our framework, we can identify the minimum number of sensors required to achieve good estimation performance, thereby minimising both capital and maintenance costs.
4. The framework was applied in an initial digital case study, in which different rainfall scenarios were simulated in a small stormwater network comprising 35 manholes. The results were promising, and further studies involving larger networks, a wider range of rainfall scenarios, and real-world water level data are recommended.