A Practical Sewer Network Digital Twin - It is not a PIPE dream

Most major municipal areas in the United States are served by a combination of sanitary sewers, separate storm sewers, and combined sewers. With growing populations across urban areas, many sewer utilities are struggling to manage stormwater-related challenges such as flash flooding and combined sewer overflow (CSOs). Furthermore, inflow and infiltration (I&I) issues can cause damage to collection systems and increase the peak flow to wastewater treatment facilities leading to untreated wastewater and storm water being released to the environment including local waterways. This could result in their impairments and risks to public health. Several states have introduced public notification requirements to promote public awareness of sewage pollution with very short timeframes (within hours) for reporting. In order to achieve desired outcomes and get a real-time view of the collection systems, many utilities have invested in sensors and hydraulic models, but it remains a challenge to utilize these investments to comply with the legislation. Proactive utilities are looking to a digital-first resilient strategy which, not only allows them to comply with regulations, but also helps assess system performance, determine effective control strategies and streamline their operations and maintenance activities. In this presentation, we will provide examples of sewer network digital twin (DT) (Figure 1) which brings in data from multiple sources (SCADA, IoT sensors, historical rainfall and weather forecasts, GIS, etc.) and uses hydraulic models and data analytics to perform anomaly detection, compare sensor and model generated information (Figure 2), forecast overflow events based on weather forecasts, comply with local regulations for public notifications and regulatory reporting, and streamline maintenance activities. This implementation of a sewer network DT platform has enabled the validation of false positive overflow readings through tying in maintenance activity data from CMMS, correct false positive notifications from flow meters by using a hydraulic model to validate overflows and provides a single dashboard combining all relevant data. The platform also uses machine learning to perform sensor data anomaly detection for short-term anomalies such as sensor drop-outs and maintenance-related issues and long-term anomalies such as sensor drift (Figure 3). The platform is able to use hydraulic models and weather forecasts to provide overflow forecasts enabling the utility to perform proactive maintenance and planning. Public notifications can be automated through direct integration to the city's website and with a reverse 911 system which includes multilingual support. Furthermore, dynamic real-time inflow and infiltration estimation enables prioritization of pipe rehab and maintenance by targeting problem areas. By providing a complete real-time and retrospective view of the sewer system, the sewer network DT can enable utilities to characterize the performance of their sewer network, prioritize maintenance and remediation needs and streamline regulatory reporting in one single platform.