Climate change has been driving the rise of extreme weather conditions in recent years, placing immense strain on urban infrastructure such as sewer systems. Compromised sewer systems can lead to leakage, overflow and even flash flooding, threatening public health and safety. To address these vulnerabilities, a research team at The Hong Kong Polytechnic University (PolyU) has developed a multi-tiered model integrating artificial intelligence (AI) and the Internet of Things (IoT), facilitating more cost-effective and intelligent sewer system management, ranging from predicting exfiltration severity and pinpointing leakage-prone zones to monitoring and forecasting overflow occurrences in high-risk areas.
Currently, closed-circuit television (CCTV) is widely used to assess pipeline conditions, but this method relies heavily on the expertise of trained inspectors; applying it to the entire sewer network is costly and time-consuming. Prof. Tarek ZAYED, Professor of the PolyU Department of Building and Real Estate, led his team in developing a smart management model that leverages deep learning algorithms to analyse sewer pipeline conditions with high accuracy. Utilising these algorithms helps in spotting ageing and defective parts to prioritise locations for CCTV inspections.
"In addition to extreme weather, population growth and structural ageing exert pressure on urban drainage," Prof. Zayed explained. "By adopting a risk-based inspection and maintenance strategy, our AI model is expected to reduce the overall time spent on sewer inspection and maintenance activities by about 25% to 30%. This is primarily achieved by deploying CCTV more precisely and reducing redundant site visits."
A core component of the system is the team's pioneering Exfiltration Severity Index (ESI) to quantify and model pipe-level exfiltration severity, allowing users to identify leakage-prone segments beforehand. "Sewer exfiltration occurs when sewage escapes from a defective sewer system into the surrounding environment. This can contaminate soil or groundwater with pollutants, such as pathogens and toxic compounds, harming both ecosystems and public health," he added.
The AI model incorporates an array of factors, including pipe characteristics, climatic conditions and environmental impacts, to predict the likelihood and severity of exfiltration, providing insights for prioritising more urgent maintenance activities. The team's research showed that the system achieved an 85% accuracy rate in severity assessment, significantly mitigating the risk of groundwater contamination. Furthermore, through optimised maintenance scheduling, the predictive model improved operational efficiency by 50% to 60% and reduced emergency repairs by 30% to 40%. The team's findings have been published in a paper, "Proactive Exfiltration Severity Management in Sewer Networks: A Hyperparameter Optimization for Two-tiered Machine Learning Prediction", in the journal Tunnelling and Underground Space Technology.
In addition to leakage, blockage is another cause of operational failure of sewer systems and even flooding in more severe cases. To tackle this problem, Prof. Zayed's team applies IoT-based technologies to simulate the performance of sewer networks and overflow occurrences under various blockage scenarios.
In collaboration with the Drainage Services Department of the Hong Kong Special Administrative Region (HKSAR) Government, the team installed water-level sensors across drainage networks in Kowloon district, collecting real-world data with various data mining techniques for case study simulations as well as for model calibration and validation. The IoT-based monitoring system yielded remarkable results, where sewer segments found with blockage issues saw an 85% improvement in their overall performance following targeted cleaning procedures.
"Utilising AI and IoT technologies, our multi-tier approach provides a reliable decision-making tool to predict the location and timing of potential sewer failures. It offers valuable insights for effective actions to reduce the number, magnitude and severity of sewer overflow events," Prof. Zayed concluded. "Overall, our innovative system successfully reduced emergency overflow incidents by 70% to 75% in the monitored areas, thanks to the rapid response enabled by timely and effective monitoring."
A related study, "Performance Assessment of Sewer Networks under Different Blockage Situations Using Internet-of-things-based Technologies", is published in the journal Sustainability. The research was supported by the Research Grants Council's General Research Fund, as well as the Environment and Conservation Fund.
