Robust Sensor Suite Combined With Predictive Analytics Enabled Anomaly Detection Model for Smart Monitoring of Concrete Sewer Pipe Surface Moisture Conditions

Globally, the water industry considers microbial induced corrosion of concrete sewer pipes as a serious problem. There are reported analytical models and data analytic models that are used to predict the rate of corrosion throughout the sewer network. Those models incorporate surface moisture conditions of concrete sewer pipes as observations. Due to the unavailability of sensors to monitor concrete sewer surface moisture conditions, water utilities use surrogate measures such as relative humidity of the air as an observation for the model. Hence, the corrosion predictions are often hampered and associated with prediction uncertainties. In this context, this paper presents the development and successful evaluation of an electrical resistivity based sensor suite for estimating the surface moisture conditions of concrete sewer pipes. The sensor was deployed inside a municipal sewer pipe of Sydney city, Australia to carry out field measurements. The post-deployment study revealed the survival of the sensing system under hostile sewer conditions and demonstrated their suitability for long-term monitoring inside sewer pipes. Besides sensor development, a predictive analytics model was proposed for anomaly detection. The model incorporates a forecasting approach using a seasonal autoregressive integrated moving average technique for anomaly detection. The model was evaluated using the sensor data and results demonstrated its effective performance. Overall, the proposed sensor suite can ameliorate the way water utilities monitor sewer pipe corrosion.