👈🏼 Back to list

Natnael Hailu Mamo

Università di Catania
combined sewer
overflow
Artificial Neural Networks
blockage
PHD school
Evaluation and Mitigation of Urban and Land Risks
PhD Cycle
39
List of Supervisors
Rosaria Ester Musumeci, Roberto Gueli
Main research approches
Field-based and/or remote sensing, Numerical analysis
Research abstract
Artificial Neural Network (ANN) based Real Time Water Level Monitoring and Blockage Detection in Combined Sewer Overflow Networks (CSO) in Palermo Province.
Background And Research Gaps
Combined Sewer Systems (CSS) are one of the common types of urban sewer drainage systems across many countries around the world. They are mainly used to convey both wastewater and storm water to the treatment plants. Blockage within a combined sewer system is one of the main concerns in populated areas since it causes overflows and consequently environmental pollution. The effect of blockage may be seen immediately or stays unnoticed for long period until debris deposit led to overflow through manholes. Different hardware and software-based solutions were provided for blockage detection. But hardware-based solutions (CCTV, Acoustic, Laser and Radar scanners) generally are expensive, intrusive, and time-consuming to implement whereas, software-based solutions (physically based numerical models) are data intensive, complex, and require model calibration, which makes them unsuitable for early blockage detection. Pre-detection of blockage within combined sewer systems will help sewer system managers to carry out immediate maintenance operations to minimize the environmental pollution due to overflows. Application of data driven models using Artificial Neural Networks (ANN) is gaining momentum because of the availability of data and the advancement in the technology. Data-driven technologies enables real-time monitoring of the Combined Sewer Networks and detect blockages as soon as they occur, thus allowing utilities to proactively remove obstructions before environment is polluted.
Research Goals
The main goal of the research project, which is co-funded by EHT a national group of companies working in the ICT field, is to develop an ANN model which can identify any anomaly in water level and detect deviation from the normal or predictable behavior within the system. Whenever a significant deviation between actual and predicted water level measurement observed, it will be an indication that blockage event is occurring in the system, and therefore, utility operators will identify it easily and have enough time to act.
Methods
The methods to be used in the research project involves the application of Machine Learning (MN) models to detect anomaly in water level and predict the occurrence of blockage within the system. The model will be trained using precipitation, sewer inflow, and water level data.
Results
Application of data driven blockage prediction method will facilitate a proactive maintenance approach and provide a solution for immediate maintenance operations that will minimize the environmental pollution due to wastewater overflows.