View Photo(s) of Award Presentation
Project Goal
Clean water and sanitation are critical issues for sustainable development globally. The Taipei Water Department, operating under the Taipei City Government, faces significant challenges in ensuring clean water and sanitation for citizens. These challenges are exacerbated by the increased frequency of extreme weather events, which makes water resource scheduling very difficult.Business Problem Solved
Previously, they assessed leakage possibilities based on revenue rates and projected leak points within a 10-thousand square meter area in District Metering Areas.In the new approach, machine learning is employed to determine pipeline leakage probabilities more accurately, resulting in significantly improved outcomes.
In this data flow process, there are four steps:
The first step involves data processing where leak points from the past five years are selected and matched with corresponding pipeline attributes to create training samples.
The second step involves selecting four key factors such as material, age, diameter, and district for use in TensorFlow. A neural network is then chosen to predict water loss probabilities based on these factors.
The third step is using the water loss possibility of variable to calculate the water loss possibility of all pipeline in database.
The final step is publishing the water loss possibility of all pipeline map service in ArcGIS Server.
Technology Implemented
Our user-friendly interface allows you to delineate areas on the map and select a water demand distribution model. The system then exports the model for use in EPANET, where you can conduct hydraulic analysis.
The automated processing for hydraulic modeling handles links, nodes, properties, demand, roughness settings, as well as reservoirs and pumps. The entire model construction process, covering 312 km of pipelines, is completed within twenty minutes, showcasing its efficiency and ease of use.
You can also import and visualize the results within our platform(ArcGIS Server). This includes information such as node pressure, flow rates, and demand. Additionally, WISE allows you to set filters for pipelines and nodes, dynamically adjusting the display based on your selections.
Furthermore, WISE offers predictive capabilities for pollutant migration, projecting outcomes after 6 and 12 hours, providing valuable insights for water quality management.
Development Team Biography
Leading our team is Terry Shiu, who has been an integral part of CECI Engineering Consultants for 18 years and currently serves as the Project Manager in the Department of Geomatics.
Our team has received numerous accolades, including the Taiwan GIS Competition Award four times in eight years and the Smart City & IoT Innovation Award in 2023.