Data Science for Water Infrastructure

Problem

The City of Syracuse experiences frequent water main breaks, and the Department of Water spends most of their time responding to emergencies rather than undertaking proactive main repairs. Additionally, when the water main breaks occur, oftentimes the valves to isolate the breaks don’t work. This forces the Department of Water to use valves farther away from breaks to turn off the water, which means that more people are affected by water system failure.

Solution

We used data to develop a model with risk scores for each water main segment. These risk scores help the water department to prioritize work and identify projects to ensure that water system failures cause as little disruption as possible. With this model, Syracuse is now 5 times more likely to identify a main that is likely to break, allowing us to deploy crews to test and repair water main valves and isolate the break to the smallest area possible, thereby reducing the number of people affected by water system failures. This not only allows us to identify priority main segments and coordinate main repairs with other departments and utilities, but also allows us to adjust operations and prioritize other types of work. We have also been able to deploy technology to help identify leaks before they become debilitating breaks.