CASE STUDY
Greater Jonestown Water Authority
Using Machine Learning To Improve The Effectiveness of Leak Detection
Background
The Greater Johnstown Water Authority (GJWA) in Pennsylvania was incorporated under the Municipality Authorities Act in 1964 by the city of Johnstown and boroughs of Westmont and Southmont as a joint municipal authority to provide potable water to the Greater Johnstown area. GJWA serves a base of more than 21,000 customers through 303 miles of water distribution pipes
Objective
When GJWA’s pumping station signals a potential water main leak, Assistant Public Work Director Jim Kukura uses the Fracta platform to search the distribution network and discover which pipe segments are most likely contributing to the NRW loss. Fracta’s holistic, data-driven approach provides Jim with a more granular, nuanced view of where water loss is likely coming from. This is done well before time and money are spent to send the crew into the field to validate the predication.
Results
Within the first seven to eight months, GJWA relied on Fracta’s Likelihood of Failure (LOF) to correctly identify 75% of the leaks in 15 pipe segments, saving the utility 20% in NRW loss.
Services
- Data Cleaning
- Machine Learning Pipe Analysis
- Predictive Modeling
- Improved Operational Planning