Premium
Machine Learning for Pipe Condition Assessments
Author(s) -
Fitchett James C.,
Karadimitriou Kosmas,
West Zella,
Hughes David M.
Publication year - 2020
Publication title -
journal ‐ american water works association
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.466
H-Index - 74
eISSN - 1551-8833
pISSN - 0003-150X
DOI - 10.1002/awwa.1501
Subject(s) - water pipe , mains electricity , key (lock) , water utility , computer science , selection (genetic algorithm) , electricity , forensic engineering , reliability engineering , water supply , engineering , risk analysis (engineering) , artificial intelligence , computer security , business , mechanical engineering , environmental engineering , electrical engineering , inlet , voltage
Key Takeaways Utilities replace water mains by responding to failures or proactively choosing pipes likely to fail. Machine learning can find fragile pipes more accurately than using age or historical breaks as indicators. More accurate and often less expensive than other condition assessments, machine learning uses hundreds of variables to find patterns most people can't see. Timely selection of the right pipes to inspect, repair, or replace can reduce breaks and optimize the pipes’ remaining useful life.