Predicting Pipe Breaks: Desktop Scoring, Advanced Statistics (LEYP), and Machine Learning

Key Takeaways Using desktop scoring to determine the likelihood of failure of water pipes should be phased out and replaced with advanced analytics, particularly machine learning. More accurate break predictions will lead to better estimates of how much to spend on pipe replacement and which pipes t...

Full description

Saved in:
Bibliographic Details
Published in:Journal - American Water Works Association 2023-05, Vol.115 (4), p.42-53
Main Authors: Raven, Annie Vanrenterghem, Campanella, Kevin V.
Format: Article
Language:English
Subjects:
Costs
Forecasting
Leak detection
Machine learning
Modeling
Pipelines
Pipes
Rehabilitation
Statistical methods
Statistics
Utilities
Water mains
Water utilities
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Key Takeaways Using desktop scoring to determine the likelihood of failure of water pipes should be phased out and replaced with advanced analytics, particularly machine learning. More accurate break predictions will lead to better estimates of how much to spend on pipe replacement and which pipes to replace. Utilities should integrate information about abandoned pipes and breaks into failure forecasting, including machine learning models. Some data issues can be cost‐efficiently rectified using machine learning and automated algorithms.
ISSN:0003-150X
1551-8833
DOI:10.1002/awwa.2089