The nD platform was used to assist a U.S. water utility and its engineering firm to pinpoint and diagnose non-revenue water issues in its drinking water system. Non-revenue water is drinking water that was treated and distributed by the utility but not billed to a customer, resulting in lost revenue. Typical reasons include water loss due to leakage during distribution, unauthorized consumption from meter tampering and meter reading errors.
Solution: nD machine learning was used to augment the utility’s standard audit process. nD was first tasked to identify anomalies across more than 180,000 meters with multiple readings per day and 15 degrees of freedom per meter. nD was able to quickly identify, classify and geospatially visualize meter reading anomalies, discovering that a small number of meters were responsible for a large number of anomalies. Diagnostics revealed data quality and categorization issues to be the primary culprit.