Gemini · Fall 2025 · Completed

Predictive maintenance for water and pipeline infrastructure

Exploratory phase of a predictive maintenance project for municipal water pipeline networks — structuring the problem and comparing survival analysis with failure classification.

Time seriesPredictive maintenance
Abstract visualization of water pipeline infrastructure

ReLU began collaborating with Gemini VA in Fall 2025 on a predictive solution for maintenance scheduling in municipal water pipeline networks. Pipe failures lead to water loss, repair costs, and service disruption, but utilities cannot inspect or replace every pipe. They need a way to prioritize where to act first, and traditional heuristics like “oldest first” ignore most of the information available in historical records.

Exploratory analysis and problem structuring

The fall work focused on exploratory analysis and problem structuring. ReLU explored two complementary solutions — a survival analysis approach to model asset lifetime, and a classification approach to predict whether a pipe will fail within a given time horizon. In Spring 2026, the project moved into model development and delivery.

Data
Pipeline asset records and operational time series
Methods
survival analysis, failure classification, exploratory data analysis
Handoff
problem framing, baseline comparisons, feature exploration notes
Back to projects