Predicting the Break: Forecasting Water Main Failures

The Skill Being Assessed

As a manager of physical and infrastructure assets, your primary responsibility is to ensure the reliability and longevity of critical systems. For a municipal water utility, this means managing an extensive and aging network of underground water mains. Historically, utilities have operated in a reactive "break-and-fix" mode, which is costly, disruptive, and inefficient.

This assessment challenges you to apply the skill of predictive modeling. You will use historical data to forecast which water mains are most likely to fail. This allows you to shift from a reactive to a proactive maintenance strategy, enabling targeted, data-driven decisions for repair and replacement schedules. Mastering this skill is essential for optimizing budgets, minimizing service disruptions, and ensuring the sustainable management of public infrastructure.

Your Task

You are an Asset Management Analyst for the City of Northwood Water District. The Director of Public Works has initiated a pilot program to move from reactive to proactive maintenance for the city's aging water main network. Your task is to use the provided data to build and evaluate a predictive model that forecasts which pipes are most likely to fail.

Your final deliverable is a two-page memo addressed to the Director. The memo must: 1. Briefly describe your methodology (data preparation, model used). 2. Summarize your model's performance using appropriate evaluation metrics. 3. Provide a prioritized list of the top 10 water main segments recommended for proactive inspection and potential replacement, justified by your model's findings.

Resources and Data

To complete this task, you are provided with a project kickoff memo from the Director of Public Works and a dataset containing the city's water main inventory and historical failure data.

Project Kickoff Memo: This document outlines the strategic goals and expectations for the predictive maintenance pilot program.

Water Main Data Set: This file contains records for 500 segments of water mains. It includes physical attributes of each pipe and a record of whether it has failed in the last five years.

City of Northwood Water Main Inventory and Failure History

pipe idinstall yeardiameter inchesmaterialsoil typepressure psihistorical breaks countfailed in period
WM-100119488Cast IronHigh7531
WM-1002195512Cast IronHigh6541
WM-100319626Cast IronMedium8021
WM-1004197010Ductile IronHigh9021
WM-1005201212PVCLow8500
WM-100619988Ductile IronLow7000
WM-100719596Cast IronMedium7810
WM-1008200510PVCHigh9500
WM-1009198516Ductile IronMedium6010
WM-101020188PVCLow8800
WM-1011196812Cast IronLow7210
WM-101219926Ductile IronMedium8200
WM-1013197510Ductile IronHigh8510
WM-1014200812PVCMedium7500
WM-101519658Cast IronHigh6810
WM-1016200110Ductile IronLow7900
WM-101719888PVCMedium9200
WM-1018194212Cast IronLow6000
WM-1019199518Ductile IronMedium5500
WM-102019528Cast IronHigh7020

Submit Your Work

Insert DropBox here

Next Steps

Excellent work applying your predictive modeling skills to a real-world infrastructure challenge. You have successfully translated data into an actionable plan, a core competency for any asset manager.

When you are ready, please navigate back to the course to continue your learning journey.