Project Summary

Project Title: “Smart Solutions for Water Transmission Planning, Asset Management, and System Performance”

Much of Southeast Michigan’s water system is decades old, and failures can leave thousands without drinking water. To prevent disasters, researchers at the University of Michigan and Wayne State University are using artificial intelligence to predict which pipes are most likely to break before they fail.

Working with the Great Lakes Water Authority, the team is developing real-time maps and models that monitor pressure, age, and soil conditions across hundreds of miles of pipelines. This technology helps water utilities fix problems before they happen—saving money, preventing service disruptions, and ensuring residents always have clean, safe water.

It’s a high-tech leap toward a smarter, more reliable water system for the millions who depend on it every day.

The Great Lakes Water Authority (GLWA) operates one of North America’s most extensive and critical water transmission systems, delivering safe drinking water to over 112 communities across Southeast Michigan, including the City of Detroit. With more than 800 miles of transmission mains, much of which dates to major 20th-century development eras, the system now faces urgent needs for modernization amid aging assets, historical failures, and rising infrastructure risk. Notable incidents—such as the 2017 14 Mile Road main failure affecting 300,000 residents—underscore the profound operational and community consequences of pipe failures. As the average age of GLWA’s primary pipe stock approaches 70 years, the urgency for a comprehensive, scientifically grounded risk management strategy has never been greater, especially as new challenges arise from climate variability and increasingly complex service demands.

This initiative unifies advanced research, practical engineering, and operational analytics in a multi-institutional partnership to support GLWA’s objective: develop and deploy an integrated, data-driven framework for risk-informed asset management, capital planning, and resilience innovation. The work leverages and improves both conventional and cutting-edge methodologies, offering real-time, asset-specific insight into network vulnerabilities and informing targeted investments that sustain the region’s water security.

The overarching goal is to guide GLWA’s Linear System Integrity Program (LSIP) in transforming from reactive asset management to a forward-looking, quantitatively rigorous program. This is achieved by:

• Building statistically robust, mechanics-based models for the probability of failure (PoF) of each major pipe class (PCCP, CI, steel, RC, DI), tailored to segment-specific operational data and inspection history.
• Introducing AI and machine learning analytics to identify key deterioration drivers, prioritize inspection schedules, and clarify uncertainty in risk forecasts at the pipe level.
• Integrating advanced Bayesian and reduced-order modeling for real-time forecasting of wire break progression and system resilience, demonstrated on high-priority assets like the 14 Mile, 24 Mile, and 120-inch Lake Huron transmission lines.
• Developing continuous risk maps and automated decision-support tools that synthesize SCADA data, transient pressure records, inspection results, and external environmental exposures.
• Partnering with external experts (Wayne State, Tennessee), GLWA’s consultants (HDR), and stakeholders through the UM Resilience Innovation Center for joint validation, field data collection, monitoring, and refinement of frameworks.

This collaborative approach aligns research contributions directly with utility practices, promoting seamless technology transfer from university research teams to GLWA’s LSIP operations and dashboards.

Project activities include:

• Detailed review and cost-benefit analysis of inspection and sensing technologies for condition assessment, including NDE methods and acoustic fiber optic monitoring.
• Construction and deployment of analytical and numerical models (from reduced-order to high-fidelity finite element) across the spectrum of GLWA’s pipe inventory.
• Continuous data ingestion, analytics, and integration with GLWA’s operational systems to produce real-time spatial risk maps and reliability indices.
• Direct in situ monitoring of prioritized transmission lines, including sensor deployment and data acquisition for validating model predictions.
• AI-driven investigation and prediction of drivers of PCCP and other pipe deterioration (age, soil conditions, vintage, operational demand, etc.)
• Regular collaborative meetings, workshops, and technology transfer sessions with GLWA, HDR, and academic partners.

The project has a number of anticipated outcomes, including:

• A scalable, mechanics-based risk assessment framework that calculates segment-level PoF and quantifies risk reduction achievable through inspection, renewal, and replacement actions.
• Unified decision-support tools, integrated into LSIP dashboards, enabling proactive capital investments and operational intervention where risk is greatest.
• Statistically robust, pipe-level risk maps for major assets, providing actionable guidance to minimize service disruption and maximize ROI on renewal investments.
• Dynamic prioritization logic that adapts as new inspection data and real-time operational pressures are acquired.
• Peer-reviewed publications, practitioner handbooks, and reproducible workflows for rapid adoption across GLWA and partner agencies.

Broader Impacts

This project marks a transformative step forward for water asset management, particularly for regions with infrastructure that, in some cases, dates back more than a century. Faced with the limitations of traditional methods, the initiative is introducing a new generation of tools—drawing on AI, structural reliability analysis, and real-time data analytics—to provide GLWA and other large utilities with more precise, efficient, and proactive approaches to pipeline health assessment and management.

One of the most significant advances lies in the development of innovative technologies capable of assessing the integrity of aging pipes from outside the pipe itself. This non-intrusive approach is critically important because inspecting pipes from the inside is costly and can disrupt water service, posing risks to system operations and public health. The new tools, now being piloted, dramatically lower both cost and operational barriers to high-quality, system-wide monitoring. With these capabilities, the project empowers utilities to make targeted, data-driven investments in replacements and repairs, reducing reliance on costly “run-to-failure” or blanket renewal strategies. As a result, operational and capital expenses can be reined in, and unplanned outages, boil water advisories, and water loss incidents can be sharply decreased. Furthermore, the integration of predictive modeling and advanced diagnostics allows managers to anticipate and address infrastructure challenges driven by age, climate change, and urban growth, rather than merely reacting after failures occur.

The frameworks now being developed offer a scalable blueprint for resilient, AI-enabled water infrastructure management at both regional and national levels. Just as importantly, collaborative fieldwork, continuous monitoring, and shared training programs are building technical and analytical expertise across the entire ecosystem—within GLWA, municipal partners, contractors, and among the next generation of water professionals. Altogether, these innovations are positioning GLWA and Southeast Michigan as leaders in securing a safe, reliable water future, and demonstrating how advanced data science and engineering stewardship can revolutionize the management of aging critical infrastructure.

Budget: More than $1 million

Partners: University of Michigan and Wayne State University

Period of Performance: 2019 – present