Mueller Water Products Uses Amazon SageMaker to Improve Leak-Detection Performance

Mueller Water Products Inc. (Mueller) wanted to continue improving its technology infrastructure and building on its sophisticated means of collecting data and delivering critical insights to customers. Thus, the company planned to consolidate many of its cloud and on-premises systems into a single, comprehensive cloud system.

Mueller accomplished this on Amazon Web Services (AWS) by building a data lake on Amazon Simple Storage Service (Amazon S3), an object storage service that offers industry-leading scalability, data availability, security, and performance. From there, the company was able to increase leak-detection performance using Amazon SageMaker, a fully managed service that provides the ability to build, train, and deploy ML models quickly.

By building its scalable software and data solution on AWS, Mueller could give its customers a more accurate view of critical water data, including leakage, enabling them to manage their infrastructures more efficiently. Now largely in the cloud, Mueller is poised to expand on its success as it reimagines technology’s role in water infrastructure.

With roots stretching back to 1857, Mueller is a global manufacturer of products and provider of water transmission, distribution, and measurement solutions. Employing more than 3,100 people worldwide, Mueller covers two segments: infrastructure and technology. On the infrastructure side, the company manufactures fire hydrants, valves, couplings, and other components. Mueller has one of the largest installed bases of fire hydrants and gate valves in the United States. To capitalize on its robust role in US water infrastructure, Mueller has been developing its technology infrastructure since the early 2000s, acquiring several technology companies to enhance its smart metering and leak-detection capabilities.

Over time, Mueller recognized that it needed a simpler method of managing internal collaboration and on-premises requirements because it did not want to be a data-hosting company. “We had numerous data-ingestion pipelines and places in which the data was hosted and managed, both on premises and in the cloud,” says Dave Johnston, director of smart infrastructure at Mueller. “A few years ago, we decided to revamp that whole strategy and centralize everything in a cloud-based server infrastructure. AWS was able to help with that initiative.” Mueller’s acquisitions greatly widened the scope of its development road map. With manual data collection becoming fully automated on AWS, Mueller would be able to collect more than 5 GB of daily Internet of Things (IoT) data with minimal time and effort—recording in a single day what could have required hundreds of hours to obtain manually.

Out of all the cloud strategies that Mueller considered, consolidating on AWS seemed most practical: “We decided it’s cheaper and simpler for us to develop on one solution,” says Kenji Takeuchi, senior vice president of technology solutions at Mueller. “Being on AWS enables a faster development process for us.”

The first part of Mueller’s infrastructure data solution involved building a data lake on Amazon S3. “We selected AWS as our primary solution for our software and data products,” says Takeuchi. “The goal was to have a scalable cloud-based infrastructure, including a data lake where we could bring our data together and glean ML and artificial intelligence insights.” Mueller uses AWS software and analytics to support its data-rich Sentryx Water Intelligence Platform, which provides a common user interface across all of Mueller’s software solutions and offerings.

Mueller now uses an Amazon S3 data lake for terabytes of water-use data related to leak detection, pressure monitoring, and metering. “In the last 9 months, we brought three major product families, including over a million connected meters, onto the AWS-powered Sentryx Water Intelligence Platform,” says Takeuchi.

Having all this data in one place enabled Mueller to more simply develop ML models using Amazon SageMaker. “We’re using Amazon SageMaker to help us be more accurate in our leak event classification,” says Takeuchi. After service deployment, Mueller cut the number of false alerts in half and maximized the potential to identify true leak events.

This use case is critical considering that leakage accounts for an estimated 1.7 trillion gallons of lost water—roughly 30 percent of all treated water—every year in the United States. That kind of loss is a major problem for both the environment and the funding and sustainability of utility companies, which see no revenue from lost water. “The loss is on the order of billions of dollars,” says Johnston, “and it converts to an unnecessarily high carbon footprint from the water industry. The numbers are staggering.”

Using its AWS-backed analytics solution, Mueller continues to improve leak-detection capabilities. A water utility in the Northeast United States has installed a Mueller leak-detection system, and based on current performance, the utility estimates that it will save $8 million in operational expenses over 5 years. The cost of water is not the only concern regarding leakages, however. If not found quickly, leaks can grow and lead to catastrophic damage and costly infrastructure repairs, as well as regulatory fines. One Mueller customer was able to use Mueller technology to mitigate exposure to environmental compliance orders that could have resulted in million-dollar fines for uncontrolled discharge of chlorinated water into protected waterways (“Small Leaks, Big Money.” American Water Works Association 2020 Virtual Summit, December 2020).

Mueller successfully built a scalable software and data solution on AWS—and the company is just getting started. “We’ve been doing a lot of work on visualizations,” says Takeuchi. “But we also want to provide full automation. As a leading player in pressure control valves, for example, we’re in a great position to use our pressure monitoring solution to automatically control pressure as needed.” Because managing pressure is a principal way to control water quality and extend the life of pipes, automating the process through technology has profound implications for saving costs and reducing water loss. As part of this initiative, Mueller plans to use AWS IoT Analytics, a fully managed service that enables users to run analytics on IoT data.

More broadly, Mueller is eager to build on its analytics and ML capabilities. Mueller has a committed software-as-a-service strategy in place to provide customers critical insights into water quality evaluation, pressure monitoring, and other aspects of water distribution. By empowering customers through increased problem-solving potential, Mueller is helping lead the market toward innovation and growth.

“The level of support from the AWS team has been amazing,” says Johnston. “Not only has it helped us troubleshoot technical problems, but it has also supported our need to develop quickly with our customers in mind. AWS appreciates our needs, and it’s always looking for opportunities to accelerate things for us.”