A paper manufacturer got more out of its AI sensors with a simple administrative fix

In Domtar's war room, workers can review the trove of data collected by its AI-assisted Waites sensors. Courtesy of Domtar Domtar's AI sensors help predict equipment issues to prevent costly machine failures. A Domtar reliability engineer kick-started an in-depth collaboration with the sensor company. Domtar's collaboration with Waites led to better data utilization for reducing machine downtime. Early in Matthew McLaughlin's tenure as a reliability engineer at Domtar, a motor at the paper manufacturer's Kingsport, Tennessee, mill failed. McLaughlin said his manager asked him to review the entire day's sensor data, collected from 450 sensors, and recommend a fix for the motor — but he knew that was an impossible task. "It would take 32 weeks for me to analyze all the data we were getting in one day," McLaughlin, who joined the company in 2024, told Business Insider. As manufacturing companies adopt promising new AI technologies, they often face the challenge of learning to do things differently, like implementing new tools and processes to get the most out of their technological investments. In the years prior to McLaughlin's arrival, Domtar implemented AI-assisted sensors from Waites Sensor Technologies to continuously monitor the vibrations coming from equipment — which signal a piece of machinery's health — and provide that data to prevent breakdowns. Domtar, armed with more data than it had before, needed a new way to manage it all so it could make the most of the Waites sensors' capabilities. That's where reliability engineers like McLaughlin come in: They are tasked with improving a company's technology and digital resources through data analytics, risk management, engineering expertise, and communication skills. As AI systems are increasingly integrated into workflows, reliability engineers like McLaughlin must develop new methods that bring humans and technology together. "This was the line-in-the-sand moment where we needed to do something different," said McLaughlin. He added, "It couldn't be treated like a legacy predictive maintenance program; it needed to be treated like the advanced system that it is." Drilling into the data on weekly partner calls To move away from manual analysis and make better use of Waites' machine-learning technology, McLaughlin said he decided to lean on the tech company for more detailed and consistent support. Domtar already had access to Waites' full-service offering, which included a Waites analyst who could learn about Domtar's facility, pain points, personnel, and staffing levels to help the company customize its sensors' AI recommendations. McLaughlin said he began putting this additional support to use, attending monthly calls with the sensor company. During one of those calls, the Domtar team mentioned the plant was underutilizing tools that could provide additional data and potential value, and the calls turned into weekly occurrences. Domtar began sharing data from their thermal imaging camera, infrared camera, and ultrasonic meter with Waites. McLaughlin said, "Let's provide as much feedback as we possibly can, because with machine learning, if you don't teach it anything, it's not going to learn anything." Within three months of taking this more active approach with Waites, McLaughlin said, Domtar began to notice significant improvements. The convergence of data and human expertise AI performs the primary analysis, but it relies on the Waites analyst, on-site vibration analysis experts, and Domtar's maintenance personnel to diagnose the issues, said Rob Ratterman, the CEO and cofounder of Waites. "It's not just a change in technology. It's not something you just plug in," Ratterman told Business Insider. He added: "You can plug in our system or any sensor system and you get some benefit, but to get it to where you have the kind of results that Matt's seeing, it takes someone who becomes a lighthouse for the entire company to follow," Ratterman told Business Insider. Shift supervisors, production superintendents, high-level managers, and the general manager have access to the Waites system to stay apprised of alerts and resolution responses. McLaughlin's team tracks each network and equipment action item, including response time, ensuring that action items are less than 30 days old. The analytics AI shares help them fine-tune maintenance, including the lubricant's viscosity grade. With lower viscosity, they can use less horsepower and draw fewer amps on the motors, driving down costs, said McLaughlin, who is now a reliability superintendent at the company. Domtar declined to disclose its AI costs to Business Insider, but McLaughlin said that the system is worth every penny. "We justify the expense — which is very reasonable compared to other options — of the Waites system because we saved 1,546.65 hours in unplanned downtime," he estimated. Gaining worker trust through daily reports and tracking McLaughlin also began sending daily email summaries about the system and its alerts to Domtar's legacy employees. Operations managers called him whenever they received an alert to ask how the reliability team was handling it, he said. "Then those calls stopped. Yesterday, I spoke with a paper machine superintendent who said he doesn't look at the alerts anymore," McLaughlin said. He also said that trust has compounded to the point where Domtar is "basing multimillion-dollar decisions on our Waites data, and it's become an integral part of our planning process." Domtar now uses 748 sensors that its staff monitors, said McLaughlin. The Waites sensors continually detect machine vibrations that signal issues — like a crack in a motor bearing, misaligned motors, or a lack of lubrication — to predict imminent part or machine failure or to flag potential issues six months out. This allows companies like Domtar to order specialized parts and keep their machines running, Ratterman said. The sensors allow Domtar to make small, consistent improvements and avoid parts failures. For example, a drive belt supplier recently told McLaughlin that his Domtar facility has not purchased fan belts in a year, which McLaughlin attributes to the regular monitoring and maintenance the sensors support. "We wouldn't be able to do that without the AI-driven computing power," he said. Read the original article on Business Insider

Read it at Business Insider ↗

← Front page