Being 135 years old certainly hasn’t slowed Umicore down. Anything but. These days the company processes more than 200 different types of raw materials and achieves an impressive 500 kt per annum. Originally established in Belgium, the company has an exceptional reputation for circularity—its mission to create sustainable value through the use of "materials for a better life.” Umicore may not be the first company to recognize its long-term business drivers: resource scarcity, increased material complexity, and eco-efficiency. But it is unique in the way it’s focused on technology to tackle them head on.
Whether it’s extracting gold and silver from mobile phones, utilizing spent automotive catalysts, or recycling photographic residues, Umicore has always recognized that the processes behind its manufacturing are the key to operational efficiency. The company has moved on a lot since the days when operators taking a one-hour lunch break inadvertently logged the shift as an hour longer.
Digitalizing massively since 1995, Umicore has been using OSIsoft PI to capture the data around its manufacturing lifecycles. This has enabled it to build an asset framework, event frames and set up fields for context types. Using templating, it can capture all events for each asset. But it was proving a lot harder to understand the data contextually: More and more Umicore’s engineers needed capabilities such as diagnostics, as well as predictive and prescriptive analytics that spanned multiple assets and simultaneous events.
The company wanted to empower its operators with a 360-degree view of operations, enable them to interact with the data and share it with others to build an operational culture around data transparency and efficiency. TrendMiner from Software AG caught its attention. Using contextual data stored in Pi and other systems, TrendMiner offers user-friendly dashboards that are both analytics and process-enabled. Since 2022, after a simple plug-and-play deployment involving data tags, its operators have been using the software to drive efficiency with no input from data scientists.
“There were numerous contexts that we wanted greater insight into” Dave explains. “From maintenance to visitor requests, temporary changes, to production, consumption and shift handovers—there were so many processes we wanted to begin automating using powerful, reliable analytics.”
Improving maintenance: Before Umicore deployed TrendMiner, operators were required to flag any abnormalities in their data sets and send a request to maintenance to review in more detail. This would take some time. Often resulting in a shift coming to an end. “With TrendMiner, process deviation has been automated,” Dave explains. “This means maintenance—who use the tool too—are automatically informed of deviations and a log made in the system that a maintenance request has been made.”
Automating shift hand overs: Umicore strives to be a data-driven collaborative organization. And ensuring that the right information is passed over when a shift comes to an end is key to this. For a shift to be successfully handed over, the TrendMiner tool triggers the operator to pass any process-related information across to their co-worker before they clock off. This includes covering off any deviations that have been detected during the shift. Dave explains: “These are automatically logged in the system when time-series data is mixed with contextual data and undergoes root cause analysis. These need to be addressed and discussed before a shift can be officially ended. If a process is supposed to last six hours and an operator completes it in five, the system automatically registers it as a success.”
Automating the digital audit trail: While all information is made available through dashboarding, operators wishing to look in more detail at the process flow for their particular asset, automatically receive an HTML attachment when the shift is handed over to them. This is a very practical way of maintaining a digital audit trail. The active directory, which underpins the access hierarchy, knows which operators can edit information, which users can see which departments, and enables administrators to delete context items if necessary.
Asked what value TrendMiner has brought Umicore so far, Dave Carpels, Senior Process Engineer, has a clear answer: “Senior Engineers like me have become experts in data. But we’re further from the day-to-day reality of our processes. It’s our operators that are the closest to them and now we can give them the intel they need to do their jobs better. Pi enables us to collect and capture our data. TrendMiner empowers us to visualize, analyze, monitor, and predict so we can trigger performance investigations, keep a track on contextual events, and make analytics-enabled enhancements. So far we can report
I definitely couldn’t put a number on what value this is unlocking across the production sites where its deployed, but it must be huge.”
