As eyewear prescription updates bring the world into sharper focus, long-standing data analysis solutions can be augmented with new and more digitized tools that identify details and opportunities that were previously unseen.
For example, Guardian Glass produces approximately 500km of glass per day using 44 plants at its global facilities. It floats raw materials on molten tin to create flat glass ribbons and coats them with microscopic layers of metal oxides and other performance-enhancing materials. These applications are remotely monitored by five engineers, including three in Michigan, one in Spain and one in the United Arab Emirates (UAE).
“Our two main processes are in different time domains of what is important to us. Float glass lines run 18-20 years. As we consume the materials we sputter coat, these lines are shut down every six weeks for maintenance,” says Jesse Karkheck, subject matter expert (SME) for remote process monitoring at Guardian Glass, during his presentation, “Reducing the Cost of Curiosity,” at the Seeq’s Conneqt event in Austin, Texas in May. “Things happen slowly with float glass and it’s like piloting a barge that can take a week to correct course. In the coating process, the devil is in the detail, seconds to minutes are important to us. Each process poses completely different challenges.”
As well as attempting to meet its usual performance goals, Karkcheck reports that COVID-19 has accelerated Guardians’ quest for remote monitoring. “We thought it was a good idea before the pandemic, but our previous state involved lots of on-site firefighting with siled data collected at each site and not being used efficiently, extensive excitement from long SME travel times, and painful data analysis,” says Kark check. “We went into the pandemic with a very small pilot PI system with about 10,000 tags. However, when suddenly no one could travel anymore, they got the idea that remote monitoring could be a big deal for conducting experiments and keeping production running. We focused on remote capabilities, accessing unlimited, real-time, widespread enterprise data through PI System software. This changed the resolution and availability of the data, going from some plants receiving data points every 20 minutes to receiving data points every few seconds.”
Choosing the right problem
“Good ideas are not always well-defined projects,” adds Karkcheck. “We have a lot of smart people in our facilities, but people are busy and we have a small remote monitoring team. Although we have continuous production lines that have batch aspects, analyzing periodic conditions really became a challenge. We may have three years of data to analyze, but we need to sift out specific parts like thicknesses, product colors, production rates and specific settings from engineers via data that is scattered in Excel spreadsheets that could solve problems.” Karkcheck also notes that Guardians Analysis approach addressed problems faced by engineers who may be hiding problem-solving data in Excel spreadsheets.
Karkheck states that Seeq makes it possible to process chemical data, use production run data that may not be in a historian, and help users with other potentially useful information. Where a given analysis test previously took three weeks with legacy tools or one to three days with other software, Seeq delivered test results in minutes with a profile search and less than 10 clicks, allowing users to learn more.
“We didn’t really know we needed Seeq until we started using it,” says Karkheck. “We started using it extensively earlier this year and it’s been organic growth ever since. I gave some people on my team the basics and it just worked. We want the cost of curiosity to be as low as possible, because if you can lower the barrier to explore, you can try discarding a lot of ideas, shifting the pivot, and finding cool things. It’s common for us to be in remote monitoring meetings with the plants and they’ll bring up problems that have arisen at some point, but they don’t know when. There may be five minutes left in the meeting, but they’ll give us a few tidbits. We try to give them an answer before the meeting ends and we can often determine when the problem started or where to look.”
Slow, big transitions
Karkheck reports that there are two main areas at Guardian’s plants where Seeq’s software is used: glass transitions and coating line operating pressures. “With glass transitions, it was difficult to compare run-to-run data for products that are rarely made, such as B. different colors that are produced about twice a year,” explains Karkheck. “Previous efforts took months to analyze, which meant they weren’t done, or not done as thoroughly as possible. The opportunity here is more than $1 million per plant for five plants.”
To further assess his glass transitions, Karkcheck adds that Guardian uses Seeq’s Chain View and Capsule View displays to compare runs, focusing on the information he cares about and omitting non-essential data. It examines data from multiple sources including PI System, CSV files and SQL databases. “We were able to analyze data from three and a half years with one person in just a few days,” says Karkcheck. “We’re new to remote monitoring and the factories are a little hesitant about outside help, but eventually their entire team showed up and asked us to look into several issues. We achieved a much better commitment to plants by showing that we care and want to solve problems. We haven’t quantified the gains yet, but one facility just had the best run in its history.”
The conventional wisdom that all production runs are the same was refuted by Chain View, which indicated temperature fluctuations in some of the Guardian’s product campaigns over the three and a half years surveyed. Meanwhile, Capsule View helps the glass manufacturer direct their process in the right direction from the start, improving the chances of their glass transition succeeding.
Fast, small plating problems
As the coating process speeds up, Guardian must maintain vacuum integrity in its vacuum sputter coaters and identify signs of failure before they can occur. “The challenge here is that with our previous toolset we had some unexpected downtime. We’ve put a lot of work into these tools and they’re wonderful. They capture most of our problems, but there are still a few that come around,” adds Karkcheck. “And as we looked more closely at the data, there were small signs of issues that would normally be thought of as noise.
“For example, on all of our vacuum coaters, our PI Vision Coater Check Engine display scans every second after a pump down and evaluates the vacuum integrity at every part of the coater. Consequently, while the regular KPI may say everything is at 100% for the few items that slip past, we used Seeq to flag certain conditions for operating pressure when everything else should be stable, although this may only be for some few counts seconds. In the end, we identified several small issues, reported them to the factory, and fixed them with no downtime or bad product. A few years ago, before we had access to PI System and Seeq, the same issue went undetected and caused more than $1 million in losses.”
Karkheck concludes that Seeq’s software found very short periods of unexpected pressure fluctuations and that Chain View clearly plotted these along with total gas flow and a running delta function of the point-to-point variability of the incoming data. “Anytime we intentionally changed the gas flow, we expected the pressure to change as well. However, when we went to the chain view, we also found brief pressure spikes that shouldn’t be there. We brought them to the factory, they adjusted their device settings with no downtime, and that made the process more stable. We have been using Seeq’s software consistently for three or four months and find many items like this and have saved about 200 hours of unplanned downtime on the coating machines which are expensive to run.”
From on-premises to servers
To extend these and other data analysis capabilities, Chris Herrera, senior principal solutions architect at Seeq, is watching cloud computing providers move related services from on-premises locations to servers. This includes data collection, edge computing, device management, model customization and other services. “Users want to change their parameters for gathering information or how they cleanse data before going to the cloud, so we see them choosing algorithms like automated machine learning tools (autoML) to brute-force the right algorithm for their specific needs. force method,” says Herrera. “This allows them to perform tasks like anomaly detection, process multiple variables, or predict signals. That’s why we recently launched the Bring You Own Algorithm strategy, which allows users to leverage Seeq to run any algorithm, compute at the edge or for a fleet, and view raw results or derived computed data in context.”
Herrera reports that cloud-based, algorithm-driven data analysis can help solve one of the long-standing problems with analytics that use comma-separated values (CSV) text files to provide results in spreadsheets. “CSV files may not contain the raw data that suggests where information came from, meaning data scientists and other users have to scroll through it to put it in context, find a pattern, gain insights, and make decisions.” ‘ Herrera explains. “Because of this, our software makes inferences so we can go to where each user’s information is and generate data from there. We initially added remote agents and accessories to our software and more recently added algorithm capabilities. This is useful because large facilities and processes might have 70 historians and 100 SQL servers, and it can be very expensive to run and maintain data pipelines from everyone to a central location.
“We can also ‘encapsulate’ information by time or other parameters, which Excel cannot do. It also facilitates fast calculation as users no longer need to implement formulas in Excel and get timely insights to positively impact performance and results. One user reported that when they open our Seeq Workbench software, the software infrastructure takes a back seat and only they and the data are needed to solve business problems.”
About the Author: Jim Montague
Jim Montague is the Editor-in-Chief of Control. He can be contacted at [email protected]