Connect dots among data sources to improve operations

Plants often have much of the information they need to reduce unplanned downtime, avoid quality problems and improve management of people, inventories and assets. But the data is hidden in siloed databases, uncollected from smart instruments, uncorrelated and left invisible, so it isn’t used.

“Lower the cost of ownership and get better utilization by using the information we have from different sources,” said Matt Giordano, learning consultant, competency team, Rockwell Automation. “Bring data out and relate it to other data, from sources like overall equipment effectiveness (OEE), process variables, asset management systems – find the right data to relate, make it visible and use it to make better decisions.”

Giordano showed attendees of his session, “Gain Meaningful Insights to Production by Associating Multiple Data Sources,” at the Rockwell Automation TechED conference this week in Orlando, Fla., the kinds of improvements that can be made, and where to find useful data.

“Look at things from the past and things happening now,” Giordano said. “Use them to predict the future.”

For example, data from the automation system, operations and raw materials may be associated with a machine failure. “If I see trends, and then see those trends happening again, I can take a more proactive approach and prevent downtime,” Giordano said.

Consider the needs of different roles and responsibilities. Management needs to know production status so they can make adjustments and manage it to maximize profits. Operations tracks operators, materials and equipment to better use, maintain and improve utilization of assets. Maintenance needs to have the right people, resources and procedures to minimize unplanned downtime.

Seek data in context

Opportunities can be found in preventive maintenance (PM), safety, batch and clean-in-place (CIP) operations, inventory management and asset management.

OEE = Availability x Performance x Quality (APQ). Most facilities track it. Use it to see where there are opportunities to improve it by looking at other data. Put OEE together with data from the PM scorecard, data from the lines, machines and their components.

“OEE of 85% is world-class,” Giordano said. “If 8% of the remaining 15% is planned downtime, let’s go after the other 7%­--the unplanned downtime.” ARC research shows that there’s $20 billion worth of unscheduled downtime every year in food & beverage alone. A study in the automotive industry placed value of a minute of downtime at $22,000.

To improve safety, take a similar approach with safety KPIs, alarms and events, and incident and compliance report data. For batch and CIP, correlate operations, line and batch scorecards with unit batch and CIP process data. Find the reasons for the “golden batch,” safety, quality, productivity and water usage variations in the data.

Make better inventory decisions by putting together the materials, equipment, workforce and location data. “As orders are made, monitor the variables and assets,” Giordano said. “See what’s going out of spec and catch it in time.” Better manage raw materials and material utilization; back flush the business systems in real time to make better decisions.

For asset management, find out who is making changes to programs and components, and when. “I worked with an automaker on a frame welding machine,” Giordano said. “The company discovered a defect and needed to know when it had started to occur—we were able to determine that only three days of production were affected. We traced the cause to a faulty sensor. The sensor wasn’t available, so the technician had jumpered it out for three days until he could put in a new one.”

Mine the right data

Find the right data, no matter the source, and bring it together to be predictive and handle problems before they occur.

Almost every device has a wealth of information—speeds, faults, counts, etc. “Most of the tags are already in the PLC but not collected,” Giordano said. Collect them and make them visible.

Intelligent drives offer data on faults, speeds, voltage, current and utilization. Controllers record machine state, part counts, energy usage and alarms. Historians show trends. “Create reports and displays that combine information,” Giordano said.

“I worked with a bottler that bought a new wrapping machine,” Giordano said. “We collected data and noticed it had many little blips of downtime. The operator hadn’t even noticed. When production sped up, the wrapper would not keep up because the heating element in the first section wasn’t strong enough – it would have to pause to get back up to temperature. We had it replaced with a bigger element.”

Five steps to get started

• First, “There’s no need to go it alone,” Giordano said. “There’s a lot of expertise at Rockwell Automation and in our partner network.”
• Second, define a well-scoped pilot with clear objectives, on one machine, “something you can do with a few kinds of data,” Giordano said.
• Third, define a clear, short-term scope. “Get an early win,” said Giordano.
• Fourth, show success. More ideas will be generated.
• Finally, find the right implementation partners, and go forward with new applications.

Download the full report: Highlights from Rockwell Automation 2017