Another chronic hindrance to effective data analytics is the ongoing exodus of veteran personnel, who are experts in collecting information from legacy devices, but leaving behind less-experienced colleagues, who don’t know how to access it.
“We were called to a petrochemical plant in Alabama in May after a worker who had been with the plant for years left the company, and took their OSI PI System’s passwords and process attributes for creating and deleting tags with him. His coworkers knew how to view their PI displays and data, but the rest of the system had always just worked,” says Brian Bolton, consultant for Aveva’s OSI PI software at Maverick Technologies, a system integrator and Rockwell Automation company. “The remaining staffers needed to learn how to manage their PI system, so we taught three shifts of operators, maintenance engineers, instrument technicians and supervisors how to find, create and manage tags, delete stale or bad tags, and fine tune the rest of their system. This was a big relief for the plant because it’s adding a large Honeywell Experion DCS in November, so we’ll go back and make sure all their instruments have the right PI connections.”
Despite its plans for a new DCS and other upgrades, Bolton adds that the petrochemical plant is running the same processes, and continues to depend on the same flow, temperature and level measurements, and controls as always. This facility also uses steam from a plant next door to make electricity, and relays the extra power it doesn’t use to Mobile’s grid. “This requires the petrochemical plant to monitor its power and distribution, use variable-speed drives (VSD) for efficiency, track what’s in the line during peak hours, and start up and run when electricity is cheaper,” explains Bolton. “All of its electrical panels have onboard microprocessors and software to monitor what’s happening and avoid drawing too much power at once. Of course, the new DCS and training its staff to use OSI PI will help it analyze and compare all these variables, optimize its processes, and show options for what it could do to generate more revenue.”
Besides upgrading instruments and retraining people, Bolton reports that data analytics can also take a page from continuous improvement strategies, which recommend picking five or six items to focus on. “I think that many of today’s analytics efforts have dropped the ball because they concentrate on costly technology investments, but skip closely examining the processes they’re trying to improve,” says Bolton. “We’ve all become technologically smarter and are adept at reading spreadsheets and using the Internet. However, we’ve stopped looking deeper in many ways. We rarely go back and check again, and don’t tear down and rebuild plant-floor processes. I believe artificial intelligence (AI) will make this situation worse.”
Bolton adds that the remedy for technology-induced inertia is encouraging and questioning the younger engineers, operators and technicians, who are still willing to go out in the field. “These questions start with the production manager asking, ‘How many batches did we run this week? What was their average time? What was the pH concentration from batch to batch? What was the standard deviation in the viscosity?’” says Bolton. “If managers ask these or similar questions often enough, their staffers will begin to anticipate them, and try to get answers ahead of time.”