Lothar Lang holds just such a position at the Baytown, Texas, plant of Bayer MaterialScience (www.bayermaterialscience.com).
“We are running several programs to improve reliability of these units,” states Lang, “including more up-time, as much production as possible, and so on. We had already implemented program monitoring for OEE (operating equipment efficiency), availability and operating rates of the units. Then we had the idea that we can contribute to performance goals by looking at the process control side,” says Lang.
The OEE monitors key performance indicators (KPIs) 24 hours a day, seven days a week. And Lang wondered what could be gained if the company also looked at performance of actual control loops on a 24/7 basis. “We are convinced that when you run a unit as much as possible in automatic, that reduces the upsets in the process and increases efficiency in the units. Plus, each upset wears on the equipment,” he says. So with 24/7 monitoring, he reasoned, the company should be able to increase availability. Plus, Lang adds, “if you have good control, you actually can run the unit with higher throughput because you don’t need all that much margin to your limits.”
So Lang decided to look for a software application that would closely monitor what each control loop in a unit was doing. He picked PlantTriage from ExperTune (www.expertune.com), based in Hubertis, Wis., and installed it on one of the plant’s eight units. “We liked the performance we got from the trial enough that we decided to put it on three more units. Now, we’re in the decision process of implementing it throughout the plant,” he adds.
ExperTune says that a typical installation of PlantTriage results in a 5 percent increase in service factor (up-time) of supervisory applications, setpoints that are 3 percent to 5 percent closer to specification due to reduced variation in the regulatory system, a 200 percent to 300 percent increase in the productivity of maintenance and engineering, because they hit the loops that make the most difference, and a 50 percent reduction in valve maintenance budget.
Lang says that he didn’t want to decide up front which control loops to start monitoring. Rather, he worked with operators and other personnel to decide on KPIs that were easily understandable by operators. “We took some time to come up with numbers we could live with, then just looked at all the control loops in the unit.” The first thing Bayer engineers looked at was just whether or not the loops were performing in design (automatic) mode. “We decided on a threshold that if the loop is more than 5 percent out of design mode, then it is considered out of mode,” Lang relates.
The second parameter monitored was the output of the control loop. It should not be fully open or fully closed more than 10 percent of the time, says Lang. The third parameter was whether the loop shows oscillating behavior. “There is a probability factor of whether the software is 100 percent certain of oscillation sliding down the scale to zero. We decided on a factor of 80 percent to send an alarm message,” Lang says.
Meetings matter
Bayer has operating teams in three of the four units using the new process. These teams—composed of engineering and operations personnel—meet regularly to discuss the KPIs and decide what to do about them. Lang reports that the three units that have regular meetings have seen good results—enough so that plant personnel are encouraged to expand the program. “To really get the benefit, you have to build teams to address the shortcomings,” concludes Lang. The fourth unit that has the software but does not have regular meetings has yet to see the benefits.
Gary Mintchell, [email protected]
