"We had a real problem," said Kimberly-Clark engineering and technical leader Bob Kilas. "Our DCS systems were at the far edge of their lifecycle, with many obsolete parts that were expensive and hard to find. We knew we couldn't continue with them, but we believed that migrating to the latest platform upgrade would be costly and not necessarily give us better performance."
The company had Rockwell Automation systems on other applications throughout its plants. "We decided to see if we could use PlantPAx on our tissue machines instead of the DCS systems. That way, we could leverage the spare parts, internal technical knowledge and technical support that we were already getting from Rockwell Automation," Kilas continued.
The tissue machine includes a stock chest where water is mixed with cellulose pulp to make the slurry; the web area where the stock slurry is spread on the 18-foot-wide web belt; a dryer where the stock becomes tissue; and the "creping" end, where the stock is scraped off and sent to the packaging plant in huge rolls.
"The machine is typically three stories high, 180 feet long and 18 feet wide. The unit we upgraded is over 50 years old," Kilas said. "This is a continuous process. We never shut it down. If the tissue on the web breaks, we just start a new roll."
Also Read: Bringing the Cloud Down to Earth
The legacy DCS architecture used proprietary controllers for the analog I/O and a proprietary network interface to older Rockwell Automation PLCs that ran the motors and on/off valves in the wet and dry ends of the machine. The DCS also used a proprietary structured text language programming system, where the PLCs used ladder logic. The legacy DCS had 80 PID loops, 180 analog inputs, 60 analog outputs and numerous HMI tag groups. It had 65 HMI process graphic screens and 30 programs written in the proprietary structured text code.
"We had to hire back a retiree who had institutional knowledge of the graphic screens to capture the programming knowledge behind these HMIs," Kilas said.
This project was done in parallel with other projects and wasn't part of the machine-down critical path. "We needed to do it while the machine was running," Kilas said. Kimberly-Clark used a common HMI application and replaced the legacy controller, PLC gateway and HMIs, and eliminated obsolete PLC5s. "We migrated the ladder logic," Kilas said. "We needed to minimize conversion risk, engineering effort and down time so we did not rewrite the legacy structured text language code, just ported it to ladder."
Describing the lessons learned from this migration project, Kilas said, "We should have re-written the legacy structured text language code, and after the project, we did do that. We needed to define and do alarm rationalization. We need to provide system status and Ethernet switch health HMI screens. We absolutely have to ensure operator and maintainer training to cope with variations in operator knowledge and experience and use off-line simulation for automation sequences for training."
The project took 26 weeks from start to the finish. Kilas said, "We improved operator and maintainer efficiency, reset our lifecycle clock from 25 years to new and going forward, we now have a single architecture and engineering system for the plant."