One measurement, multiple meanings: hidden risk in process control
Key Highlights
- Consistent signal scaling is critical to reliable control.
- Legacy systems create hidden maintenance risks and can turn a seemingly simple instrumentation change into hours of downtime.
- AI-assisted troubleshooting helped restore consistent measurement scaling.
As a controls specialist serving the large process industries, Ruth rarely needed more significant figures. What she needed was for the same number to mean the same thing everywhere it appeared. For example, if the compressor was moving 100,005 pounds-per-hour it wouldn’t be the least bit impactful if it was 100,006 PPH. Nonetheless, Ruth valued the “bussed” I/O world where the measured values and outputs were communicated as real numbers with minimal analog-to-digital or digital-to-analog (A-to-D, D-to-A) conversion.
During an outage, the compressor OEM’s representative, Sam, who was a veteran of the last century, sought her out to explain how the interlocks for protecting his machine were changing. With his present-century understudies in tow, Sam inhabited a world where a few thousandths of an inch could change the conversation. Indeed, this machine circulated high-pressure flammable gas at high speed, with only a microscopically thin gas film keeping its rotating and stationary seal faces apart. Sam was earnest, “When we set the new shaft, the settings for alarms and trips will have to be adjusted.”
Ruth though changing settings in the emergency shutdown (ESD) system was easy. “I can’t believe the high trip was 10 mils,” remarked another machinery guy who was familiar with the compressor. It was changing to the number seven. That’s 0.007 inches versus 0.010 inches, expressed as “10” and “7” in the ESD system. But the inactive shoe displacement alarm should now be -27 mils, with the trip at -28 mils. This was outside the configured range that had been established decades ago, which stopped at -25.
The displacement measurements were all connected to Ruth’s ESD system, which was a triple-modular-redundant (TMR) system, with a “voting” architecture designed to keep both spurious trips and dangerous failures to trip exceedingly rare. The early process superintendent participating in the design phase had endured the vagaries of conventional PLC interlocks in a previous life and would tolerate only the seemingly bulletproof TMR system for this compressor shutdown. It was arguably not a safety interlock, but very consequential, as either a spurious trip or a failure to trip had multimillion-dollar penalties and extended downtime.
Get your subscription to Control's tri-weekly newsletter.
All such measurements came into the ESD system as 4-20 mA. Even though the digital age had already dawned when the system was designed, there was little impetus for end users to employ digital communication, e.g. fieldbus, for safety interlocks. Fieldbus for safety—Foundation SIF—was demonstrated at Shell Global Solutions nearly 20 years ago. However, whenever Ruth brought up digital comms to her safety instrumented systems (SIS) consultant, it was as if she was lighting up a cigarette in church. Even a whiff of digital communications seemed to end any consideration for additional layers of protection where safety was a concern. Meanwhile, her TMR vendor supported Modbus RTU and Modbus TCP/IP but never offered a fieldbus I/O card.
“Let’s change the range to -30 to +20 mils” they agreed. It sounded simple enough, but changing the scaling in the TMR system also meant changing it in the dedicated HMI for interlocks, as well as the DCS, which redisplayed the values polled from the TMR over Modbus. After downloading the changes to the TMR, the various displays were befuddling. The machine at rest showed -29 mils, apparently below the new trip setting, but there was no vote to trip displayed. Ruth re-checked the scaling in the DCS and in the dedicated HMI database.
Everyone knew the scaling of the 4-20 mA signal from the compressor’s machinery protection rack would need an in-kind adjustment. What if no one remembered how to connect to and configure it? Last time, it was done by the machinery protection vendor’s field engineer, who wasn’t answering her calls. The retired machinery specialist had the software on his old laptop, but he couldn’t recall his password, or the password for accessing the rack.
With a little advice from her company-issued AI assistant, Ruth figured out how to access and configure the rack from the centralized machinery-health server and bring the scaling into agreement with the receiving systems. The restart waited only a few hours. But the lesson lingered: we had one shaft position and several places to tell the system what it meant.
About the Author
John RezabekRelated
Leaders LogoLeaders relevant to this article:




