Confessions of a Looney Tuner
Switching to Manual Captain.
March 22, 2018
Control Valves. You get what you pay for
In this issue, I will talk about control valves purchased for a greenfield paper mill site. I was part of the instrumentation and controls commissioning team, working for the mill directly.
I will not name the make of the valves for reasons that will become obvious. As is the case in many projects, these valves were purchased on the basis of price. And like most things, you get what you pay for. The quality of these brand new valves was very poor.
For example, almost every single butterfly valve failed. The pin that connects the shaft to the disk part of the valve sheared off, leaving the valve fluttering in the wind. Perhaps they took the term “butterfly” too literally. In several other cases, the bolts securing the actuator to the valve came loose. Consequently, when the valve was bumped, the entire actuator would teeter back and forth like Frankenstein taking his first steps. The end result was backlash. Some of these valves were quite large, up to 18 inches. You can imagine these swaying behemoths “Franken-actuators” were certainly a safety hazard!
And it got worse! What became most distressing during the startup of the mill was the large number of valves that suffered from excessive friction. They were afflicted with a condition commonly referred to as stiction – short for STatic frICTION. Essentially, they would stick and respond very slowly, often in a “jerky” fashion and overshoot. You could say that these valves were constipated… in spite of the pulp fiber flowing through them!
Stiction can occur in many places. It can live in the actuator – notably piston actuators. It can also live in the valve packing, within the valve seating and the positioner itself. In our case, the stiction was caused by the positioner… the “smart” positioner.
“How can this be?” you ask. This smart positioner has many software parameters (as most smart positioners do). In this case, there was one parameter in particular that was quite perplexing; it was called the dead band parameter. This parameter essentially builds a dead band in the positioner. For example, if the dead band was set to 2%, the positioner would be happy if the position feedback was within 2% of the signal received by the positioner from the DCS. Moreover, output changes that were within the dead band would not cause the positioner to move the valve. As Spock would say (with a raised eyebrow): “this dead band parameter is most illogical”.
Richard asked me why anyone would implement such a dead band parameter. I responded: "Damn it Richard… I’m a control engineer, not a smart positioner software programmer!"
The default setting for this parameter was “Auto”. In this case, the positioner automatically sets the dead band during its auto-calibration procedure. The actual “automatic” value set by the positioner can not be viewed. However, this dead band feature can be set to manual, and a value can be entered. The theory is that by setting the dead band to manual and entering a minimum value of 0.1%, the precision of the valve would improve.
After accidentally discovering this, the mill’s E/I supervisor commissioned me to audit and document the performance of every single control valve. This entailed bump tests and observation from the DCS and simultaneous direct observation of the valve itself (the proverbial “finger test”). I was also told to change the dead band parameter from “auto” to “manual” and set it to 0.1%. The valves were tested before and after the dead band change.
So off I go with my trusted help, one of the mill’s instrument technician – let’s call him Richard - to audit those valves. The first one I pick was a flow control valve with an astounding 7% stiction. This stiction was clearly evident in automatic, the flow cycled and the controller output had the telltale “saw-tooth” pattern. The “before” bump tests confirmed the stiction in the valve.
The next step was to change the dead band parameter from Auto to Manual and set the value to 0.1%. As he was doing so, Richard tried to put on his best impersonation of Scotty, from Star Trek, and jokingly quipped on the radio: “Switching to manual, Captain!” We re-tested the valve, and miracle of miracles, the 7% stiction was gone! The valve tracked every change with a crisp response. In fact, every major case of stiction initially observed was cured by this software change.
At one point, Richard asked me why anyone would implement such a dead band parameter. I responded: “Damn it Richard… I’m a control engineer, not a smart positioner software programmer!”
Throughout the rest of the day, Richard was relentless with his “Scotty’isms”. His continuous chatter eventually drained his radio battery. The last thing I heard him say was “She’s gonna blow, Captain”. During the radio silence, I was left with the curious notion that in the 23rd century, they will still be putting stuff in manual.