Take Mystery Out Of Control Loops

Some operators think so, observes Terry Blevins, principal technologist with Austin, Texas-headquartered Emerson Process Management’s (www.emersonprocess.com) DeltaV Product Engineering group. “They’ll tune a loop using some procedure and get good performance. But the next day, the same loop
operates differently.”

That may be reasonable, because process dynamic and process gain may be functions of operating conditions, indicates Blevins, co-creator of the Weblog “Modeling and Control: Dynamic World of Process Control” (www.modelingandcontrol.com). For example, when the manufacture of different products requires control valve position changes, loop gain may change. So he expects variable performance of loops functioning over wide dynamic ranges, especially in batch operations.

Process performance stems from tuning, states Greg McMillan, an Austin-based consultant with Emerson’s DeltaV Research & Development group. “Therefore, if you can have a tool that identifies the tuning, then you have the potential for doing better and getting closer to what [the process is] actually capable of,” notes McMillan, Modeling and Control’s other co-creator.

He and Blevins helped develop a tool, Emerson’s DeltaV InSight, that lets end-users define dynamics to better understand performance. “It automatically identifies process gain and dead time. It [also] gives recommended tuning—but the operator still must take action,” Blevins explains. The tool can be set to show only those loops needing tuning, McMillan adds. While this recently released software currently finds use only with Emerson devices, a new release next year will be for use with others’ equipment.

洞察力还提供了数据如何调优更改with time. That’s important, McMillan stresses, because “one of the most critical things an end-user can do is to stay apprised of the variability in process conditions. As a process input changes—
for example, flows through a valve—the operating point changes, as do most process dynamics.” Therefore, observe gain, time constant and dead time—all three main process factors that are functions of process load, he advises.

Another critical item that McMillan believes requires oversight is a loop’s objective, which typically is reducing process variability. Sometimes, that’s achieved by slowing a loop to reduce noise as well as interaction with other control loops, he notes. This provides smoother, more coordinated responses.

But in the chemical industry, reducing process variability may mean accelerating the loop to reject unmeasured load disturbances at the process input, McMillan explains. He emphasizes that critical to this industrial sector is maximum disturbance rejection, particularly in exothermic reactor control. To tune this rejection, the most common tuning methods “come down to about the same formula for controller gain with a tweaked coefficient,” he adds.

Do what you can

While tuning for higher controller gain can reduce dead time caused by valve deadband and stick-slip, this tuning won’t eliminate the limit cycle, which is loop output changing constantly and repetitiously, observes McMillan. “The only thing you may change is the cycle’s period of oscillation.”

Whatever and whenever tuning is done, however, it requires skillful optimization. For example, with changing process dynamics, an operator “may tune too conservatively or sluggishly—more than what is actually needed to provide stable operation for all operating conditions,” Blevins observes. This type of tuning, generally means poorer process performance, he notes.

Such sluggish tuning may not be much of an issue for surge tanks, but for temperatures on critical products? “Any sluggishness in responding to a deviation of one degree above the target temperature could cost you,” McMillan says. Still, Blevins remarks, “ultimately, the best you can do is determined by what the process allows you to do.” No mystery there.

C. Kenna Amos,ckamosjr@earthlink.net,is anAutomation WorldContributing Editor.