McMillan, an Austin, Texas-based private consultant who develops and tests new, advanced control algorithms for Emerson Process Management (www.emersonprocess.com), also based in Austin.
Myth No. 1 is that it is always better to use only one tuning method. This is impractical, McMillan says. That’s because of the diversity of practically everything involved: the processes themselves, and valves and control algorithms, as well as the objectives.
Myth No. 2 is that you can precisely compute controller settings. McMillan says that if you attempt to get more than one significant digit, you’re likely wasting time. That’s because the match to the plant system is transitory. Further, for each test conducted, expect to get an answer unique to only that test, he says.
McMillan lays out these myths, as well as 34 rules of thumb for PID loop tuning, in a publication from The Instrumentation, Systems and Automation Society (ISA, www.isa.org): “Good Tuning: A Pocket Guide.” First published in 2000, a second edition is slated for publication this spring.
The most common problem occurs in more critical control loops, he says. Those loops are slow, meaning they have large time constants and long dead times—that interval from the start of a disturbance until the controller makes a correction that arrives at the same point in the loop at which the disturbance entered.
Tuning loops associated with large distillation columns is particularly problematic, he believes, because the time to achieve steady-state exceeds the length of a work shift in a continuous process environment. For batch operations, that’s either because there is no steady-state, or the time to reach steady-state approaches or exceeds batch cycle time. McMillan advises that the most reliable manual procedure for these processes is using a shortcut tuning method. It should look only at the initial change in ramp rates and should not need steady-state condition.
There are three key points to consider when tuning loops, he advises. One is that processes are nonlinear. “The process gain, dead time and time constant will change with operating point and load,” he notes. That means any attempt to get more than one or two significant digits in tuning settings is an exercise in frustration, McMillan asserts. His second point is that the step size in output or the set-point for a tuning test should be at least five times the noise band, valve dead-band (or dead time) and unmeasured load upsets.
The final point is that unmeasured disturbances will deform the response. “And stick-slip will cause a sustained sawtooth oscillation in the controller output of a self-regulating loop and in the controlled variable of an integrating loop,” explains McMillan. Stick-slip, which is commonly created by excessive packing friction in pneumatically actuated valves, arises when the controller continues to increase output but the valve position does not change immediately.
Using common sense is important, too. McMillan’s guide suggests that “before entering new settings or trying tuning methods, ask the operator for permission to tune the loop. Also ask him or her what the maximum allowable size can be for each direction of a step change in the controller set-point and output without causing a problem.” And understand this: There are practical limitations to the great expectations you may have about tuning loops.
C. Kenna Amos, [email protected], is an Automation World Contributing Editor
