Why are PID loops so difficult to master?

Getting PID loops to operate well is a major concern in most process industries. Loop tuning related topics are the most heavily researched subject on the Control Engineering Website, year after year. Tuning articles are always heavily read. One would think that by now, loop tuning techniques would be mastered by middle school, but such is not the case.
By Control Engineering Staff June 1, 2008

Getting PID loops to operate well is a major concern in most process industries. Loop tuning related topics are the most heavily researched subject on the Control Engineering Website, year after year. Tuning articles are always heavily read. One would think that by now, loop tuning techniques would be mastered by middle school, but such is not the case. Conventional industry wisdom suggests that one-third of loops work reasonably well, one-third are running fully in manual, and one-third limp along with periodic fiddling. Other sources suggest perhaps half of loops perform adequately.

So why are loops so hard to tune? At the ABB 2008 user group meeting, several reasons were suggested:

  • Process units frequently operate at production levels that are simply not suited to the equipment. In the real world, process hardware is designed and built for specific production levels and feedstock characteristics. However, production usually deviates from these, which makes a plant harder to control. The greater the deviation, the more control suffers.

  • Some processes, due to the dynamics and interactions of the process itself, are simply more difficult than others.

  • Process units frequently have multiple loops that interact, and one or two that misbehave can throw others off. The tricky part is identifying which are the most strategic to fix first. It can be difficult to differentiate cause from effect in the real world.

  • One size does not fit all. Parameters can be sensitive to operating levels. When a plant is running at full capacity, it may behave. When running at lower capacity, loop parameters may need to change to remain stable.

  • Hardware problems will frustrate you every time. If instrumentation, valves, and so forth do not perform reliably, the process unit will never perform reliably. The data may tell you what’s happening, such as having valves that stick or hysteresis problems, if you can spot the signs.

Software is available that may provide a solution or at least mitigate the problem. In the May 2008 issue of Control Engineering, our loop expert, Vance VanDoren, offers an in-depth analysis of control loop management software that may help point you in the right direction.