Greg McMillan and Stan Weiner bring their wits and more than 66 years of process control experience to bear on your questions, comments, and problems. Write to them at email@example.com.
Stan: We conclude our tour of PID world with Mark Congiundi. There are plenty of other attractions, so we encourage our readers to take a road trip, and share their travel experiences in a future "Control Talk" column by contacting Control's managing editor, Nancy Bartels.
Greg: We have seen over and over again in the last three columns how external feedback has been used by Mark to greatly extend the capability of the PID. One point perhaps overlooked in the discussion is that the external feedback signal is added to the output from the proportional and derivative modes after the signal goes through a filter whose time constant is the reset time in the positive feedback method of integral action. This method of implementing reset action is the key that opens the door to an incredible creative spectrum of possibilities.
Stan: What do you do for low controller gain applications or when the external feedback from the secondary loop or final control element (e.g., valve position or drive speed) is inaccurate?
Mark: I simply subtract a "leeway" from the external feedback signal when this feedback is greater than the primary PID output. Otherwise, I add a leeway to the feedback. The normal allowed differential in percent between the PID output and the external feedback is one repeat (percent error multiplied by gain). This technique increases the allowable differential by the leeway. The PID has room to roam with the enhanced external feedback, but is still limited.
Greg: I can see how this would help when valve position feedback is slow from the use of second, third or fourth HART variables for feedback, or inaccurate due to the location or type of positioner feedback mechanism. Actuator shaft rotation for positioner feedback on rotary valves with high friction in the sealing surfaces or backlash in the linkages is not indicative of internal control element (e.g., valve disc or ball) position due to dead band, stick-slip and shaft windup as discussed in "Improve Control Loop Performance." (See the article in our sister publication, Chemical Processing, www.chemicalprocessing.com/articles/2007/200.html.)
Stan: What about abnormal conditions?
Mark: I provide sustained or one-shot easy valve trip action by use of the PID output track mode. Two simple examples are for high level and high pressure. For high level, the flow loop PID output tracks a trip signal until the condition clears. At a lower level, the output is released to the PID for flow control. For high pressure, the flow loop PID tracks the trip signal for a specified pulse time, and is automatically released to flow control.
Greg: I have used a similar one-shot strategy for flow control of phosphorous. When the control valve position would start to drift upwards due to plugging, the PID would track a wide- open signal for a few seconds to sweep out the solids accumulating around the ball and seal, and then go back to flow control. This strategy is useful whenever solids can build up in a valve.
Stan: How do you get smooth control of a cyclic measurement?
Mark: I configure a synchronized, moving average. The cyclic measurement is an input to a series of dead-time blocks. The total dead time of the series of blocks is the period of the cycle. The number of dead-time blocks used depends on the resolution required. The moving averages of each dead-time block for a segment of the period are added together and divided by the number of dead-time blocks to get the synchronized moving average. A perfectly synchronized moving average will reflect the true average accurately in the shortest possible time. The dead times for each block can be updated if the period changes. When the total dead time matches the cycle time, the noise is completely eliminated. In comparison, a filter attenuates, but does not eliminate an oscillation, necessitating the reduction of PID gain to keep the PID output oscillation amplitude from proportional action within the dead band and threshold sensitivity of the final control element.
Greg: This technique would have solved the finisher level control application I faced decades ago in polymer manufacturing. The teeth on the internal rotating drum would pass between the nuclear level source and detector, creating a predictable oscillation in level measurement. Tight level control meant tight residence time control and, therefore, better polymer quality control.
Stan: It is interesting how things change yet remain the same. Despite the PID being the workhorse of the process industry with expanding opportunities, courses and books just scratch the surface of what the PID can do. Advances in the industrial application of the PID by talented individuals are generally lost as they retire. We feel extremely fortunate to have captured some of Mark's knowledge for the benefit of the profession.
Greg: The IFAC conference on Advances in PID Control 2012 in Brescia, Italy, (http://pid12.ing.unibs.it/) is a once-in-a-decade opportunity to learn and share the wealth of PID opportunities. Also check out Mark's and my presentations at ISA Automation Week 2011 in Mobile, Ala. (www.isaautomationweek.org/). I will be doing tutorials on reactor control in the automation track, pH in the environmental track, and wireless control in the wireless control track. All of them will show the power of the PID. Many of the techniques in the tutorials are documented in the chapter "Industrial Applications of PID Control" in the soon-to-be published book by Springer, PID Control in the Third Millennium: Lessons Learned and New Approaches, as noted in my blog post, "PID Unleashed" (http://modelingandcontrol.com/2011/03/pid_unleashed/). This chapter and the ISA tutorials show how external feedback and the enhanced PID can be used for optimization, equipment and environmental protection, and riding out measurement and communication failures.