Loop Control

Intoxicating answers

Control Talk columnists McMillan and Weiner describe the period of time between when you first take a drink and when you first recognize the effect and bypass the next round for coffee as dead time.

McMillan & WeinerBy Greg McMillan and Stan Weiner, PE

Stan: Have you ever been asked a question on process control that leaves you at a loss for words? Here is your chance to amaze friends and relatives (or at least baffle them) by learning some control concepts and analogies. If you understand the ideas and their significance conveyed in the answers to the following devastating questions from students in Greg’s class during his brief but illustrious career as a professor, you might just become famous by Friday. The answers never made it into a university text for obvious reasons.

Student: What is loop dead time?

Greg: Dead time is that period of time from the start of a disturbance until the controller makes a correction that arrives at the same point in the loop that the disturbance entered. The controller needs to see the upset, react to it, and send the correction to the right place. To appreciate dead time, consider when you go to a party and start drinking. The period of time between when you first take a drink and when you first recognize the effect and bypass the next round for coffee is dead time. As the dead time approaches zero, the portion of the open-loop error that appears as the closed-loop error approaches zero, and nonlinearities and high process gain (high proof and small body) become unimportant.

Student: What is the open-loop error?

Greg: Well, if you continued drinking, your state of intoxication would exponentially increase and reach a steady state with you passed out on the floor. The open-loop error is the alcohol concentration at this point. It is the result of having the controller (your mind) in manual or disconnecting it from the loop (your body). A smart party animal knows when to say "when," according to a famous seer.

Student: What is the closed-loop error?

Greg: The peak excursion of the alcohol concentration until the coffee first enters the bloodstream is the closed-loop error. For the well-tuned PI and PID controllers, the best you can do is limit the closed-loop error to the maximum deviation from set point that occurs at 150% and 110%, respectively, of the loop dead time. Often, the party mind does not behave like a well tuned controller, and the closed-loop error reaches an unacceptable level for driving, especially if there is a non-self-regulating process.

Student: What is a non-self-regulating process?

Greg: If your liver break down and remove alcohol at a fixed rate, the alcohol concentration ramps, which is characteristic of an integrating process. If an increase in intoxication causes a decrease in your restraint, you might start drinking faster. This is positive feedback, and the acceleration of the concentration corresponds to a runaway process. People with such non-self-regulating processes are easy to spot at parties; they are the ones with the lampshades on their heads.

Student: How can you reduce loop dead time?

Greg: When I first left home, my father said to me, "Be as honest as the day is long, don't talk when you should listen, and don't be fooled into thinking that dead time compensators can eliminate dead time from the loop." I thought this was extremely strange in that I only planned to walk around the block. Some professors must not have gotten the same words of wisdom because they are convinced they can mathematically cancel out the dead time term in a loop by an advanced control algorithm. Dead time cannot be reduced without violating the basic principles of classical physics. While the theory of relativity concludes that you could contract distances and dilate time and hence shorten transport delays as you approach the speed of light, unless you have Scotty and warp drive on your loop you are stuck with dead time caused by the vessels, piping, and instruments within your loop. If the loop dead time is well known and considerably larger than the largest time constant in the loop, a dead time compensator, such as the Smith Predictor, can make the loop perform better if you tune the controller with a smaller reset time. In general, your best bet is to look for the biggest sources of dead time and work on a change in the equipment, piping, or instrument design to reduce their contribution to the total loop dead time. If the disturbance is much slower than the loop dead time, the closed loop error is small. Have you ever had guests stay too long? Just start showing a video of you at work and simulate dead time by freeze frames. Some guests have hurt themselves in the rush for the door.

Student: What if you can't change the plant design?

Greg: Then your best bet is to use feedforward control for disturbances that can be measured when the dead time in the disturbance path is smaller than the dead time in the correction path to a common point in the process; use cascade control for disturbances that can be isolated by an inner loop when the inner loop dead time is several times smaller than the outer loop dead time. With the help of Mark Sowell, we conclude with a top ten list, which doesn’t have much to do with dead time but is still timely on the road to fame and fortune …. well maybe just fame.

Top Ten Signs You Have Too Many Alarms

(10) The console is more exciting than a pin ball machine.
(9) The only time there are no alarms is when the console is unplugged.
(8) Operators are deaf to the high pitch tones.
(7) The operator can play along with his favorite rap song on the acknowledge button without missing a beat.
(6) A woodpecker is being trained to keep pecking the acknowledge button.
(5) The alarm password is going for $50 on the Plant’s black market.
(4) Visiting the control room for a doughnut is not worth the noise.
(3) The alarm management consultant wants to showcase your alarm log in his training class.
(2) Hearing protection is required when a total download activates all your suppressed alarms.
(1) “Because they are free” is the basis for your alarm design strategy.

  About the Authors
Greg McMillan and Stan Weiner, PE bring their wits and more than 68 years of process control experience to bear on your questions, comments, and problems. Write to them at controltalk@putman.net.

Free Subscriptions

Control Global Digital Edition

Access the entire print issue on-line and be notified each month via e-mail when your new issue is ready for you. Subscribe Today.

controlglobal.com E-Newsletters

Biweekly updates delivering feature articles, headlines with direct links to the top news stories that are critical to staying up to date on the industry — company news, product announcements, technical issues and more. Subscribe Today.