Peter WelanderPillar to Post: Peter Welander's Blog


News and comment from Control Engineering process industries editor, Peter Welander

PID trials and errors

February 12, 2010


After working on the story Understanding Derivative in PID Control with Vance VanDoren, I was interested in some extensive comments from Larry Trammell, a software engineer in Bellevue, WA. He apparently took issue with one of Vance’s comments in the article sidebar about early work on PID control development. He said, in part, “Ziegler and Nichols did not ‘discover by trial and error.’ Read their paper. In fact, their contribution was providing a mathematical basis for systematically analyzing response loops. I never disparage their work, but consistently point out that it applies only under severely restrictive assumptions.”

I thought Vance’s use of the description “trial and error” was an interesting one, but I also know that he is intimately familiar with Z&N’s work so he would know whereof he spoke. I have also seen situations where mathematical models grow out of observed behavior, so it isn’t always clear after the fact which came first. In any case, here is some additional thinking from Dr. VanDoren:

“Anyone who has studied analytical control loop design knows that the underlying mathematical principles required to predict the behavior of a closed-loop system can be daunting. Even with a ‘simple’ PID loop, it’s not always obvious how adjusting any one term will affect the performance of the controlled process because a change in the P, I, or D action affects the behavior of the other two. This problem was even more acute back in the late 1930s and early 1940s when Zeke Zeigler and Nate Nichols were trying to crack the loop tuning problem because the mathematical tools we use today were not fully developed back then. They don’t say in their famous 1942 paper exactly how they ended up with their tuning formulas, but later comments by Mr. Zeigler hint that there was a lot of trial and error involved. Nonetheless, their ‘whatever works’ approach has proven remarkably effective and continues to be used today.”

Whatever works, indeed. By the way, Vance’s article Loop Tuning Fundamentals may be the most heavily read article ever on our Website.