Last week, I taught a couple of classes at the Pacific Energy Center that were focused on control system (and controlled system) commissioning. As you might expect in a class on controls that focused on building control systems, the subject of PID or Proportional plus Integral plus Derivative control came up.
Controller Tuning 101
As a part of the class, we did a little lab session using a booklet and related software resource that I have been using for years.
Both the software and booklet are self-published by David St. Clair, who is a retired DuPont process control engineer. At one point in his career, David realized there was a need for some sort of entry level instruction book that described proportional closed loop control theory and tuning practice in layman’s terms and without the math. So, he set out to do just that. Via his publications, we are the benefactors of his efforts.
I first ran into David’s work when, as a fairly young (and inexperienced) engineer, I encountered the reality of proportional error in the field. You can see proportional error in this enlargement of the previous screen shot.
The black line is the response of a proportional control loop to a set point change from 50% to 70%. Notice how it stabilizes, but not at the desired set point. The difference between the control point, which ends up being about 64-65%, and the set point of 70% is the proportional error, which is an inherent characteristic of any proportional control loop.
As a young engineer, I did not really comprehend this until confronted with the reality of it while making a punch list for one of my projects. The work had modified a constant volume reheat system in a hospital with one of the goals of the project being to save energy. Missing your desired set point can be especially costly in terms of energy in a constant volume reheat system.
Recognizing that, I thought I could ensure that the air handling system controlled at exactly the temperature I wanted by specifying a very high quality controller. Upon inspecting the system in the field, I discovered that my expensive controller was not maintaining set point. And, worse yet, the control point wandered all over the place over the course of the day I was there.
While its true that precise control requires quality equipment, its also true that no matter how good a proportional controller is, it will still be subject to proportional error. It took me a (subsequently embarrassing) letter to the local branch office (written with all of the passion of someone trying to save the planet and the arrogance of uninformed youth) and another day in the field to realize the error of my ways (as well as the error inherent in any proportional control loop). The details of that story, along with an overview of PID are in a paper I wrote for ICEBO a while back if you are interested.
The silver lining to my somewhat devastating comprehension of proportional error was that it led me to investigate this new (to the HVAC industry at the time) “thing” called PID control, which in turn, led me to David’s work.
The red line in the picture above illustrates the benefit rendered by PI control. It is the response of the same control loop as the one that generated the black line, but with integral gain added. Notice how the loop takes a bit longer to stablize, but when it does, it stabilizes at the set point and the proportional error is eliminated.
The blue line in the picture below illustrates one of the benefits rendered by acquiring David’s publications.
The blue line is what happens if you start with a stable proportional control loop (the black line), add a little integral gain (the red line) with a little improvement, and then, being impatient, decided if a little is good, then a lot will be great (the blue line). If you make that mistake learning about PID using David’s software, then you say things like “Wow” and “I need to be careful adding integral gain”.
If you did that in the field with a working system, you probably would be saying other things. One of them might be “is anyone looking to hire an (experienced) control engineer?”
The fact that you can experiment with out fear of doing harm is one of the things that makes software tools like Loop Tuning 101 so handy. What’s particularly nice about Loop Tuning 101 is that a set of exercises are built in that walk you through all of the concepts in the booklet. When you pick the exercise of interest from the drop down “Exercise” menu, a help file with instructions opens up in a separate window.
Next, you type the exercise number in the box provided below the process parameters, which loads the exercise starting point into all of the variable locations. Having done that, you can experiment to your heart’s content and if things get too out of hand, you just reset everything and start over.
Unlike some of the other resources I write about, David’s products are not free. But at $40 total for both, they probably represent one of the best values for your money in terms of increasing your understanding of PID control (and as I write this, Christmas is just around the corner). Plus, you’ll be providing David with a little cash flow in his retirement, which I think is a nice way to say thanks to someone for sharing what they know with those who follow in their footsteps.
Visualizing Proportional Error by Studying a Float Controlling Level in a Tank
For me personally, a float controlling a valve to maintain a level in a tank was the window into really understanding proportional control and proportional error. In fact, I use that very analogy in classes when I attempt to explain the concept. Here is one of the slides from the current set on proportional control.
If you think that such an explanation would be beneficial to your understanding, I have loaded the slide set that I use currently when I discuss PID control onto my Google Documents account with public access rights.
Everything You Wanted to Know about PID (But Were Afraid To Ask)
If the discussion to this point has tweeked your interest in PID, then you may want to down load a collection of articles that was put together by the editors of Control Engineering.
The collection includes a number of articles of interest. My favorites include the original Ziegler – Nichols paper on PID tuning techniques, an interview with John Ziegler, a historical look at how pneumatic instruments led to automatic control, and the article that led me to David St. Clair’s stuff in the first place.
You used to be able to download these articles as a series of three. pdf documents from the Control Engineering web site. But currently, it seems like all of those links are broken, probably the result of the publication closing its doors for a while, just like CSE did. The have re-emerged and are up and running again and I have an inquiry in to some folks I know at CSE (they are both published by CFE Media these days).
I’ll let you know what I find out in a subsequent post. Meanwhile, I see that you can get a copy of the paper version on Amazon for about $15. To me, that would be worth it just to have the original Ziegler Nichols paper and the interview with John Ziegler in my library.
But I’m getting old (as someone in a class pointed out the other day). And perhaps, as a result, my interest in where the industry came from in addition to where its heading seems to be constantly growing. So the historical stuff appeals to me.
Who knows, maybe someday, when I’m put out to pasture, you’ll find me in volunteering at the National Building Museum talking to kids about how “back in my day we used to draw on Mylar and control things with air”.
Senior Engineer – Facility Dynamics Engineering