Using automation and controllers to eliminate unintended variability and errors in any manufacturing environment is certainly one way to stabilize both process or discrete production operations. Even if your production line consists only of hand assembly, without a doubt, there exists a controller of some sort in your plant. The process industries that rely on continuous operations, like refineries and chemical plants, also have an extensive network of automation and control.
Ignoring the trivial on-off approach, there are three modes of smoothly continuous control. The first is proportional control (P), wherein the magnitude of the corrective action the controller impresses on the system is directly proportional to the magnitude of the error — the difference between what you are getting and what you want. Increase the proportional gain and this controller provides a more intense corrective action. The second mode, integral (I), sets the magnitude of the corrective action proportional to the product of the error and the time the output has been deviating from the desired value. The last mode is called the derivative mode (D) and it generates a correction that is proportional to the rate at which the measured variable is deviating from the set point. The three modes can be used separately or in combination to provide more precise control action, hence the existence of a PI controller, a PD controller or a PID controller.
High-powered technology should eliminate variability problems, but that is not always the case. One Web reference claims that 80 percent of process control loops are causing more variability running in the automatic mode than in the manual mode. If that is true, I guess someone out there in readerland could use a little help. Another reference claims that proposed legislation in California would jail refinery officials found guilty of negligence after a refinery accident. Getting control of your control system seems to be the order of the day to avoid being the designated corporate felon.
Because control loops are ubiquitous, one would wonder if there are appropriate automation resources available on the Web. It’s a good question, but wonder no longer. Such is the case — in spades. We dove into the morass we call the Web in search of solid, non-commercial content that supports this month’s focus on instrumentation and control. Because the material to be found on the Web is stored in an inherently chaotic manner, locating pearls of wisdom is sometimes difficult and time-consuming. That is why we search the Web — so you don’t have to.
Eliminate the Tower of Babel
Every field of endeavor has its own jargon, argot, acronyms and other so-called communication-enhancing devices. It would be nice to have some sort of decoder ring to be sure that we are all speaking the same technical language. Malcolm Robins, a member of the Institute of Automation and Control Australia, Inc., provides a “Quick Reference to Instrumentation and Control Terminology”. Actually, the listing covers far more than straight instrumentation definitions. But that is beside the point. Point any browser to http://www.iica.org.au/dict/quikref.htm . If someone tries to baffle you with techno-babble, just go to this site. Then click on the appropriate letter of the alphabet to get instant access to the meaning of the mysterious gibberish. With a command of automation language, your journey into the Web of instrumentation will be easier.
They wrote the Web book
VT Mechatronics, an enterprise at Virginia Tech, aims to fuse mechanics and electronics in the design of intelligent machines. This organization publishes an online book dealing with the PIC16C84 processor and other devices. Go to http://www.mechatronics.me.vt.edu/book/contents.html to see the full contents of the book. Section 5, “Digital PID Control Theory”, is probably most relevant for the matter at hand. The first part provides an overview of PID controller theory and the second part delves into the details of PID theory.
Section 4 deals with sensors and data acquisition. It covers light, ultrasonics, strain gages and load cells. The individual pages are highly linked and the site will take you some time to explore fully.
Speaking of books, the University of Texas at Austin posts an online book on control theory to accompany a class in chemical engineering—ChE 360: Automatic Process Control. Head on over to wysiwyg://40/www.che.utexas.edu/~che360 to access 21 chapters of the text book used in this course. The chapters cover topics such as mathematical modeling, first and second order systems, fitting models to data, control loop analysis, multivariable control and much more. The graphics are of a high quality and the complex formulas that control theory requires are easy to read without ambiguity. This is a good site to use for continuing education or for refreshing skills.
Frequency response and Mr. Bode’s diagram
When set up as a regulator, a control loop responds to cyclical upsets by trying to maintain a constant output. The Bode diagram quantifies how the output follows the cycling by showing the relationship between the output and the frequency of oscillation. At low oscillation frequencies, the controller responds in one way. However, when the frequency of the oscillating upset increases, the controller reacts differently.
One weak analogy that might make sense of this is the suspension system in your car. Its function is to keep your chassis a fixed distance from the ground. Go over a speed bump ever so slowly and the entire car, suspension system included, rises slowly and settles down on the far side of the speed bump. However, hit the bump faster and the chassis remains at a constant elevation while only the suspension system retracts and extends. The point is that the control system responds differently to slow and rapid oscillation.
There is a lot of material about Bode diagrams authored by Tom Bullock of Industrial Controls Consulting in Fond du Lac, Wis. at remote.control.com/tutorials/bode.html . His article, posted to a Web site operated by Control Technology Corp., lets you visualize how well the controller output will follow the input signal. Getting a controller to function well involves adjusting the gains for each of the three control functions. Another Bullock article at remote.control.com/tutorials/adjus.html discusses the effect that changing the individual gains will have on the performance — a process called “tuning the loop.”
As an aside, if you want to get your book published, the Control Technology Corp. site also has a list of contact information for publishers of technical books.
Back to tuning a control loop
The degree to which a control loop provides minimum variability is a function of the interactive action of the three modes of control. If some level of control intensity is good, more is not necessarily better. For instance, excessive gain in a proportional controller leads to process instability. Minimizing variability requires tuning and optimizing the controller to achieve the best balance of proportional, integral and derivative control.
Fortunately, Jukka Lieslehto, at the Tampere University of Technology in Tampere, Finland, might be able to give you everything you need and more. Model your control loop using the information provided at http://www.ae.tut.fi/~juke/pidtuning/modidx.html . Jukka offers four models of a step response and one for an impulse response.
Jukka provides no less than 30 Java applets for tuning a wide range of models, including time constant (with and without delay), two time constants (with and without delay), three time constants, integrator (with and without delay) and more. Go see the list yourself at http://www.ae.tut.fi/~juke/java/pidtuning/tunidx.html.
The Department of Chemical Engineering at the University of Edinburgh offers something called “The ECOSSE Control HyperCourse—The Virtual Control Laboratory.” The site features simulation experiments to illustrate process control principles. The three main parts of the site cover an introduction to control, controller tuning and multiple loop control systems. The use of Java applets at various locations in the section on controller tuning makes for an interactive experience. You enter values for the control variables and the applet redraws a graph that displays the response of the controller to those input variables. This is definitely a good use of Web resources and your time. Found at http://www.chemeng.ed.ac.uk/ecosse/control/course/map, these pages are rich in links and provide a decent educational experience.
You can get your hands on a spreadsheet that calculates PID gains for common low-order process models. It is available as a zipped file and as a self-extracting zip file. It is brought to you by the good folks at CICS Automation and is accessible through the CICS Web site at http://www.cicsauto.com.au/unac_members/pid_calculator.html.
Thorn Technologies, Inc, developer of chemical process simulators, offers its base system and a PID module that allows users to develop skills in learning to adjust the PID variables. Go to the bottom of the Web page at http://www.thornti.com/index.html and enter your request.
The folks at ExperTune will let you load their demo software if you go to http://www.expertune.com/xpform.html . The name of this company gives you a good clue as to what it brings to the marketplace.
Some of these download sites require you to register if you want the software. It is only fair that you give them something — demographic information — in exchange for something they think is of value — software. When you get to the part of the online forms that ask where you heard about them and their Web site, make sure to tell ’em Russ K. at Plant Services sent you. That is demographic information that is more reliable and stable than anything you will get from a PID controller.