Why do we call it loop tuning?
If a process can stick to the setpoint, isn't that enough?
Dear Control Engineering: Why do we call it loop "tuning?"
The word tuning can be used in various contexts. We talk about tuning a guitar or piano (but not a fish) so the pitches are correct and harmonious. Back in more analog days, we had to tune the radio to get it on a given station for highest fidelity and reduce interference. We also use the term to describe getting a controller to make a process run in the way we want it to. It suggests an action that requires finesse and ongoing adjustment.
Making a process run the way we want it to requires some thought as to what is supposed to happen. Consider this illustration: My sister used to live near Pasadena, CA. I remember visiting her and spending some time in the part of town where the Tournament of Roses Parade passes through. There was a colored line painted down the street, and she told me that line was put there so the parade float drivers know how to steer. Since most can’t see out the front, they watch the line through the bottom and steer so as to keep the float centered over the line. This always gives them the optimal position in the street.
The line takes corners very gently with the widest possible radius so as to preclude any violent maneuvering. I suspect the turning radius of a parade float isn’t very tight. It isn’t something that could make a u-turn or parallel park easily. Steering actions are probably precise but minimal, however that is all that is required to drive the parade route. The driver’s objective is to make the float move gracefully.
Contrast that with a different image. Picture a narrow, winding mountain road with craggy rocks on one side and a sheer cliff on the other. The line down this road was painted by a man who’d had too much to drink and tends to wander even more than the road. If you had to drive down this road at 40 mph and all you could see was the line, you probably wouldn’t want to do it in a parade float. The float’s limited steering would not allow you to make turns sharp enough to avoid hitting the rocks or going off the cliff. Here you need a driver that can make very fast reactions with a vehicle that can make sharp turns.
When we say we want a controller to regulate a process, it’s like saying we want the vehicle to stay on the line. However some processes operate like the parade and others are more like the mountain road. Some are stable due to process inertia. Let’s say you have a one-inch pipe coming out of a tank that can hold 100,000 gal. of water. Pressure is provided only by gravity. The pipe has a flowmeter and a control valve configured such that you can set it for a constant flow of 5 gpm. Water comes into the tank from a source that can vary from 2 to 20 gpm. Even if the input to the tank is behaving erratically, the tank’s volume will dampen those changes and the control loop regulating the output will see very slow changes in pressure. It can operate like the parade float and should be tuned in a way that will make adjustments slow and precise.
As an alternative, consider the same scenario where the tank is 50 gal. If the supply is erratic, the level in the tank could change very quickly which will make the pressure feeding the valve change quickly. Here the controller regulating our flow loop would have to make more violent adjustments to compensate and keep output stable. Theoretically, the same controller, flowmeter, and control valve could suffice in both situations because they can be tuned to change the nature of the response. Understanding how the P, I, and D factors interact to control the action in a specific process context is the skill of a loop tuner. Process conditions and objectives change from situation to situation, so each requires its own adjustment for optimal performance.
Vance VanDoren has written extensively on the topic of PID tuning. One of my favorites is The Three Faces of PID.
Peter Welander, firstname.lastname@example.org