A question often asked by novice sales persons and other technical personnel is: what is PID, and how do I tune my control to match my process conditions? I have been asked this question many times by people new and old to the wonderful world of process instrumentation. This question is very often answered in mathematical equations and/or technical doubletalk. Tuning a process loop can be easily performed without the mathematical equations as long as you have a basic understanding of what each of the PID values do.
There is one additional requirement in being able to tune a process loop, and that requirement is patience. Being able to tune a loop requires making one tuning parameter change at a time, documenting the results, then if necessary, making another parameter change. Remember that all control loops are a function of time, and time is what you have to invest. After making each tuning parameter change, disrupt the process by changing the controller’s setpoint by at least 10% of the controller’s input range and then monitor the process’s new controlled response.
First things first. The term PID stands for Proportional, Integral and Derivative. The controller parameter for Proportional is referred to as Proportional Band or Gain. The Integral Parameter will be Reset, and the Derivative Parameter will be Rate. There are often additional tuning parameters that the controller has available. Some of these will include Cycle Time, Dead Band, Load Line, and On/Off Control.
In the beginning of process control, the controllers were proportional only. For basic control loops, not requiring precise control, this was a great innovation. The development of proportional control didn’t require a man having to turn a valve handle a little more open or closed to maintain the temperature of a steam chamber. As requirements for tighter control became more demanding, Foxboro Corporation invented the Integral or Reset tuning parameter. The biggest problem with Proportional only control was "Proportional Offset." Proportional Offset would cause the controlled process to overshoot or undershoot the process. It would look something like this:
The Proportional Offset would vary in size from process to process or even from setpoint to setpoint. Some individuals would try to eliminate the Proportional Offset by fudging the controller’s setpoint. However, as outlined above, the Proportional Offset may vary from setpoint to setpoint.
After The Foxboro Company invented Integral (Reset), controller setpoint stability was greatly improved. Reset is used to eliminate the proportional offset caused by Proportional only controllers. Reset is a function of time. The longer the process variable deviates from the controller’s setpoint, the more the Integral adds or subtracts from the output value of the controller. In short, Integral, when properly tuned, will allow the measured process variable to match the controller’s setpoint (see illustration below).
Now enters Derivative (Rate). Derivative was developed when it was noticed that on some processes, even with good PI tuning, the response to process upsets was very slow, and sometimes out of control. Derivative helps the Proportional and Integral tuning functions respond in relation to the rate of speed that exists in a process upset. A good example of a need for Rate is when a door is opened on an oven causing the process temperature to drop more quickly than normal. The faster the process variable moves away from setpoint, the more the Derivative adds or subtracts from the controller’s output.
