Tuning a proportional-integral-derivative controller (PID controller) is a common activity for engineers specializing in process control. In this case, "tuning" refers to changing the parameters relating to the controller proportional band, integral action and derivative action. There are several methods for calculating tuning parameters by hand and numerous software packages that can be used to automatically tune controllers in a chemical process. Before any tuning can begin, it is crucial for the engineer first to investigate the control loop being tuned and the impact of the control loop on the overall system.
The performance of an automatic controller can be adjusted and changed by altering the controller's tuning parameters. When tuning a PID controller, there typically are three settings that can be changed: the proportional band, the integral action and the derivative action. These are represented by the first, second and third terms in the classic PID algorithm, respectively u = KP e + KI ∫ e dt + KD de/dt.
The term u represents the return signal; KP is the proportional gain; e is the error or offset term, which represents the difference between the present value and the controller setpoint; KI is the integral gain, KD is the derivative gain; and t is time. The Laplace transform of this equation can be stated as KP + KI/s + KDs.
Before tuning a PID controller, an engineer should first examine the process to be tuned to determine if improper tuning is causing upsets or if there is another assignable cause, such as malfunctioning or broken equipment. Tuning changes will mean very little if the true cause of variability is found to be a sticking control valve, broken instruments or errors in control system logic. Only when the process has been thoroughly examined and the field instruments' functionality has been verified should tuning be considered.
There are multiple methods used by chemical, electrical and instrument engineers in tuning a PID controller. The Ziegler-Nichols method is one such example that uses the ultimate gain and the ultimate period of the process to calculate aggressive tuning parameters for P-only, PI-only and PID control schemes. Other control schemes, such as the Tyreus-Luyben method, are formulated to reduce system oscillation. The method used for tuning a PID controller may be dictated by the nature of the control loop itself.
In general, increasing the gain term of a controller will make the controller act more aggressively. More integral action will help reduce the offset between the steady state value and the desired setpoint but may lead to oscillations if too much is used. The derivative term is used to help stop rapid movement of the controller's present value. These are only heuristics that provide a general sense of the effect of each of the classic tuning parameters.
Many distributed control system (DCS) packages include software that can be used to automatically tune control loops. These software packages often will tune processes by examining past performance or by automatically performing the test methods described by established tuning procedures. As with most procedures, fine tuning and small adjustments must be made by the engineer to suit the process after the major tuning procedure has been completed.