Manipulated variable based PI tuning and detection of poor settings: an industrial experience

A new method, based on the physical interpretation of the manipulated variable, is developed for tuning of proportional-integral-derivative (PID) controllers and detection of poor settings. This variable is used to set specifications and tune controllers for set-point changes. Here, these specifications are to have a closed-loop time response that is similar to the open-loop time response and a no static error to nonzero mean disturbances acting at the process input. The method is applicable to self-regulating first- and second-order models (overdamped) having stable or unstable zeros as well as delays to integrating first-order processes. For second-order processes, an approximation of equal time constants is used in order to facilitate the identification with step changes and tuning of PI controllers. Results show that the tuning is near optimal even when the time constants are very different. The method has been taught to people with different backgrounds: process engineers, maintenance technicians, production operators, as well as managers. Results have shown better understanding of the possibilities and limitations of process control and improved communication between the different groups.