A new PI-PD smith predictor for control of processes with long dead time

Abstract Good control of processes with long dead time is often achieved using a Smith predictor configuration. Typically a PI or PID controller is used, however, it is shown in this paper that for some situations improved setpoint and disturbance responses can be obtained if a PI-PD controller is used. Several methods are possible for selecting the parameters of the PI-PD controller but when the plant transfer function has no zeros the use of the standard forms provides a simple algebraic approach, and also reveals why difficulties may be encountered if a PID controller is used. Some examples are given to show the value of the approach presented.