The Smith predictor function block has six inputs and one output. The implemented predictor structure contains a FOPDT model which means that actual process dynamics are approximated with first order plus dead time dynamics. In addition to the estimated process gain, time constant and dead time and cycle (sample) time, the controller output and the process variable are needed in order to calculate that output that is going to be used as the feedback signal in the closed loop system. They cycle (sample) time must match the time interval at which the function block is executed i.e. time interval of OB processing the Smith predictor.
The Siemens S7 300/400 series is chosen as the implementation platform. The Smith predictor is based on first order plus dead time model of a process. This fact doesn’t exclude it from use with more complex processes.
As dead time increases, the process is harder to control. Dead time generally decreases gain and phase stability margin. For this reason, a smaller controller gain must be applied which decrease loop performance. The Smith predictor function block is used first with a FOPDT process, where it improved loop response and allowed use of higher controller gain. Tests were also made on an S7 simulation model of a system with second order plus dead time - SOPDT dynamics with under damped response. In this case, the FOPDT Smith predictor provided much better response compared to the loop response without the dead time compensator.