A review on modified Smith predictor schemes for delay dominated processes

Abstract Prolong dead time is quite common for industrial processes due to transportation lag, cascaded time constant elements like distillation column, recycler, demineralization tank etc. In addition, presence of computational and communication delay, data conversion phase, time taking measurements also contribute considerable time delay. In such cases, Smith predictor (SP) based control scheme is recommended to ascertain desirable close-loop behaviour. However, SP performance suffers considerably with model uncertainty and measurement noise. Consequently, different SP augmentation schemes are proposed, referred as modified Smith predictor (MSP). A good number of publications on MSP designs are reported by the researchers over the last two decades. Various model based MSP schemes are explored here to point out their suitability especially with delay dominated integrating and unstable processes due to their challenging behaviour. To ascertain improved set point tracking as well as efficient disturbance rejection multiple number of controllers are being utilized in MSP design. Choice of appropriate filter plays crucial role towards performance enhancement of MSP methodologies. Close-loop performance analysis among the widely reported MSP schemes are made with integrating plus time delay (IPTD), unstable first-order plus time delay (UFOPTD), integrating first-order plus time delay (IFOPTD), double integrating plus time delay (DIPTD) and unstable second-order plus time delay (USOPTD) processes. In addition to graphical responses, quantitative performance indices along with stability margins are also computed. In addition, efficacy of selective MSP schemes is also validated through real-time experimentation.