Set point weighted modified Smith predictor for delay dominated integrating processes

Abstract

A process having any pole at origin is said to be integrating in nature. In practice, it is difficult to obtain the desired output from such integrating processes with significant time delay. Pure integrating processes have inherent non self-regulating feature and hence if they are disturbed from their equilibrium condition, process output continuously fluctuates over a considerable period of time. In practice a number of industrial processes like combustion chamber, distillation column, chemical reactors are well-known integrating processes with considerable time delay. Smith predictor based control technique is an established methodology for controlling such processes with considerable dead time. But, this technique fails to perform satisfactorily for pure integrating processes with considerable time delay. Modified Smith predictor method by Majhi and Atherton may be considered to be a good alternative. However, its performance is not found to be quite satisfactory due to undesired overshoot and sluggish recovery. In addition, complexity lies with the tuning of three controllers involved in multi-loop structure. To overcome this limitation, a simple tuning methodology is proposed here for modified Smith predictor with two controllers only. Efficacy of the proposed mechanism is substantiated through performance evaluation of pure integrating processes with significant time delay in comparison with well-known modified Smith predictor tuning reported by Majhi and Atherton.