Pi Controllers Performances for a Process Model with Varying Delay

Varying delay systems represent a serious challenge in many facets of process control. A frequent issue that arises in practice is introduced by transportation delays in fixed lengths pipes at speed which varies with setpoints. Many classic control techniques can be used to deal with constant delays systems but they do not specifically address this structural delay variability. In this paper we present a process model (Diesel Hydrodesulfurization) that features this delay variability and explore robustness properties of a wide panel of PI controllers. A conclusion is that the recent method proposed by Tavakoli and Fleming compares favorably with all others, including Smith predictors, when the delay variation is not known.