Nonlinear model based distillation control

Abstract

In a fractional distillation process which performs a binary or limited multicomponent separation, and in which it is desired to maintain composition set points for dual product streams, a computer is programmed for applying model based control. Multivariable, noninteracting control of the distillation process is achieved by utilizing a nonlinear model, which calculates process vapor and distillate flow rates required to maintain product set points. The model, which applies tray-to-tray calculation essentially derived from a McCabe-Thiele analysis, is periodically updated by adjusting tray efficiency based on steady state composition measurements to keep the model in pace with changing process conditions. Guards to exceeding distillation process constraints, while keeping the distillation column running are provided by detecting the constrained condition, specifying which of the dual products is more valuable and then substituting constrained values for certain normally measured process variables used in the model calculations. In this manner, normal control actions are overridden when a process constraint is encountered.