Steady-state-gain identification for multivariable process control

Abstract

The steady-state gains of a multivariable process contain much information about the control properties of the process. Identification of the process gains when the process operates in closed loop has advantages to open-loop identification. The process may be more linear in closed loop, the experiments may be designed to give more well-conditioned data, and the ordinary process operation may be less disturbed by the identification experiments. Closed-loop gain identification is illustrated for the distillation process. Open-loop gains for any control structure can then be calculated through control structure transformation rules from closed-loop gains, which are independent of the controlling structure used for the identification. The method is illustrated by results obtained from a nonlinear distillation column model and by experimental data from a pilot-plant column