REDUCED ORDER MODELLING AND DYNAMIC SIMULATION OF DISTILLATION COLUMN

Abstract

The simulation of dynamic model of a distillation column is used for design of control schemes, online identification and optimization purposes. The response to any change in the input is obtained from the model in a reasonable amount of computer time. In order to integrate dynamic model equations in time, a steady state solution is required. This can be obtained from Napthali-Sandholm steady state method or from parameter estimation, by optimization. The dynamic simulation model can be classified into three categories named as (1) rigorous (2) semi rigorous and (3) simple dynamic model. Optimal design with full order stage wise models requires the solution of mixed nonlinear programming problem. The problem is compounded by the fact that the number of equations varies with the number of stages. The solution of such a system of differential equations is not straightforward because of the algebraic equations being nonlinear and differential equations being stiff. One scheme is being the possibility of reducing the total no of equations to be solved. Reduced order model is an outcome of research efforts in that direction. Reduced order model is developed to achieve the saving of considerable amount of computation time and hence facilitates its usage's for online identification, optimization and control with a little loss in accuracy compared with full order model. It preserves the stage wise structure of the model and also predicts the state variables at any stage