THE SIMULATION OF OPTIMISATION OF A 3- TERM CONTROLLER ACTION IN A CRUDE OIL DISTILLATION UNIT

Abstract

The optimization of a digital computer configured in a direct digital control systems to work as a proportional- plus- integral- plus derivative controller subject to a quadratic constraint in the Warri Refinery and Petrochemical Company is presented. The design uses state- space techniques, operating on measurements of the plant inputs and outputs. A multivariable self- tuning controller analysis with inputs and output values in the state- vector is adopted. Self-tuning control includes two substantial steps, which are executed sequentially in each control cycle. One is parameter identification and the other is the control law construction based on the defined goal which is the optimum gain setting for the controller. The algorithm developed for the system is based on the transfer function in the Z domain with z-1 defined as a unity delayer. The controllability and observability of the control system is investigated and hence the development of a state observer is considered. The observer design criterion is based on the assumption that the transfer function matrix from the input to the observer state be equal to that from the input to the system state. The discrete-time plant is assumed to be stabilizable (controllable) and detectable (observable), linear and time invariant; hence change of state does not depend on initial time but depends only on the length of time during which the control force is applied. This is the sampling period. The input responses and the output responses obtained at each performance condition are presented. The optimum performance condition is determined by the least square technique. This is shown in the appendix.