Adaptive control of a chemical process system

Abstract

Modification of parameters i n the conventional controller equation usually i s necessitated by uncontrollable process parameter drifts which occur during the operation of the large scale continuous processes typically found i n the chemical industry. T o this end, a normalized version of the model reference adaptive control system, including suitable procedures for adjusting the adaptive loop gains, was developed and demonstrated t o provide excellent adaptive performance for a single concentration control loop for a simulated stirred-tank chemical reactor. The three constants i n a conventional PID controller were simultaneously adjusted t o accomp l i s h this adaptation. Equally excellent adaptive performance was achieved i n an interacting control loop scheme simulated by control of both the concentration and the temperature i n the same stirred-tank reactor.I n both cases adaptation was achieved i n a time interval which was sufficiently short to make t h i s method of adaptive control feasible for use i n a real processing situation. Models needed for implementation of the model reference adaptive control scheme were easily developed. I n a real application these models could be easily developed from studies which could take place during prestartup or early operations of the process; hence no a priori knowledge would be required. A wide range of adaptive loop gains was demonstrated to provide a stable overall control system. A high degree of stability was demonstrated for the overall system despite the initiation of large extraneous load upsets occurring during the operation of the adaptive control system.The method presented has not been rigorously proven to be applicable t o every real control problem requiring adaptive control. However, the method was demonstrated t o be highly effective over a wide range of required controller adjustments i n cases similar to those encountered i n large continuous processes. Thus the method merits serious consideration for possible implementation whenever adaptive control is required.The availability of the conventional PID (proportional integral derivative) controller has permitted the chemical engineer to design control systems satisfactorily for most complex plants despite incomplete knowledge. During prestartup or early operation of such a plant, the parameters available for adjustment in the PID controllers are set, often by trial and error, to provide the plant with good dynamic performance in the presence of the inevitable process upsets. However, uncontrollable parameters which can vary slowly with time, such as catalyst activity, may exist. These parameters could vary in such a way that as the process i s operated, the overall dynamic performances of the control systems might be adversely affected if some technique to retune the controller parameters i s not employed. Adaptive control addresses itself to this particular problem. An adaptive control system can be defined as an on-line systematic technique for modifying the values of available adjustable paramet...