Frequent real-time optimization using dynamical disturbance observers

The presence of disturbances and uncertainties calls for real-time optimization (R;TO) of chemical processes, such that the operational optimality is restored. The classical formulation of the two-step RTO rehes on steady state measurements for model updates and re-optimization, which is often slow in the optimizing performance. To accelerate the RTO speed, this paper proposes a new frequent RTO scheme that can be peformed in the trnsient phase. In the control commumty, the dynamic disturbance observers have been widely adopted for setpomt tracking. In this study, the disturbance observers are employed t? identify and compensate the plant-model mismatches for for the RTO purose. Efficient dyamic disturbance estimation makes 1t possible to extract information that is required for performmg the static optirmzations, without the need of reaching steady states. A CSTR example is studied to illustrate the proposed approach.