Dynamic optimization integrating modifier adaptation using transient measurements.

• Real time optimization, Modifier Adaptation and Model Predictive Control is integrated in a single layer to improve the application in industrial processes. • The convergence time is speed up using a direct estimation of modifiers with transient measurements. • The new framework is applied to Williams-Otto reactor where the model and process have significant parametric and structural mismatch. • The results show that the proposed framework reaches closely the real optimal economic operation point. In this work, a dynamic optimizer (DO) is enhanced with elements taken from the Modifier Adaptation (MA) methodology for Real-Time Optimization (RTO). The objective is to use the capability of MA methodologies to reach the real optimum of a process despite the existence of process-model mismatch. The modifiers are computed during a process transient, without waiting for several steady states to occur, speeding up the time required to reach the plant optimum. The architecture proposed includes a modified dynamic optimization, a module for estimating the model states and an additional module for direct estimation of the modifiers during transient. The proposed approach was applied in a well-known benchmark example: the Williams-Otto reactor, where the model and process have significant parametric and structural mismatch. The results show that the proposed integration is capable of bringing the process closer to the real optimal economic operation point. [ABSTRACT FROM AUTHOR]