An improved iterative real-time optimization scheme for slow processes

Abstract

Iterative Real-Time Optimization (RTO) has gained increasing attention in the context of model-based optimization of the operating points of chemical plants in the presence of plant-model mismatch. In all these schemes, it is necessary to wait for the plant having reached a steady-state to obtain the required information on plant performance and constraint satisfaction, which leads to slow convergence in the case of processes with slow dynamics. This works addresses this issue by considering both parametric, and structural plant-model mismatch. First, a simple approach to determine the type of plant-model mismatch with the use of transient data is discussed. An approach for dealing with parametric mismatch based on a sensitivity analysis of the nominal dynamic model is presented, and its performance is evaluated with the case-study of a Continuously Stirred Tank Reactor (CSTR), where fast convergence to the optimum can be obtained, even with noisy measurements. For the case of structural mismatch, nonlinear system identification is integrated with iterative RTO. The identified models are used to predict the steady-state of the system, thus reducing the total optimization time. The performance of the strategy is illustrated by simulation studies of a CSTR and a hydroformylation process. It is shown that a mixed scheme, where both a linear and nonlinear model are used for steady-state prediction, results in fast convergence to a neighborhood of the true optimum, even in the presence of measurement noise. The use of taylored nonlinear models for dynamic system identification is shown to be a promising approach for reducing the time necessary to reach the optimum of a process.