Economic analysis of different structures of on-line process optimization systems

This paper presents a method that estimates the likely economic benefit of implementing a steady state on-line optimizer on a plant. This benefit depends on the structure of the on-line optimizer. The structural decisions considered are the selection of the estimated parameters, the process measurements, the set points used for the control system and the optimization model. Based on first and second order information of the process and the optimization model the average deviation from the true process optimum can be calculated for every structure dependent on the measurement error, the parametric uncertainty and the structural modelling error. The method of the average deviation from optimum is demonstrated on a simple case study. The analytical results are compared against nonlinear Monte Carlo simulations.