Optimal Sensor Selection for Successful Real-Time Optimization

In Real-Time Optimization (RTO) systems, the perfomance of the RTO depends strongly on the selection of measured variables and of sensors used for their measurements. Many studies have investigated the selection of variables (e.g., Krishnan, et al., 1992; Fraleigh, et al., 1998). This study extends previous work by considering the error in the implementation of the RTO results due to imperfect sensors used in process control. A practical optimal sensor design strategy based on nonlinear closed-loop Monte-Carlo simulations is developed. Case study results demonstrate that the implementation error can be essential for determining the appropriate sensors