Results analysis for trust constrained real-lime optimization

In real-time optimization (RTO), results analysis is used to ensure that RT O predictions can be implemented and are not the result of the unnecessary variance transmission around the RTO loop. Miletic and Marlin [I. Miletic, T.E. Marlin, On-line statistical results analysis in real-time operations o ptimization, Ind. Eng. Chem. Res. 37 (1998) 3670-3684.] proposed a statisti cal framework for analyzing RTO results; however, their method cannot effec tively deal with inequality constraints. Many industrial RTO implementation s include bounds on the changes that the RTO system can make to the process operation (i.e. trust-region constraints). Such trust-region constraints c an seriously degrade the performance of existing results analysis methods. In this paper, a results analysis procedure is proposed that incorporates s tatistical testing on both the primal and dual variables of the optimizatio n problem to effectively analyze steady-state RTO results in the presence o f trust-region constraints. The proposed method is illustrated using two sm all case studies, one of which is the same Williams and Otto [T.J. Williams , R.E. Otto, A generalized chemical processing model for the investigation of computer control, AIEE Trans 79 (1960) 458-473.] reactor example used in Miletic and Marlin (1998). (C) 2001 Elsevier Science Ltd. All rights reser ved.