RESULTS DIAGNOSIS FOR REAL-TIME PROCESS OPERATIONS OPTIMIZATION

Past work has shown that results analysis based on; knowledge of the c ommon cause measurement variability and the transmission of this varia bility in the real-time optimization (RTO) loop to the recommended ope rating conditions can be applied to an RTO system to increase operatin g profit and to decrease unnecessary operations changes. An extension to results analysis which detects and diagnoses the cause of ill-condi tioning in parameter updating, via singular value decomposition, is de veloped here and is. applied to a case study.-The case study has shown that it is possible to evaluate conditioning on-line. With an appropr iate selection of an updater control limit a hypothesis test can preve nt ill-conditioned parameter estimates from being used for plant optim ization. (C) 1998 Published by Elsevier Science Ltd. All rights reserv ed.