SOME KEY ISSUES ON A SUCCESSFUL APPLICATION OF REAL-TIMEOPTIMIZATION TO AN ETHYLENE PLANT

摘要

Lyondell and Applied Manufacturing Technologies (AMT) have implemented a real-timeoptimization (RTO) system on the ethylene plant at Lyondell's Chocolate Bayou Facility.The RTO system contains an integrated plant model that is a combination of rigorous andsimplified plant section models covering both the olefins and aromatics units. The decision ofwhether or not to model a plant section rigorously or simply is based on whether the model isrigorous enough to capture the majority of benefits that an RTO system can bring, whilesimplified to an extent that the plant model represents the needed relations about the plantproduction, constraints and optimal moves. As a result, the RTO system can run very robustlysolving 99.9% of the time and implement setpoints at least every 30 minutes (including waitingtime after setpoint implementation). This type of RTO system responds to plant operatingcondition changes and active constraint changes more quickly than traditional RTO systemsthat run less frequently. An olefins RTO system based on this approach requires less effort inthe design, development and implementation phases, and since it is more robust, it requiresless effort to maintain.The RTO system working consistently with the advanced process control (APC)system is a key factor that directly affects the performance and success achieved by the RTOapplication. The interface between the RTO system and the APC system* covers setpointimplementation, constraint handling, and steady state target coordination. Additional factorsaffecting the success of an RTO system include: an effective way to acquire up-to-date feedstock composition and pricing information and a reliable operator interface.This paper will discuss this new approach to online optimization as well as the keyfactors determining the overall success of the application.