Data Analysis and Modelling of a Fluid Catalytic Cracking Unit (FCCU) for an Implementation of Real Time Optimization

Abstract

Abstract In the Fluid Catalytic Cracking Units (FCCU) large hydrocarbon molecules are cracked into smaller molecules, generating high value products such as diesel, gasoline and useful petrochemical olefins. The control of these units is fundamental to maintain a satisfactory operation. Hence, the Real Time Optimization has proved an interesting strategy. A dynamic simulation of a FCCU was developed using a phenomenological industrial validated model. A Dynamic Neural Network (DNN) was trained with data from the FCCU model and gross and systematic errors were added to employ this system as a virtual plant. Data from this virtual plant were used to study strategies of online data processing, considering steady state identification (SSI) and gross error detection (GED), in order to eliminate measurement noise, as the initial steps into an RTO implementation.