With an aim to optimize propylene production,
petroleum oil is catalytically cracked by contacting the oil with catalyst mixture consisting of a base
cracking catalyst containing an stable Y-type
zeolite and small amounts of rare-earth
metal oxide, and an additive containing a shape-selective
zeolite, in an up-flow type
fluid catalytic cracking apparatus having a regeneration zone, a separation zone, and a stripping zone. According to this
fluid catalytic cracking process, the production of light-fraction olefins such as propylene is maximized by applying appropriate
process control, monitoring, and optimizing systems. The present invention provides mathematical process models, including: neural networks, statistical models and finite impulse models. These various mathematical process models are used in conjunction with advanced controllers and optimizing routines to calculate optimal settings for various process parameters. A
microwave based
system can also be employed for optimizing the performance of a stripping zone, which further optimizes
catalyst regeneration. Process model and historical data to test a predictive
system can provide early warning of potential performance degradation and equipment failure in the FCC unit, which decreases overall operating costs and increases
plant safety.