Development And Industry Applications Of Cracking Severity Control System For Ethylene Plant Of Jilin Petrochemical Company

Cracking severity of is a very important index for cracking reaction in ethylene cracker. In the operation of crackers, the cracking severity must be controlled in a suitable range to obtain a given ethylene and propylene product rates. But it is hard to get an effective mechanism models because the cracking severity dynamics is a complicated and nonlinear process. In this thesis, the soft sensor method was introduced into cracking severity control system. Firstly, BP neural network technique was applied to model the relationship between the cracking severity and the operation variables of the cracker. The model can predict the cracking severity according to the feedstock properties and the operation condition. After that, the controller responses the change of the outputs of the model. Secondly, the fault diagnosis programs are designed based on the certain familiar faults and operational experience. Finally, an application software was developed which had a friendly human-machine interface and was convenient and reliable for daily operation and maintenance. The practical applications of the cracking severity control technique in crackers F0801 and F0901 of Jilin Petrochemical Company showed that the cracking severity control system of ethylene cracker can be adapted the change of the feed composition with good reliability and stability, and bring good economic and social efforts for the Enterprise.