Study On State Simulation Of Catalytic Cracking Riser Prediction Model

Fluid catalytic cracking unit(FCCU)is the main unit of heavy oil processing in refineries which is mainly used to produce light oil such as liquefied gas,gasoline and diesel oil.In the past half century,with the increasing demand for more clean energy such as liquefied petroleum gas and olefins in the market,FCCU simulation technology has great importance for the optimization and research.The main difficulties in the construction of riser model for catalytic cracking lie in:(l)the characteristic expression of oil gas in feed oil and riser outlet products used in secondary cracking process;(2)There are huge reactions occur in the catalytic cracking reaction system in riser,so it is very difficult to construct a continuous reaction network in accordance with the reaction mechanism;(3)the whole riser model mechanism modeling process contains a large number of partial differential equations,and there are non-linearities among the reaction variables.Because of the complexity of the riser mechanism model,there are many parameters involved in the whole process,which makes it difficult to optimize and identify.Based on the above problems,the purpose of this paper is to optimize the original operation mode and process conditions in the modeling process of riser mechanism model.The specific work is as method was used to characterize the feed oil,which were considered the effollows:(1)The special pseudo-components fects of feed characteristics and molecular structure on the reaction system.The concept of catalytic cracking yield function from Sildir et al.is introduced,and a new reaction network form is proposed:LSPCj??pkj·(1-j)LDCk+?pi+dj(1-j)LSPCk.Only light special pseudo-components(LSPCs)and true components(LDCs)participate in catalytic cracking reaction,while heavy special pseudo-components(HSPCs)didn't participate in the reaction.The reaction rate formula rL,i for LSPCs is futher given,which simplifies the huge reaction network system.Experiments show that the model has good fitting and predictability for riser outlet temperature and oil gas.(2)Based on the original riser hydrodynamic model,11 parameters are identified by parameter optimization method,and eight groups of riser industrial data are introduced to verify the fitness and predictability of the model.The results show that under the same riser conditions,the parameters of riser model needn,t be re-identified,and good simulation results can still be achieved.(3)In order to further verify the reliability of the riser model proposed in this paper,the riser mechanism model is used as the first principle model for production data.The simplified model is trained by BP and RBF neural network algorithm,respectively.The results show that the model has higher accuracy and faster response speed,and the reliability of the riser mechanism model is well validated.At the same time,the simplified model of riser doesn,t need to construct complex mechanism equation,but also can efficiently generate experimental data,which provides data support for advanced control field.