Study On The Method Of Predicting The Distribution Of Slurry-bed Hydrocracking Products

In this paper,the hydrocracking of coal tar vacuum residue(CTVR)was studied in the autoclave at the different conditions(the reaction temperature,initial pressure,reaction time,catalyst concentration)in order to simulate the residue slurry-bed hydrocracking process.Based on the experimental data,lump kinetic and BP neural network was used to predict the distribution of slurry-bed hydrocracking products.The results show that slurry-bed hydrocracking products of CTVR is influenced by reaction temperature,initial pressure,reaction time and catalyst concentration and the degree of influence vary with reaction depths.Based on the experimental data,the applicability of the 9-lump kinetic model was studied.According to the reaction characteristics and rules of CTVR optimized 9-lump kinetic model and developed the 8-lump kinetic model which was applicable to the CTVR.Based on this model,gasoline and diesel were combined into a lump and regarded coke as a important component of the condensation and cracking reaction.According to the reaction rules,the kinetic equation was derived and the kinetic parameters were solved.The results showed that all the errors were less than 10.0% and most of the errors were within 5.0%.In particular,the error of coke had fallen markedly.The model was verified through the experimental data under the condition of 385? and 9MPa.The relative error of prediction of each product is within 7.0% which prove that the established 8-lump kinetic model is suitable to CTVR.Using BP neural network predicted the product yield of slurry-bed hydrocracking of CTVR and established a BP network model of 8-12-5 type.The result of prediction show that all the errors are less than 10.0%.However,the extension of the BP neural network model is poor and lead to the error of prediction is large.