Study Of Process And Dynamics Model For Hydrogenation Of Coking Naphtha And Low-quality Diesel Fractions

Making coking naphtha react with low-quality diesel can take the advantage of heat fromthe coking naphtha after hydrogenating. This can reduce the temperature of the entrance ofreactor so as to avoid temperature’s rise quickly when the coking naphtha reacts alone andavoid some other problems, such as catalyst’s coking. This process has been widely used inindustry. In order to predict the hydrogenating reaction results according to the nature of feedsand the reaction conditions or select the appropriate process parameters according to thenature of materials and the requirement of hydrogenating reaction, the research of processcondition and kinetic model for the mixed hydrogenating reaction of low-quality naphtha anddiesel cut is imperative.In order to achieve a complete simulation for the hydrogenating process in industry, alaboratory fractionation unit for naphtha and diesel was established in this experiment. Thenits number of theoretical plates for fractionation was determined as9.9. And the conditions forthe separation process of naphtha and diesel were determined.The tests of the hydrogenating reaction of five different low-quality naphtha and dieselover Ni-Mo-P/Al2O3catalyst were carried out in different conditions such as temperature,space velocity, hydrogen to oil ratio and pressure. The effects of different conditions on thedesulfurization and denitrification ratio were studied as well as the density, the distillationrange distribution and the bromine value of the produced oil. The optimum conditions of thefeedstock used in the experiment were as follows: temperature was360℃, space velocity was1.5h-1, hydrogen to oil ratio was450, pressure was7.0MPa.N-order kinetics models for the low-quality naphtha and diesel were established. Theyhave been corrected by space velocity, hydrogen to oil ratio and pressure and their index have been obtained. The activation energy of the desulfurization and denitrification for the fivedifferent feeds were determined. The inspection found that the sulfur content and nitrogencontent between simulated and experimental values of the five feeds fit good performancethrough the kinetic model. The applicability of the kinetic model was strong.The models used to predict the desulfurization and denitrification ratio of the feeds wereestablished by means of momentum BP neural network, LMBP neural network and RBFneural network. The average relative errors of the three artificial neural networks forpredicting desulfurization ratio were0.57%,0.55%and0.69%, which shows good predictionperformance. The average relative errors of the three artificial neural networks for predictingdenitrification ratio were3.42%,3.11%and2.58%. They were all less than5%. The threemodels were able to achieve the requirement of predicting the denitrification rate in industry.RBF neural network was used to study the effects of feed properties (density, brominevalue,90%distillation point, nitrogen content, sulfur content and kinematic viscosity) andreaction conditions (temperature, space velocity, hydrogen to oil ratio and pressure) on thedesulfurization and denitrification ratio. This provided a basis for optimizing the operation ofthe device.