1.Introduction
Since the environmental regulations are increasingly strin gent than ever, the production of clean automobile fuel is a vital target in China, especially after China has become a WTO member. However, th...1.Introduction
Since the environmental regulations are increasingly strin gent than ever, the production of clean automobile fuel is a vital target in China, especially after China has become a WTO member. However, the sulfur content in gasoline pool of China is still very high. Over 95% of the sulfur in gaso line pool comes from FCC naphtha. It is necessary to remove sulfur in FCC naphtha.展开更多
According to the characteristics of FCC diesel, a technology of liquid-phase hydrodesulfurization of the diesel in tubular reactors was proposed and lab-scale experiments were carried out. A kinetic model for the hydr...According to the characteristics of FCC diesel, a technology of liquid-phase hydrodesulfurization of the diesel in tubular reactors was proposed and lab-scale experiments were carried out. A kinetic model for the hydrodesulfurization process was developed and verified. The model was utilized to predict the sulfur content of products under different operating conditions. The effects of temperature, space velocity, pressure, and hydrogen concentration on the dcsulfurization rate were investigated.展开更多
Based on the Residual Oil Hydrodesulfurization Treatment Unit (S-RHT), the n-order reaction kinetic model for residual oil HDS reactions and artificial neural network (ANN) model were developed to determine the sulfur...Based on the Residual Oil Hydrodesulfurization Treatment Unit (S-RHT), the n-order reaction kinetic model for residual oil HDS reactions and artificial neural network (ANN) model were developed to determine the sulfur content of hydrogenated residual oil. The established ANN model covered 4 input variables, 1 output variable and 1 hidden layer with 15 neurons. The comparison between the results of two models was listed. The results showed that the predicted mean relative errors of the two models with three different sample data were less than 5% and both the two models had good predictive precision and extrapolative feature for the HDS process. The mean relative error of 5 sets of testing data of the ANN model was 1.62%—3.23%, all of which were smaller than that of the common mechanism model (3.47%— 4.13%). It showed that the ANN model was better than the mechanism model both in terms of fitting results and fitting difficulty. The models could be easily applied in practice and could also provide a reference for the further research of residual oil HDS process.展开更多
文摘1.Introduction
Since the environmental regulations are increasingly strin gent than ever, the production of clean automobile fuel is a vital target in China, especially after China has become a WTO member. However, the sulfur content in gasoline pool of China is still very high. Over 95% of the sulfur in gaso line pool comes from FCC naphtha. It is necessary to remove sulfur in FCC naphtha.
基金the financial support from the SINOPEC(No.2014310031600599)
文摘According to the characteristics of FCC diesel, a technology of liquid-phase hydrodesulfurization of the diesel in tubular reactors was proposed and lab-scale experiments were carried out. A kinetic model for the hydrodesulfurization process was developed and verified. The model was utilized to predict the sulfur content of products under different operating conditions. The effects of temperature, space velocity, pressure, and hydrogen concentration on the dcsulfurization rate were investigated.
文摘Based on the Residual Oil Hydrodesulfurization Treatment Unit (S-RHT), the n-order reaction kinetic model for residual oil HDS reactions and artificial neural network (ANN) model were developed to determine the sulfur content of hydrogenated residual oil. The established ANN model covered 4 input variables, 1 output variable and 1 hidden layer with 15 neurons. The comparison between the results of two models was listed. The results showed that the predicted mean relative errors of the two models with three different sample data were less than 5% and both the two models had good predictive precision and extrapolative feature for the HDS process. The mean relative error of 5 sets of testing data of the ANN model was 1.62%—3.23%, all of which were smaller than that of the common mechanism model (3.47%— 4.13%). It showed that the ANN model was better than the mechanism model both in terms of fitting results and fitting difficulty. The models could be easily applied in practice and could also provide a reference for the further research of residual oil HDS process.
