Prediction Of Sulfide Distribution In FCC Diesel Fuel And Its Hydrodesulfurization Using BP Neural Network

With the crude oil quality is becoming inferior and the environmental protection regulations are being strict gradually, and there is also a tendency for the feedstock to FCC equipment that to be blended with residual oil or refine heavy oil entirely, it is important to study and develop the process for producing diesel oil with low sulfur content. The sulfur compounds in FCC diesel oil include benzothiophenes derivatives and dibenzothiophenes derivatives. The study indicated that the sulfur content and sulfide distributionin FCC diesel are different with the different source of diesel and the different catalytic cracking process. In order to select the more efficient catalyst and process, it is necessary to make it clearly before hydrodesulfurization.The sulfur content and sulfide distribution in different FCC diesel distillate were studied. The results showed that the sulfur content increased with the boiling points. In addition, the benzothiophenes derivatives concentrates in the distillate that its boiling points were lower than 290℃, and the dibenzothiophenes derivatives concentrates in the distillate that its boiling points were higher than 290℃. BP artificial neural network (ANN) was used to establish a prediction model for the sulfur content in FCC diesel. The biggest relative error of the prediction is 3.28%. BP artificial neural network method is also used to establish a prediction model for the sulfide distribution in FCC diesel in FCC diesel, and the biggest relative error of prediction is 4.95%.The deep hydrodesulfurization of diesel was studied. The influences of reaction conditions were investigated. BP artificial neural network method is used to establish a prediction model for the temperature of hydrodesulfurization. The biggest relative error is 1.3%. The model was used in the prediction for the temperature of hydrodesulfurizationthird of the other batch of FCC diesel, and the relative error is 1.6%.The model of deep hydrodesulfurization for the third batch diesel was established, and the influences of reaction conditions were studied. The prediction model for the reaction temperature was established by BP artificial neural network method, and the biggest relative error is 1.5%.