Improvement And Application Of Narrow-lump Chevron Model For Hydrocracking Of Petroleum Fraction

Chevron model has always been the basic model for the mechanism study of narrow-range lump model of hydrocracking of petroleum fractions.But the accuracy of the model is limited because of some flaws,including the hypotheses of constant reaction heat(21MJ of heat will be released for every 1kg of new hydrogen consumed in all reaction systems regardless of the nature of raw materials,product scheme,reaction temperature and pressure)and constant reaction activation energy(i.e.,the reaction activation of all lumps on the k series of catalytic beds),and the way of fitting the model parameters according to the principle of minimizing total errors of temperature distribution and product distribution.Thus,the dissertation has implemented the improvement from the following aspects:1.Based on the equilibrium of carbon and hydrogen elements in the reaction process and the method of the thermodynamic state function,the reaction heat is calculated by the standard combustion heat of reactants and reaction products of petroleum fractions,so as to help the treatment of reaction heat rise from experience to theory.2.In view of the logic of the differences of catalyst packing,temperature distribution and composition of reaction system evitably lead to the difference of reaction heat.To change the assumption that the reaction activation energy of all the catalytic beds are identical,different reaction activation energy Ek for each catalytic bed are specified,and the model parameters are increased from 6 to(5+k).3.The multi-objective genetic algorithm(NSGA-II),which is based on the simultaneous minimum of reaction temperature error and product distribution error,is used to replace the single objective genetic(GA),which is based on the weighted sum,in order to overcome the error caused by the empirical distribution of weight factors and improve the fitting accuracy of model parameters.4.Based on the minimum error between the restored flow and the actual flow,genetic algorithm(GA)is used to optimize the lump distribution coefficient.A method is proposed to minimize the errors between the restored product and the real product,because of the opinion that the flow and proportion of each lump forming the same product are totally identical with those calculated from reaction system in the process of restoring the calculated lumps to products,The improved model is applied to a 6-stage reactor of a 200×104 tons/year hydrocracking unit of VGO.The result shows that the average errors of the calculated temperature and product distribution between calculated ones and actual ones are only 1.34% and 8.54%,which are enhanced by 0.81% and 2.31% compared to traditional model,with the first and second correction.And the average errors of the calculated temperature and product distribution between calculated ones and actual ones are only 1.57% and 1.85%,which are enhanced by 0.58% and 9.02% compared to traditional model,with all the correction.The model with correction,new parameters fitting and new restoration method can be well applied to actual production.