Abstract: The main fractionation system is the primary energy-consuming unit of the hydrocracking unit, where the feed over-vaporization rate and product separation degree directly affect unit operation and energy consumption. Based on this, a rigorous mechanistic model constructed using PRO/Ⅱ software was employed to simulate the entire process of the main fractionation system in a hydrocracking unit. The results revealed that the feed over-vaporization rate determines the top cooling load of the main fractionation column, the furnace heat load, the bottom steam injection rate, and the heat load of the jet fuel/diesel side-line column. The heat load of the diesel side-line column and the furnace load showed weak correlation with the separation degree δ₁ between jet fuel and crude gasoline. The heat load of the jet fuel side-line column was largely unaffected by the separation degree δ₂ between diesel and jet fuel. Subsequently, a 4-5-7 structure BP neural network surrogate model was established to optimize the main fractionation system, with the objective function being the minimization of annualized total cost. The genetic algorithm was used to solve for the optimal combination of feed over-vaporization rate and product separation degree. The results indicated that the optimal feed over-vaporization rate is 40.67%, and the optimal product separation degrees δ₁ and δ₂ are -0.06 ℃ and -0.64 ℃, respectively. Under these optimal conditions, the main fractionation system can increase economic benefits by 6.669 million Yuan per year.

Key words: hydrocracking unit, main fractionation system, over-vaporization rate, separation degree, rigorous mechanistic model, BP neural network, surrogate model