【作者】 张金利；

【导师】 胡宗定；

【作者基本信息】 天津大学 ， 生物化工， 2003， 博士

【摘要】 添加柴油降凝剂生产低凝柴油，可使部分蜡油馏分深度切割成为柴油馏分，拓宽了柴油馏程，节约了资源；同时具有操作灵活简便、费用低、见效快，保留高十六烷值的蜡组分等特点，是一种增产高品质柴油的有效方法，对于提升我国炼油企业的国际竞争力具有重要的作用。本文首先对乙烯—醋酸乙烯（EVA）类柴油降凝剂进行了合成与降凝效果研究，获得了该类聚合物的合成规律，初步考察了其降凝效果，为实验研究该类降凝剂作用机理提供了实验样品。通过对柴油低温析蜡过程中蜡晶形貌观测、柴油降温过程的DSC测量和蜡晶的XRD研究，从实验的角度对该类降凝剂的作用机理进行了系统的研究，确定了该类降凝剂的降凝机理为共晶和吸附共同作用，为该类降凝剂的分子设计与优化提供了基础。采用密度泛函、半经验量子化学、分子力学和分子动力学等方法直接从微观上对EVA类降凝剂分子与烷烃分子之间的相互作用规律及其在蜡晶晶面上的吸附能和吸附规律等进行了模拟计算。计算表明，其分子中的极性基团增加了分子的局部刚性，而分子的非极性部分与烷烃分子保持了很好的亲和性，从而使其可以进入蜡晶格形成共晶。在各种乙烯—醋酸乙烯—丙烯（EVAP）降凝剂中，当EVAP酯含量在30％左右，且在每个链段含有一个靠近酯基的甲基支链时容易在蜡晶的成核和生长过程中发挥作用。根据分子模拟和初步降凝实验结果，降凝剂EVAP是降低冷滤点效果最好的降凝剂。因此，对分子设计的EVAP降凝剂进行了实验合成，并进行了放大实验和工业生产；将工业产品在我国部分炼厂进行了工业应用，获得了较其它降凝剂更好的降低冷滤点的效果。为了使工业应用EVAP进行加剂低凝柴油的生产更为方便，针对燕山石化公司炼油厂的工业应用情况，以五种组分柴油的质量百分组成、粘度、折光指数、馏程宽度、冷滤点及降凝剂加入量共26个参数为输入节点，隐含层50个神经元，调和油冷滤点为输出层节点，构成神经网络，且设定MSE为0.5℃时，对空白与加剂调和柴油冷滤点的预测结果最好，该神经网络模型可以较好地指导工业柴油调和。更多还原

【Abstract】 Pour point depressant (PPD) additives are very useful to produce diesel fuels with excellent low-temperature flow properties. Components of the paraffin distillate can be partly extended to produce diesel oil by using the PPD additives, which has been practically applied in industrial processes by taking advantages of improving the flexibility and the profits of the product, as well as maintaining the high values of cetane components. Firstly, the synthesis of ethylene-vinyl acetate-based (EVA) PPD were investigated and the synthetic principles for the copolymers were obtained. According to the characterization of the relevant functions, it was suggested that the copolymer of ethylene, vinyl acetate and propylene (EVAP) had the optimal efficiency to decrease the cold filter plug point(CFPP) of the diesel oil. Then the functional mechanism of this kind of PPD was intensively studied by using the inverted microscope, the differential scanning calorimeter, the X-ray diffraction, etc. The mechanism was determined as the function of co-crystallization and adsorption, which provided the basis for the exploring of new PPD through the method of molecular design and optimization. Further it was investigated that the mutual molecules actions between the EVA and the alkanes by using density function theory, semi-empirical quantum chemistry, molecular mechanics and molecular dynamics, etc. Adsorption energy was calculated at the planes of paraffin crystals and the adsorption principles were simulated. It was indicated that the polar groups enhanced the rigidity of the corresponding molecules and the non-polar groups kept good affinity for the alkane molecules, which made it possible for the PPD molecule to invade into the crystal lattice and form co-crystllization complexes. The simulation results suggested that the EVAP showed significant effects on the nucleation and growth of the paraffin crystal when the acetate concentration was around 30 % and there existed one branched methyl group near the acetate group in each block. Based on the above molecular design, the synthesis of the optimal PPD additives <WP=4>were investigated and the operation conditions were optimized. The scale-up effects were studied and the industrial produce of the PPD additives was accomplished. Cooperating with some oil refineries in China, the practical applications of the PPD products obtained effective decrease for CFPP, superior to other commercial PPD additives. Finally, an artificial neural network (ANN) model was established, according to the industrial requests of the oil refinery in the Yanshan Petroleum Chemical Co., to estimate accurately the CFPP of the blended diesel oil with and without the PPD additives. The ANN model consisted of 26 input nodes including the component fraction, viscosity, refraction index, distillate range, and CFPP for five classes of diesel oil, and the PPD additive amounts, together with 50 hidden nodes and one output node of the CFPP of the blended diesel oil. When the mean square error (MSE) was set as 0.5 oC, the ANN model showed the optimal accuracy to estimate the CFPP of the blended diesel oil with and without the PPD additives. The obtained ANN model has promising applications in instructing the industrial process of blended diesel oil manufacture.更多还原