石油炼制与化工 ›› 2015, Vol. 46 ›› Issue (7): 15-22.

• 基础研究 • 上一篇    下一篇

基于结构导向集总的渣油分子组成矩阵构建模型

倪腾亚,刘纪昌,沈本贤,孙辉   

  1. 华东理工大学化学工程联合国家重点实验室
  • 收稿日期:2014-12-01 修回日期:2015-01-16 出版日期:2015-07-12 发布日期:2015-06-26
  • 通讯作者: 刘纪昌 E-mail:liujc@ecust.edu.cn
  • 基金资助:
    国家自然科学基金项目;中央高校基本科研业务费专项基金资助项目

MOLECULAR COMPOSITION MODEL OF RESIDUE BASED ON STRUCTURE ORIENTED LUMPING

  • Received:2014-12-01 Revised:2015-01-16 Online:2015-07-12 Published:2015-06-26
  • Contact: Liu Ji-chang E-mail:liujc@ecust.edu.cn

摘要: 渣油组成信息对重油轻质化加工工艺的选择具有重要的指导作用,但由于渣油组成的复杂性,传统方法难以对其组成进行分子水平的描述。采用结构导向集总方法,设计了包含烃类结构、杂原子结构及重金属结构的21个结构单元,构建了代表渣油分子组成的55类共2 791种典型分子的结构向量。采用模拟退火算法计算渣油的分子组成矩阵,使烃类组成信息和平均分子结构参数的计算值与仪器测定值相吻合,构建了基于结构导向集总的渣油分子组成计算模型。结果表明,采用该模型对渣油进行分子组成矩阵构建后,渣油的残炭、密度等性质指标和芳碳率、芳环数等结构参数的计算值和实验测定值吻合较好。表明基于结构导向集总方法可以对渣油组成进行分子水平的定量描述。

关键词: 结构导向集总, 分子组成模拟, 结构参数, 模拟退火算法

Abstract: The composition information of residue is critically important to the selection of heavy oil cracking process. Due to complex composition of residue, it is difficult to describe the composition of residue by traditional method at the molecular level. Based on the Structure Oriented Lumping (SOL) method, 21 structure units (metal structural groups, hydrocarbon groups and heteroatom groups) and totally 2 791 kinds of typical molecular structural vector were built to describe the molecular composition of residue. This paper built a molecular composition calculating matrix of residue calculated by simulated annealing algorithm, which makes the hydrocarbon composition information and the calculated average molecular structure parameters in conformity with the instrument measurements. The results show that simulated values of residual properties such as CCR, density and structural parameters (fA, CA) are in good coincident with experimental values. This model, based on SOL could be used for quantitative description of other residual composition at molecular level.

Key words: structure oriented lumping, molecular composition simulation, structural parameters, simulated annealing algorithm