Abstract:

In the construction of smart factory in petrochemical enterprises, the establishment of molecular level device models is an important element in the transformation of enterprises to a refined and intelligent production model. In this paper, based on the naphtha composition and molecular type-homologue series (MTHS) matrix model, a deterministic molecular library of naphtha containing 270 molecules was established and a naphtha molecular reconstruction model was constructed, whose simulation values matched well with the actual values, achieving the goal of predicting the detailed composition of naphtha fractions by using macroscopic physical properties as input information. The reaction network simplification rules were established based on the reaction mechanism of catalytic reforming process, and rule input network generator (RING) was innovatively applied to a complex mixture such as naphtha to construct a catalytic reforming process reaction network containing 865 molecules and 6616 reactions. The genetic algorithm was used to estimate the reaction kinetic parameters and construct a kinetic model of catalytic reforming reaction at the molecular level, and the absolute error between its simulated and actual values for the product composition of the plant was 0.85%, and the product prediction at the molecular level can be realized. The molecular level model can be used to guide the operation of catalytic reforming plants and to build an intelligent model, the model construction method and ideas can be used to build molecular level models for petrochemical companies.

Key words: catalytic reforming, reaction kinetics, molecular type-homologue series matrix, reaction network, smart factory, genetic algorithm, computer simulation

摘要：

在石化企业智慧工厂建设中，建立分子水平装置模型是企业向精细化、智能化生产模式转变的重要内容。本文以国内某石化企业催化重整装置为背景，基于石脑油组成和分子类型-同系物（MTHS）矩阵模型，建立含有270种分子的石脑油确定性分子库，构建石脑油分子重构模型，其模拟值与实际值吻合良好，实现了通过宏观物性作为输入信息来预测石脑油馏分详细组成的目标；基于催化重整的过程反应机理建立反应网络简化规则，创新性地将基于规则的自动网络生成器（RING）应用到石脑油这类复杂混合物中，构建包含865个分子和6616个反应的催化重整过程反应网络，并采用遗传算法估算反应动力学参数，构建分子水平催化重整反应动力学模型，其用于装置产品组成的模拟值与实际值的绝对误差为0.85%，可实现分子水平上的产品预测。本文的分子水平装置模型可用于催化重整装置的操作指导和智能化建设，模型的构建方法及思路可用于石化企业相关装置分子水平模型的构建。

关键词: 催化重整, 反应动力学, 分子类型-同系物矩阵, 反应网络, 智慧工厂, 遗传算法, 计算机模拟