【作者】 徐欧官；

【导师】 苏宏业；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2007， 博士

【摘要】 芳烃是重要的基本有机化工原料，尤其是苯、甲苯和二甲苯（BTX），在现代国民经济中有着极其重要的地位和作用。芳烃的大规模工业化生产是通过现代化芳烃联合装置实现的，其加工流程复杂，伴有芳烃间的相互转化过程。碳八芳烃临氢异构化反应和甲苯歧化与C₉芳烃烷基转移过程是芳烃联合装置中重要的两个芳烃转化工艺，其目的分别是将异构体邻二甲苯（OX）、间二甲苯（MX）和乙苯（EB）转化为价值更高的对二甲苯（PX）和将直接用途较少、相对过剩的甲苯和C₉芳烃转化成用途广泛但供应不足的苯（B）和二甲苯（X）。过程建模是石化工业过程一个重要的研究课题，基于过程模型的工业综合自动化技术如软测量、先进控制、流程模拟与优化以及优化调度与计划等在石化领域越来越受重视。本论文以石化过程工业装置——芳烃联合装置中C₈芳烃临氢异构化单元和甲苯歧化与烷基转移单元为背景，探讨了俩个芳烃转化过程建模及模型应用中涉及的理论和工程应用实际问题。论文的内容分成两个部分，第一部分为两个芳烃转化过程的建模，即C₈芳烃临氢异构化和甲苯歧化与C₉芳烃烷基转移过程的建模；第二部分为两个过程模型的应用，包括模型的离线模拟和在线应用。本论文的研究为专用流程模拟与优化软件功能模块的划分提供了依据，主要包括以下研究内容：1．针对某实际工业异构化和工业歧化装置，分别建立了适用于工业生产的碳八芳烃临氢异构化反应动力学模型和甲苯歧化与C₉芳烃烷基转移反应简化机理模型，用于反应过程的监控、优化及反应器内物料组成分布的预测。所建立的过程机理模型结合了催化剂结焦失活模型，基于时间的经验失活模型考虑了氢分压（P_h）和重时空速（WHSV）的影响，能够合理地解释长期使用的催化剂的失活问题。针对过程机理模型参数数目过多且难以同时估计频率因子和活化能参数的难题，采用同系列催化剂的活化能和经验活化能，使所需估计的参数数目减少了一半。所采用的简化处理方法大大降低了参数估计难度，但引入了误差并将累积到频率因子上。采用Runge-Kutta法对模型方程进行数值求解，基于多套稳态平衡数据采用差分变尺度优化算法（BFGS）对频率因子进行估计，进而在不同操作条件下对模型进行验证，结果表明模型估计值与工业标定值相当吻合，达到了工业应用的模拟精度要求。所建立的模型形式简单，参数估计方便，适用于工业装置的离线仿真和在线估计。2．在已开发的异构化反应过程机理模型和歧化与烷基转移过程机理模型的基础上，讨论了过程模型的应用，包括模型验证、过程模拟和灵敏度分析。采用所建立的模型，对大批工业数据进行模拟，详细比较了关键组分浓度和工艺指标的模型估计值和实际观测值之间的偏差，以它为依据，对过程模型进行维护，以保证模型的估计精度；模拟了反应器内部各组分浓度的分布，然后详尽地分析了组分浓度模拟情况的合理性；最后讨论了操作条件包括反应温度、液时空速、氢烃比和反应压力对反应性能的影响。异构化反应和歧化与烷基转移过程的离线模拟充分体现了模型验证和过程模拟这两个功能，而反应过程分析则包含了反应器内部的模拟和灵敏度分析。这些内容的讨论和分析为异构化和歧化反应过程专用流程模拟软件功能模块的划分提供了重要依据。3．针对国内工业异构化装置和工业歧化与烷基转移装置，研究了过程机理模型的在线应用。文中讨论了在线软测量技术实施过程中所涉及的关键问题，且采用模型参数在线更新策略修正模型参数，以保证模型的预测精度。应用APC—Sensor软测量软件实现过程模型在工业装置上的在线计算，计算结果表明模型的预测值能够很好地跟踪实际值的变化。机理软测量模型具有同时计算多个生产指标的特点，模型在线计算的成功应用为先进控制和优化的实施提供了保障。最后，在总结全文研究工作的基础上，对需要进一步完善和深入研究解决的部分问题进行了探讨和展望。更多还原

【Abstract】 It is well known that aromatics are key raw materials and important starting materials for many intermediates and petrochemicals, playing an important role in the modern national economy, benezene, toluene and xylenes in particular. The commercial production of aromatics is achieved by modern aromatics combination unit with complex processes accompanying transformation between aromatics. The modern aromatics combination unit is assembled properly with many processes together including process of C₈-aromatics hydroisomerization and process of toluene disproportionation and transalkylation with C₉-aromatics. The process of C₈-aromatics hydroisomerization is to transform isomers such as o-xylene, m-xylene and ethylbenzene to more valuable p-xylene and the process of toluene disproportionation and transalkylation with C₉-aromatics is a convenient way to get more valuable and widely used benzene and xylenes from surplus toluene （C7） and trimethylbenzenes（C9）.Modeling is a key issue in petrochemical industry, and more and more attentions are payed to the integrated automation technologies based on process model such as soft sensor, advanced process control, process simulation and optimization, process scheduling and planning in modern petrochemical industry.The dissertation takes the practical commercial unit, unit of C₈-aromatics hydroisomerization and unit of toluene disproportionation and transalkylation in aromatics combination unit, in petrochemical industry as the research background, focusing on several typical theory and engineering issues in petrochemical process modeling and its applications. Two main parts are involved in the dissertation, one relating to process modeling of two aromatics transformation processes, namely process of C₈-aromtics hydroisomerization and process of toluene disproportionation and transalkylation with C₉-aromatics, and the other dealing with applications of the two established models, off-line simulation and on-line application. The contents discussed here are the foundation of functional modules for special process simulationand optimization softwares and the details are arranged as follows,1. Based on the reported reaction networks, new kinetic models for commercial unit of C₈-aromtics hydroisomerization and unit of toluene disproportionation and transalkylation with C₈-armatiocs are developed for process monitor, process optimization and prediction. A time based empirical catalyst deactivation function considering hydrogen partial pressure （P_h） and weight hourly space velocity （WHSV） is incorporated into the models, accounting for the loss in activity because of coke formation on the catalyst surface during the long-term operation. As to the difficulty of simultaneous estimation for frequency factors （k₀） and activation energies （E_a） as well as the difficulty of many parameters to be estimated at the same time, the activation energies of the same kind catalyst and empirical activation energies are adopted, hence the number of parameters is reduced to half. The kinetic parameters are benchmarked with several sets of balanced plant data and estimated by the differential variable metric optimization method （BFGS）, while the ordinary differential equations of the models are solved by the Runge-Kutta method and the errors introduced are inevitably accumulated to the frequency factors. The kinetic models are validated by industrial units with sets of plant data at different operating conditions and simulation results show a good agreement between the model predictions and plant observations, which means that the developed model are suitable for off-linesimulation and on-line soft sensor application of industrial units.2. The off-line applications of the developed models about process of C₈-aronaticshydroisomerization and process of toluene disproportionation and transalkylation with C₉-aromatics are discussed in detail including model verification, process simulation and sensitivity analysis of operating variables. First of all, large number of balanced plant data are simulated with the developed models, and the comparisons of key components concentration and technologic indexes between model predictions and plant observations are made for the warranty of modelmaintenance; then the distribution of the components inside the reactor is simulated by the models and its rationality is showed in detail; at last, the effect of operating variables such as reaction temperature, weight hourly space velocity, molar ratio of hydrogen-to-hydrocarbons and reaction temperature on the process performances is investigated. In fact, off-line simulation involves model verification and process simulation, and process analysis incorporates simulation of the reactor and sensitivity analysis. The above-mentioned discussion and analysis point out functional modules of the special process simulation and optimization softwares for two aromatics transformation processes.3. The on-line applications of the first principle process models are studied for domestic industrial units about aromatics transformation processes, C₈-aromatics hydroisomerization and toluene disproportionation and transalkylation with C₈-aromatics. The key problems about practical operation of on-line soft sensors system based on process models are discussed and the strategy of on-line model rolling-revision is proposed for the warranty of model prediction accuracy. On-line computation is applied to the industrial units with APC-Sensor soft and the results show that the model predictions agree well with plant data observed. The high performance of the model predicting many indexes simutaneously is presented and its successful on-line application guarantees the operation of advanced process control and optimized process control. The dissertation is concluded with a summary and perspectives of someimportant issues for further investigation.更多还原