Simulation And Optimization Of Naphtha Catalytic Reforming Process And Its Application

As the basic of integrated control technologies in modern petrochemical industrials, the technology of process simulation and optimization is widely applied and brings large profits for related plants or componies. This dissertation takes the importantsecondary process of refinery-----catalytic reforming as the research background, andfocuses on the simulation and optimization of this process and its typical industrial applications. The lumped kinetics model and corresponding process model of catalytic reforming is firstly developed in this dissertation. Based on this process model, the industrial applications comprise developing special simulation software, devolping secondary simulation software on ASPEN PLUS platform, on-line computation, constraint optimization and multi-objective optimization. The detailed content is arrangend as follows,1. A new kinetics model involving twenty lumped components and thirty-one reactions is developed for catalytic reforming. The deactivation of the catalyst in semiregenerative or continuous catalytic reforming process is modeled by two different methods. Sequential modular approach is then implemented for the process modeling of catalytic reforming, which is composed of reformers, heaters, heat exchanger and separator. Aiming at industrial application, the traditional and steady algorithms are selected to solve process model equations and unconstrained optimization problem deriving from parameter estimation. The process model is the foundation of all the following industrial applications.2. The special process simulation and optimization software for catalytic reforming is developed with C++ language based on the 20-lumped kinetics model. This software possesses good structure of functional modules, integrated model library, steady and fast algorithms and powerful input and output system. As commercial software, it has been applied upon one domestic industrial catalytic reforming unit successfully. The software is proved to be effective and convenient for application. In addition, the sensitivity analysis is also performed with this software to guideprocess operations.3. By developing the process model as a user module, a whole industrial continuous catalytic reforming process is simulated on ASPEN PLUS platform. Fair agreement between the calculated and actual operating data is obtained. Based on the process model and the built-in SQP algorithm, process optimization is then studied and the calculated optimization results are also tested on the actual industrial unit about one month. The testing results show that the aromatics yield increases about 0.49 wt% averagely, which is close to the calculated result and makes a profit of about six million yuan annually.4. Based on the process mechanism model, linear PLS regression model and first-order TSK fuzzy neural network model, the on-line computation of production index for one industrial continuous catalytic reforming process is studied. The process model is proved to have such performances of stronger robustness and generalization, fewer samples for modeling needed and computing more than one index by using the same model and model parameters. By adopting an on-line prediction and correction strategy, the process model is used to on-line calculate aromatics yield, yield of each aromatics lump, and RON (Research Octane Number) of the process. The on-line prediction trend and precision are very good.5. A constrained optimization strategy of industrial catalytic reforming process is proposed based on the process model. A genetic algorithm based on infeasibility degree (IFDCOGA) is selected to compute this problem. This genetic algorithm is proved to have shortcomings of premature and dead state convergence during optimization computation. Aiming at overcoming its shortcomings, a new optimization method is proposed by integrating IFDCOGA with traditional algorithms. The feature of this method is its faster convergence speed and convenience because of not using the genetic algorithm each time in industrial applications.6. A new hybrid genetic algorithm, named as HNAGA, is proposed by integrating a genetic algorithm based on neighborhood and archived operation (NAGA) with traditional algorithm such as SQP or LM. Based on the process model, this hybridalgorithm is then applied to compute a multi-objective optimization problem of industrial catalytic reforming process. The optimal objectives include maximizing the aromatics yield and minimizing the yield of heavy aromatics. Four reactor inlet temperatures, reaction pressure and hydrogen-to-oil molar ratio, are selected as decision variables. The hybrid algorithm HNAGA is proved to be more excellent than NAGA in obtaining Pareto optimal solutions.The dissertation is concluded with a summary and perspectives of some important problems to be solved in the future research and application.