The P-graph-based Combined Method Synthesis And Optimization Of Seperation Systems

Process synthesis is about searching the systemic frame and the function of each subsystem in line with systemic properties, and optimizing system on the basis of target. It is the core of process systems engineering and the key of process system design. Recently, the idea of sustainable development and environmental conservation introduced the concept of clear production, and recycle economy with properties of resource-saving and recycling introduced the principles of reducing, reusing and recycling. These made process synthesis more important, and new theory method of process synthesis was needed to.address increasingly large-scale industrial problem.Process synthesis mostly includes the generation of super-structure and the search of optimum solution. In recent years, how to automatically generate the exact super-structure and to achieve optimization was always the development trend of process synthesis. Among the approaches used in process synthesis, heuristics had deeper possibility of missing the optimal solution, while it was a powerful selection means at the beginning of process synthesis. The other classical method, algorithmic method still had the possibility that the superstructure can not be definitely created in complicated process, while it can automatically achieve optimization. P-graph approach was just the algorithmic approach improved by Frieldler, which can optimize the complex large-scale problem automatically. But some necessary conditions ― the known scopes of sources, productions and operating units, must be achieved, as the method was used. However, the scope of operating units was commonly unknown as practical problems were treated. So gathering each advantage by combining different methods was the development trend of process synthesis.A new combined method of process synthesis, P-graph-based combined method, was brought forward, after the status quo and development trends of process synthesis were studied sufficiently. The new combined method can be briefly described as: the necessary conditions, the scopes of feeds, products, and potential operating units, were determined by hierarchical decomposition approach, and then the rigorous superstructure was produced by the P-graph approach. Differing from others, the new combined method took advantage of the insights provided by physicochemical properties, while simultaneously utilizing the power of mathematical programming for addressing complex interactions and trade-offs. Moreover, the method provided a transparent design procedure, which was achieved by adopting a stepwise methodology in the spirit of hierarchical methods. Since the combined method could reduce the complexity of the problem by considering only relevant design options without sacrificing optimality, so that the design method can be applied to large practical problems. Thereafter, the new method was verified with the separation-network synthesis, and the practical implementation steps were listed.After the super-structure was generated, its explicit expression of Mixed Integer Nonlinear Programming(MINLP) should be optimized, which was still a challenge. In our research, the solving method of process synthesis optimization was also studied. The method mainly referred to the procedure of Branch-Bound algorithm(BB), and combined super-structure analysis and Accelerated Branch and Bound algorithm(ABB). Its idea and solving process were demonstrated in detail by example, and the result provided the evidence for the efficiency of the arithmetic. Moreover, randomized algorithm of MINLP and Particle Swarm Optimization(PSO) were also studied, and an improved PSO, beneficial to the optimization of MINLP, was proposed.