Study On Chemical Process Integration For Multi-objective Optimization Based On Fuzzy Set Theory

Economic maximization is the optimization objective for chemical engineering at first. With the development of sustainable development policy, environmental impact of the chemical process has been gained much attention on both of academic and industrial circles. In this dissertation, the process safety and equipments reliability are optimized as objectives with economics and environmental impact, and the research on this multi-objective optimization problem is a useful exploration on chemical process multi-objective optimization.The optimization strategies are studied in this dissertation. With the knowledge of process integration technologies and fuzzy set theory, hybrid process integration for multi-objective optimization based on fuzzy set theory is established. Detailed studies concerning this strategy in the dissertation are as follows:(1) Environment impact assessment (EIA) is a multi-decision-making, multi-objective and multi-criteria problem with uncertainties and ambiguities, and environmental impact index (EII) of chemical compound is difficult to be calculated appropriately by traditional methods. A methodology to determine chemical compound EII based on fuzzy optimal selection theory is suggested to reflect relativity environment impact of compounds in the dissertation, and is tested by sample from literature. The satisfactory result demonstrates feasibility and effectiveness of the suggested method.(2) There are internal relation among optimization objectives. Weightiness and precedence of objectives are different, so priorities of optimization objectives should be calculated which would affect the optimization result greatly.Fuzzy decision-marking 'analytical methodology is applied to calculate objectives priorities instead of AHP which is a traditional method, and it is a comparison course among qualitative factors according to strict mathematic theorem and logic discursion.(3) Process integration technology provides the strategy for process optimization, and combination of process integration methodology is necessary for this multi-objective optimization problem. In addition, the increase on number of optimization objectives would cause more uncertainty and fuzzy phenomena, and fuzzy set theory should be utilized to solve the problem.In this dissertation, hybrid process integration strategy based on fuzzy set theory is established, and the framework of multi-objective optimization is build. Process integration techniques including combination of expert systems, hierarchy of decision procedures and pinch analysis are considered to generate retrofit alternatives. Process simulation is the basisof optimization. Simulators such as Aspen plus and ProII can simulate the process strictly to get detailed observation of mass and heat distribution. What is more important, relations among operating and equipment factors can be provided by simulation to set modeling by black box model while process mechanism is too complex to be known.With the simulation data or by mechanism analysis for alternatives containing possible features, multi-objective optimization (MO-O) model is formulated as MINLP type, and would be solved by fuzzy optimization algorithm based on relative importance degree (RID).(4) Conventional techniques have short-coming to solve MO-O model. Genetic algorithm (GA) is thought well suited to MO-O because multiple solutions can be searched in parallel, but sometimes it is difficult to converge to the feasible solution in practical case. In my work, fuzzy optimization algorithm applied fuzzy sets theory based on RID is proposed to transfer MO-O formulation into a GA suitable single objective optimization (SO-O) framework, and determination of RID reflects both of problem characters and decision-maker subjective idea.(5) A new arithmetic of stable non-dominated sorting genetic algorithm based on fuzzy operators(SNSGAFO) is put forward in this dissertation, and there are two improvements compared to stable non-dominated sorting genetic algorithm(SNSGA).Restrictions are treated by integration of fuzz...