Models And Solutions Of Fuzzy Programming For Oil Blending Optimization Problems

Oil blending is an important production working procedure in a refinery and the great benefit of refineries can be achieved from the optimization of oil blending.Considering the uncertain constraints of oil blending such as market demand, blending ability, capability of tank and so on, this dissertation presents an optimization method of fuzzy programming to solve these problems and establishes the fuzzy programming models of oil blending. By analyzing the example of oil blending in one refinery, the outcomes of the blending optimization show that dealing with the fuzzy problems of a refinery with fuzzy programming method can increase the production profit. So it validates that the method is validity. The main work and innovation of this dissertation is as follows.( I ) Based on the analysis of the fuzzy programming method and oil blending technology, some key technological qualities of the main oil production such as gasoline and diesel oil are expounded and the fuzzy programming models of oil blending are put forward.( II ) Applying the theory of fuzzy programming to the oil blending, the models of fuzzy single goal linear programming and fuzzy multi-goal linear programming are presented for the conventional technology of oil blending. And the solutions of the models can be found by simplex method after transforming the fuzzy programming to the determinate programming. (III) The model of fuzzy nonlinear programming is presented because of the nonlinear characteristic exists in the oil blending such as octane number and flash point. After transforming the fuzzy model to determinate model, the solutions can be found by common nonlinear optimization method such as Successive Linear Programming and Penal Function Method and so on.(IV) Because the change of oil characteristic and the blending schedule can not been reflected from the models of conventional single-period optimization technology well that the model of multi-period fuzzy programming is presented by importing the multi-period optimization technology to the oil blending.(V) In order to increase the production profit and decrease the blending cost, the model of fuzzy mixed-integer programming is presented with the goal of the maximum profit and the minimum blending times. And the solution of the model can be found by the method such as Enumeration Method, Branch and Bound, and Genetic Algorithms combining with fuzzy programming method.