Modeling And Optimization Of On-line Blending Scheduling For Oil Product In Refinery

Blending is one of the most important refinery processes affecting the quality of oil product,which directly influences the economic benefits of refineries and the quality of product oil.In the process of oil product blending,the factors such as production capacity of refinery,operation rules and key properties of oil product should be taken into consideration.The optimization for blending scheduling of oil product can help refineries upgrade the quality of oil product,improve production efficiency,reduce the consumption of raw materials and increase the income of the enterprises.Therefore,there is a great theoretical and practical value to study the blending scheduling problem for oil product.This thesis focuses on the on-line blending scheduling problem for oil product.Based on the features of on-line blending process,the on-line blending scheduling problem is proposed,and then a MINLP model and a GDP model are formulated.An improved logic-based Outer Approximation algorithm is presented to solve the GDP model,which can get the near optimal solution guiding engineering practice.In addition,an improved global logic-based Outer Approximation algorithm is designed to solve the GDP model for the global optimal solution.Finally,the decision support system of blending scheduling for refinery oil product is developed.Details are as follows:1)Based on the features of on-line blending process,the product oil on-line blending scheduling problem is defined.A MINLP model of the product oil on-line blending scheduling problem is established.The optimized scheduling scheme and blending recipes can be obtained,and the properties of oil product affecting environment are ensured to meet the national emission standards.2)According to the characteristics of product oil on-line blending scheduling problem,a GDP model is established.Compared with the experimental results of the MINLP model,the correctness and effectiveness of the model are shown.3)Based on the idea of engineering optimization,an improved logic-based Outer Approximation algorithm is presented to solve the GDP model,and near optimal blending scheduling solution can be obtained for guiding engineering practice.The experimental results are compared with those of the MINLP model and the GDP model solved with GAMS solvers.The experiments show that the solving efficiency of the proposed algorithm is obviously increased compared with GAMS solvers.4)An improved global logic-based Outer Approximation algorithm is designed to solve the GDP model,which also used to verify the effectiveness of the algorithm for nonconvex GDP model.Compared with the experimental results of LOGMIP solver,the experiment results show that the proposed algorithm can globally solve the GDP model with few iterations and solution time.5)The decision support system of oil product blending scheduling is developed to achieve the digital platform management of on-line blending process.The product oil blending scheduling can be optimized with the embedded algorithm.Moreover,the optimized blending recipes,scheduling scheme and blending costs can provide an important reference for refinery management.