Short-Term Scheduling Optimization Of Crude Oil Operations In Refinery

Oil refining industry plays a significant role in national economy. Oil refining production planning includes long-term and short-term production planning and scheduling. Long-term production planning can be modeled and solved by mathematical programming. At present, by using linear programming-based technique, methods for the long-term production planning problem of a refinery is well-developed. For short-term production scheduling of a refinery, one needs to optimally sequence not only the discrete operations but also optimally determine the value of the continuous variables in the operations to improve the profit of a plant. However, the operations to be scheduled are not known in advance and they need to be generated in the scheduling process. Therefore, its complexity and difficulty is much greater than that of the scheduling of discrete and batch processes. It has been shown that the short-term scheduling problem of crude oil operations is NP-hard such that an exact solution method is not applicable to practical short-term scheduling problems of crude oil operations. Furthermore, at the beginning of a scheduling horizon, the tasks to be scheduled are unknown. Thus, heuristics and meta-heuristics are not directly applicable to this problem. The short-term scheduling problem of crude oil operations is one of the most difficult problems in scheduling an oil refinery. This is the problem addressed in this thesis.Because heuristics and meta-heuristics are not applicable, and mathematical programming methods are unable to solve real-life practical application problems for scheduling crude oil operations, to search for efficient techniques, our team studies this problem from a control theory perspective and a two-layer hierarchical solution is proposed for it. At the upper level, one finds a realizable refining schedule to optimize some objectives. At the lower level, a detailed schedule is obtained to realize a given refining schedule. So far, linear programming-based approaches are presented to efficiently solve the upper level scheduling optimization problem by our team. Thus, to obtain an optimal schedule for crude oil operations, the lower level detailed scheduling optimization problem needs to be solved urgently. Therefore, in this thesis, attempts are made to deal with the detailed scheduling optimization problem at the lower level under the assumption that a target refining schedule is known in advance. The following studies are conducted:(1) Because heuristics and mathematical programming models cannot be directly applied to the short-term scheduling problem of crude oil operations, and it is necessary to do it in other way. This is the motivation of this thesis. First, a mathematical programming model for the problem is built. Based on this model, we analyze the characteristic of this problem. Then, the short-term scheduling problem is defined as a series of operation decisions. Finally, based on a control-theoretic perspective, the problem is innovatively transformed to a problem of assigning charging tanks to distillers such that heuristics and mathematical programming models can be applied and the problem can be efficiently solved. Thus, the computational complexity problem is overcome.(2) Two heuristic algorithms for the detailed scheduling at the lower level are proposed. Both of them can ensure the realization of a realizable target refining schedule given at the upper level. Although the heuristic method is simple and cannot guarantee the optimality of the solution, the results show that the proposed heuristic algorithms are effective.(3) The detailed scheduling problem of crude oil operations involves multiple optimization objectives. The multi-objective optimization problem is converted into a single one by using weighted sum. A genetic algorithm approach is developed to solve the problem. Based on a set of existence conditions of a feasible schedule, methods are presented to guarantee that each chromosome corresponds to a feasible schedule. Finally, illustrative examples are given to show the application of the proposed method.(4) A multi-objective optimization algorithm for solving the detailed scheduling problem of crude oil operations is proposed. Because those multiple optimization objectives may conflict with each other, a way for such an optimization problem is to use Pareto-optimality without the selection of preferences. In this thesis, we aim to obtain a set of Pareto-optimal schedules. This thesis develops a chromosome with variable length and dual genes, which can always describe a feasible detailed schedule by ingenious decoding. Then, we innovatively adopt an improved nondominated sorting genetic algorithm to solve the problem. An industrial case study is used to test the proposed solution method. The results show that the method makes a significant performance improvement and is applicable to real-life refinery scheduling problems.In summary, this thesis transforms the short-term scheduling of crude oil operations into a problem of assigning charging tanks to distillers such that heuristic and meta-heuristics can be applied. In this way, the difficulty in computational complexity resulting from mathematical programming models is overcome. It provides an efficient way to develop a practical optimization method and solves one of the important difficulties.