In discrete manufacturing, just in time schedule is pursued so as to respond better to the market. As a typical process industry, oil refinery is also required to do so. However, in practice, the existing techniques for short-term scheduling in oil refinery are based on push production mode. Linear programming based techniques for long-term planning and scheduling are well-developed. Mixed integer programming methods are also developed for short-term scheduling. However, they are not applicable in practice due to the computational complexity. Furthermore, these methods create a short-term schedule in a push production mode. Thus, up to now, there is no efficient tool for short-term scheduling in oil refinery and it is done manually. This motivates us to develop an efficient software tool to help planners for short-term scheduling so that a short-term schedule in pull production mode can be created.In the thesis, we address the short-term scheduling problem for crude oil operations. After brief introduction of the processes and constraints, the short-term scheduling problem for crude oil operations is defined. Based on the description of the problem, algorithms for the creation of various operation decisions are proposed to realize a given refining schedule that is obtained from market demands. Because of the complexity of the problem, we use "heuristic + simulation" algorithms. In this way, it is computationally efficient and applicable to real-world problem.With the algorithms proposed, a short-term scheduling system for crude oil operations is developed. By using this system, short-term schedule can be created with less human interference. It is tested in a real oil refinery plant, and it works well. |