This thesis addresses the problem of operation scheduling for oil movement in refinery processes where crude and processing equipments are interconnected with complexity. This operation scheduling involves vessel docking, crude unloading to storage tanks, crude mixing in charge tanks, and crude feeding to crude distillation tower(CDU). The whole problem is subdivided into several module problems to implement the decentralized scheduling algorithms. For each module, the intelligent scheduling combining an OR technique and expert systems generates the data for movement amount and movement time. For a given scheduling horizon, QP and LP optimization models are proposed to compute movement amount to be converted to the demanding amount of products, and expert rules are incorporated to give us the detailed movement time accommodating practical operation constraints. A real time expert system, G2 is used for representing rules and aiding graphic user interface(GUI), and GAMS for modeling optimization problems. For a practical size refinery domain, this scheduling system is illustrated to automatically generate scheduling data for oil movement and simulating trends for the inventory and sulfur composition of each tank unit.
