In discrete manufacturing, a just-in-time schedule is pursued so as to respond better to the market. It is also required in oil refinery. However, the existing techniques for short-term scheduling in oil refinery are based on the push production mode. This paper addresses the short-term scheduling problem for crude oil operations in a pull production way. A target refining schedule resulting from production planning is given as a constraint to make an executable schedule. The system is modeled by a timed hybrid Petri net. This model is structurally simple and can describe the dynamic behavior and all the constraints of the system without any difficulty. Based on the model, an efficient heuristic algorithm is proposed to test the realizability of a target refining schedule. If it is realizable, a feasible short-term schedule realizing it is created. A real-life industrial case study is presented to show the industrial application of the proposed method.