A rescheduling problem of steelmaking-continuous casting with tapping tardiness was studied in this paper. And a dynamic constraint satisfaction model was established to minimize the difference of the starting time, processing time and processing machine and waiting time of the heat between the adjacent equipment. According to this model, a local repair algorithm based on dynamic constraint satisfaction techniques and interrupted-cast repair rules was put forward. In this method, variable selection and value selection rules were used to assign one value to a variable. The conflict identification and elimination principles were used to identify and eliminate the conflicts in the assignment. The interrupted-cast repair heuristic rule was used to repair the interrupted-casts in a continuous casting machine. In this experiment, three groups of random data with a uniform distribution were generated. Target values were 0.15, 0.28 and 0.51. The results demonstrate that the size of the delay time has a certain influence on target value and the algorithm can satisfy the needs of real-time and stability as much as possible.