A wind turbine (WT) should be shut down as fast as possible to minimize its own damage in a storm-driven situation. Shutdown of a large wind farm requires a power grid to have a ramp-up capability large enough to balance between generation and consumption of electrical energy. This paper proposes a supervised shutdown algorithm of a wind farm to meet a required ramp-down rate in a grid code in the case of a storm-driven situation. The information on the speed and the direction of wind is measured at a wind mast (WM) installed around a wind farm. If the wind speed exceeds a cut-out speed, the number of WTs to be shut down simultaneously is decided to meet a required ramp-down rate of a grid-code. Arrival times to each WT from the WM are calculated and sorted in the order of time. Then a sequence of groups is generated. The shutdown start/end times of each group are decided to avoid superposition between adjacent two groups. The performance of the proposed shutdown algorithm is verified under various storm scenarios. Results indicate that the proposed algorithm can not only protect the wind farm in the case of a storm-driven situation but also meet the required ramp-down rate. In addition, the algorithm can produce more energy than that of a conventional shutdown algorithm.