The rapid increase in demand for ecofriendly ships fueled by liquefied natural gas (LNG) has led to the proposal of floating large-capacity bunkering terminals at sea. Although computational fluid dynamics (CFD) simulations on preventing gas accidents on LNG-fueled ships and bunkering facilities have been carried out, research on evacuating humans in the event of disasters such as fire and explosion due to gas leakage has been insufficient. In this study, an algorithm was derived that considers the interaction between workers and gas accidents for evacuation to the open deck of a floating bunkering terminal, and a simulation platform was developed for visualizing gas accidents. The developed algorithm and simulation platform were applied to analyzing the evacuation characteristics considering both the gas accident and evacuation destination, and the results were compared with those of existing commercial software under the same conditions. The results demonstrated that considering the human characteristics of evacuees and their interaction with risk factors is important for simulating crowd evacuation more realistically.