Application of Fault Tree Analysis and Swiss Cheese Model to the Overspeed Derailment of Puyuma Train in Yilan, Taiwan

Abstract

On October 21, 2018, a Puyuma express train went overspeed through a sharp curve and derailed in Yilan, Taiwan. This accident resulted in 18 fatalities and 267 injuries. Although such accidents occur once in a while worldwide, this case of an overspeed derailment from a train-set equipped with an automatic train protection (ATP) system (similar to the function of Positive Train Control (PTC) in the U.S.) is rare. A temporary investigation team was appointed by the Executive Yuan, the highest administrative organ in Taiwan, and the investigation was completed within 2 months. This paper presents the process, analysis, findings, and recommendations from the accident investigation. The accident was first analyzed using fault tree analysis to identify potential causes and contributing factors of this derailment. The results were then categorized into layers of defenses by using a Swiss cheese model. We further extended the original Swiss cheese model to a “time-dependent Swiss cheese model” to demonstrate how the barriers were penetrated at different times by incorporating the timestamps of important events. Another modified Swiss cheese model called “causal relationship Swiss cheese model” was presented to further demonstrate the causal relationships. With the proposed process and models, the immediate causes and contributing factors were quickly identified and presented in a way that could be easily understood by the general public. The results showed that the ATP system (or the PTC) cannot guarantee 100% safety. A review of the safety culture and corresponding procedures is important to ensure the safety of railway operations.