Enhancing railway operational safety: A quantitative risk assessment of signals passed at danger

Abstract

Unauthorized ‘Signals Passed At Danger’ (SPADs) are common accident precursors in the Portuguese railways. Critical events in the past highlight the need to assess the risk posed by SPADs. This study proposes a quantitative risk assessment of SPADs in Portugal as a decision support tool, utilizing a three-step procedure, and data from 2016 to 2021. The first step involves statistically modeling the SPAD frequency, including the spatio-temporal and operational factors using Generalized Linear Models, such as Binomial and Poisson regressions. The Binomial models exhibit the best goodness-of-fit statistics. The second step involves statistically modeling the severity of SPADs and creating a severity prediction model based on incident-related data, with Ordinal and Multinomial models compared for prediction performance. Finally, the risk is assessed by estimating the cost magnitude for one year. Additionally, the influence of the Automatic Train Protection (ATP) system on SPAD risk is evaluated. The findings suggest that a fully operational ATP system could further reduce SPAD risk by 19%. This study highlights the importance of considering SPAD risk for safe and efficient railway operations. The results of this risk assessment can support decision-makers in prioritizing prevention and mitigation measures to improve railway operational safety.