Excess probability of dwell time delays from train meets and passes

Abstract

We introduce a novel measure of how a factor influences train delays: the excess probability of delays. Using some 50 million observations across two years of railway operations in Sweden we then investigate the role of train meets and passes in explaining dwell time delays, which are known from previous research to be behind most train delays. About 4% of observed train movements include either meets or passes at stations, with the former being much more frequent. On average, they are associated with a 30%-points higher probability of delay, more than double the baseline rate of 25%. Accounting for their frequency, this adds about 1%-point to the probability of dwell time delays, or about 4% of existing dwell time delays. As expected, these delays are mostly concentrated on single-track lines with relatively dense operations. Considering that the railway sector is mature and under pressure to cost-effectively improve both capacity and punctuality, these results suggest an encouraging potential for improvement. Reducing the probability of delays even by a single percentage point would be valuable, and quantifying this contribution is helpful when prioritising resources and measures across issues and geographies.