Simulating railway punctuality in three Swedish metropolitan regions

Abstract The railway transport system has seen an increasing demand in recent years, while delays and missed punctuality goals continue to be problems in many countries. To improve punctuality, the amount of primary delays needs to be decreased, and knowledge of the shares of the different types of delay – entry, run, and dwell – is important to identify the most efficient countermeasures. This paper investigates the punctuality in the three Swedish metropolitan regions around Stockholm, Gothenburg, and Malmö, applying Circumscribed Central Composite Design in the calibration of a macroscopic simulation tool. The results show that the ratio of primary to secondary delays varied from 1:1.8 to 1:3.8 between the geographical regions, and that the mix of primary entry, run, and dwell delays can vary substantially between regions. In all cases, the primary delays need to decrease by well over $$40\%$$ to reach the goal of $$95\%$$ punctuality, highlighting the importance of reducing primary dwell delays to improve punctuality. For simulation purposes, our results highlight that large networks require careful calibration, as key parameters and inputs vary significantly between different parts of the network. Future work includes experimenting with iterated calibration runs, modelling the prediction error between simulated and observed punctuality, simulating a larger network while allowing for regional differences in input settings and distributions, investigating the potential relationship between capacity utilisation and the occurrence of secondary delays, and exploring the possibility of automating the calibration process.