Train rescheduling for a metro circle line: a comparison of different response rules

Within the metro networks of agglomerations all over the world, circle lines play important roles with respect to network connectivity and passenger transfer. The train rescheduling problem of a metro circle line under small disturbances is considered in this paper, including response rules for inserting standby trains, using storage tracks, and holding. A mixed-integer nonlinear programming model is proposed, aiming to minimize timetable deviations, headway variations, number of cancelled stops, and number of rolling stocks, which is then transformed into a mixed-integer linear programming (MILP) model. Furthermore, we enrich our optimization model by including the practical 'snowball effect', depicting the phenomenon that the first delayed train behind the gap is likely to accumulate more delays due to longer dwell times caused by more waiting passengers on platforms. In particular, our solutions suggest that the impact of small delays can be heavily underestimated when ignoring the snowball effect.