Peak-Hour Rail Demand Shifting with Discrete Optimisation

John M. Betts, David L. Dowe, Daniel Guimarans, Daniel Damir Harabor, Heshan Kumarage, Peter J. Stuckey, Michael Wybrow

Abstract: In this work we consider an information-based system to reduce metropolitan rail congestion in Melbourne, Australia. Existing approaches aim to reduce congestion by asking commuters to travel outside of peak times. We propose an alternative approach where congestion is reduced by enabling commuters to make an informed trade-off between travel time and ride comfort. Our approach exploits the differences in train frequency and stopping patterns between stations that results in trains, arriving within a short time of each other, to have markedly different levels of congestion, even during peak travel periods. We show that, in such cases, commuters can adjust their departure and arrival time by a small amount (typically under 10 min) in exchange for more comfortable travel. We show the potential benefit of making this trade-off with a discrete optimisation model which attempts to redistribute passenger demand across neighbouring services to improve passenger ride comfort overall. Computational results show that even at low to moderate levels of passenger take-up, our method of demand shifting has the potential to significantly reduce congestion across the rail corridor studied, with implications for the metropolitan network more generally.