Evaluating critical lines and stations considering the impact of the consequence using transit assignment model -case study of London's underground network

Journal of Advanced Transportation,  Fall, 2008  by Hiroshi Shimamoto,  Fumitaka Kurauchi,  Jan-Dirk Schmocker,  Michael G.H. Bell

• E-mail
• Print
• Link

As the problem of full transit vehicles is encountered daily by passengers in most of the big cities, previous research evaluated the consequence of overcrowding in terms of on-board crowding and passengers not being able to board with full vehicles. The impact of overcrowding in the real world is, however, not necessarily proportional to these numbers. This paper attempts to specify the critical lines and stations of a network by considering the number of passengers failing to board and attempting to evaluate its impact on service quality and safety risks. The hypothesis is that larger stations with wider platforms can often cope better with overcrowding than smaller stations. Therefore a station size dependent satisfaction function is proposed, which takes values from 0 to 1. The method is applied to London's underground network with a number of scenarios which show critical stations in the network if delays occur.

1. Introduction

The problem of full transit vehicles is encountered daily by passengers in most of the big cities of the world. For example, many commuters during the morning peak-hours fail to board the transits in London's underground. This overcrowding leads not only to journey unreliability through having to wait for emptier services but also to stress, delays through passengers trying to push on overcrowded trains and even health and safety risks. The costs related to crowding and delays of its public transportation network are estimated at 230m [pounds sterling] per year for the City of London alone (Oxford Economic Forecasting, 2003). In London crowding will occur during the morning peak-hour at any working day even if all services run as scheduled, however extreme congestion leading to significant numbers of passengers not being able to board will only occur if transits are delayed. This is however frequently the case; London Underground Limited publishes on their homepage that they cancel in average 5% of all service almost every day and the rate is even higher during the morning peak period (London Underground Limited, 2006). Moreover, more than half of the services are cancelled if a major accident occurred.

The U.K.'s Rail Safety and Standards Board (RSSB, 2005) discussed how to control the crowding on the platform and recommended to take measures with following steps; (1) measures to exclude passengers from certain parts of the stations, (2) closing pedestrian access to the station, (3) keep trains away from busy platforms, (4) keep trains away from stations which are overcrowded, (5) suspension of trains services, and (6) evacuation of station. However, if the service frequency is reduced or the station is closed, there is a possibility that passengers change their route choice behaviour and therefore that the location of overcrowding stations/platforms varies. Hence, it is very important for operators to comprehend the influence of reducing frequency of a certain line or that of closing a certain station: i.e. which line or station is critical against reduction of service level.

The authors have evaluated the effect of overcrowding on public transit and its mitigation measures using the transit assignment model described in for example, Shimamoto et al, 2006. In this and other papers, only the consequence (for example, how many passengers failed to board) is evaluated. The impact of overcrowding in the real world is, however, not necessarily proportional to the number of passengers who fail to board. For instance, suppose the same number of passengers fail to board at a major station with a large platform and at a minor station with a small platform. Then, even if those who fail to board affect the service little (in terms of safety hazard or delay) at the major station, they might affect the service much more at the minor station. Lam et al (1999) investigated the effects of crowding at the Light Rail Transit (LRT) station platforms in Hong Kong. They classified five Levels of Service (LOS) on LRT platform crowding and examine the crowding effects of passengers discomfort at the vehicles and the platforms throughout the SP survey. They focused on how passengers perceive the congested level, but the transit operators would also concern the congestion level since they have to take some measures as congestion gets worse.

Based on these backgrounds, this paper evaluates the effect of congestion using the reliability analysis technique, in which the critical lines and stations are specified considering the scenarios with reduction of service frequency of each line. In the context of network reliability, several researches have been carried out to address several concerns over the reliability of the transportation network. In particular, in the field of network analysis against disasters the main aim has been to evaluate the robustness/vulnerability of the network or identify critical components of the network. For instance, Du and Nicholson (1997) evaluated the connectivity reliability of degradable transport networks in which the connectivity reliability is inferred to the probability of all OD pairs to still be connected after different possible road closure patterns. Bell (1999) proposed a game theoretic based approach between the "evil entity" aiming to degrade the network so as to maximise the total travel time and the travellers re-routing to minimise their travel time. The result of this game will be the most critical link in the network. D'Este and Taylor (2003) defined the concept of node vulnerability whose accessibility index decreases significantly with a small number of links degraded. Kurauchi et al. (2007) evaluated the road network and identified the critical link from the network capacity indicator; i.e. the network is vulnerable if a minor link degradation reduces the network capacity, and such links are defined as critical.