All transit operators face delays and disruptions resulting from incidents. The incidents include those more directly within the operator’s control, such as equipment failures and train/bus or staff availability, as well as those less within the operator’s control, such as suicides or power supply problems. The delays resulting from incidents not only affect the reliability of the transit service but also have a direct link to customer satisfaction. Furthermore, major delays are the subject of media attention in most countries, and delays can affect an operator’s public perception and its relationship with the government, potentially affecting decisions on funding.

Incident frequency ranges from about once per week for very small or reliable metros up to approximately 50 or 100 times per day for very large or unreliable metros (data from the CoMET and Nova metro benchmarking groups).

The primary focus for transit operators has long been to try to prevent such incidents from occurring, and many operators have significantly reduced the frequency of incidents over time. However, it is also critical that operators dedicate attention to reducing the duration and effect of incidents (such as the time that the service remains stopped) and reducing the time it takes to restore normal operation after incidents occur. This is especially true for operators with capacity constraints, because the consequences of any incident will be great, as well as for those operators who have already reduced the frequency of incidents, because the remaining incidents will more likely be larger incidents with greater effects (Barron, Melo, Cohen and Anderson, 2013).

Service reliability is a key success factor for mass transit systems and is often identified by passengers as the most important aspect determining service quality (Lombart and Favre, 1995). Service reliability depends on the variability and predictability of travel times (Van Lint, 2008).

The first step in improving service reliability is to develop appropriate indicators. There is evidence showing that an increasing number of mass transit operators around the world are using indicators to monitor service reliability (International Transport Forum, 2010).

Operator-oriented measures tend to focus on the vulnerability of the network to disruption and the operating performance of the network as compared with some agreed level of service (e.g., number of train cancellations, number of incidents, average punctuality). These measures tend to provide an aggregate view of the network’s performance and, as a result, generally fail to reflect the user’s experience (D’Este and Taylor, 2001). In contrast, passenger-oriented indicators focus on the users’ experience. Passengers are concerned mostly with the variability and uncertainty of the travel times of their journeys, not the average network performance. Typically, operator-oriented indicators include measures of service availability (e.g., number of train cancellations), average punctuality or regularity (e.g., percentage of trains arriving on time), network vulnerability (number of incidents by cause), and total and average train delays (e.g., train hours delay). On the other extreme of the spectrum of indicators of train service reliability, there are passenger-oriented measures that describe the degree of travel time variability and uncertainty. Some of the most common indicators include the travel time index, planning time index, and buffer time index. The travel time index measures how much longer travel times are during peak compared with off peak. The planning time index and the buffer time index measure the total time and the extra time, respectively, travelers should allow to ensure on-time arrival (Lomax, Schrank, Turner, and Margiotta, 2003).

The empirical evidence shows that passengers attribute a higher value to travel time reliability than to travel time. This means that passengers prefer to trade off some additional journey time by a reduction in the variability around journey times (Vincent and Hamilton, 2008). The two main parameters used to quantify this relationship are the reliability multiplier and the reliability ratio. The former is a measure of the value of the mean delay time and is expressed as the value of delay divided by the value of scheduled journey time. The second parameter consists of the ratio of the value of the standard deviation of journey time over the value of journey time. In a review of the empirical literature conducted by Preston, Wall, Batley, Ibanez, and Shires in 2009 it was shown that 1 min of delay is valued at 1.25 to 3 times the value of 1 min of journey time, and 1 min of standard deviation of travel time is valued between 1 and 2.8 times more than 1 min of journey time.

Barron et al in 2013 wrote that the ability to disaggregate incident data by cause is fundamental to managing incidents and their effects because it underpins any effort to address common causes. Service reliability measures such as the MDBF tell how often incidents occur, but nothing about their effect. Total service reliability should also consider the length of delays and how many passengers are affected. To understand total service reliability one needs measures that capture the effect of incidents on trains and customers, such as train hours of delay and passenger hours of delay. Moreover, these measures need to be disaggregated by incident cause.

According to the CoMET and Nova metro benchmarking groups, (Barron, 2013) which ranked severity of incident types depending on performance indicator, rolling stock is the type of incident with the highest severity when “number of incidents” is considered as the preferred indicator. Signaling is the type of incident with the highest severity when one considers “number of trains affected,” “total train delay,” and “average number of passengers delayed,” because signaling incidents are harder to resolve and therefore tend to last much longer than other incidents. Passenger-related incidents have the most severe effect on “total passengers affected,” which is believed to be related largely to the fact that these incidents tend to occur at busy times and locations. Other equipment-related incidents have the largest contribution to “initial delay or resolution time,” which is likely due at least partially to the inclusion of power-related incidents in this category. Finally, staff-related incidents have the least severe effects according to all of the indicators.

References

Barron, A., Melo, P.C., Cohen, J.M., Anderson, R.J, 2013. Passenger-Focused Management Approach to Measurement of Train Delay Impacts, Transportation Research Record: Journal of the Transportation Research Board, No. 2351, Transportation Research Board of the National Academies, Washington, D.C., 2013, pp. 46–53.

D’Este, G. M., and M. A. P. Taylor, 2001. Network Vulnerability: An Issue for Regional, National and International Strategic Transport Networks. Proc., 1st International Symposium on Transportation Network Reliability (INSTR), Kyoto University, Japan.

International Transport Forum, 2010. Improving Reliability on Surface Transport Networks. Organisation for Economic Cooperation and Development, Paris.

Lomax, T., D. Schrank, S. Turner, and R. Margiotta, 2003. Selecting Travel Reliability Measures. Texas Transportation Institute, Cambridge Systematics, Inc.

Lombart, A., and M. Favre, 1995. Global Quality of Metros. Presented at 51^st UITP Congress, Paris.

Preston, J., G. Wall, R. Batley, J. N. Ibanez, and J. Shires, 2009. Impact of Delays on Passenger Train Services. Evidence from Great Britain. In Transportation Research Record: Journal of the Transportation Research.

Van Lint, J. W. C., Van Zuylen, H. J. and Tu H., 2008. Travel Time Unreliability on Freeways: Why Measures Based on Variance Tell Only Half the Story. Transportation Research Part A: Policy and Practice, Vol. 42, No. 1, 2008, pp. 258–277.

Vincent, M., and Hamilton, B. A., 2008. Measurement Valuation of Public Transport Reliability. Land Transport New Zealand Research Report 339.