Why Wellington Trains Perform Better Than Ours
One of the puzzling questions for Auckland rail commuters is why there is – perceived at least – to be so many delays and breakdowns and monthly performance figures are not as good as those for Wellington.
Last time I was brave enough to do a post comparing the latest Auckland and Wellington performance figures, Veolia blew a fuse and said you could not compare the two systems.
Now Auckland Transport officers have done a paper comparing the two cities’ train performances and has come to Veolia’s conclusion that the two can not be fairly compared.
The report says the Wellington system is operationally less complex than Auckland’s.
Rail service punctuality has varied in Auckland over the last two years (November 2009 to October 2011) between 65% and 85% of services arriving at the destination station within 5 minutes.
On average 20.1% of scheduled trains were delayed by five minutes or more over the two year period.
The report says that on the basis of the delay minutes attribution, this can be apportioned as follows:
- People 8.6%: Passenger factors 6.8% Staff 1.6%
- Network 8.6%: Network control 1.1%, Track & signal faults 2.7%, speed restrictions 2%, track protection 2.8%
- Train faults 2.2%: Engine faults 1%, door faults 0.4%, others 0.8%
- Other causes: 0.8%
The period that this analysis covers includes two Christmas/New Year periods during which the network was closed for major upgrades following which there was a significant increase in network-related failures. Specifically these included the Newmarket station works, that caused disruptions for an extended period from January 2010 through to March 2010, and the electrification clearance works that caused disruptions from December 2010 and January 2011. Train delays caused by network issues are highly variable and are a direct consequence of the network upgrades to complete the DART Project and preparations for electrification that have been on-going since 2004.
The delays from train faults is described as being relatively small and does not vary significantly by month. However the highest proportion of service cancellations are caused by train faults (53% of the cancellations in the sample period).
Over this period, 2.2% of scheduled services were cancelled in part or in full. This includes services that commenced their scheduled trip but did not reach their scheduled destination because of a fault that developed during the course of the trip.
The following shows an apportionment of train cancellations:
- Operations Actual 0.3% Proportion 12.4% (Mainly as a result of staff error or crew unavailability due to a previous delay)
- Network Actual 0.5% Proportion 22.9% (Mainly as a result of track or signal faults
- Train Faults Actual 1.2% Proportion 53.9% (Faults with trains that result in their removal from service)
- All Other Causes 0.2% 10.9%
The report argues that while Wellington currently operates a higher number of scheduled services per week than Auckland, the track infrastructure provided in Wellington has fewer network pinch-points and provides for greater resilience to recover from delays.
There are four (soon to be five) major junctions on the Auckland network where conflicts can occur between train movements on different lines – Quay Park, Newmarket, Penrose, Westfield, Wiri – compared to two junctions on the Wellington network – Kaiwharawhara and Petone.
Critically for the Auckland network, the major junction (Quay Park) is less than 1 kilometre from the main station Britomart and has two tracks in a tunnel linking five platforms. Due to emergency evacuation requirements, the approach tunnel limits the number of trains that can be “stacked” between Quay Park Junction and the Britomart platforms.
In Wellington, the main junction at Kaiwharawhara where the Kapiti and Wairarapa lines converge is 2.6 kilometres from the main Wellington Station and provides three tracks linking nine platforms (the Johnsonville Line operates in and out of Wellington Station on its own dedicated track). Train movements through the junctions on the Wellington network also minimise the requirement for trains to cross tracks ahead of another train travelling in the opposite direction.
In addition, in the Wellington operation the maintenance depot and the main daytime stabling facility are both in close proximity to Wellington Station which means empty stock movements that increase the total number of train movements on the network are limited.
The report says that Wellington has a mature and stable system with low patronage growth. That allows the operator to make better informed decisions about the allocation of rolling stock to meet the demand based on known historic loading profiles.
The mixed fleet operated in Auckland can also lead to delays as the different train types have different operating characteristics and door configurations. The diesel-multiple units accelerate and brake quicker than the locomotive-hauled trains, and there are operational performance differences between various configurations of locomotive-hauled trains. All of these factors can lead to a variation in the station dwell times and/or sectional run times.
There are differences in timetabling concepts between the two operations reflecting the geographical and travel demand differences: The Wellington rail network was designed for the operation of a high frequency suburban service while Auckland’s network was designed primarily to meet the needs of the freight operations.
A freight network does not require the same signalling or track crossings and the locations of these can constrain commuter operation flexibility.
Wellington has a lower frequency of operation at the extremities of the network compared to Auckland.
During peak travel, Wellington trains operate a layered stopping pattern meaning that the longer trains originating from the stations furthest from the CBD operate as express over the inner part of the network leaving these stations to be serviced by short running services.
This type of operation leads to uneven headways (intervals between services) at intermediate stations and at the terminal stations. With the constraints at Britomart, arrivals need to be evenly spaced in order to maximise the use of the tracks and platforms which is best managed by having even headway services with standardised stopping (all stations)
The Auckland rail performance measurement is calculated as the proportion of services that were not cancelled (reliability measure) that arrived at their destination station within five minutes of the time shown in the published timetable (punctuality). The measures are made irrespective of the cause of any cancellation or delay and represent the experience of customers. The five-minute threshold was selected as it represented the baseline used in the measurement of on-time performance for many international suburban railway systems, for example, Melbourne.
The Greater Wellington Regional Council measures performance against departures from the originating station within 3 minutes of the advertised time, or arrivals at Wellington Station within 3 minutes of the same. Failures that are attributed to non-operator causes such as network faults are not included in this measure.
AT concludes these two measures are not comparable as, using the Wellington measure, a train that departs Wellington station “on-time” but is subsequently delayed en-route is not recorded as being delayed. Trains that are delayed because of a non-operator fault such as a network infrastructure fault or rolling stock maintenance are excluded. The proportion of trains that are cancelled also appear not to be reported.
The report’s conclusion: “ Due to the above differences in performance measurement methodology and exclusion of non-operator faults, and other differences including the age and performance of infrastructure and rolling stock, network stability and upgrade activities, timetable stability and change, passenger volume change and network configuration and characteristics it is difficult to compare Auckland and Wellington performancestatistically.”
But things will improve when we get the new EMUs.