Much has been written and said in recent times about efforts to improve the throughput of trains on metro and inner suburban railways. Using CBTC has enabled 36 trains per hour (tph) to be achieved in each direction on London Underground’s Victoria Line. A dramatic improvement to passenger comfort has resulted and much of the severe overcrowding has been eliminated. Similar predictions are made for other LU Lines and also the full Thameslink and Crossrail (Elizabeth line) services when they reach full fruition.
However, a crucial factor in all of this is the ‘dwell time’ at stations to allow travellers to alight and board the trains. If the time taken for this overruns by more than a few seconds, then very quickly the delay to following trains builds up and the intended throughput becomes unachievable.
A service gap of more than three or four minutes means that crowds increase on the platform, extending the dwell time at every station as they attempt to board the first train, compounding the problem.
Although, in theory, drivers are not supposed to initiate door closure until everyone is safely on board, in practice, they occasionally have to start the closure process whilst people are still squeezing in, otherwise the train would never get underway.
A further factor is now influencing the process, this being to take account of the needs of disabled people, with legislation potentially increasing the dwell time period. Whilst the Rail Vehicle Accessibility Regulations (RVAR) of 2010 (its forbear being the Disability Discrimination Act) is intent on allowing additional time to board, the basis of this prescription may not have been scientifically derived, with perhaps a less than optimum situation developing.
London Underground was concerned that a negative impact could result and initiated a trial to establish exactly how passengers behave when boarding tube trains. Rail Engineer went along to learn the facts.
Door closure sequence
When a train arrives at a station, providing it is proved stationary and at the right location, door opening is initiated by the driver. After passengers have alighted and boarded, a door closure alert signal (known as a Chime) sounds for a period before the doors begin to close. Providing nothing is trapped in the doors (see later paragraph), the driver’s door close pilot light illuminates and the train start buttons can be pressed for ATO equipped lines or the driver engages traction power if driving manually.
The chime signal time is crucial. The LU standard is 1.75 seconds ± 0.25 seconds, whereas the RVAR (aligned to the Equality Act) requirement is three seconds. This difference, whilst small, can accumulate to several seconds for an end-to-end train journey and, if applied to every train, can significantly reduce the overall service throughput. More importantly, however, does changing to three seconds make any detectable difference to either non-disabled or disabled passenger behaviour?
One important element is the ‘hustle’ effect. When the door chime sounds, a regular occurrence is for passengers to hurry into a nearby door so as to avoid waiting for the next train.
Regular commuters are adept at knowing which door is nearest to the exit at their destination station, so they will not always board a train when it arrives but instead continue to walk along the platform to the preferred door. If the chime sounds before they get there, they jump in through the nearest open door.
Equally, travellers who are not regular underground users and who may be tourists from another country, on hearing the chime, will rush to the nearest door, often with large amounts of luggage.
Any instance of incomplete boarding will result in a door obstruction situation and potential train delay. As a door obstruction takes a minimum of five seconds to resolve, the delay impact can be significant if compounded along the route. On the Victoria line, any door obstruction incident is automatically flagged within the train software and sent by Wi-Fi to the server, available for review by engineers within 20 minutes.
Different marques of Underground train have differing solutions to this situation. The most modern trains (Victoria line 2009 stock and the S stock units on the sub-surface lines) have obstruction detection whereby, should an obstruction occur, the door will reopen part way to allow the obstruction to be pulled clear.
They also have sensitive door edges where any deformation of the door edge will cause an emergency brake application should the train have started to move. This deformation can be caused by even very thin items which, when caught, would be pulled on as the train begins to move.
Older trains are designed to ensure doors are fully closed before traction power can be applied. However, this system is not foolproof and thin items such as bag straps or coat belts can be trapped between the doors and are not always detectable. Unfamiliar users often expect the doors to re-open if an obstruction is detected, much as they do on lifts, but this is not true for Underground trains or indeed any UK train with sliding doors.
All of this presents a complex set of circumstances that, when combined with the differing views on chime time, meant that a comprehensive trial was necessary to understand more completely the impact on passenger behaviour and minimising the ensuing risks.
To be meaningful, plans for any trial must define what it sets out to do, the way it will be measured and how the results will be analysed. The intent of this exercise was to assess the impact of different door chime timings with respect to the following aspects of door usage and passenger behaviour: safety, accessibility and capacity (in that order).
The trial, which took place on the Victoria line, consisted of a number of investigations within the context of both a 1.8 sec and 3.0 sec chime duration, covering:
- Door obstruction data, collected from the rolling stock;
- Platform observations;
- Reported safety incidents;
- Passenger survey;
- Service data (dwell times, lateness).
The door obstruction data ended up being the most meaningful in terms of assessing safety: it was used as a proxy for the number of passengers being struck by the doors, an indication of items that could be trapped (risk of dragging), and an indication of the number of passengers running (risk of slips and trips). If LU’s concerns about the hustle effect were correct, an increase of door obstructions would be observed.
To be effective, the trial needed to ascertain the before and after situation, so data and observations were obtained prior to August 2017 after which the chime duration was changed to three seconds for a period of six months. The whole Victoria line fleet of 37 trains, each of eight cars, was altered so as to get consistency and to accurately observe passenger behaviour. The results have proved interesting.
The door open and close sequence was found to be well disciplined at peak hours, with regular travellers standing clear of the doors before getting on in order to allow passengers getting off to disembark more quickly. During off peak periods, however, unfamiliar travellers tended to block the door egress, thus slowing the whole process. A longer chime duration did allow more time to get out of the way of the doors, but equally gave more time to try and board.
The sounding of the chime has always been known to prompt late boarders to run for the nearest door. With 1.8 seconds, the time was insufficient for this to be successful unless very close to a door. Extending the time to three seconds saw an increase of the hustle effect with the result that noticeably more door obstructions occurred. This worsened the safety risks as more doors were striking or trapping passengers, the numbers of trips/falls increased and more pushing/obstruction of other travellers took place.
Clearly there is a linkage between door obstructions and passenger demand, with the number of obstructions using a 1.8 sec chime remaining fairly constant throughout the year, rising slightly between October and the year end. Introducing the three second chime saw a marked increase in obstructions during the lead up to Christmas, during the January sales and at public holiday weekends.
The overall finding was that off peak, more people ran for a door once the chime sounds as the platforms are less crowded, whereas in the peak more people tried to squeeze in.
Not surprisingly, the highest number of door obstructions occurred at the busier stations. The northbound platform at Victoria was by far the worst, with an average of 245 obstructions happening each day during the three-second chime period, an increase of 60 (32 per cent) over the 1.8 sec time. Oxford Circus and Kings Cross also recorded high numbers of around 150, in all cases the longer chime time being marginally worse.
At less-busy stations, the effect was more prominent – at Highbury and Islington southbound in the morning peak, the obstructions rose 80 per cent with the extended chime time and an overall increase was noticed throughout the day.
The door obstructions were also markedly different down the length of the train. At Oxford Circus (northbound), the problem mostly occurred at the doors in the second and third front cars, close to the exit for the concourse and interchange for the Bakerloo line, and at the rear end where people change for the Central line. The number of door obstructions were significantly increased with the three-second timing.
Analysing the trial
Clearly the increase in chime time had a detrimental effect on obstruction occurrences and, due to the safety impact, the Victoria line fleet has since reverted to a 1.8 second timing.
That said, it has been necessary to submit the findings to vested interest groups. Presentations have been given to the London Underground hierarchy (DRACCT – Director’s Risk Assurance Change Control Team), to Transport for London (TfL), the Department for Transport (DfT, in effect – the government) and the Office of Rail and Road (ORR – the rail regulator). The DfT has forwarded the results to the Disabled Persons Transport Advisory Committee (DPTAC) and other stakeholder groups.
A passenger questionnaire has been conducted in an attempt to establish what passengers believe are the safety risks associated with their journey. Approximately 150 responses were received, which is considered sufficient to gain some understanding but is a very small sample compared to the approximate 75 million passenger journeys made on the Victoria line during the period of the trial.
Of these respondents, 60 per cent considered themselves to have a disability. Oddly, the time to board and door closure time feature less than the fear of interaction with other passengers who might, in their urgency to board, push people both on the train and on the platform. There was little difference in the responses from both disabled and non-disabled passengers.
Overall, the trial findings have been well received, since there is now hard proof that the three-second chime offers little improvement to accessibility and has a negative impact on safety. The recommendation from LU is that the standard should remain at 1.75 ± 0.25 secs, and this is being considered by the aforementioned bodies.
London Underground stresses the point that it is totally committed to improve accessibility across the entire network, with step-free access being provided at an increasing number of stations and large projects underway to improve accessibility.
Factors for the future
It may be asked how this trial impacts on other metro/light-rail operators and, indeed, mainline suburban services. Main line operation is different in that it provides timetabled departures rather than a high-frequency, turn up and go service, and as such passengers have more time to plan their journey and associated timings at a station. That said, the likes of Thameslink and Crossrail (Elizabeth line) may be more akin to LU operation in the central London sections.
The eventual adoption of fully automatic trains (Unattended Train Operation – UTO) may well become reality. They exist already on the Paris Metro (Lines 1 and 14) where door operation is programmed automatically, dependent on the particular station and the time of day. When the programmed time has elapsed, the doors will begin to close regardless of whether boarding is still taking place or not. Centralised CCTV monitoring of conditions takes place both on platform and train so that remote intervention can happen should anything untoward occur. Travellers have got used to this and it is now part of normal life.
The current modernisation of the Glasgow Subway will adopt UTO when completed in the early 2020s.
In all, this trial demonstrated the many complex interactions that take place at the platform-train interface. As the population of London increases, so the pressure to provide more and more public transport services will mount and the need to be up-to-speed with technology and optimum routines will become ever more vital.
Thanks to Zoë Dobell, LU’s project engineer for the trial, for this fascinating insight.