Select Committee on Transport Written Evidence

Memorandum by London Underground Ltd (OPT 18)



  1.  London Underground recognises the importance of the issue of crowding on the Underground network. The level of crowding has implications for customer comfort and journey times, and of course for the safety of customers and staff. We seek to manage all safety risks, including those associated with crowding to a level where the risk is as low as reasonably practicable.

  2.  We recognise that parts of the network are heavily crowded at various times of day, usually the commuter peaks, and that this is a cause of concern for our customers. In a recent survey 19% of those asked to suggest improvements to London Underground cited Crowding. (This made reduced crowding the 4th most frequently cited improvement after service frequency and reliability—both factors that directly impact on the level of crowding.)[4] Accordingly, customer satisfaction with levels of crowding is low compared to satisfaction with other aspects of the service. For example, satisfaction with Crowding averages around 70, whereas satisfaction with Safety and Security tends to average around 78. Table 1 presents customer satisfaction scores for train and platform crowding over the last five years.[5] Figures for demand (customer journeys) and service volumes (train kms operated) are included for reference.

Table 1: Customer Satisfaction Survey Results and Customer Journeys


CSS Service attribute (score 0-100)

Platform Crowding
Train Crowding

Demand and Supply (millions)

Customer Journeys
Train kms (operated)

  Sources: LU Customer Satisfaction Survey: Quarter 2 results 2002-03, October 2002, LU Annual Reports and Business Plans 2002-03 figures are year to date actuals for CSS and full year forecast for Customer Journeys and Train Kms.

  3.  Table 1 shows that demand increased dramatically in the late 1990s[6] and this is reflected in falling platform crowding scores. Train crowding scores have however been maintained and improved, reflecting the increase in service volumes introduced to meet the higher levels of demand. Recent scores may indicate recognition of the effect on crowding of the increase in peak service volumes resulting from new timetables on the Central and Northern lines and the significant improvement in reliability, benefiting customers on all lines. Just as crowding is an important issue for our customers, so it is an important issue for London Underground in planning and operating the service.

  4.  This document discusses the safety risks and operational management of crowding, and briefly outlines our plans to tackle crowding through service improvement. In presenting this, we generally refer to levels of `crowding' rather than `overcrowding'. This is not because we do not recognise overcrowding as a problem, rather, it is simply because `overcrowding' refers to a situation of excess over a defined limit, and that limit can be defined in a variety of ways, which will inevitably include an element of subjectivity and semantic argument. Definitions of `overcrowding' could be based on performance against a certain percentage of design capacity, or ratios of standees to sitting or standees to crush capacity. Some customers will consider a train heavily crowded even when there are only a few standees, while others will only think a train highly crowded when they are unable to board and left waiting for the next one.


  5.  The level of crowding is essentially the result of the extent to which supplied capacity can accommodate customer demand. Actions that impact on demand or supply will have consequences for the level of crowding.

  6.  Demand can be seen as having three dimensions: total, spatial and temporal. The total level of demand is partly a function of price but primarily a function of exogenous economic factors, notably central London employment, tourism, income and population. Total demand is not uniformly spread: land use patterns determine the spatial dimension, while seasonal, economic and lifestyle factors determine the temporal dimension. LUL/TfL have a limited ability to influence demand. The Mayor's fares policy is the only real tool available, but raising fares in real terms is regarded as socially undesirable. Limited forms of price discrimination can be used to influence the temporal demand spread, such as the discounted off-peak Travelcard, and the new smartcards will permit further options, `early-bird' or reverse flow discounts for example. While these options can influence demand at the margin, the power of exogenous drivers means that overall ability to affect crowding through the demand side is limited.

  7.  There are two dimensions to supply: capability and reliability. LUL has far greater ability to control these supply factors than demand. Capability depends on the volumes of service scheduled, while reliability is the ability to deliver that schedule. When the railway is running reliably there are fewer cancellations (hence more trains available for customer use and shorter intervals between them) and the intervals between trains are consistent. This means that loadings on both trains and platforms will be more even; a reason for the importance we place on reliability. At present most lines operate central area peak services to the maximum the infrastructure constraint will allow, and reliability levels have improved over the last two years as a result of various reliability initiatives. (It should be remembered that a `side-effect' of these service improvements is an increase in demand, which absorbs some of the additional capacity.)

  8.  This suggests that pursuing a reliable high-frequency service is the best way to influence crowding. Day to day tactical management of crowding therefore combines seeking to maximise reliability over a line as a whole and active local management of stations.


  9.  London Underground employs sophisticated models to assess levels of safety risk associated with 16 top-level safety events. Crowding contributes 8% of the overall network risk of a customer fatality arising as a result of LUL's operations. The main causes of crowding arise from stations being very busy as a result of:

    —  Special events, eg sporting events.

    —  Peak customer rush hours.

    —  Delays to the train service.

    —  Trains held in platforms.

    —  Nearby station(s) closed.

    —  Platform access problems.

  These situations can lead to increased crowd levels which can lead to:


    —  Falls down stairs and escalators.

    —  Falls off the platform.

    —  Crushing on the platforms.


    —  Falls from the train when boarding/alighting or between train and platform.

    —  Heat exhaustion in the event of a crowded train being stalled within a tunnel.

    —  Limbs becoming trapped in the doors.

  The main safety measures in place to reduce these risks are discussed below.

  10.  In the last five years, there have been no fatalities associated with crowding, and very few reportable major injuries.[7] The following is a summary of incident statistics associated with crowding issues which resulted in injury over the last five years:

    —  1 reported incident resulting in a major injury on stairs.

    —  12 reported incidents of minor injuries on or around escalators.

    —  23 reported incidents of minor injuries on platforms due to crowding (eg involving station furniture/structure) and one incident resulting in major injury.

    —  17 reported incidents resulting in minor injury from falls between train and platform whilst boarding or alighting and one incident resulting in a major injury.

    —  56 reported in-car incidents where congestion may have contributed to customers receiving minor injuries when their limbs (arms/hands) were caught in the doors and one reported incident resulting in a major injury.



  11.  Each station on the network has produced an emergency and congestion plan, which ensures that all issues relating to emergency planning and control of congestion are covered. Each plan details the pinch points on the station, and defines the measure by which each individual station would be classified as congested. LUL's Safety Quality and Environment department have validated all these plans, and the Health and Safety Executive have validated LUL's overall safety case.

  12.  If at any time a station reaches the predetermined definition of congestion, the crowd control procedures within our operational rulebook are put into operation. These are tried and tested procedures for cordoning off or evacuating stations. All staff have training and familiarisation in these areas.

  13.  The general rule of thumb that staff are trained to observe is that once free movement along the back of the platform is lost at any point on that platform, crowd control measures should be introduced. A graded response applies. For example, if it is known by the station staff that a train is due on that platform within the next few seconds, the best response is to wait for that train to clear the platform congestion. Otherwise, staff may take an escalator out of use to slow the flow of people to the platform. Alternatively, some of the ticket gates may be reversed to favour exiting flows and slow the flow into the station until the level of crowding can be alleviated by the passage of trains. If necessary, staff will actually close parts of the station, or some or all of the entrances. These measures are supported by announcements at relevant stations to inform customers that delays are likely and to advise of alternative routes.

  14.  Almost all stations on the network have fully comprehensive CCTV facilities, all have radio communication and public address systems, and many stations are equipped with dedicated station control rooms to direct crowd control operations, and liase with line control and neighbouring stations. All stations have staffing levels appropriate to the level of station usage, and we have recently introduced SATS (Station Assistant Train Services) to many of the busiest platforms. These are dedicated platform staff who manage the boarding and alighting of trains and encourage customers to use the full lengths of platforms to avoid crowds building up in parts of the platforms. Crucially, SATS are supported by radio based public address (PA) systems that do not require them to remain in a fixed place on the platform, thus they are able to move freely along the platform and make announcements at the same time.

  15.  Crowd control procedures are implemented on a daily basis at some of our stations (Victoria being a notable example) and staff at those stations are experienced in managing these situations. A consequence of that experience and training is that staff tend to be able to predict the build up of congestion, and proactively take appropriate actions to slow flows.

  16.  Overall, we believe that the process is both procedurally robust and in practice works well on the ground. This is not to say that we have got it right every time—there have been a few occasions where customers, or other informed observers such as railway inspectors, have fed back to us that crowding has got beyond the point we would normally want it to. (This is often because crowding has suddenly arisen at a station that is not used to it and less experienced staff have misjudged the point at which crowd control should be introduced—eg national rail network problems causing a surge at a quiet suburban station.) These incidents are treated seriously with investigation and appropriate action, such as refresher training.


  17.  Our engineering and safety standards ensure that trains, their braking systems and the signalling system are designed such that a fully loaded train is a safe train. This essentially means that the control on train crowding is train capacity itself (the alternative, placing a member of staff at every door to count on customers, would be highly impractical). It should be understood that even though trains are built to be safe at design capacity, in practice, even on the busiest sections of line, actual train loadings do not reach design capacity.

  18.  Loading and crowding on individual trains will greatly depend on the regularity of the service. Consistent intervals between trains tends to ensure even loading, whereas the first train following an irregularly long gap is likely to become more crowded, which in turn will lead to longer boarding/alighting times, slowing the train even more. Meanwhile a train immediately behind may well be much less crowded.

  19.  In addition the centre cars on a train tend to be more crowded than those at the front and back. Customer behaviour (clustering in the door vestibules, not standing down to allow access, anticipating platform exits, etc) means that crowding levels can vary within a car and between cars, as well as between trains.

  20.  Even though loading is well below design capacities, it is nevertheless true that some trains can get unpleasantly crowded. In normal operation however, exposure to these conditions is usually only for a short period of time. Nonetheless, if customers do feel unwell for any reason, it is possible to alert the Train Operator using the passenger emergency alarm, which on most new and refurbished trains includes a facility to talk directly to the Operator. (However, while this facility is available, we would advise customers feeling unwell to alight at the next station and seek help, and only use the alarm if it is a real emergency because of the delay alarm activations could cause to following trains.)

  21.  We recognise that the effects of high levels of crowding are likely to be more felt if exposure is prolonged, particularly if a train is stationary in a tunnel for longer than usual. On a very crowded train, humidity and temperature will rise and some customers may experience breathing difficulties. Our first priority is always the safety of our customers and staff. In the first instance this means we will seek to avoid trains becoming stuck in tunnels, eg at the first sign of an incident controllers will try to bring all trains behind the incident/failure train into a platform and hold them there. On those odd occasions where we are not able to achieve that and a train is trapped in a tunnel section, we will wait no longer than a quarter of an hour before commencing detrainment procedures leading to walking people out through the tunnel. We will not just leave customers on a train until the incident is resolved. Staff have a quarter of an hour to rectify the cause of the delay, but then we start making arrangements to take off the current, get staff to the trapped train, and then walk people out.[8] Where a train is trapped for an extended period, ambulances will be summoned to the nearest station(s) to be available should any customers require medical attention.

  22.  As with our procedures for managing station crowding, we believe that our approach is effective in minimising the risks to customers. While train crowding in normal conditions is generally self-regulating, controllers do seek to maintain regular intervals between trains and avoid impacts of earlier incidents reducing peak services. Occasionally more direct intervention will be used, such as holding a train spare for large expected crowds (eg following football matches) or turning trains short of their booked destination to provide relief in the other direction. There is also close interaction between train and station controllers to ensure stations can clear platforms before a heavily loaded train arrives.

Planning context: actions to address crowding

  23.  As discussed above, the level of crowding is a response to the service supplied and the level of customer demand. We are most able to affect crowding through the supply side, and the same measures intended to meet customers' demands for a frequent and reliable service, also address crowding. The strategic management of crowding rests on increasing the capacity of the Underground and the London public transport system as a whole.

  24.  In the short term this means continuing the focus on reliability to maximise performance within the existing infrastructure constraints. As noted above we have increased operated service volumes by 3 million train kilometres (5.5%) in the last five years,[9] with measures to address the causes of unreliability a significant contributor. (For example the number of peak cancellations due to non-availability of train operators averaged 8 per peak in 2000-01, but is now regularly less than 3 per week).

  25.  The reliability focus will be taken forward under the Public Private Partnership (PPP) through improved maintenance regimes and the renewal of assets, as asset incidents currently account for approximately 70% of delays over two minutes.[10]

  26.  The PPP will also deliver a significant increase in capability through a programme of line upgrades combining renewal and enhancement of assets, taking capacity to the limits of what is possible with proven technology and 19th century tunnel structures. Line upgrades involve some combination of more trains, faster trains, and more capacious trains with the exact solution depending on the characteristics of the line. (For example the Jubilee line upgrade is likely to involve increasing the total fleet size and adding an additional car to each train in addition to signalling work, while the Northern line upgrade is likely to involve resignalling and improved control but no changes to rolling stock.) We anticipate the line upgrade programme to deliver a peak capacity increase of roughly 15% over the next 15 years.

  27.  At the same time however, projections of a strong London economy as set out in the London Plan mean that Underground demand is likely to increase by 15-17%.[11] Our projections also show that the increase in demand over the next 30 years will be roughly matched by increases in the service resulting from the line upgrades and improved reliability. In overall terms, this means that while peak crowding levels are unlikely to worsen; they are not expected to show significant improvement from today's levels. This being said, it should be understood that land use patterns will create pinch points of heavy crowding at particular locations at particular times of day.

  28.  Much of the overall demand increase is expected to be in off peak travel. Here there is much more flexibility to increase service volumes to meet demand since infrastructure constraints are less evident. We will continue the successful policy of the last decade to increase off-peak frequencies in a manner sensitive to demand.

  29.  However, dramatic reductions in crowding levels ultimately can only be achieved through the construction of new lines (though even then, new demand generated by infrastructure will absorb some of the new capacity.) LUL supports the Mayor's advocacy of the Crossrail schemes; proposals for intermediate modes; and improvements to Bus services and we look forward to all these schemes being progressed. It is important to recognise that the different schemes will have different impacts on the Underground network: Crossrail 1 will particularly relieve pressure on east-west routes, notably the Central and Hammersmith & City lines. Crossrail 2 will relieve pressure on the central sections of the Victoria and Piccadilly lines, while Thameslink 2000 and the Cross-River Transit scheme would particularly assist the Northern and Bakerloo lines.

  30.  In addition to the line upgrade programme, LUL will be embarking on a major programme of station and interchange congestion relief projects. These projects involve major rebuilding of existing stations to widen passageways, provide additional escalators, provide step free access, improve circulating space and expand ticket halls. This type of work is currently underway at stations including Kings Cross St Pancras (to meet CTRL demand) and Brixton, and are planned for many central London stations including Tottenham Court Road, Camden Town and Victoria among others. Providing increased station capacity not only relieves congestion and delay at the stations in question, but also benefits the whole line by ensuring platforms are cleared quickly to allow trains to arrive and unload/load in quick succession. The emphasis of these, and other journey time related projects, is to minimise customer travel time: crowding is modelled and appraised as part of this.[12] The station congestion relief programme will be taken forward as quickly as funding and particularly planning procedures will allow.


  31.  We recognise that crowding is a feature of travel on the London Underground, and this is part of the nature of operating a high frequency urban mass transit system. We are also very conscious that at certain times and in certain locations, crowding can make journeys unpleasant for customers. However, the safety risks are managed to be as low as reasonably practicable and we will continue to ensure we have the best procedures in place to safely meet the daily challenges of crowding. We also have clear plans to increase network capability and reliability. Delivery on these plans will mitigate the effects of the projected increase in demand on crowding and deliver on our aim to help customers move swiftly around London.

December 2002

4   Source: London Underground (2002) Londoner's Survey: Wave 4, March 2002. Back

5   In addition to using the Customer Satisfaction measure, we also measure crowding levels on trains and at busy stations on a quarterly basis, and customer usage data is updated annually. Back

6   Overall growth of 25% was experienced between 1996-97 and 2000-01. Back

7   As defined by RIDDOR (Reporting of Injuries, Diseases and Dangerous Occurrences Regulations 1995). Back

8   It should be understood that this does not mean a train is completely emptied a quarter of an hour after the start of an incident, it means that detrainment procedures should commence after a quarter of an hour with the last customer walked out as quickly afterwards as it is safely possible to do so. Back

9   Excluding increases due to the Jubilee Line Extension (approximately 6.8 million km increase including JLE). Back

10   Average based on periods 6-8 2002-03, includes rolling stock and temporary speed restrictions as well as other asset incidents. Back

11   GLA (2001) The London Plan: The Mayor's Draft Spatial Development Strategy. Back

12   Crowding parameters are based on quarterly samples of platform crowding, which can be expressed as the percentage chance of boarding a train with all seats full etc. These figures, broad averages across whole lines, are reported quarterly and submitted to the TfL Rail Transport Advisory Panel. Back

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2003
Prepared 5 November 2003