Summary of Meeting Paper

The 1996 Annual Meeting of the Society for Risk Analysis-Europe

Improving Passenger Safety at Platforms. T.A.W. Geyer, C.P. Chapman, and M.I. Morris, Four Elements Limited, 8 Cavendish Square, London, UK; and P.R. Chislett, London Underground Limited, London, UK

BACKGROUND

A number of passenger accidents at the platform/train interface, some resulting in fatal injuries, has led London Underground Limited (LUL) to consider what additional measures might be adopted to improve passenger safety in this area still further. In fact the London underground system is very safe at the platform/train interface -- over two billion passenger boardings and alightings take place each year without incident. Nonetheless, whenever a passenger accident does occur, there is justifiable public concern as to whether LUL has taken, and is continuing to take, all reasonable steps to avoid passengers being exposed to undue risk; and LUL itself recognises the need for continuous monitoring and improvement where practicable.

For these reasons, LUL, in conjunction with Her Majesty's Railway Inspectorate commissioned Four Elements Limited to consider various measures intended to improve passenger safety at platforms, to evaluate their benefit using the technique of quantitative risk assessment (QRA), and on this basis to recommend for further consideration those measures which seem cost effective. This paper describes the approach and findings of the study.

HOW SAFE IS SAFE ENOUGH?

The regulation of safety in the United Kingdom is based upon the principle that risks must be reduced to a level which is "As Low As Reasonably Practicable" (the so called ALARP principle). The meaning of "reasonably practicable" is well established in English case law:

"Reasonably practicable" is a narrower term than "physically possible" and seems to me to imply that a computation must be made by the owner in which the quantum of risk is placed on one scale and the sacrifice involved in the measures necessary for averting the risk (whether in money, time or trouble) is placed in the other, and that, if it be shown that there is a gross disproportion between them -- the risk being insignificant in relation to the sacrifice -- the defendants discharge the onus on them. "(Judge Asquith, Edwards v. National Coal Board, All England Law Reports Vol. 1, p. 747 (1949)).

Thus the ALARP principle allows cost to be taken into account in determining how far to go in the pursuit of safety, so that if a risk reduction measure involves "grossly disproportionate" cost, it is not "reasonably practicable". This principle was adopted in the Health and Safety at Work Act (1974), and is the basis of the approach adopted by the Health and Safety Executive in their regulation of the major hazard industries, such as the offshore oil and gas industry. It has also been adopted by BNM in its application of the Railways (Safety Case) Regulations (1994).

It is against this background that LUL, as a major railway operator, aims to provide a safe, efficient and quality service to its many millions of customers. As such it requires effective decision making support to assist in the management of safety and the prioritisation of capital expenditure, ultimately to discharge its legal obligation to ensure that risks are managed and controlled to levels which are as low as reasonably practicable.

LUL believes that the techniques of risk assessment can help managers make better decisions on safety issues by providing a more systematic basis for appraising safety options. The study presented in this paper is therefore a key part of the decision making process within LUL, whose aim is to achieve an optimum level of safety for passengers at platforms, by identifying and prioritising opportunities for investment in measures which can effectively improve safety, whilst avoiding expenditure on measures whose cost is grossly disproportionate to their benefit.

ASSESSING THE EFFECTIVENESS OF MEASURES TO IMPROVE SAFETY

In order to demonstrate reasonable practicability, or gross disproportion, it is becoming increasingly the norm in the railway industry, as well as in the major hazard industries, for the extent of the benefit of a safety measure to be evaluated using the technique of quantitative risk assessment. In essence this involves firstly identifying a comprehensive set of possible accident scenarios, and then for each scenario deriving quantitative estimates of both how often it may occur (eg, once in ten years) and how severe the consequences may be (eg, one person suffers fatal injuries). The effectiveness of the safety measure is then assessed by considering the extent to which it can be expected to reduce either the frequency or the severity of each accident scenario. Finally, the results of this calculation are incorporated into a cost benefit analysis to determine the balance of practicability.

In this case the relevant accident scenarios and the current level of risk at the platform/train interface have been established from the record of passenger accidents for the last few years. The accident categories considered in the study are person under train, person on track, caught in doors, and fall between train and platform (or between cars).

The next stage involved assessment of each safety measure, to determine its effectiveness in reducing the risk. Consideration has to be given to each category of accident separately, as some safety measures are effective only against certain types of accident. In addition, both the frequency and the severity components of the risk must be addressed. This is important because the various measures address these components differently.

For example, a measure such as the use of closed circuit television (CCTV) is of benefit because, once the driver becomes aware that a passenger has fallen between the train and the platform, he will not start the train, and so the injuries suffered by the passenger will be much less severe. However, the measure is not expected to influence the frequency of incidents -- passengers are still just as likely to fall between the train and the platform. Alternatively, the benefit of a measure such as platform edge doors is that it would reduce the frequency of accidents, rather dm the severity of their consequences.

The risk reduction assessment process was essentially a judgemental exercise, based on the experience and expertise of LUL operations personnel familiar with day to day operations, and facilitated by a risk assessment consultant with considerable experience of the issues involved. In particular, a meeting was held where estimates of the risk reduction achievable by each measure were agreed. In some cases human factors specialists with experience in railway procedures provided additional input where the improvement envisaged relies on assumptions about staff and passenger actions.

As regards the cost estimates, LUL provided general background data on the number of stations, trains, platforms, etc., and prepared a preliminary specification of each of the potential risk reduction measures.

In assessing the net cost to LUL, it is necessary to take into account the capital cost (eg, for hardware and its installation), any change in on-going operating costs, for example, in maintenance costs and/or staff costs, and any operational benefits (revenue from increased usage due to passenger benefits). In terms of passenger benefits, there is the societal benefit associated with the reduction in accidents, the reduction in the costs associated with delays (as a result of reducing the frequency of accidents), but also any potential disbenefit associated with disruption caused by passenger mis-use of the safety measures (eg, mis-use of emergency stop plungers).

A standard cost benefit calculation was then performed for each option. LULs prefer-red measure for distinguishing the merits of each of the candidate risk reduction measures is the benefit to cost ratio, calculated as the sum of passenger benefits divided by the net cost to LUL.

Although there is always a margin of uncertainty associated with the results of a quantitative analysis, experience has shown that the envelope of uncertainty is often swamped by the much larger variation in the effectiveness of safety measures themselves. Thus the benefit to cost ratio is generally a useful discriminator between upgrade options because some options are usually found to be clearly worthwhile, and others clearly ineffective, even allowing for uncertainty in the risk and cost estimates. In fact in this study, for two of the measures the overall benefit was found to be negative.

THE SAFETY MEASURES CONSIDERED

Guards on trains. A guard is positioned at the rear of each train and has access to an emergency brake device with which to stop the train in an emergency. His primary role is to observe the doors closing and the departure of the first passenger carriages from the platform. Trains are also fitted with standard passenger emergency alarm equipment.

Platform attendants and emergency stop plungers. A platform attendant with access to an emergency stop plunger device is positioned in a prominent position on each platform with maximum available visibility of the length of the train and platform. Operation of an emergency stop plunger will illuminate red signal lights at the trackside ahead of train to warn the driver to stop immediately whilst entering or leaving a platform. On Automatic Train Operation (ATO) lines, the Victoria Line for example, the device will interrupt track circuit signalling to bring the train to a halt automatically.

Emergency stop plungers on platforms. A series of platform wall mounted alarm devices are positioned at predetermined intervals along the platform for use by passengers and staff.

Emergency stop plungers on trains. An alternative to installing emergency stop plungers on the platform wall would be to install them on the outside of the train. This would be a simpler device and quicker acting as it could directly apply the train brakes, and would not have to illuminate red signals or interrupt ATO codes.

Train borne passenger emergency alarm automatic braking system. Fitting all trains with a (Central/Jubilee line type) modified passenger emergency alarm system which, if operated whilst leaving the platform area, will automatically apply the emergency brakes and bring the train to a halt. Beyond the platform area, the system will function as normal, alerting the driver to stop at the next station.

Reducing gaps between platform edge and doors. Platform gaps modified to comply with a maximum gap criteria (150 mm both vertical and horizontal).

Inter car barriers. A simple, robust and flexible barrier (either a gate/screen type or a simple rope) is installed between all cars to prevent passengers from falling between cars or, in the case of the visually impaired, mistaking the gap for a door aperture.

CCTV monitors in cabs. A Central Line type CCTV facility is provided in all operational cabs to enable the driver to observe the approach and departure of a train from the platform.

Platform edge doors. Doors are provided which run the full length of the platform and cannot be readily climbed by passengers. The doors can only open if a train has properly berthed and must be proved to be closed before a train departs. A door may need to be opened manually from the train side if normal power falls.

Platform surveillance using CCTV. Sufficient station staff available to monitor all departing trains from a central point in the station. This measure would necessitate enhanced communication systems to ensure that the station staff monitoring the CCTV have a means of immediate contact with the driver in the event of an incident, or some means of automatically stopping a train for example, access to an emergency stop plunger.

Improved door engineering/control mechanisms. Reducing the doors closed tolerance to the latest standard - 6 mm. All rolling stock would be fitted with door closing alarms, and all door system tolerances improved. Door edge rubbers would be stiffened to the maximum practical extent.

FINDINGS AND RECOMMENDATIONS

Each of the measures described above has been assessed, and the findings and recommendations are described below.

  1. Overall a negative benefit to passengers is found for platform emergency stop plungers for use by passengers (due to the high frequency of misuse) and also for emergency stop plungers fitted on the outside of trains (due to the likelihood of injuries to passengers attempting to use them on a moving train).

  2. Reduction of gaps between platform edges and trains, and improvements to train door engineering and control mechanisms generate only very small improvements in safety (because they each address essentially only one category of accident) and are not cost effective.

  3. Platform edge doors, guards, platform attendants with access to an emergency stop plunger, and increased surveillance of platforms by CCTV, all go some way towards improving safety, but are disproportionately expensive for the benefit delivered, so money would be better spent elsewhere. In fact guards have been killed or injured carrying out their duties in recent years.

  4. Passenger emergency alarms and inter car barriers also target only certain specific accident scenarios. However, because they are relatively inexpensive they have the highest benefit to cost ratio of all the measures considered, and therefore merit further consideration.

  5. Targeted use of platform attendants with access to emergency stop plungers during peak hours contributes a significant net benefit and is also cost effective. (Platform attendants can also provide non-safety benefits, such as assisting with directions at interchange stations). The benefit from cab CCTV monitors is greater still, and although the costs are also somewhat higher, this option is fairly cost effective overall. Both these measures merit further consideration.

It should be noted that some of the measures recommended for further consideration address the same accident scenarios. The overlap between different measures therefore needs to be taken into account if more than one measure is taken forward for implementation.