The high off-peak frequency and flexibility to meet demand surges which unmanned trains offer may lead to conversion of existing lines in Paris or Berlin. Tom Parkinson and Ian Fisher believe the caution displayed by operators and safety regulators is no longer justified

DEVELOPMENTS through the last century have made the operation of high capacity driverless metros technically and economically possible. Ample experience has been accumulated over 15 years to show that they are safe, and it is time to move on to the next stage. Several metro administrations have announced plans to convert existing lines to driverless operation, including Berlin and Paris.

It was a modest step to combine multiple-unit control, automatic acceleration and coded track circuits generating in-cab speed displays to automate the driving function, notwithstanding difficulties in achieving final stop accuracy that often required complex wayside equipment.

Demonstrations of automatic train operation (ATO) were carried out on New York’s Times Square shuttle, Barcelona’s Horta Line and London’s Hainault Loop in the early 1960s. The focus was less on driverless operation than on providing more consistent driving and the potential economies of one-person operation.

These trials matured into ’attended ATO’. London’s Victoria line was the first into revenue service on September 1 1968, with Philadelphia’s Lindenwold line close behind in January 1969. Many other systems adopted ATO, including most of the new US metros. Established operators had labour contracts to contend with, but even new systems were reluctant to dispense with an attendant at the front whose duties were limited to door closing and observation. Only Lindenwold saw the merit of a wider role, placing the attendant in an open position accessible (as in a bus) to passengers in the first car, and providing 24 hour service.

It was left to new-technology automated guideway transit to pioneer driverless operation. Westinghouse Electric Corporation’s 1965 Transit Expressway demonstration line in South Park, Pittsburgh, was the first. Conceived as a way to run frequent service economically on urban systems, it failed to gain acceptance for public operation. Instead it evolved into the smaller scale unattended peoplemover, of which there are now over 30 in service from several suppliers.

Most operate in a protected environment, notably at major airports where skilled personnel are on immediate call and personal security is tight. However, Japan has several New Technology Systems of significant length and passenger volume.

Others are planned or under construction, including three lines feeding the Singapore metro network. The first of these, opened last year at Bukit Panjang, sees the latest variant of the Westinghouse (now DaimlerChrysler) system following its first public domain application for the Miami MetroMover in 1985.

Driverless metros

Driverless operation in the tough environment of urban public transport thus evolved from these small institutional applications. However, the first North American public driverless operation was an anomaly. Morgantown Personal Rapid Transit, a demand-response (off-peak) small-car system, opened in 1969. It still operates successfully but has not been replicated.

Automotive rather then traditional railway vehicle technology was frequently used, and the scale of operations gradually increased as confidence in the ability to deal with safety issues such as fire and evacuation of trains was gained. Such systems are often referred to as intermediate capacity transit or mini-metros.

The dividing line between metro-scale operations and more specialised applications is bound to be arbitrary, but we have listed seven systems in Table I that clearly fall into the metro category.

VAL in Lille, developed by Matra Transport (now 95% owned by Siemens), was the groundbreaker. This was the first city in the world to adopt driverless operation for a new metro, and the 13 km initial section of Line 1 opened in April 1983. The capacity is 15000 passengers/h in one direction, lower than a conventional metro, and all stations are equipped with platform edge doors. Two lines are in operation, and planned expansion will see the network reach 45 km with 45 stations.

Vancouver’s SkyTrain was the first large-scale, driverless operation in North America. It was the first to use steel wheels, although the linear motor drive and steerable axles were unconventional. SkyTrain uses ALRT technology developed by Canada’s Urban Transportation Development Corp, now part of Bombardier. The initial 21·4 km line from downtown Vancouver to New Westminster opened for revenue service on January 3 1986 after 10 days of free rides. Extensions have seen the line grow to 29 km, mostly elevated. A 21 km expansion is to open in 2001-02.

SkyTrain uses Alcatel’s Seltrac moving block signalling for train control, a system prototyped on Berlin’s short U-Bahn Line 4 where train and platform attendants were retained. While SkyTrain is driverless, roving attendants and security staff are assigned to patrol stations, check fares, provide information, and resolve minor technical faults. After considerable debate, platform edge doors were omitted. Intrusion into the guideway is detected by vibration sensing plates in stations and optical sensors at platform ends. Fencing along the at-grade section has intrusion detection where road vehicle incursion is possible.

In the same period, ALRT technology was used with attendants for a feeder to the Toronto Subway, and for the driverless Detroit Downtown People Mover. Like the concomitant Miami MetroMover, Detroit has neither platform doors nor intrusion detection. Much later, the Putra light metro in Kuala Lumpur was the second driverless application of Canadian ALRT technology to a major urban line. Opened in two sections in 1998-99, the 29 km line with 24 stations is elevated at each end with a tunnel under the city centre.

Although the other metro lines in Lyon have attendants, the 12 km underground Line D was built for driverless operation using Matra’s Maggaly ATO. Problems with automation led to the line being opened in stages in 1991-92, with one section operated manually for a time. The automation is derived from that used for VAL systems, but the larger cars are classic French metro pneumatique. Lyon followed Vancouver’s lead in omitting platform edge doors in favour of a detection system below the platform edge.

Toulouse opened its first metro line on June 26 1993 using VAL technology. The mainly underground 10 km Line A serves 15 stations and carries 150000 passengers a day. Platform-edge doors are used to prevent unauthorised access to the tracks. Construction of a 17 km Line B is due to start this year, with opening expected in 2006.

VAL subsequently appeared in Taipei, where a 10·9 km segment of the 24 km Mucha line opened in March 1996. This elevated line has been beset with contractual and technical problems, and the extensions are under review. Matra also installed the VAL package at Chicago O’Hare and Paris Orly airports, and on a short line in Jacksonville.

To date, all driverless systems have been proprietary, leaving the operating authority with sole sourcing for new cars, extensions and spares. In Jacksonville the demanded cost of a short extension resulted in the entire system being replaced by automated monorail with Bombardier technology.

Lyon apart, all driverless systems had used small cars revealing their automated guideway transit antecedents. For Kuala Lumpur and the new Vancouver cars, Bombardier developed the 2·65m wide, 35m long, eight-axle articulated Mk II car, which takes linear motor technology to its limits.

Breakthrough in Paris

Météor is the newest metro line in Paris, and it marks a major step forward for unattended operation. The 7 km initial segment opened on October 15 1998 with five intermediate stops, having been threaded into the dense fabric of a 100 year old metro serving one of the world’s great capital cities.

Operator RATP had already gained experience of driverless trains on OrlyVAL, a feeder connecting the city’s second airport to RER Line B. Météor is in an altogether different category in terms of the numbers of people to be moved. It includes a 2·8 km non-stop run between Gare de Lyon and Châtelet, the longest on the Paris metro, and is designed to relieve the parallel RER Lines A and D. It is expected to carry almost 100 million passengers a year when a Fr870m extension from Madeleine to Gare St Lazare opens in December 2003.

Further extensions are planned: on December 17 the Ile de France region approved a grant of Fr325m for a southern extension to Olympiades, costed at Fr730m, which should be ready for service by the end of 2005. In the longer term the line would be projected north from St Lazare to take over the Line 13 branch to Port de Gennevilliers (RG 12.98 p839).

The ATO system builds on Matra’s experience with VAL and Line D in Lyon. Together with the wider station spacing on the line, it lifts capacity by one-third compared with manually driven lines. Line 14 also has the highest number of daily train trips and the shortest off-peak headways. RATP sees Line 14 as a test case for unattended operation with a view to automating other metro lines as rolling stock is renewed.

Service is provided by a fleet of MP89 cars marshalled into 21 wide-gangwayed trains of six cars each; they can be extended to eight cars as demand rises. While no personnel are assigned to trains, car interiors are equipped with CCTV monitored in the control centre.

Mini-metros in Japan

In addition to these seven metros, there are five rubber-tyred NTS lines in Japan that operate some or all of the time without drivers, and are of sufficient length and ridership to merit mention. They include two lines in Greater Tokyo and two in Kobe, the Portliner (6·4 km, 1981) and the Rokko Island line (4·5 km, 1989). In Hiroshima, the 18·4 km New Tram (1994) uses 132 bus-sized cars to carry 90000 passengers/day. There are also two installations in Tokyo and Tama.

There are other driverless systems in Japan and around the world serving airports or other institutions. Atlanta Airport is believed to have the highest ridership at 120000 per day. In Newark the airport monorail is being extended to a new Northeast Corridor station near the airport, and New York’s JFK airport is building the AirTrain system with Bombardier’s ALRT technology to link a loop around the terminals with the NYCT subway and Long Island Rail Road.

The case for automation

Although driverless metros saved on-train staff, they are still required to sort out technical and passenger problems, and respond to emergencies within a reasonable time. Vancouver aims at a 5min maximum response time, and has roving attendants allocated to groups of stations. Except in the rush-hour there are as many attendants as trains.

Fig 1 compares direct operating costs per passenger-km for selected operators in North America. While Vancouver is indeed the lowest cost, other well-managed metro and light rail systems with attended trains are close. In contrast, the driverless Miami MetroMover is almost off the scale!

Vancouver SkyTrain attendants represent 18·9% of direct costs which is 2·5cents/passenger-km. Patco drivers represent 10·3% of annual costs or 1·7cents/passenger-km, but SkyTrain has a minimum evening headway of 5min whereas Patco maintains 12 to 20min. In truth, operating costs are more related to intensity of use and the sensible deployment of technology than to whether or not trains are staffed.

In Vancouver, attendants play a first line security and fare-checking role but have no enforcement powers, so separate security staff are needed - and their number has grown with recent increases in graffiti and vandalism. However, the attendants’ mobility and training does make them more effective than a similar number of drivers, and passengers appreciate the assistance and information available from such staff.

Trains can be rapidly added to service to cope with surges. At all off-peak hours Vancouver positions spare trains in pocket tracks throughout the network, occasionally to augment service but mainly to substitute for trains with technical problems - or more commonly, vandalised or soiled trains in accordance with the zero tolerance graffiti policy. In this way service gaps or disruptions are held to a very low level.

Handling special events efficiently is a major advantage of driverless systems. For major events in Vancouver, evening service can be expanded from the base 17 trains to the maximum 35 easily and inexpensively. There is no totally free lunch; this has to be pre-planned, and extra staff are still required for security, crowd control and manual fare collection. Nevertheless, handling crowds after football games or other events is a profit centre with additional revenue two to three times marginal costs.

By far the largest advantage of driverless systems is the ability to offer frequent service throughout the service day. Passenger surveys indicate that this is highly regarded and a contributor to higher ridership and revenue. Where peak throughput is moderate, as in medium sized cities, increased frequency means shorter trains and stations. Halving the length of a station cuts the capital cost by about one-third - particularly significant for underground routes.

Safety (as distinct from personal security) is no longer a major issue; it is certainly not a public concern. A cab-bound attendant’s ability to deal expeditiously with passenger illness or crime is poor, and separate security staff or police are invariably required. Driverless systems have established an excellent safety record, marred only by the 1993 accident on Kobe’s Portliner when a train overran a terminal station injuring 60 passengers.

Platform screen doors

Even with widespread ATO, the number of accidents due to human error is significant, often related to alcohol, drugs or mental illness. Nevertheless, convincing rail safety regulators that leaving the driving to a computer is safer than putting a driver or attendant in the cab remains difficult. The most contentious issue is that of platform edge safety.

VAL and the first driverless systems in airports used platform screen doors. These are expensive and difficult to accommodate in simple stations. They are also likely to reduce overall reliability - it is well known that train doors are the single largest cause of delays on many metros.

Because SkyTrain was the first major driverless system in North America, particular effort was made during design to protect the tracks from intrusion. It was judged that on a smoke-filled train, stopped out of alignment with platform gates, elderly or handicapped passengers, possibly in a state of panic, would be unlikely to be able to actuate emergency releases and manually open both the train and platform doors and exit within the time required by safety codes. Consequently, open platforms were selected, along with a system that detects objects on the track and applies the brakes faster than the most vigilant driver. This philosophy was followed on Lyon Line D and at Kuala Lumpur’s above-ground stations.

Suicides are an unfortunate fact of rail transit operation, and delays due to trespass on the right-of-way and into tunnels appear to be increasing. Platform screen doors help if all stations in a city are equipped, but it is difficult to prevent access through the platform end gates intended for emergency train evacuation between stations.

In general, the case for platform screen doors applies equally to driverless and driven systems. In hot climates, platform doors are used primarily to permit station air-conditioning. Platform screen doors were installed on London’s Jubilee Line Extension in order to keep the maximum air velocity in stations below 5m/s with safety a secondary factor.

To assist in the transition to one-person train crews, Toronto’s Sheppard line was to be built with platform screen doors, although they were eliminated in a budget cutting exercise. A description of Tokyo’s Line 12 states that platform screen doors were provided to improve the appearance of station platforms.

Next step is conversion

In an increasingly impatient world, long waits for public transport drive away all but the captive passenger. Frequent, economic service at all hours was the original rationale for the development of the Westinghouse system in 1963, and it remains the most compelling advantage of driverless operation.

While staff savings have proved to be modest, roving staff are generally better placed to deal with problems and provide passenger assistance and information.

Dispensing with a driver or attendant is still difficult to accept, and is opposed by labour unions on most established metros. The slow but continuing move to one-person operation is a sign of change.

Meanwhile, with seven driverless metros and over 30 small driverless systems in the world, the industry is still waiting for the first traditional metro line to be converted to driverless operation.

TABLE: Table I. Driverless metro lines

City Type 1 Train First Route- Daily

control open km riders 2

Lille (2 lines) VAL Matra 1983 45 210 000

Vancouver ALRT Alcatel 1985 29 150 000

Lyon Line D Metro Matra 1991 12 200 000

Toulouse VAL Matra 1993 10 150 000

Taipei Mucha VAL Matra 1996 11 50 000

Kuala Lumpur ALRT Alcatel 1998 29 90 000

Paris Line 14 Metro Matra 1998 7 200 000

1. Both metros and the VAL lines are rubber-tyred2. Latest estimates for 1999 ridership

* Tom Parkinson is a consultant involved in Vancouver SkyTrain expansion, having worked on rail transit planning and design for 35 years. Ian Fisher is a Transport Planner with the City of Vancouver.

CAPTION: The Westinghouse Electric automated guideway transit system, shown here on Miami’s Metromover, was the antecedent of all driverless systems. The platform has no screen

CAPTION: Platform screen doors on Lille’s VAL, shown open. When closed, the space between the doors is an emergency exit from a mispositioned train

CAPTION: Vancouver’s SkyTrain crosses the 616 m cable-stayed Fraser River Bridge between New Westminster and Surrey

CAPTION: Lyon Line D at Vieux-Lyon; an intrusion system below the open platform edge detects fallen objects or persons

CAPTION: Fig 1. Direct operating costs for various metros in 1997; US$ per passenger-km

CAPTION: Paris Line 14 (Météor) at Madeleine. Platform edge doors are provided at all stations. The screens do not extend to the ceiling. While not fully controlling air circulation, they do eliminate train arrival air surges

CAPTION: The red plates between Vancouver’s SkyTrain tracks and the linear motor reaction rail are the Passenger Intrusion Emergency System that uses microprocessor analysis of vibrations

CAPTION: Kuala Lumpur’s Putra light metro uses Bombardier’s MkII ALRT vehicles which will also be supplied for Vancouver’s second SkyTrain line

Driverless metros poised to expand

While automatic driving is now commonplace, there is still reluctance among heavy metro operators to run trains with no staff on board. Experience over 15 years shows the benefits lie not so much in overall staff reduction as in the ability to maintain high off-peak frequency, and bring trains into service quickly to cover failures and demand surges. Driverless metro lines are now operational in seven cities. Tom Parkinson and Ian Fisher believe their advantages will soon lead to the conversion of existing lines, as the undue caution of established operators and safety regulators is no longer justified.

Les métros sans conducteur prêts à prendre de l’expansion

Alors que le pilotage automatique est désormais monnaie courante, les exploitants de métros lourds sont encore réticents à faire circuler les trains sans un seul agent à bord. L’expérience de ces 15 dernières années a montré que les bénéfices ne se situent pas tant du c?€?té des réductions de personnel, mais plut?€?t dans la capacité d’offrir des fréquences élevées de circulation en dehors des heures de pointe et de mettre en marche rapidement des trains à la suite de pannes et d’augmentations de la demande. Des lignes de métros sans conducteur fonctionnent à présent dans sept villes. Tom Parkinson et Ian Fisher pensent que leurs avantages mèneront bient?€?t à la conversion des lignes existantes, étant donné que la prudence excessive des exploitants existants et des commissions de sécurité n’est plus justifiée

Fahrerlose U-Bahnen bereit zur Expansion

Obwohl das automatische Fahren jetzt schon verbreitet ist, herrscht immer noch grosse Zurückhaltung bei den U-Bahn-Betreibern, ihre Züge ohne Personal an Bord fahren zu lassen. Über 15 Jahre Erfahrung zeigen, dass die Vorzüge nicht so sehr in der generellen Reduktion von Personal liegen, sondern viel mehr in der M?€?glichkeit, einen dichten Betrieb zu Randzeiten zu erm?€?glichen, sowie Züge bei St?€?rungen oder Bedarfsspitzen rasch einsetzen zu k?€?nnen. Fahrerlose U-Bahn-Linien existieren jetzt in 7 Städten. Tom Parkinson und Ian Fisher glauben, dass diese Vorteile sehr schnell zur Umwandlung von bestehenden Linien führen werden, da die übertriebene Vorsicht von Betreibern und Sicherheitsorganen sich nicht länger rechtfertigen lässt

Habr? m? s metros sin conductor

Pese a que la conducción autom? tica es ya común, existe todavía resistencia entre los grandes operadores de metro a poner en operación trenes sin personal a bordo. La experiencia de los últimos 15 años demuestra que las ventajas no son las de una reducción general del personal, sino en la capacidad de mantener una alta frecuencia en períodos de menor afluencia, así como poder poner los trenes en servicio con rapidez para cubrir fallos y puntas de demanda. En la actualidad ya hay líneas de metro sin conductor funcionando en siete ciudades. Tom Parkinson e Ian Fisher creen que sus ventajas pronto conducir? n a la conversión de líneas existentes, ya que la innecesaria precaución de los operadores establecidos y de los reguladores de las normas de seguridad ya no est? justificada