Original Docklands Light Railway Signalling
The first signalling system installed on the Docklands Light Railway in London (UK) was designed with an early form of the "distance to go" system of Automatic Train Operation (ATO) using an on-board computer containing a line speed profile. This edited paper was originally written by Nick Cory.
More articles on this site about signalling are listed in the Signalling Index
Introduction to ATO
Many ATO systems rely entirely on track-mounted equipment to provide the "current target" and "current maximum" speed information to the on-board train controller. Typically, the ATP codes are used. So, if an "80/80" code is received, the ATO accelerates the train towards 80 km/h and attempts to maintain that speed, subject of course to the limits of the motor characteristics. When, later, an 80/60 code is encountered, the ATO begins braking to 60 km/h while ATP continues to ensure that 80 is not exceeded. To stop smoothly, the code sequence must be maintained down to zero, the final code being 25/0 or similar. Note there is also the 0/0 code (which is usually, in fact, no code at all) in the overlap or buffer zone that is always provided in coded track circuit systems. A correctly-functioning ATO of course, never drives the train into this zone. See the Metro Signalling and ATP page on this site for more information.
The Docklands ATO System
When the Docklands Light Railway in London, England, was being planned, it became apparent to GEC-General Signal, the signalling contractor, that a conventional coded track circuit system would not be suitable. The alignment of the railway included many tight curves with speed restrictions of 20 to 30 km/h, with straight sections in between where up to 80 km/h was possible. Had coded track circuits been used, it would have been necessary to have many codes and thus many track circuits to give the level of fine control needed to yield the minimum journey time. GEC-General Signal opted instead for a very simple block system using, in the main, just one track circuit from station to station. This track circuit was overlaid with a single code, the "ATP safety tone", used to hold off the emergency brake. Safety tone was only transmitted along block sections that were clear, the route locking being proved where necessary. This arrangement was more than adequate to meet the initial headway requirement of 240 seconds (s).
The ATO on-board computer was provided with a database of "route speed profiles". A profile was simply a table of distances from the station and the speed value applying at that position. So, if the track from one station to the next was straight and the platform-to-platform distance was 560m, the speed profile would be:
Ref x v
The ATO software included a "look ahead" calculation that allowed it to apply brakes smoothly at the appropriate distance from the position at which the new speed value came into effect. In this way it was possible to provide infinitely-variable speed control of the trains and run at the highest performance that the adhesion could support. In fact, this was not as high as initially expected. The original aim was for trains to brake at an average of 1.0 m/s2 from 80 km/h to standstill. This proved to be impracticable in service due to the vagaries of wheel-rail adhesion and, eventually, a value nearer to 0.6 m/s2 was found to give satisfactory performance on all but the most slippery rail. The ATO was soon found to be useful for other purposes: extra data was added to the route profiles to control the flange lubrication equipment, so that it only applied lubricant in curves. "Gap flags" were also added, which caused the ATO to smoothly decrease the traction current on the approach to a third rail gap and increase it again afterwards, thus minimising the arcing that occurs as the third rail shoes enter the gap.
The ATP system had to provide for infinitely-variable speed monitoring. This was achieved by using an inductive loop laid along the track. The "go" and "return" wires of the loop were crossed over ("transposed") at spacings corresponding to 1 s of travelling time at the prevailing maximum speed. Each loop was fed with 6A at about 400Hz, the "ATP speed monitor tone". A circuit on board the train detected this tone and also the phase change that occurred every time the detector passed by a transposition. A timer was reset to zero at each transposition. As long as the timer always ran to 1 s before being reset, a vital relay driver kept the emergency brake loop closed. But if another transposition was detected less than 1 s later, it meant that the train was "overspeeding"; the timer was reset before it could send its pulse to the relay driver and the emergency brake was applied. The original plan to lay the "go" and "return" sides of the loop in the webs of the running rails had to be abandoned when it was found that the rail inductance prevented the transpositions from being detected, so one wire was laid along the track centre line.
At points and crossings, complex switching circuits had to be used (controlled by the interlocking) to ensure that there was always the correct loop energised beneath a train and no energised speed monitor loop behind the train that could be misread by a following train. With each loop through a junction having its own unique transposition pattern, there could be up to four loops, i.e. 8 cables required to be threaded along the rails in turnouts. Although designing the loops and programming the interlocking was complex, the loops were not the maintenance headache that might be imagined. However, they became a problem soon after commissioning when it was realised that the capacity of the line had to be raised, with extensive track remodelling. New loops had to be threaded beside the old ones in advance of commissioning an installer's nightmare.
The ATO was able to cope with additional "signals" between stations. In reality, the DLR did not have any signals, but the start of each block was marked with a trackside signboard. The ATO would drive from a station towards the intermediate block and, by default, stop there. To circumvent the stop (or restart the train), a third tone, the "ATO go tone" (408 Hz), was applied to the rails together with the ATP safety tone. However, the system could not cater for a change of route being demanded by the dispatcher after the train had left a station. If this became necessary, which was rare, the train captain (the travelling member of staff) had to open up the "emergency driving position" and drive manually.
At stations, the ATO had to be given information about which profile to use for the next journey. The ATP also had to be given data about the location of the current platform (left or right) for release of the correct doors. The ATP was responsible for selecting the direction of travel for the next journey. All this was achieved using a device called a Docking Data Link, or DDL. The DDL consisted of a vital computer and a 1.8m long inductive coil laid alongside one rail. The train had to stop with a pickup coil aligned over the trackside coil, otherwise data transfer would not take place. In the download direction, the DDL was eventually used to prove a train stationary at a platform and obtain a faster release of route locking than was possible using the traditional "track occupied for time" method.
The DDLs were all linked via a "wayside Automatic Train Supervision (ATS) " processor to a central ATS computer, which not only set routes via the interlocking but also accurately monitored the arrival and departure of every train at every platform. As a train left a platform, the ATS calculated the "dwell" or waiting time the train would need at the next platform in order to leave exactly on schedule. If the train were already late, the ATS would command the ATO to run at the "all out" performance level, meaning that ATO would ask for the maximum possible tractive effort demand throughout the route (observing speed limits, of course). If the train was on time, the ATO would use a more economical style of driving, coasting for a major part of the journey to save energy and taking about 10% longer in the process.
When the tunnelled extension to Bank was opened in 1989, the original "P86" trains were prohibited from using it since they did not conform to post-1987 fire protection regulations. They were confined to the original Tower Gateway branch. As part of the safety system, spare bits in the DDL message were used to send an "P89 allow" bit to any train when a route was set to Bank. P89 trains were identified as such by a wire loop in the on-board DDL and, with both bits set to 1, they could go to Bank, whereas a P86 would just sit there at Shadwell staring westbound. Ironically, DLR eventually sold the P86 trains to the German city of Essen.... where they operate in a tunnel! British safety paranoia strikes again. DLR would not even allow the "P86 Farewell" railtour a one-off trip to Bank!
As with any ATO system, accurate distance measurement during the journey is essential if accurate stopping is to be achieved. As is normal, an axle-driven pulse generator was used to obtain 20 pulses per revolution. Three phases with an offset of 120° gave a direction dependent signal and one redundant channel. The GEC-GS system used the transpositions of the ATP speed monitor loop to calibrate the ATO distance measurement. Any number of transpositions could be identified in the ATO route profile data with their true position on the ground, which had to laboriously measured after installation using a measuring wheel: attempts to collect "true" position using a train had to be given up. It was found empirically that, on sections where adhesion was always poor, up to 30 transpositions had to be logged in the route profile just between the brake application point and the platform, otherwise wheelslide caused the train to overshoot the DDL. GEC-GS also used transpositions in non-braking areas to automatically recalibrate the system for wheel wear. The distance measurement was initially very prone to interference from British Rail electric trains on a parallel track near Stratford, eventually solved with some "desensitising" of the signal processing hardware and "sanity checks" in the software.
The on-board ATO and ATP was based around VME bus and 68000-series processors. Trainborne and wayside DDL racks were identical. A few cards, such as vital relay drivers, were specially designed but, generally, off-the-shelf cards were used. The tones were transmitted and detected using standard "reed" track circuit units and the relay logic was implemented using BR930 series vital relays. The interlocking was implemented using British Rail SSI, the very first combination of this technology with ATP, ATO and ATS. The ATS was GRSs "Traffic Master II" running on a single DEC MicroPDP11/75 machine with another of the same type as communications server. This ATS had served the American freight roads well but was totally out of its depth with the heavy traffic levels of the little DLR. Upgrading to 11/83s helped some but did not cure all the problems. Wayside ATS and the SSI/ATS protocol converter were implemented on GEC-GS "Mark III" computers. One wayside ATS could serve three DDLs and provide them with default values for dwell in the event that no update was received from central ATS. The ATS and ATO were designed to control the vehicle destination indicators, but these were not implemented. ATS controlled only the platform destination indicators.
The GEC-General Signal system was commissioned in August 1987 when the railway opened. It was upgraded in 1990 to handle the growing traffic and train operating headways of as little as 67 s were measured during controlled trials. A number of initial reliability problems had been addressed and the contractor had developed a strategy to eliminate the remaining niggles. However, by then, the railway management had already committed to resignalling the line with "Seltrac" moving block (a system which lost out in the original tendering process for the line in 1985). The GEC-General Signal system was decommissioned in 1996.
Information about the present signalling system can be found on the (unofficial) Docklands Light Railway site.
More articles on signalling:
The Development and Principles of UK Signalling - Metro Signalling and ATP - ATP Transmission and Moving Block Systems - ATO - Route Signalling - ATC - UK Warning & Protection Systems - UK Signalling - What the driver sees - Single Line Signalling - US Signalling