Author: Chris Turner, www.cycom.co.uk. Copyright Cycom Limited, 2002.
This scheme eliminates any possibility of train-train collision at any speed while also allowing the number of trains running on existing tracks to be quadrupled. The legacy discrete signalling system is fundamentally flawed by its own discreteness and so therefore any train protection system that relies on signalling is also flawed (and therefore a wasted investment). Implementing this scheme is likely to be cheaper than alternatives or even cheaper than maintaining legacy signalling.
Until now, yes, it has been impossible. Now it is possible and even easy. The technologies that now make this possible include computers, mobile telephone networks and satellite global positioning systems.
Yes. While people are understandably dubious about trusting a computer with their life (especially given that many peoples experience of a computer comes from using Microsoft PC technology) it is still a fact that real enginneers are able to design for reliability (that greatly out-performs humans) using techniques such as redundant independent systems. i.e. the system continues to function even if parts of it fail and intrinsic safety, i.e. if the sytem should fail it also removes all dangers.
Trains are heavy, fast moving objects. They cannot be steered to avoid obstacles in their path. Their only option when faced with an obstacle is to stop. The stopping distance is large. This distance exceeds the visible range of the driver. The problem is to get from A to B without hitting anything. The principle obstacles are other trains.
The track is segmented into discrete segments of track. These segments are protected by signals (red lights) at the entrance to the track segment. Various means are employed to detect that a train occupies a track segment e.g. an electrical short circuit between rails.
The protection points are only at a few discrete points, i.e there is not a red light every 10 metres.
The protected segments of track are relatively long so that large lengths of track are unoccupied at any point in time.
The time between a danger signal and an actual collision is variable. (the obstacle may be right at the beginning of a segment or near the end of a segment, the signalling system does not distinguish between them.
The time between a danger signal and an actual collision is not made known to the driver.
There is no predictive element to the system.
Both the signalling and the train detection mechanisms can be unreliable and expensive. The expense is proportional to track length and desired safety factor.
The effect of the above features is to:-
limit track occupancy (reduces the number of trains)
causes trains to stop even if a collision is not imminent (wastes fuel, slows journeys, ripple effect).
reduces driver confidence in the accuracy of the danger (leading to frustration and complacency).
give very little warning time in some cases of imminent collision.
there is a financial pressure to reduce safety margins.
The discrete, quantized nature of the legacy system is completely replaced by a continuous, non-quantized system. The non-predictive nature of the legacy system is replaced by a predictive system. The detection of the position of trains, their length, their direction and speed is continuously monitored by two satellite GPS systems fitted to the front and back of the train. This information, together with the driver profile and journey plan for the train is broadcast via the mobile telephone network to central office computers and also via radio to any trains ahead or behind on the track. Computers both on the trains and at central office model the trajectory of the trains near each other to predict the probability of collision and to suggest corrective action.
The predicted future scenarios will be transmitted to the drivers both as data and as synthesized voice using the richest, most expressive sounds designed to induce an appropriate response by the driver and providing quantitative information that will permit him/her to respect the advice.
Examples might be:-
"Mad Jake Brown is 600 metres behind and gaining. Probable collision in 63 seconds, possible collision in 32 seconds. Maintain speed.
"Hugh Jones is 600 metres ahead and slow. Probable collision in 63 seconds, possible collision in 32 seconds. Slow to 105 mph over 6 seconds.
"Hugh Jones is 300 metres ahead and braking hard. Probable collision in 29 seconds, possible collision in 21 seconds. Stop over 13 seconds.
(in an urgent tone of voice) "Possible collision in 13 seconds. Braking time 9 seconds. Braking sequence will be automatically applied in 3 seconds. Stop now.
(in a relaxed tone) "Clear ahead for 2.3 minutes. You are 30 seconds behind schedule. Suggest accelerate to 135 mph over 60 seconds."
(interim period) "Red light 90 seconds ahead. To avoid braking I suggest no throttle for 61 seconds then maintain 74 mph.
(for proud drivers) "Red light 90 seconds ahead. "..... (silent pause to assess driver action) ......... "Nice adjustment!".
The continuous nature of the danger assessment and the continuous application of the control messages to the driver means that:-
the maximum possible warning time is given to avert the danger.
a minimum warning time can be set and ensured.
the track occupancy can be greatly increased without reducing safety.i.e. 4 times more trains can be run.
The safety margin can be set at given level even if train operating speed is increased.
Reliability through redundancy. The front and rear systems act as backup for each other. Direct train-train radio and computation backup the central office and mobile network. Europe will have a second GPS system soon and dead-reckoning will confirm even these signals.
Predictive and so early application of control gives smoothest ride, lowest running cost, and maximum throughput. Effectively, the lights are always green and there is no braking except at stations.
Drivers will feel empowered by the information. Non-urgent driving suggestions might be made only on driver demand when the drivers actions are assessed as sufficiently compatible with the advice. i.e. the computer will only offer advice to the incompetent.
Cost is independent of track length or train speed and is proportional to the number of trains.
Maintainance of the legacy signalling can be phased out.
The alternative train protection systems which rely on the legacy discrete sigalling system are made obsolete.
Even improves performance and safety in presence of legacy signalling (by predicting and avoiding red lights). You cannot pass a signal at danger if you never reach a signal at danger!.
This scheme eliminates any possibility of train collision at any speed while also allowing the number of trains running on existing tracks to be quadrupled. The legacy discrete signalling system is fundamentally flawed by its own discreteness and so therefore any train protection system that relies on signalling is also flawed. Implementing this scheme is likely to be cheaper than alternatives or even cheaper than maintaining legacy signalling.