HomeSignalling and TelecomsCloser running: is it feasible?

Closer running: is it feasible?

Listen to this article

Railways around the world are facing demands to transport more passengers and freight, but constructing new tracks is both expensive and unpopular. This leads to an ambition to run more trains on existing tracks, and challenges to the fundamental principles of present day railway operations are being proposed to achieve this goal.

One such principle is that trains following one another on the same track must be separated by a sufficient margin to ensure every train is capable of braking to a stand before reaching the last known position of the train in front. With braking rates limited by the adhesion between steel wheel and steel rail, the required separation increases dramatically with speed – on a high-speed line trains must run several kilometres apart. The migration from fixed block to moving block signalling (such as CBTC or ERTMS/ETCS level 3) is driven by an ambition to maximise capacity within this constraint, but demand for ever increasing capacity means that the principle itself is now under challenge.

The argument for an alternative approach starts with the assertion that it is unrealistic to assume the train in front will stop instantly. Provided the two trains have similar braking rates, the trains can run much closer together. A communication link between the trains can ensure that if the leading train starts to brake, the following train will do the same and maintain separation as the two trains slow together. This concept has been described by its proponents as “ERTMS level 4”, “Train Convoy”, or “Virtual Coupling”, and it features in national and international research agendas such as the UK Railway Technical Strategy and the European  Shift2Rail research initiative.

The precedent for this mode of operation is on the roads, where vehicles routinely run with a spacing based on the distance travelled while the driver reacts to the brake lights of the vehicle in front, not the full braking distance to a stop. While the safety levels achieved are orders of magnitude lower than on the railways, this is primarily due to the reliance on human drivers. Autonomous road vehicles are now a major research topic, and this is likely to deliver an automated version of the existing style of motorway driving that will be considerably safer than at present and could have technology spin-offs for railway applications.

So is this an idea that railway signal engineers should be taking seriously? Three questions arise:

1. Is the technology feasible?

2. Is it safe?

3. Does the concept create useful additional capacity?

Is the technology feasible?

The answer to this is undoubtedly yes. The odometry systems that report train location for CBTC or ETCS level 3 also report train speed, and braking characteristics are a known parameter within the system. Sensors developed for automated road vehicles may also play a part. A direct communication link between the trains would be needed and there are various technologies that can be used for this purpose.

Automatic driving of the trains would be essential, but this is already required to exploit the maximum available capacity from our existing signalling systems, such as on the Thameslink project in the UK. The greatest technical and operational challenge is probably in the management of transitions – how to insert or remove a train from a convoy?

Is it safe?

Safety is a much bigger question. The instinctive reaction of every signal engineer will be that it is an intolerable risk for trains to operate closer together than the braking distance. This is confirmed by a simple risk assessment – a train coming to a stand in less than its normal braking distance is a credible consequence of a derailment due to a train or track defect, or hitting an obstacle, and the collision of a following train that is running too close to stop could have catastrophic consequences with multiple fatalities.

However, think about it another way. Instead of running two trains as a convoy, they could be coupled together. Then, if the combined train derails, twice as many passengers will be affected by the initial derailment.

With two trains running separately in a convoy, there is at least a chance that the second train would have been able to stop in time – in this situation the convoy option would actually be safer.

Train Convoys article - figures-1 [online]

Depending on the separation of the trains, and how quickly the leading train stops when it becomes derailed, there will be a spectrum of consequences. The key parameter is likely to be the speed differential when the following train catches up with the derailed train in front. If this is significant, then it is very likely that the loss of life from derailment of the first train followed by collision of the second will be greater than from derailment of a double length train.

The mechanical and structural design of the vehicles could have a significant effect – modern rolling stock is designed to crashworthiness standards that consider collision with a fixed obstacle, and to allow coupling with a stationary train. How will these features perform in the derailment followed by collision scenario described above?

Paradoxically, with the right vehicle design, trains running very close together could be relatively safe, with the risk increasing as they become further apart. When the first train derails, a second train very close behind will catch up before the first train slows significantly, and the speed differential may be within the crashworthiness and coupler capabilities. The consequences would then be the same as for derailment of a double length train.

Accident scenarios involving a train on a parallel track also need to be considered. At present we are happy to allow trains to run at full speed in opposite directions on parallel tracks when a derailment on one track could block the other, but historically the accidents with greatest loss of life are those where a derailment as a result of two trains colliding on one track has resulted in a third train running into the wreckage (Quintinshill, Harrow and Clapham for example). Would this type of accident become more likely, or could the technologies used for train convoys provide additional warning to the train on the parallel track?

As well as the probability of occurrence, the other key issue for the adjacent line derailment case (and to some extent for ‘sudden stop’) is that, whilst the causes remain essentially unchanged, the consequences for an event involving convoys, rather than single trains, might be significantly greater and thus there may be an increase in risk.

From this very superficial safety analysis, it is apparent that there is no absolute answer, and the safety of a train convoy, or convoys, compared to single trains with equivalent capacity will depend on a number of parameters. The mechanical and structural design of the trains needs to be taken into account as well as the mode of operation that is envisaged and the possibility of designing in additional controls and/or mitigations.

Useful additional capacity?

There is no doubt that the train convoy principle would improve plain line capacity, but is this really the factor that puts limits on the useful capacity of the railway system?

The experience of metros that have adopted moving block CBTC signalling is that the achievable capacity depends largely on station dwell times for passengers to board and alight at the busiest stations, and turnaround arrangements at termini. If this is the case, abandoning the principle of absolute braking distances for the train convoy concept will not deliver any further improvement.

Longer platform dwell times are required in mainline applications, but this can be compensated for by the provision of additional platforms, so that the dwell time does not necessarily determine the route capacity. However, this introduces the need to switch successive trains into different platforms by moving points between trains.

Train Convoys article - figures-2 [online]

If the train convoy concept was adopted, this would mean moving points between trains in the convoy. This introduces another safety risk, that of the points failing in position that would derail the train. This is a very credible scenario, much more frequent than a train derailment, hence the universal practice of proving that the points are detected in the required position before granting a movement authority to an approaching train – so trains in the convoy would have to be spaced out to braking distance on the approach to any diverging junction. The same factor limits the benefits from existing moving block signalling technologies on complex track layouts – they become fixed block at the very places on the network where capacity is most critical.

This implies that, on its own, the train convoy concept cannot deliver a real increase in capacity. A whole- system approach is needed that starts by identifying the real limiting factors in complex areas of the network, and then seeks out relevant technical and operational concepts that will deliver an improvement in these critical locations.

Realistic option?

So should the concept of train convoys be taken seriously? Going back to the original questions at the start of this article:

1. Is it technically feasible? Technically, it can almost certainly be made to work, but there will be significant operational challenges to make it useful in practice.

2. Is it safe? This will depend on a number of factors so it should not be ruled out entirely on safety grounds but the safety argument will be difficult.

3. Will it deliver real benefits? Not without addressing other factors limiting practical capacity at stations and junctions.

Nevertheless, no doubt more will be heard on this topic in the future. ‘Virtual Coupling’ is one of the topics to be covered in the European Shift2Rail research project, and ‘Closer Running’ is the title of an RSSB research project in the UK.

Written by Ian Mitchell, a member of the International Committee of the Institution of Railway Signal Engineers.


  1. Well to my mind the answer is in the second part of Q1 – “there will be significant operational challenges to make it useful in practice”
    It will work fine for a squadron of trains leaving Euston, Kings Cross or Paddington with the last train dropping out of the flight as they approach the relevant junction or in the case of Stevenage and Watford Junction stations. But where on the inward journey would you start to marshal trains for the dash into London. With the possibility of beginning the process at Doncaster (Leeds, Newcastle and Hull services) you could complete the process at Peterborough with a service which has served Retford, Newark etc. Strikes me there are too many variables and just one late train could cause a very large (delay) ripple.
    On HS2 you’d get a squadron formed up to depart for York, Preston, Newcastle, Edinburgh, Glasgow, Liverpool, Meadowhall, Leeds, Manchester and Curzon Street. Not so sure that Euston could actually hold (and board) this number of trains to depart at frequent intervals; besides they all need to stop at Old Oak so again platform occupation and re-occupation would be the key. It would indeed be an impressive sight but the logistics of getting this flight of trains assembled close to Curzon Street would be an interesting challenge with so much capacity wasted as to *actually* reduce overall journey times.

  2. CBTC or ERTMS/ETCS level 3 Signalling could be tested out on the Great Eastern Main Line as the picture shows a Class 360 train passing by which I believe it’s Kelvedon level crossing. But it could be tested on that line between London and Colchester and as far as Ipswich, Clacton and Southend Victoria.

  3. So, we move from fail safe signalling design principles to a reliance on the hope that the train behind can brake to a stand before hitting an unplanned obstruction (whatever form that takes and taking the multitude of variables affecting train the ability of a train to brake effectively into account) and it is justified it on the principle of luck that derailments on adjacent lines haven’t fouled the line at a time when a train is approaching on the other line? That’ll reassure the users I’m sure.

    • That principle seems to be how many car drivers drive much of the time, so whether they understand it or not, that is the decision they’ve already taken.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.