Signals are Key

Dutch Infrastructure operator ProRail has been looking at Japanese railway practices for some time now, in an effort to understand the „secrets” behind their efficiency and see how these could be applied at home. In fact they had their „man in Japan” spend a year there and posted regular reports on his findings on their intranet. In a recent presentation to the IRSE Dutch section, Vincent Weeda (ProRail) and David Koopman (DHV) explained a concept they call „Kort Volgen” which literally translates as „rapid following” and is all about shorter headways and less complicated lay-outs. Something that should be very familiar to those of us involved with rapid transit systems.

To start it off, some observations about Japanese performance were given:

On many (sub-)urban lines a slow train and an intercity service, both at 10 minute intervals is operated

In and between cities large numbers of passengers are transported

97% of trains operate at less than 5 minute delay

In busy spots a train stops at a station every 2 minutes

Signals passed at danger are extremely rare

In most railways fares have not been raised since 1987

Railway companies operate both infrastructure and trains and are profitable

No wonder ProRail, which has enjoyed less than favourable reviews of its performance by both passengers and parliament off late, are asking themselves „How do they do it?”. The answer it turns out is surprisingly simple and has nothing to do with our traditional perception of white-gloved drivers, miraculous quality management practices and rigid discipline. The secret formula is keeping things simple and focussing on the essentials. Not necessarily as an integrated railway, but as a business. If you want to make more money, you operate more trains and transport more passengers. That in turn requires more capacity and an unobstructed flow of traffic.

1. One of ProRail’s first observations was that „Signals are Key to Everything”, always something that goes down well with an IRSE audience.

2. Optimum traffic flow requires signals to be spaced at short distances, if needed 100 meters apart, so every train releases its route for a next one quickly.

3. A smooth flow of traffic even when trains bunch up, and speedy recovery of perturbations also requires short signal spacing.

4. In designing a line, signal placing should be as restriction free as possible.

5. Speed is essential, but avoiding slower speeds is more important than high maximum speeds.

6. Short dwell times are essential, as are rapid turnaround times at terminal stations.

Signal Placement

That’s really all there is to it. Of course the Japanese also have their stabling yards, depots and workshops and the tracks and point work to get to and from them. But first and foremost, the principle is to keep it simple and concentrate on the essentials. All frills cost money, require maintenance and are a source of disturbances. „If it’s not there, it can’t break”. And therein lies the lesson for our railways it would appear. In contrast, Vincent points out that Dutch requirement specs for a layout read like long wish lists with endless restrictions. Some of the examples quoted are:

Additional point work for various scenarios to handle disturbances.

Additional crossovers to cater for possible future timetables.

Bi-directional working on all tracks, even when four tracks are available, to overtake defective or slow trains.

Braking distances based on worst-case performance of each and every train (as designed decades ago).

Maintaining existing functionality.

Future proofing of everything.

Of course all these requirements have their backgrounds and historic justifications. But together they result in overly complex systems, the consequences of which are seldom analysed. In short it leads to a very expensive infrastructure, which paradoxically is hardly able to deliver the required performance anymore.

So how does this affect signalling? The engineers sent on the study trips to Japan noted that the Japanese had no qualms about placing a signal almost everywhere, including places where Dutch engineering rules would never allow this to be done. Examples are directly in rear of the platform, just in advance of a level crossing, after a set of points etc. Secondly the layout appeared much less complicated, with almost no bidirectional working, much less points and crossovers and much simpler station throats.

When they discussed our engineering rules with their hosts, the simple question returned was „but how can you possibly position a signal anywhere then?”

In adopting the Japanese principles, the first rule of thumb was “just position all the signals you need to optimize traffic first, and let the other technical systems follow from there” (as opposed to today where e.g. an overhead sectioning or level crossing can prevent optimal signal spacing). And „don’t worry about trains stopping at inconvenient locations too much. Just make sure they can move on quickly instead”.

Short signal spacing in the direction of travel, is another example. In Holland, bidirectional working is required virtually everywhere. But what does it really get us, is the question Vincent asks. In disturbed situations, following an accident, more often than not, both tracks are blocked. If we want to drive past a stranded train, we disturb the traffic in the other direction and defective trains can (or should be able to) be removed fairly quickly anyway. It is an eye-opener to see how much simpler the layout becomes if we abandon bi-directionality. Signal placement is simplified dramatically, control systems and algorithms become simpler, drivers and signalmen have better overview and situational awareness etc.

A typical Japanese layout has much less points than ours. Points can give flexibility in disturbed situations, although „ only to a degree” as pointed out in the discussion of the benefits of bidirectional working. But this flexibility comes at a price. Points eat capacity, as they limit signal placement options and usually introduce speed restrictions. Not just when traversed diverging, but also due to induced cant limitations, standardization of point types leading to requirements to apply only in straight sections of line etc. They impose restrictions on timetables, as they have to be traversed and require de-rusting movements to be made regularly, offer the possibility for misrouting, require inspections and scheduled maintenance. They are an inherent safety risk, subject to wear and tear themselves but also cause wear to rolling stock etc. And in the Japanese philosophy one should avoid the need for, rather than employ the option of rerouting.

In employing these different „system philosophies”, the potential for simplification and cost saving appears to be enormous. Compare Tokyo and Utrecht Central stations. Both have a similar number of platforms and a comparable role in the network. Tokyo Central has 28 points, Utrecht Central has 280! Both stations have uncoupled the main traffic streams to eliminate crossing moves as much as possible. At present Utrecht handles 60 train movements per hour, Tokyo handles 180.

In Holland most stations have a generic speed limitation of 40 km/h within the station limits. This is due to the nature of the Dutch ATP, but also to limit the number of point type and space taken up by them in the complex station throats. If we could raise the lower speeds on the network first, or instead of concentrating on higher speeds, that would optimise traffic flow as well as benefit both slow trains and InterCitys alike. And it avoids the capacity penalty usually incurred when the speed differential between train types is increased.

Early traffic simulations by the study group showed that this concept has promise and now the studies are being extended to the SAAL (Schiphol-Amsterdam-Almere-Lelystad) corridor upgrade project.

Simulation results

As an example, one simulation showed that the simple addition of a signal directly in rear of a platform on a station on the open line and leaving out the cross-overs traditionally found in such locations, still respecting all other design constraints like braking tables etc. would increase theoretical line capacity by some 29%. Most readers will note that this is a very common practice in rapid transit systems as it allows train to ease up to the platform and the train they are following is departing, usually referred to as platform re-occupation time. Further studies taking into account optimised braking tables etc. show even greater promise but are likely to require some more thought on the implications for safety cases when combined with the limitations of existing ATP systems. This of course led to a debate on the benefits ERTMS level 2 might deliver in such system concepts.

Any signal engineer worth his salt will undoubtedly be able to point out many potential issues with the „kort volgen” concept and the Dutch section audience was no exception. „How many Yellow signals will a driver encounter in rear of a red signal”? „Where do we get the additional ATP codes”, etc. But isn’t the real challenge to our profession to investigate under which conditions a number of these benefits can be had, as they apparently are in Japan? And why not examine how we can pick the „low hanging fruit” quickly instead of waiting for ERTMS to be rolled out as an enabling technology. And „adopting 1954 engineering rules to ERTMS design has already been shown to deliver just about no capacity benefit at all”, as one of the members in the audience remarked.

We are perhaps left with the question „can it really be that simple”. And if it is, why hasn’t it been tried before. And the obvious answer might be that it has, and successfully as well. In rapid transit and urban railway settings. So the real issue might be to adopt these principles on mixed traffic lines and break out of the heavy rail design paradigm. It will be interesting to see how this develops and observe a possible application to the SAAL project.

Level 3 from High Speed Vision to Rural Implementation

Written and edited by Wim Coenraad on behalf of the International Technical Committee of the Institution of Railway Signal Engineers


This article is part of a series of articles by the IRSE-ITC on the subject of “what is preventing ERTMS Level 3 from entering into service?” and was published in IRSE News 173.

ERTMS/ETCS (hereafter called “ETCS”) is a train control system designed to replace all existing national systems on the Trans European Rail Network.  It enables trains equipped with on-board units from different suppliers to operate freely over track equipped by the same/different suppliers.  It consists of both on-board and trackside subsystems.  ERTMS is specified to allow implementation in three functional and performance levels.  The levels range from a “conventional” but interoperable ATP system using spot transmission (Level 1), a level adding continuous communication via radio, allowing to increase line performance and the elimination of wayside signals (Level 2) and a level replacing wayside train detection with on-board localisation, allowing the elimination of much of the trackside train detection equipment (Level 3).

Whilst Levels 1 and 2 have been developed and are in operation, for main line railways, Level 3 exists only on the drawing board.  For main line railways, Level 3 promises significant benefits for the Infrastructure owners, but adds extra systems and complexity to the train-borne equipment.  On top of that some difficult technical issues remain unsolved, limiting the potential field of application.

A trial of ERTMS Regional (a specific implementation of Level 3 for regional lines) is being implemented on the Västerdalsbanan (Repbäcken – Malung) in Sweden and is about to enter into service, extending ERTMS applications to Regional Lines.   This system is described in more detail in the November issue of IRSE NEWS.


ERTMS Level 3 is characterised by the fact that a train in Level 3 determines its own location, using position references transmitted by fixed Eurobalises and its on-board odometry.  It transmits this location data to the Radio Block Centre, which issues movement authorities to the trains under its control.


Railway Undertakings

For a railway undertaking or infrastructure operator, Level 3 is attractive because it dispenses with the requirement for trackside train detection equipment.  This not only saves investment and Maintenance costs, but also reduces the exposure of staff to working in or near the tracks, which reduces Health and Safety risks.

Of course this can only be accomplished if the robustness of the systems and the maturity of the ERTMS implementations allows deployment of Level 3 without the use of fallback systems.

Train Operators

Train operators are unlikely to see direct benefits of investing in Level 3 train-borne equipment rather than Level 2.  In fact a solution to the issue of train integrity proving (refer to the article in this series in IRSE News of September 2011) is likely to require additional equipment to be installed on some, mostly locomotive hauled, trains.

An indirect benefit of Level 3 might be the availability of additional train paths and/or reduced cost of such paths, if Infrastructure access charges were to become subject to the laws of supply and demand.

On the other hand operators are likely to have further requirements for functions to be delivered by future train-control systems that are not within the scope of the present ERTMS product set.  Such requirements might include train service regulation (conflict and delay detection and resolution), energy optimisation driving support, Automatic Train Operation.


Recently, virtually every railway around the world has been confronted a dramatic rise in the occurrence of copper and cable theft.  In some cases this has led to a wrong-side failure and where return conductors are removed, electrocution is a possible hazard as well.  It might be argued that ETCS Level 3 mitigates these risks, to an extent (on an electrified railway obviously return conductors will still be required).  On the other hand, as a centralised, communications base IT system, ETCS Level 3 will inherently require more protection against cyber attacks, jamming etc.  These IT related hazards have traditionally not been taken into account in the design of railway signalling systems and it remains to be seen whether or not the level of protection offered by the Euroradio links over GSM-R and the networked Radio Block Centres will prove to be adequate.


Most protagonists of Level 3 claim capacity benefits over and above those delivered by Level 2 of up to 10%-20%.  However, some doubt this, as Level 2 when implemented as a high performance block system, possibly using virtual intermediate block signals to improve train spacing, seems to be able to deliver the same performance levels, although at much higher equipment cost.  In addition these studies usually compare theoretical headway calculations in isolation, where in reality network capacity is determined by many other factors such as station layout, and platform capacity, timetabling constraints dictated by fleet and commercial constraints etc.  It might be significant that Computer Based Train Control systems delivering traffic densities of around 24 trains/hour on a given track are usually found only in rapid transit applications.  They benefit from uniform train performance characteristics and uniform timetables/diagramming in those types of railways.


Some studies estimate the capital savings on infrastructure equipment for ETCS Level 3 to be in the order of 25% as compared to Level 2 and even up to 50% or 60% when compared with a multi aspect signal block system (Transport Research Laboratory report Chapter 4.3, based on Network Rail’s Signalling Equivalent Unit cost model).  This of course would be offset by a cost increase in the required communication systems on board, train integrity proving systems etc.


Once the closed loop control system that is inherent to Level 3 is in place, other options emerge, such as tighter network control, conflict detection and resolution; energy optimised driving leading to “even greener” mobility.  However none of these functionalities are part of the ERTMS/ETCS specifications (in fact the “M” in ERTMS seems to have been “lost in space” during the development), increasing the risk of non-interoperable bespoke implementations emerging.

Engineering Flexibility

The design and engineering for pure ETCS Level 3 lines does not need to take into consideration all aspects of signal sighting, block length based on train / train category braking characteristics etc. and should therefore be more cost effective and flexible, e.g. with regard to changes in timetabling, train fleets etc.  On the other hand the pre-requisite of course is that all trains are equipped and/or special provision be made for the movement of non-equipped trains such as maintenance vehicles.


Whilst on the one hand and at present, the requirement to fit all vehicles with an EVC capable of ETCS Level 3 functionality seems to be an obstacle and cost disbenefit.  On the other hand, vehicle route acceptance will no longer be complicated by complex Electro Magnetic Comptability and shunting compatibility issues, or other detection system induced restrictions, shunt assistors and / or complex traction current filters would no longer be needed.  Of course in practical terms, such benefits can be reaped much earlier and realistically in “closed environments” such as mass transit railways and may never be obtainable on a larger network where the network wide roll-out may not likely to be achievable.


In ETCS Level 3 trains become virtual self locating and reporting objects and this property can be used to extend the same level of protection from “conflicting movements” to entities such as track gangs, maintenance plant etc.  This makes possession management simpler and inherently safer, examples of this have been implemented on the Betuweline using Hand Held Terminals.  In fact this feature is applicable to Level 2 as well, because the only requirement is to have a centralised Movement Authority issuing system, i.e. the Radio Block Centre (RBC) and can in fact be overlaid on all signalling systems that employ an “interlocking machine” capable of communicating with an RBC.


An inherent benefit of a Level 3 system would be enhanced resilience and faster recovery from disturbed situations, as trains can move closer together as they approach a pinch point in the network, or a location where e.g. a points failure exists.  Likewise resuming speed and recovering normal operating conditions could benefit from this “harmonica effect”.  Of course the absence of many track-circuits at least in non-point areas will itself lead to fewer equipment failures and delay minutes.  This may on the other hand be offset by a higher dependence on communication systems and the potential for very dramatic and disruptive failures if the communications were lost entirely.  Providing redundancy and fallback systems to address these issues will of course detract from the predicted cost savings.


Train Integrity Proving

The issue of train Integrity proving does not have a generic solution yet, making Level 3 an option available only in such cases where a solution exists and/or the risk would be deemed acceptable (trainsets, fixed formation rakes, rural lines).  For trainsets, loco-hauled passenger trains and fixed formation trains a practicable solution seems feasible, but for freight trains made up of individual cars, a solution that is affordable, practical and logistically manageable does not exist.

Radio Bearer Service

The near total dependence on radio if Level 3 is ever to become a reality must raise questions re the capacity of radio networks to cope with the volume of data to be exchanged on a typical busy main line or suburban area.  It is doubtful whether GSM-R would be able to handle the traffic and therefore Level 3 may be dependent on having a new radio standard in place.  The so called LTE (Long Term Evolution) system as advocated at the Indian Convention thus missing out the 3G standard, may become a pre-requisite for Level 3


ERTMS Regional is a development of Trafficverket, the Swedish infrastructure operator, Bombardier and the UIC.  It is aimed at providing a signalling and control system for (very) lightly used regional lines, which account for 21% (2116 km) of the Swedish network and builds on the principles employed in an earlier version of the Swedish Radio block system, based on Sweden’s ATC2 technology, in operation since 1995.  Its aim is to reduce the cost of equipping the line with a signalling system by 50%, which we were told is achieved, albeit on the assumption that the cost of providing GSM-R coverage had already been absorbed by the national requirement to provide full GSM-R coverage on the network.  Interestingly, the system is based on an open architecture with the interface specifications owned by Traffickverket.

In short, ERTMS regional is based on a combined interlocking and RBC processor which sets routes and issues command to local trackside elements which are connected to object controllers.  These object controllers communicate with the IXL/RBC computer either over GSM-R (using General Packet Radio Service), cable or in the trial even an ADSL (Asymmetrical Digital Subscriber Line) internet connection.  Interestingly the ERTMS part of the specification is unchanged ERTMS Level 3 with fixed blocks, based on the Class 1 specification (baseline 2.3.0d), so trains equipped for running on ERTMS regional lines have no problem continuing onto the main line, or into main stations equipped to “standard” ERTMS Level 1 or Level 2.

Since the lines are so lightly used (typically in the pilot line 8 passenger trains and 8 freight trains per day) and the passenger trains are single car DMUs anyway, the risk of train separation is accepted, so the unsolved problem of train integrity proving is left for another day.

Level crossings at stations are controlled through the object controllers. On the open line autonomous level crossing installations can be left as is, but there is an option to control them using the functions to be integrated in ERTMS baseline 3.  In short a brake curve is established to any level crossing controlled by ERTMS, the strike-in point is marked by a balise, the train reports being in the strike-in zone and waits for confirmation of the level crossing’s closure before extending the movement authority beyond it.

The ITC was pleased to be taken on a demonstration run of the system between Repbäcken and Mosbjäck and observe its functioning first hand.  In fact this turned out to be the final night of formal testing in shadow mode, before a commissioning decision was to be taken.


Whether or not Level 3 really delivers benefits other than savings on train detection and Occupational Health & Safety hazard reduction remains to be seen.  The proof of the pudding is in the eating, but for main line railways, the chef is still hiding in the kitchen.  Early implementations such as ERTMS regional certainly provide an interesting appetiser!

© Wim Coenraad 2019