ERTMS remains the great hope for train control in the future, but costs must come down. Richard Hope talked to Clive Kessell, who believes that current trends which compromise true interoperability must be challenged if the real benefits are to be achieved


Clive Kessell BSc CEng FIEE FIRSE is a member of the IRSE Technical Committee, Chairman of the Railway Engineers Forum, and a member of the IEE Railway Professional Network Executive

AFTER MORE than a decade of development, the technology required to support the European Rail Traffic Management System is largely available today. Railways within Europe are signing up to implement ERTMS on prestige high speed lines, and will do so on other routes as signalling renewals fall due.

Doubts remain over reliability, the very high cost of retrofitting rolling stock, and the ability of GSM-R to handle continuous communication with trains in high-density areas such as the approaches to London termini. However, Kessell says 'the UK seems willing to commit to ERTMS because of the benefits it will bring in the longer term'.

ERTMS embraces three components: the European Train Control System, which at Level 2 can replace lineside signalling and provides full automatic train protection including speed supervision; GSM-R for voice and data communication by radio; and the European Traffic Management Layer - the dispatching or regulating function.

ETML is intended to provide automatic train routing and take intelligent regulation decisions, such as optimising the movements through busy junctions. It has made little progress, and in fact is little more than a concept. Debate continues as to whether this level of activity would benefit from a harmonised European solution anyway.

GSM-R performing well

In contrast, GSM-R is in full production and is being rolled out across several countries, notably Germany and Sweden. Nortel and Siemens are competing to meet this demand, while four or five suppliers are producing handsets.

In the UK, Network Rail is committed to replacing old analogue radio systems with GSM-R by 2013. This is being done in parallel with the installation of a new optic fibre communications network that will, among many other tasks, link GSM-R base stations to the Radio Block Centres required for ETCS. With the necessary expenditure already agreed by the Office of Rail Regulation, this is effectively a sunk cost.

Most European countries have allocated the required spectrum of frequencies in the 800 to 900MHz band, and GSM-R is being used for a range of ancillary functions, such as communication with trackside workers and yard shunters.

There is one caveat. So long as GSM-R is only used for voice communication the base station spacing can be 10 to 12 km, and this is what is being provided to serve most British lines. When used as a data bearer for near-continuous contact with trains, however, a stronger radio signal is required, reducing the spacing to 5 to 8 km. This closer spacing will be provided on the UK's designated TEN routes from the outset. Radiating cables and local aerials ensure continuous coverage, especially in tunnels.

Even so, in very busy areas ETCS Level 2 requires more channels than are available from GSM-R. The continental railways appear to accept the concept that ERTMS will not be applied in such areas. Conventional lineside signalling will be used instead.

Current UK practice is to take ATP right to the buffer stops, as is done with the Train Protection & Warning System. The intention is to do the same with ETCS. Kessell says this is where the industry will get into trouble, because in several countries ETCS is also being permitted to have local rules built in. 'Really and truly we are making huge trouble for ourselves, because interoperability looks as though it's going out of the window.'

The intended solution in the UK to this perceived deficiency of GSM-R is to adopt the General Packet Radio Service in congested areas, but this is taking longer than anticipated to be defined and specified because the potential use is so limited. One enterprising manufacturer has demonstrated that GPRS is feasible for rail use on the metre gauge line between Nice and Digne, using a public GSM network.

Slow progress with Level 2

So far as the UK is concerned, TPWS has proved very successful in preventing collisions caused by passing signals at danger since national roll-out was completed in December 2003. This effectively rules out investment in ETCS Level 1, which bolts on ATP to existing lineside signals.

The focus is firmly on Level 2 without lineside signals, and a programme exists that would see conventional signalling (including renewal schemes that have not yet started) progressively replaced from 2011 to 2038 (RG 8.05 p472). Preparations are in hand to award contracts this year for the Early Deployment Scheme to trial ETCS Level 2 on the single-track Cambrian lines in mid-Wales, expected to cost £60m.

The principal criticism of the Cambrian EDS is that trains at 2h intervals, a few passing loops, and one junction consisting literally of a single set of points will fail to test the ability of Level 2 to cope with a busy multi-track line carrying mixed traffic. This should make a second trial essential, but there is no agreement as yet on where it should be.

There is another weakness of the Cambrian trial: only a small fleet of just over a dozen Class 158 DMUs will be retrofitted with ETCS, plus three or four diesel locomotives for engineering work on the line. The biggest concern in the UK, as elsewhere, is the high cost of retrofitting existing trains. 'At least the Cambrian EDS should give us some measure as to what the train fitting costs are, and how difficult it is', says Kessell.

The latest available estimates were published in 2004 by Unife, representing the supply industry (Table I). These put the cost of the equipment required for one cab at £40000, and if this is installed during manufacture of a new train the total cost per cab becomes £70000. Retrofitting estimates, however, range up to a total cost of £220000 per cab - and even this assumes that the necessary space in the cab and under the floor actually exists.

Unife says the cost of fixed equipment for ERTMS ranges from £100000 to £200000 per route-km, depending on the number of tracks and how busy they are. This is expected to be less than renewal of lineside signals and track circuits, and the cost of maintaining lineside equipment should be lower too.

Risks for the UK

Kessell identifies two remaining risks peculiar to the UK. Firstly, compared to other European countries Britain is considered by many engineers to have an over-zealous safety regime. 'Installing ERTMS and getting it through a safety case could prove to be quite a challenge'.

Fortunately, passive balises are the only hardware to be installed on the track. The West Coast Main Line has been equipped with Eurobalises to carry the TASS data messages which regulate the tilting and permitted speed for Virgin's Pendolino and Super Voyager trains.

Balise installation methodology, the installation safety cases and safety data management regimes have all been prepared and approved, and the work was incorporated into possession windows for routine track work. Cross-acceptance of work elsewhere in Europe should also help to reduce the cost of introducing new ERTMS equipment.

The second risk is reflected in concerns as to whether all train operators and rolling stock owners would willingly participate in fitting their trains, especially those that would only be a minority user of an equipped route. Fortunately, the Association of Train Operating Companies has taken a positive lead to achieve co-operation, and is now a leading voice for ERTMS co-ordination in the UK.

The regime for cost allocation is already in place through regulatory network change procedures. A pragmatic approach to train fitting would see all new trains designed and wired for ETCS as part of a national fitment strategy, regardless of whether or not the actual equipment is installed at that stage.

Obsolescence threat

A risk common to all countries is the long implementation period and the projected operating life of ERTMS. Technology moves on at a daunting pace, and the same components will not exist in 10 years time. Thus future ETCS and GSM-R hardware and software builds will have to be backwards-compatible with what is being designed now.

ERTMS will inevitably evolve, and network upgrades will be required from time to time. The big challenge will be for train equipment to be able to read variations in infrastructure equipment.

However, the current specifications - developed by manufacturers in line with EU recommendations - only require the hardware to be capable of supporting two versions of software simultaneously, with the expectation that new versions would only be introduced every threeyears. Given the typical minimum life for railway equipment of 30 years, and the requirement for continuous interoperability, this issue needs to be addressed with some urgency.

Kessell is still convinced that ERTMS remains the great hope for train control of the future. Much has gone into specifying and designing it, and more work has still to be done. But it is expensive, and the costs will have to come down. He questions the level of interoperability being achieved, and says problems here must be overcome.

Roll-out across Europe seems destined to happen, but UK deployment may be at the tail end of the programme. Kessell believes this may be no bad thing. The goals of eliminating trackside signals and much of the lineside cabling are there to be had. But only then can the promised downturn in train control costs be realised.

Table I:Unife ERTMS/ETCS price indicators (€000 at 2004 prices)

Onboard equipment (per trainset) Level 1 Level 2

On new rolling stock 92 125

On pre-fitted rolling stock 105 135

Retrofitting where space is easily available 175 215

Retrofitting where space is at a premium 270 320

Infrastructure equipment (per route-km) Level 1 Level 2

New line - low density of traffic 45 160

New line - high density of traffic 80 210

Existing line - low density of traffic 82 200

Existing line - high density of traffic 85 260