The players

First let's establish the players.

The Channel Tunnel infrastructure itself, and the shuttle trains carrying road vehicles across it, are operated by private company Eurotunnel. The high-speed trains are operated by an independent company, Eurostar, which is a joint subsidiary of French state railways SNCF, its Belgian counterpart SNCB, and a (changing) British partner. The latter also control the infrastructure other competitors have to use to reach the Channel Tunnel. The first of those is former German federal railways Deutsche Bahn (DB).

Channel Tunnel safety is regulated by the Channel Tunnel Intergovernmental Commission (IGC), which is staffed by the British and French governments. It was the IGC that launched an inquiry in 2009 into revising Channel Tunnel safety rules. The recent disputes also involve Germany-based rail industry giant Siemens, which will build the new trains for both high-speed competitors, and France-based rival Alstom, which lost against Siemens in Eurostar's tender.

The trains

When looking at how well trains fulfil safety rules, how strict those rules are, and how much sense they make; at least seven train types have to be considered:

• Truck Shuttles: 800 m Eurotunnel trains consisting of two locomotives at either end, a passenger car for truck drivers, and 33 semi-open wagons with grid-like sidewalls to carry trucks without their drivers.
• Passenger Shuttles: 800 m Eurotunnel trains consisting of two locomotives at either end, and in-between two 14-car sets of hermetically closed wagons to carry cars and buses. Passengers remain in their vehicles.
• Class 373 "Three Capitals" sets: these are the standard high-speed trains presently operated by Eurostar between London, Paris and Brussels. The 20-car, 394 m trains consist of two traction heads at both ends and two nine-car articulated sets in-between.
• Class 373 "North of London" sets: a 14-car, 320 m version originally planned to run direct services to northern England bypassing London, but never used as such.
• Deutsche Bahn Velaros: 8-car, 200 m high-speed trains with distributed traction which DB will operate in pairs on Frankfurt–London runs.
• Eurostar Velaros: 16-car, 400 m version of the former, ordered by Eurostar.
• Alstom's AGV: the articulated distributed-traction high-speed train offered in a 400 m version to Eurostar in Alstom's losing bid.

Now let's look at the individual safety principles.

Driving through

In the original safety concept of the Channel Tunnel, the default reaction to fire was driving the train out of the tunnel to fire-fighters at the portals, leading to the requirement that trains be fast enough to drive through in under 30 minutes, and for fireproof doors within the train able to contain the fire for the same amount of time. Quoting from the IGC Concession Agreement, the 1986 document setting out basic rules (download from here):

Shuttle rakes must be provided with intermediate fire doors or curtains with a fire resistance time of at least 30 minutes to prevent the spread of smoke along the train in the event of a fire occurring on board. As far as is practicable, the design of shuttle rakes should be such as to enable, in appropriate circumstances, a train to continue on its journey to clear the tunnel in the event of an outbreak of fire.

This principle is fulfilled by all but one of the present and future trains: Truck Shuttles have no separating doors. But apparently that wasn't seen as relevant, because there were no passengers on those cars who could have choked on the smoke.

However, this principle ignored something basic: winds feed fires, and the faster you go the stronger the relative winds. This is not an issue when the fire is on-board in a closed train, or if it is external but can't catch on or damage essential equipment. Relative wind is, however, a big issue when you have open cars, and successive trucks with flammable cargo in them.

Indeed both big Channel Tunnel fires involved Truck Shuttles, and in both cases, the burning trains had to stop in the tunnel, resulting in total burnouts and massive material damage. In spite of this, Eurotunnel was unwilling to invest in closed truck shuttle cars. Instead, they opted for the installation of fire suppression stations one-third into the tunnels. This is an improvement, shortening the distance to travel; however, it doesn't work in cases when there is a derailment, or some damage to the catenary (causing power loss).

Mobility with reduced power

For various reasons, long mountain rail tunnels are usually built with a slight grade from both portals until a summit within the tunnel. If a train loses power, normally it can still roll out by gravity. The Channel Tunnel is, however, a sub-sea tunnel, so trains must climb out with their own power, thus a special rule is justified.

This principle is fulfilled by all present and future trains. However, all the accidents showed that this principle is of limited applicability.

In the Channel Tunnel fires, once the trains stopped, the fire soon damaged the catenary, thus there was no power for any part of the train to leave on its own. In the 2009 snow chaos, all motors of five successive trains broke down at the same time, and diesel shunter locomotives were needed to pull out the trains. Eurotunnel is now buying more shunters for this task.

Separability

The original Channel Tunnel safety concept assumed that damage to trains serious enough to stop them mid-tunnel is typically localised. For such cases, the idea was to evacuate passengers from the damaged half of the train into the intact one, and decouple the latter to drive out of the tunnel.

This theory can, in principle, be implemented with all trains. However, while the past accidents lessened the applicability of the previous principle, they completely negated this one: in all cases when the option of half-train evacuation would have came handy, the entire train couldn't move any more.

Powerheads vs. distributed traction

As operating rules were originally drawn up, both of the previous two principles were implemented by prescribing two locomotives per train, one at both ends:

Except where operating rules otherwise permit, all trains shall be equipped with two locomotives, one situated at the head and the other at the rear of the train enabling the train to be split and to reverse direction. This arrangement may be modified for certain types of trains (particularly in the case of freight trains) and in certain operating conditions provided that the operating rules are followed.

If taken literally, this rule is fulfilled neither by the Deutsche Bahn and Eurotunnel Velaros, nor by Alstom's AGV: all three trains lack locomotives and have distributed traction instead (underfloor motors in multiple cars), and the end cars are unmotorised.

Was the two locos rule based on any safety advantage of trains with locomotives over trains with distributed traction (as claimed by Alstom lawyers and French government officials thinking no one will notice that Alstom offered a distributed traction vehicle)? No. It's just that in 1986, all high-speed trains that were in people's minds – the French TGV, Britain's own HST and then upcoming IC225, Germany's experimental ICE – had traction heads (permanently coupled power units with drivers' cab only at one end), while Shuttles and freight trains had locomotives. The emphasis is quite clearly on redundancy and bi-directionality (having two traction heads rather than one).

Distributed traction can, in theory, mean a difference in fire safety: fires connected to electrical equipment would be possible along the entire length of the train, rather than concentrated at passenger-free ends.

However, on one hand, this is not something untested (as claimed by Alstom lawyers). Trains with distributed traction routinely cross tunnels with more than half an hour travel time for decades now – in subway tunnels and commuter railway tunnels (f.e. Paris's RER). In Japan, all high-speed trains have distributed traction since 1964, and cross long tunnels, including the 18,713 m Shin-Kanmon Tunnel under the Kanmon Straits, and a succession of long tunnels altogether 102 km long on a 130 km section along the Joetsu Shinkansen.

On the other hand, consider what was flammable in traction equipment: carbon brushes in DC motors with commutators, and cooling oil for transformers and power electronics. However, Siemens's Velaros have modern contactless AC motors, and water-cooled transformers and converters.

Train length

A key element of EU-induced rail reform is the separation of infrastructure and train operations. Infrastructure managers are now obligated to prepare so-called "Network Statements", which contain basic information and rules for any train operator to access their network.

With its 2007 Network Statement, Eurotunnel (not the IGC) introduced a new rule (not present in the 2004 version):

The availability of emergency exits every 375m into the protected environment of the service tunnel is a main feature of the safety arrangements for occupants of the Channel Tunnel. One of the preconditions for efficient and safe evacuation of passengers in an emergency is to stop the train, and more specifically a coach carrying passengers or directly accessible by passengers, alongside an emergency exit. In order for this condition to be systematically achieved, irrespective of stopping conditions in particular, passenger trains are normally required to be at least 375m long (excluding power cars, unless passengers can easily evacuate from them) and passengers have to be able to pass from one end to the other. This base configuration provides the optimum conditions of safety if evacuation is necessary.

The 375 m rule persisted until the 2011 Network Statement, but was dropped in the 2012 version.

From the viewpoint of going practice, this was a rather bizarre rule: at present, only the Passenger Shuttles fulfil it! In Truck Shuttles, passengers – the truck drivers – are in one single car, and can't walk along the train (at least not safely) due to their trucks. Without traction heads, Eurostar's "Three Capitals" version of the Class 373 is only 348 m long, the "North of London" version much shorter. Of the Velaros, the Eurostar version would have fulfilled the 375 m rule.

There is a good rationale for such a rule: consider the situation that a train comes to an uncontrolled stop somewhere in the tunnel, and the tunnel is smoke-filled due to an external fire or some similar problem, ruling out evacuation to the emergency walkways on the tunnel sides. In that case, it would be best to evacuate the whole train across a single door, the door that came to a stop closest to an emergency exit. Note that the rule was shabby there: what counts is not the length of trains minus passenger cars, but the distance of extreme doors. (For the current Eurostar trains, that's 338 m.)

However, this consideration ignores another factor: evacuation time. When DB held an evacuation drill with an ICE3 train in the Channel Tunnel last October, it took 13 minutes to evacuate 300 volunteers across a single door. Eurostar's Class 373 "Three Capitals" sets have a capacity of 750 seats, its on-order Velaros are planned to seat more than 900. And the tunnel's one person wide gangways won't allow much faster speeds, either. Thus IMHO fast evacuation of a 400 m train in a tunnel towards a single emergency exit is just unrealistic.

Here it is again what I called the nasty little secret of new long tunnels: the spacing of cross-passages (which function as emergency exits). The 375 m spacing on the Channel Tunnel suits going EU regulations, and many recent tunnels have a spacing significantly longer than that. The best is 200 m, on the recently opened Perthus Tunnel on the Catalan/French border. I would make less than 200 m (which is also the new standard length of European high-speed trains) compulsory – for the Channel Tunnel, that would mean a doubling of emergency exits.

Staff preparedness

If you want to evacuate lots of people across a narrow cross-section, organisation and psychology become major factors. Train staff have to direct the flow of people, inform people and act in a way to prevent panic, they have to prevent logjams by helping elderly, young or disabled persons; and have to have the self-confidence that they can handle the situation, so that they don't panic themselves. In this field, both Eurotunnel and Eurostar showed deficiencies. As unrealistic its evacuation drill was, DB at least held one. (Then again, the behaviour of DB's staff during last year's air conditioning problems was deficient, too.)

A particular field of staff action which could be regulated explicitly is the direction of the flow of people in the case evacuation is done using multiple doors, to prevent dangerous logjams forming where multiple lines of evacuees meet. By default, drivers of 400 m trains could be advised to try to get the front of the train stop just at an emergency exit, so that evacuation at both ends is possible. Evacuation at both ends should be what the staff organises in the case there is simultaneous internal and external damage blocking the path of passengers (for example bomb damage). In case there is no such damage but there was an uncontrolled stop with the emergency exit somewhere in the middle, in 400 m trainsets, it would be best to open the next door ahead of the exit for passengers in the front, and the next door behind the exit for passengers in the rear. In the case of two coupled 200 m trainsets, the train next to the emergency door should be evacuated across the door next to the exit in the opposite direction from the other trainset.

:: :: :: :: ::

Check the Train Blogging index page for a (hopefully) complete list of ET diaries and stories related to railways and trains.