France - railways, public transport, services and incidents (merged posts)

[whole diary]

[Lee]

stuving - I see your high speed rail riposte, and raise you...a bunch of driverless trains!

From DW:

France's SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) rail network plans driverless trains by 2023

Germany's Bosch engineering company is part of the consortia developing driverless freight and passenger trains over the next five years. The project is part of the modernizing drive for the French state railway.

French state railway concern SNCF announced the project for mainline, long-distance trains which is to be financed by the SNCF, the French state and commercial partners. French cities, including Paris, already have driverless metro trains.

The €57 million ($66 million) first phase of the development is divided into two consortia:
A prototype for driverless freight trains which includes the commercial Alstom, Altran, Hitachi-Ansaldo and Apsys (Airbus) companies
A passenger train prototype for regional TER routes which will be developed in partnership with Bombardier, Bosch, Spirops and Thales.

SNCF said it is in discussions with Deutsche Bahn to promote a European standard for driverless trains.

Director of the autonomous train project Luc Laroche said: "These human and technological issues are considerable ones. We are mobilizing the collective intelligence, competence and savoir-faire of our partners from the varied world of industry."

"These partners constitute a world first for ambition, approach and combined skills," Laroche said.

Read more: French Senate approves SNCF overhaul, cementing Emmanuel Macron's rail reforms

President of Bosch France, Heiko Carrie said "As a provider of leading mobility solutions, the Bosch Group is pleased to be part of this consortium and to be able to contribute, thanks to its technology, to this important new step towards autonomous trains."

Major changes for SNCF

In its statement, the SNCF said the benefits of autonomized trains for passengers and for freight clients would be greater capacity, better circulation of trains and improved timekeeping. It also said the trains would use less energy and therefore have ecological benefits.

"The digital transformation of the network and signal system will make it possible for more and better trains to circulate," SNCF Network President Patrick Jeantet said. "It is essential to develop autonomous trains. It will make us a champion of digital industry."

SNCF has 17,000 trains and transports 4 million passengers every day. It is currently running an annual deficit of €3 billion and attempts to change working practices have been met with a series of strikes by many of its 146,800 employees.

Earlier this year, Finance Minister Bruno le Maire said: "For 30 years we have shied away from making the necessary transformation of SNCF and for 30 years we've seen the service deteriorate... We can't go on like this. We're going into the wall." The current annual government subsidy to SNCF is €14 billion.

The plan is to roll out the first semi-autonomous trains by 2020 and completely-autonomous trains by 2023.

Or if you prefer, The Guardian:

France’s national railway operator SNCF has announced plans to introduce prototypes of driverless mainline trains for passengers and freight by 2023.

“With autonomous trains, all the trains will run in a harmonized way and at the same speed,” SNCF chairman Guillaume Pepy said. “The train system will become more fluid.”

The operator hopes the switch will allow it to run more trains on France’s busiest main lines, and cut energy consumption.

Many French cities, including Paris, already run driverless metro trains but driverless long-distance travel presents a new set of challenges, Pepy said.

“Railways are an open system, and the unexpected is the rule,” he said.

SNCF will be partnering up with rolling stock specialists Alstom and Bombardier who will be heading up consortia for freight and passenger traffic, respectively.

Pierre Izard, who runs SNCF’s rail technologies division, said the shift to driverless trains was to happen in stages, “up to the most extreme of automatisation, when there is no human presence onboard”.

Pepy said autonomous trains were “clearly the future”, but he added it may take time before passengers accept boarding them.

Although Australia, China and Japan are already experimenting with driverless trains, France is not coming too late to the game, said Carole Desnost, head of innovation at SNCF.

The French rail operator said it was talking to German operator Deutsche Bahn about promoting a European standard for driverless trains.

[TaplowGreen]

It's not a new thing in the UK^▸ (United Kingdom). GWR^▸ (Great Western Railway) have scores of driverless trains, mostly on Sundays.

[stuving]

Le Parisien is reporting what's in a report (expertises techniques) prepared for the juge d’instruction, which is a leak not a publication. As we perhaps expected, the evidence all points to the barrier being down in time but coach driver having braked too late to stop before it.

The report is really a collection of separate experts' analyses of the evidence, including witness statements, but the one called "automobile" uses the full detail of the bus's tachograph. That shows that "after her delicate manoeuvre onto the D612, and having accelerated to 12 km/hr, the driver became aware of an obstacle in front of her and stamped on the brake pedal. The only possible obstruction at that point is the barrier of crossing PN25. By the time she had decided to brake, the coach was already too close to the closed  barrier to stop before it, and it went onto the railway line bending the barrier out of the way." (Remember, it is only 17 m from that junction to the crossing.)

The main subject of comments is now the medicines the driver was taking, in particular hypnotics. (No English report so far.)

[stuving]

Le Parisen reported this morning that one of the three SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) (then RFF) employees implicated in the case, and previously "witnesses under caution", has been mis en examen - charged with a criminal offence, roughly. The prosecutors' office at Evry has confirmed only that step about an unnamed SNCF employee.

The newspaper report says more, that it was the local track maintenance manager (the other two being his direct reports). He did the last inspection before the accident, and the independent engineering reports said (based largely on the state of the fracture surfaces) that the relevant trackwork damage was present at that time.

[stuving]

Today's not quite a peak travel holiday - that was last Saturday and next too (for winter sports) - but there was a demonstration from SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) of what happens when an overhead wire goes ping on your busiest high-speed line (Paris-Lyon). The answer, of course, is that having cancelled as many trains as possible, the rest (enough for the number of passengers) have to squeeze onto the surrounding "classic" network. This is 1500V DC^▸ (Direct Current) territory, but being on half power isn't likely to matter given the lower line speeds anyway. Typical delays on these trains, between Paris and all points south of Lyon, have been 3-4 hours.

Repair work started at 13:00, but rather than keep both tracks closed for a full repair they chose to reopen the intact one for alternating one-way use and will rewire the other one tonight. Obviously the longer-distance services thin out from late afternoon, helping with recovery, but arrivals in Paris are still very delayed. SNCF are warning of further disruption tomorrow, presumably due to out-of-place stock and out-of-hours staff.

I guess much the same would happen anywhere.

[stuving]

The canicules are out early this year, so ... there are delays due to precautionary measures for hot rails and the like. SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) also have a problem with droopy catenaries, which is a bit odd - I've not heard of it as a significant issue here. It arises with gravity tensioners, where the weight hits the ground and tension is lost. there has been a programme of modifications; presumably one more turn of wire round the pulleys and a bigger weight. But obviously there are still some places to do.

So, about Thursday night. An intercités train Paris-Bercy to Clermont-Ferrand was delayed for more than two hours by an attempted suicide, and finally set off at around 9 pm. At 22:00, near Montargis, it stopped for want of juice, and obviously the chillers went off. Initial reports from passengers said they were left alone with no food, water, or other help for hours; when rescued by another train it stopped everywhere and then finally arrived the next morning 11 hours late. Reports of the trouble said the wire broke and that it melted, and it wasn't clear if the heat was a factor. The rescue train was delayed because the other track was closed for works, and had to be de-possessed.

SNCF's story was different, saying that water and food were supplied by the local rescue services and the stranded were picked up as soon as possible. Today, as part of an apology, they add that driver error at a point of voltage change, combined with slack wires, was the cause. I'm not sure that's a whole lot clearer. They also say they have stopped selling TGV^▸ (Train a Grande Vitesse) and Intercités tickets to allow for train cancellations.

There is a voltage boundary near Conflans-sur-Loing, a few km south of Montargis. Oddly, it's from 1500V DC^▸ (Direct Current) to 25 kV AC (which is newer OLE^▸ (Overhead Line Equipment, more often "OHLE")) when going south.

[IndustryInsider]

Thanks, Stuving.  As usual we would probably not of heard about this latest incident on the foreign railway network had you not alerted us to it.

[stuving]

Last night there was even an overnight TGV^▸ (Train a Grande Vitesse): Paris to Grenoble, 11 hours late. It spent the night at Lyon-Saint-Exupéry (so it does have a use!), and even then couldn't use the direct route. While that's due to reopen today, the Chambéry-Grenoble line (and other minor ones) will take several days to clear.

All due to trees felled by last night's storm, which was apparently quite spectacular - there was even a near-shipwreck on Lake Annecy. So most people have been have been understanding of SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways)'s problems, with the obvious exception of the passengers involved.

[stuving]

The BEA-TT report into the Millas accident was released on 17th May. I'm surprised to have missed it then, but looking back it got surprisingly little mainstream media coverage, for such a high-profile incident. It is the usual highly detailed, technical, dispassionate, document providing an interpretation of the events as close  to the truth as is possible. maybe that's just not what news is made of.

My summary of the report has got a bit long - it'll be the next post - but these are the key points from the report's introduction:

• the crossing worked correctly
• the direct cause was that the bus didn't stop despite the barrier, lights, and bell
• the most likely explanation is that the driver just did not see or notice the warnings

The factors found to have played a role include:

• the side road the bus arrived along was very close to the crossing
• the visibility of the warning lights from that side road was poor
• the bell stopped once the barrier was down, so wasn't a back-up to the lights
• the turn into the main road was very demanding of the driver's attention
• these factors were not included in risk assessments of the crossing

[stuving]

More on that BEA-TT report:

There were always just two possible narratives: either the school bus (really a coach) drove through the barrier after it came down, or the crossing wasn't closed when it should have been. The driver's conviction that the crossing was open, combined with the (sometimes noisy) support from her neighbours, never looked like good evidence given how traumatic her experience had been. The driver and passenger in the light van waiting at the barrier on the other carriageway had a much clearer view and said all along that the crossing close sequence happened just as it should and the coach did drive through the barrier. So it's no surprise that the BEA-TT adopt that as highly likely to be the case.

They dismiss the idea (which had so excited the press previously) that being a long-term user of sleeping pills affected the driver's performance; reporting that only normal therapeutic traces were found, of medicines for which no warnings about driving were required. But they did need to explain how she could have driven up to the closed crossing and through the barrier without noticing. As always, it's a story of many small contributory factors. The two tachographs both recorded the impact, so the events could be reconstructed working backwards. There was no CCTV^▸ (Closed Circuit Tele Vision) at the crossing or in the train's cab, and fitting more of these is one of the recommendations.

The barrier was down when the coach reached the junction, 20 m south of the crossing. The bell isn't audible inside a vehicle from there, and in any case had stopped once the barrier was down. There were three flashing lights, one of which was aimed down to be seen by pedestrians. The other two were aimed along the nearly straight main road, and the far one would have been out of its main beam. The last one, on the near side, could have been hidden behind a pole (for domestic power and/or telephone wires) - its presence was never picked up in assessments. The first light was also hidden from exactly the same viewpoint. Of course the pole blocking a driver's sightline just when they look had a low probability, but even such unlikely things can and do happen.

The main element of the narrative was based partly on the driver's evidence about turning into the main road. The road at that point was rebuilt in 2000/01, when the D612 was upgraded to a designated "route avoiding  Perpignan", so as to have two single-lane carriageways with raised kerbs at each side and at the central island (which is unusual for French rural roads). Getting a coach to turn into this narrow (3.1 m) slot, only just long enough for it, required a lot of attention and looking in the mirror to check the offside read wheel was off its kerb (the nearside front wheel being invisible). The theory was that this task so occupied the driver that she didn't look with attention at the barrier until it was so low down in front of the coach she was unsighted. The nearest flashing light might have been seen, just, once the coach was straight but its hood came so far down at the sides as to block that view. The barrier arm itself was designed to bend easily enough that the effect on the coach wouldn't be felt.

A minute examination of the regulations finds a number of gaps, and contradictions between the railway and road rules. The performance of warning lamps isn't clearly specified, not in brightness nor in angle. The pointing of the lights is defined as if there is only a single road to consider, and while junctions are considered (e.g. in the review checklist), that is mainly for queueing back.

The school (Christian Bourquin) at Millas was built only in 2015, so the route this bus took - and the demanding turn manoeuvre at PN25 - was not used by such large vehicles before that. The road from the school (Chemin du Ralet) was widened at the time specifically for school buses. There wasn't a review of the crossing safety triggered by that, but from the comments it probably wouldn't have considered this aspect anyway. In fact, these reviews are clearly seen as inadequate - the 2009 assessment that the crossing was far enough from any junction or other noteworthy point being labelled "manifestly wrong". That 2009 national review programme of level crossing risks was, of course, itself a response to the Allinges accident in 2008.

There's no mention of the other point sometimes made about single flashing lights, that a quick look may only see them dark between flashes. Presumably that's the rationale for the twin lights of a wig-wag, though I've not seen proof of that. If looks are never that quick, so single lights are not an issue, then twin ones would avoid the "post in the way" problem - but the superiority of wig-wags isn't mentioned either.

One final point that surprised me is that the railway line has never been reopened since the accident, eighteen months ago, being replaced by buses throughout its length.

[SandTEngineer]

Thanks for that excellent summary.  I too missed the issue of the report.  I'll have a good read over the coming weeks as there are always lessons for everybody when an accident like that happens.

[stuving]

And it's getting worse ... the main line from Lyon to Turin, via the Fréjus tunnel, has been closed for "several weeks" by a mud slide (near les Sordieres). Does it really look such a huge heap as that? It wasn't even big enough to cross the railway and invade the autoroute right next to it. This is originally an SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) image:

[eightonedee]

Does it really look such a huge heap as that?

....but it might be because of what's still about to slide down the hill/mountainside above the track that they will need to stabilise first!

[stuving]

Does it really look such a huge heap as that?

....but it might be because of what's still about to slide down the hill/mountainside above the track that they will need to stabilise first!

I think you're right - having just seen some overhead footage, there's a fence further up the slope that's holding back a lot more of that ...  stuff ... so what's on the track is just the overflow. And it has now oozed onto the near carriageway of the road (A43). And there's more rain forecast for the Alps; not on the same scale but it only needs to top up the lubrication.

This of course is the main rail link between France and Italy: three TGVs^▸ (Train a Grande Vitesse), a limited TER service, but a lot of freight (per day each way).

[stuving]

This of course is the main rail link between France and Italy: three TGVs^▸ (Train a Grande Vitesse), a limited TER service, but a lot of freight (per day each way).

Coincidentally, one of the other France-Italy lines was in the news today: Nice-Tende. French trains used to run onward to Turin but no longer do, but Italian ones still run from Ventimiglia into France to join the line and then through to Turin. I imagine displaced TGVs will use the coast line, which might also take goods traffic, though it involves flogging along the full length of  the Riviera. But the gradients, loops, and wiggles of the line to Tende (Train des Merveilles) would put off any heavy train.

The trouble with the line is the usual one - most of the year it has only a handful of passengers. Fares only cover about 7% of costs, and both the region and their audit body think the subsidy has to be reduced. Currently it's the usual SNCF^▸ (Societe Nationale des Chemins de fer Francais - French National Railways) approach of no closures, but more trains will be buses.

That's all happening across France, but this is PACA* - and the region has been having a long bout of arm-wrestling with SNCF over the cost and performance of their services. They threatened an open competition, but in the end found a legal process that let them requisition SNCF in some way rather than negotiating a contract. I've no idea how that works, but after three years they have a new agreement.

Politically PACA is a mixture of extremes - left-wing strongholds in Marseille and Toulon, but right (very in parts) elsewhere. Its transport policy is a mix too, and used to include 1€ bus tickets across the region (but only if you asked for one).

In passing, I note that Nice airport is to be multimodalised, with on-site buses (which used to stop on the road outside), trams, and trains at a station moved from St Augustin.

*Provence-Alpes-Cote d'Azur