TGV derailed Paris - Strasburg (test train not in passenger service)

Status
Not open for further replies.

LNW-GW Joint

Veteran Member
Joined
22 Feb 2011
Messages
14,128
Location
Mold, Clwyd
Apparently there were 7 people in the cab at the time.
The 2 drivers and the traction inspector have been suspended.
Test runs have also been suspended (not surprising, as the test train has been largely destroyed).
Looks like a very lax regime was allowed on the train.
Railway Gazette's piece: http://www.railwaygazette.com/news/...-braking-caused-tgv-derailment-says-sncf.html
Speaking at a press conference on November 19, Pepy outlined a series of measures to be implemented immediately. These cover four main areas: the management of test runs, organisational and human factors, disciplinary procedures against the individuals involved, and wider safety improvements.

Specific steps include the suspension of high speed test runs pending completion of an inquiry into how these are managed, and much stricter controls on access to test trains. Pepy said that the ongoing investigation would examine the ‘irresponsible’ presence of children on the train, as well as the interaction between people in the cab and those in the trailer cars
 
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R

RailUK Forums

33Hz

Member
Joined
2 Dec 2010
Messages
483
I'm still puzzled as to why the signaling system was turned off, as opposed to programmed to allow the higher speed running.
 

AlexNL

Established Member
Joined
19 Dec 2014
Messages
1,349
Reprogramming is not as simple as it may sound. The entire line and its signalling system are designed for a commercial speed of 320 km/h. Changing that means that block length and calculation of headways and braking curves are also calculated based on that parameter, which has to be severely revised when the maximum speed is raised.

Furthermore, the system will then have to be (re)tested before it can be used to protect a test train. And how do you test the signalling system if the trainside equipment has not been certified for use at that speed?
 

Murph

Member
Joined
16 Feb 2010
Messages
725
Reprogramming is not as simple as it may sound. The entire line and its signalling system are designed for a commercial speed of 320 km/h. Changing that means that block length and calculation of headways and braking curves are also calculated based on that parameter, which has to be severely revised when the maximum speed is raised.

Furthermore, the system will then have to be (re)tested before it can be used to protect a test train. And how do you test the signalling system if the trainside equipment has not been certified for use at that speed?
The obvious thing would seem to be to adjust the train's software, not the entire signalling system, increasing the threshold before it steps in with a brake application. I'm not suggesting that it's necessarily easy to do that, but it seems like a goal that could be looked at for future overspeed testing. Being safety critical software, if adjustments to the code itself would be needed, that's going to make it a non trivial task, but fundamentally some sort of flag saying "allow 10% overspeed for testing" should really be possible (and the flag well protected so that it can never be casually or routinely enabled by driver or depot staff).
 

notadriver

Established Member
Joined
1 Oct 2010
Messages
3,296
Why would a driver casually or routinely operate such a switch ? They can already disable it completely if they want to (!)
 

Murph

Member
Joined
16 Feb 2010
Messages
725
Why would a driver casually or routinely operate such a switch ? They can already disable it completely if they want to (!)
They wouldn't, but I said that on the same basis that various safety-relevant isolation switches are commonly protected by a seal or similar. Those seals are there to prevent casual or routine use.
 

notadriver

Established Member
Joined
1 Oct 2010
Messages
3,296
They wouldn't, but I said that on the same basis that various safety-relevant isolation switches are commonly protected by a seal or similar. Those seals are there to prevent casual or routine use.

Really ? I didn't know that. Thanks for enlightening me.
 

Hophead

Member
Joined
5 Apr 2013
Messages
942
I would be interested to know the time or distance that the overspeed test is intended to operate. Surely, once you've gone, say, 5km - 10 max, the job's done and it's back to linespeed?

Was this incident a one-off, or evidence of a fairly laissez-faire attitude to safety during testing?

I also wonder whether British-style route knowledge is expected for the LGVs (that's what the computer is for), so perhaps the driver was not sufficiently aware of the impending tight curve (although one of the crowd who were also present in the cab could possibly have piped up).
 

AlexNL

Established Member
Joined
19 Dec 2014
Messages
1,349
The overspeed testing is required for the entire line, because the entire line is designed for a commercial speed of 320 km/h and the effects of going harder than that should be tested along the entire line.

There are sensors all over the train that measure everything related to how the train and the track operates. If the test at 320+10% would only be performed on a short section of track, potential flaws or defects in other track sections might go unnoticed.
 

LNW-GW Joint

Veteran Member
Joined
22 Feb 2011
Messages
14,128
Location
Mold, Clwyd
I also wonder whether British-style route knowledge is expected for the LGVs (that's what the computer is for), so perhaps the driver was not sufficiently aware of the impending tight curve (although one of the crowd who were also present in the cab could possibly have piped up).
From the report (p14), the driver had been driving TGVs for 7 years and had been running test trains over the new section of the LGV for the 6 weeks since the tests started on Sept 28.
The second driver had 9 years on TGVs and had done 2 weeks driving on the new line.
It's not likely they did not know the geography of the line.
 

JonathanP

Member
Joined
1 Aug 2008
Messages
214
Location
Berlin, Germany
The SNCF test plan called for 200 runs at gradually increasing speeds between September and December, so if they were already running at Max+10% I imagine a lot more running at these speeds was planned.

I don't know how the rules work in France, but in Germany there is still a requirement for drivers to have route knowledge under normal circumstances, but because speed rather than route signalling is used drivers are allowed to drive over routes they do not have knowledge for at a reduced speed, for instance if there is an emergency diversion. In any case it would be very surprising if they would use a driver who didn't know what he was doing for something like this.

The report talks of there being 7 people in the cab, of a loud noise level in the cab and of obstructions to the view of the driver. In those circumstances it's not surprising the braking was a few seconds late, which was sadly all it took.
 

33Hz

Member
Joined
2 Dec 2010
Messages
483
Surely 10% over-speed testing of the system includes the safety systems?

What happens if a temporary speed restriction is in place, thus requiring reprogramming the other way?


What was done for the Eurostar overspeed testing on HS1?
 

MarkyT

Established Member
Joined
20 May 2012
Messages
4,285
Location
Torbay
I'm still puzzled as to why the signaling system was turned off, as opposed to programmed to allow the higher speed running.
Had the signalling been finished and fully commissioned? The track circuits can't be connected and set up for example until the civil engineering and tracklaying activities are complete so it's certainly possible train testing on a 'one engine in steam' principle could have begun before signalling was complete. In the absence of the final signalling to enforce speed limits I wonder if they could have set up a temporary emergency speed trap at the crucial point, perhaps using beacons of the legacy French train protection system, KVB. I don't know enough about the system as to whether it could be used or configured for the speeds concerned but very similar ETCS balises are definitely capable of being read at full TGV speed. A train at a transition point must be running in KVB in order to switch to TVM so all TGVs are equipped with it, and it's also necessary for running on the classic network, which all services do to an extent at their extremities. Even Eurostar uses KVB at St Pancras because that area with its low speed and complexity was more appropriately signalled using colour lights rather than cab signalling. On the connecting lines between the UK classic network and TVM430 fitted HS1 and Eurotunnel lines, the first 'foreign' infrastructure encountered transitioning from UK signalling is a KVB beacon, necessary to initialise the TVM. I think the KVB system and balise reader remains active when running in TVM and the system is used for switching TVM off again at the end of high speed infrastructure when joining a classic route.
 
Last edited:

Kentish Paul

Member
Joined
25 Apr 2012
Messages
454
Location
Ashford Kent
Surely 10% over-speed testing of the system includes the safety systems?

What happens if a temporary speed restriction is in place, thus requiring reprogramming the other way?

What was done for the Eurostar overspeed testing on HS1?
I assume the same as 186mph + 10% = 204.6mph which is near enough the UK rail speed record set by a Eurostar set prior to service introduction.
 

edwin_m

Veteran Member
Joined
21 Apr 2013
Messages
18,147
Location
Nottingham
Surely 10% over-speed testing of the system includes the safety systems?
If fully operational the safety systems will intervene before the train gets anywhere near a 10% overspeed.

Presumably the intention was to run round the curve at 10% above the speed permitted there, but the only practical way of doing this would be to isolate the speed supervision system completely so the driver had full control over the speed of the train. The train would probably not be capable of going fast enough to risk overturning on the true high-speed sections of the route, but the sudden transition to a much lower speed curve would be a different matter.

This is in many ways a similar situation to Santiago de Compostela. Both have a long high-speed section, leading directly to a curve where the safe speed is very much lower. In the Spanish case train protection was not provided because the curve was not considered to be part of the high speed line, and in this accident train protection was probably provided but was suspended for test purposes. In both cases a delay of a few seconds in braking resulted in the curve being entered far faster than was safe.
 
Last edited:

AlexNL

Established Member
Joined
19 Dec 2014
Messages
1,349
Surely 10% over-speed testing of the system includes the safety systems?
In normal situations the safety systems do not allow overspeeding by 10%. On HSL-Zuid (the Dutch HSR), the ETCS system intervenes when a Thalys goes faster than 305 km/h.

What happens if a temporary speed restriction is in place, thus requiring reprogramming the other way?
TVM works by means of track circuits: a frequency is placed on the rails and receivers on board of the train pick up this signal and interpret it. One of the bits of the transferred information is the "network code", an identifier for the characteristics of the infrastructure.

This network code allows the on-board system to identify what the maximum permissible speed is (300 on LGV Nord and High Speed 1, 320 on LGV Est, 270 on LGV Sud-Est, 160 in the Chunnel, and so on) and it influences how the frequencies emitted by the track circuits are interpreted.

TVM supports temporary speed restrictions, as far as I know this can be done without reprogramming. The TVM system can enforce speeds of 320, 300, 270, 230, 170, 80 and 0 km/h.

To have a different speed limit enforced by the TVM system (such as 360), a new network code should be introduced on the trackside equipment and also in the onboard equipment. This is a significatn amount of work, not worth the effort.

With ERTMS it would have been easier to change these parameters, I still doubt that the effort would have been taken to do so as this too would require commissioning - while the speeds are only reached during testing.

What was done for the Eurostar overspeed testing on HS1?
Probably the same as on LGV Est: with disabled safety systems.
 

Polarbear

Established Member
Joined
24 May 2008
Messages
1,588
Location
Birkenhead
I understand that the extension where the accident took place was due to open on 3 April, but was deferred. Is there any indication when the extension may open?

I'm planning a European trip in late April & journey planners are currently showing the faster schedules, which given the circumstances would seem to be a bit optimistic.
 

30907

Established Member
Joined
30 Sep 2012
Messages
8,686
Location
Airedale
I've seen references to early July (SNCF change to summer timetable then) but I don't think that's official.

Bookings aren't open yet on the SNCF site, so there's nothing you can do really.

I'd suggest planning your journey on the current timetable, but check that onward connections will still work if the new one, or a version of it, is introduced (IIRC there are differences greater than the basic 30 minute acceleration in some cross-border and cross-country services).

That way you're covered!
 

LNW-GW Joint

Veteran Member
Joined
22 Feb 2011
Messages
14,128
Location
Mold, Clwyd
I understand that the extension where the accident took place was due to open on 3 April, but was deferred. Is there any indication when the extension may open?
I'm planning a European trip in late April & journey planners are currently showing the faster schedules, which given the circumstances would seem to be a bit optimistic.
The LGV test programme has been cancelled until the investigation into the derailment is completed (and the test train was destroyed).
So there is no date for introducing the new service.
 

33Hz

Member
Joined
2 Dec 2010
Messages
483
The LGV test programme has been cancelled until the investigation into the derailment is completed (and the test train was destroyed).
So there is no date for introducing the new service.

Is there a source for that, as others are still saying first half of this year. I've been looking for news but can't see anything in the English or French press.
 

Murph

Member
Joined
16 Feb 2010
Messages
725
Is there a source for that, as others are still saying first half of this year. I've been looking for news but can't see anything in the English or French press.
Do those saying it will open in the first half of this year have a reliable source post-accident. It's very clear that the unique test train was almost completely destroyed before the testing was complete, you should not have any difficulty in finding reports which satisfy you of that fact. The specialist test equipment inside the test train was also likely destroyed. The line can't be opened until testing is complete, and specially equipped test trains are not quickly thrown together in a few days. Taking all that together, and the complete closure of the line in the immediate aftermath of the accident, it seems very unlikely to me that there would be less than a 3 month delay in the opening (anything more than a 3 month delay pushes it out of first half of the year).
 

eastwestdivide

Established Member
Joined
17 Aug 2009
Messages
1,849
An article (in French) in La Tribune, dated 1 December, is quoting a delay until summer 2016, probably July, according to "internal SNCF sources"

http://www.latribune.fr/entreprises...g-mise-en-service-en-juillet-2016-531946.html

Prévue initialement le 3 avril, la mise en service du deuxième tronçon de la LGV Est devrait être décalée à l'été 2016 selon des sources internes à la SNCF

En interne à la SNCF on table aujourd'hui sur l'été prochain, probablement en juillet selon certaines sources, même si la date restera à confirmer
(my bold)
 

Taunton

Established Member
Joined
1 Aug 2013
Messages
4,551
It's very clear that the unique test train was almost completely destroyed before the testing was complete
I don't know whether the fatalities involved the key test engineers, but in other comparable incidents the loss of a number of such staff has been the most important aspect. The German Maglev accident was an example of this (it was pretty much abandoned afterwards) and there have been some in aviation as well where the loss of a number of main team members in an accident is a major issue.
 

Gordon

Member
Joined
11 Feb 2011
Messages
951
Location
Surrey
.

July would be the earliest date I suspect. Quite possibly later.

There is a lot to sort out as various operators in several countries will be affected by the delay. The delay affects timetable planning, stock utilisation plans (e.g. traditional trains that were to be replaced by new TGV links, connections planned to be introduced to connect with the new TGV trains that will not now start running), reservation systems etc etc

As has been reiterated in recent posts the main test unit specifically created for the testing, a lot of equipment, and key staff were all lost on this incident. Much less serious losses or railway incidents have led to projects being put back much longer than a few months.



.
 

k-c-p

Member
Joined
22 Jan 2013
Messages
146
An article in "L'Express" has some more details
- Timetables are currently in the making and will be published in April
- gain in travel time will be less than the planned 30min
- the track were the accident took place will remain closed. Trains will single track
- when the track is cleared for repairs by the authorities, the works will take about 7 month

Complete article in Frensh:
La LGV Est-Européenne entre Paris et Strasbourg sera mise en service le 3 juillet
 
Status
Not open for further replies.

Top