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

[Taunton]

The westbound line (which the train was running eastbound on) appears to have notably sharper curves, in fact it looks like two curves separated by a straight, whereas the more northerly eastbound track appears to be a continuous curve. I suppose it makes sense that trains starting out from Strasbourg do not need the highest speed provision, compared to those coming from the LGV.

I wonder what the differential speed limits are, and how any such differential is applied to a train running wrong line. We might also wonder why they were running wrong line in the first place, and how a train at that speed running wrong line was going to be handled onto the conventional layout just a very short distance ahead.

Curious to know why the design splayed the two tracks so far apart. It must have caused significant extra land purchase costs.

 

[edwin_m]

I would guess the splay is to allow any future eastward extension of the high speed alignment, in which case the curve would become the connecting tracks to another flying junction.

I'm also pretty sure the high speed lines are signalled for bi-directional operation (certainly judging by the frequency of crossovers seen from the Eurostar!) and the permitted speed is likely to be close to the geometric limit in both directions. The classic line may also be bi-directional (normal operation on the right in this part of France). However normal operation of the LGV is on the left and I doubt extra money would be spent to change the curvature if the acceleration limitations of trains in the normal direction meant they would not benefit from it.

 

[LNW-GW Joint]

The 176km/h speed limit seemed rather odd, but reading the Railway Gazette version of the accident, it would be the 10% overspeed test for the 160km/h limit at the end of the LGV line.
All lines are bi-directional, so they would have to test all combinations of running.
The train was running eastbound on what is normally the westbound track.
http://www.railwaygazette.com/news/...w/view/test-train-catastrophe-on-lgv-est.html

 

[jon0844]

Has foul play with regards to the track been ruled out, or is the investigation still at a very early stage?

 

[Taunton]

The classic line may also be bi-directional. However normal operation is on the left

Just for clarity of others, SNCF normally run on the left, as in the UK, and unlike French roads (or the Paris Metro). However, in Alsace in eastern France, as here, which used to be Germany until 1918, the old German right-hand running has always been retained, with changeovers where the border used to be. LGV runs on the left throughout, including through Alsace, but has a flyover rollover where it connects with classic Alsace routes to align it correctly, which the TGV here was approaching.

 

[33Hz]

LGV runs on the left throughout, including through Alsace, but has a flyover rollover where it connects with classic Alsace routes to align it correctly, which the TGV here was approaching.

Which you can see in the picture I posted above.

 

[amcluesent]

Was the intended travel as shown in green and the derailment shown in orange? The article says is was the rear power car which ended up in the canal and the disturbed ground suggests the train carriages crossed the track, over the canal and scoured the grass.

I'm perplexed why there are fires at both ends of the bridge, mind.

Just the kinetic energy to propel most of the train clean over the canal and across the field seems at odds with the driver saying he was within speed tolerance.

 

[Taunton]

Was the intended travel as shown in green and the derailment shown in orange?

I presume so as well. I'm therefore not quite certain how the fires at both ends of the canal bridge are accounted for (or how a fire even happened), and whether that is vehicles from the train still on the line at the far side of the bridge.

 

[Murph]

I'm perplexed why there are fires at both ends of the bridge, mind.

Shorted OHLE, perhaps?

 

[Groningen]

With the old line is was so that somewhere between Paris and Strasbourg the left running became right and right became left. And after long searching from Bar le Duc to Strasbourg i finally found it at 48.703223, 7.012582 (the town seems to be Imling) near the moterway N4.

 

[edwin_m]

Just for clarity of others, SNCF normally run on the left, as in the UK, and unlike French roads (or the Paris Metro). However, in Alsace in eastern France, as here, which used to be Germany until 1918, the old German right-hand running has always been retained, with changeovers where the border used to be. LGV runs on the left throughout, including through Alsace, but has a flyover rollover where it connects with classic Alsace routes to align it correctly, which the TGV here was approaching.

I have amended my post to reflect this - it became rather misleading when I edited to add an extra sentence!
--- old post above --- --- new post below ---

I presume so as well. I'm therefore not quite certain how the fires at both ends of the canal bridge are accounted for (or how a fire even happened), and whether that is vehicles from the train still on the line at the far side of the bridge.

This is perplexing me too. I can only think that something near the front of the train fetched up hard against something solid such as the bridge parapet, and this caused the rest of the train to jack-knife off to the left.
--- old post above --- --- new post below ---

Shorted OHLE, perhaps?

While low-voltage DC can arc for some time if there isn't a hard short, 25kV should cut out almost instantly and a fire will only ensue if the momentary arc has ignited something (as it appears to have done on the most recent HGV Channel Tunnel fire reported by RAIB). But a modern electric passenger train shouldn't have anything on board that can sustain that sort of fire. I suggested a diesel generator some posts back but I don't think anyone has come back to say whether one was fitted.

 

[Peter Mugridge]

I presume so as well. I'm therefore not quite certain how the fires at both ends of the canal bridge are accounted for (or how a fire even happened), and whether that is vehicles from the train still on the line at the far side of the bridge.

I count 10 carriages including the power cars in the field so I should think none of the train is still on the track at all - a RGV DAYSE is a 2+8 formation; it looks like it's all come off the line before reaching the bridge and passed to the left of the bridge. Pretty much as per the orange arrow drawn on the picture.

The two fires are puzzling me as well, if that's the case.

But a modern electric passenger train shouldn't have anything on board that can sustain that sort of fire. I suggested a diesel generator some posts back but I don't think anyone has come back to say whether one was fitted.

They don't still use oil bathed transformers anywhere these days do they?

 

[Murph]

While low-voltage DC can arc for some time if there isn't a hard short, 25kV should cut out almost instantly and a fire will only ensue if the momentary arc has ignited something (as it appears to have done on the most recent HGV Channel Tunnel fire reported by RAIB). But a modern electric passenger train shouldn't have anything on board that can sustain that sort of fire. I suggested a diesel generator some posts back but I don't think anyone has come back to say whether one was fitted.

I'm thinking that it almost certainly will have brought the knitting down as it derailed, then ignited miscellaneous debris (either something already there, or fragmented bits of the train). The supply will only trip out if the fault current is large enough and sustained. 25kV laying on the deck could easily dance around sparking for more than an instant if it doesn't sustain a current to earth and trip the circuit.

 

[richw]

From the aerial photo, did the train even make it to the bridge?

 

[Emblematic]

I'm thinking that it almost certainly will have brought the knitting down as it derailed, then ignited miscellaneous debris (either something already there, or fragmented bits of the train). The supply will only trip out if the fault current is large enough and sustained. 25kV laying on the deck could easily dance around sparking for more than an instant if it doesn't sustain a current to earth and trip the circuit.

Looking at the aerial photos posted earlier, it seems all the stanchions on that curve are on the inside, which by good fortune may have minimised the casualties; the northern curve has the stanchions on the outside. I haven't seen any photos of the OHLE being involved.
But not all of the train has derailed clear towards the canal, as this photo shows that part of the train has collided with the bridge parapet, with clear evidence of the immense forces at work. Obviously some evidence of fire as well, quite easy to see how anything combustible there could be ignited.

 

[LNW-GW Joint]

With the old line is was so that somewhere between Paris and Strasbourg the left running became right and right became left. And after long searching from Bar le Duc to Strasbourg i finally found it at 48.703223, 7.012582 (the town seems to be Imling) near the moterway N4.

You are right. http://www.bueker.net/trainspotting/lines_interfaces.php
The actual border was at Avricourt about 15km further west, where frontier formalities were made.
This is still the border between départements in France. http://wikimapia.org/#lang=en&lat=48.647995&lon=6.807747&z=13&m=h&search=baudrecourt

I hadn't realised until researching this, that the Paris-Strasbourg railway was built for left hand running throughout, with the eastern section converted to right hand running by the Germans after 1870.
The French version of the Wiki page on the line is a lot more detailed than the English.
https://fr.wikipedia.org/wiki/Ligne_de_Paris-Est_à_Strasbourg-Ville

 

[Murph]

Looking at the aerial photos posted earlier, it seems all the stanchions on that curve are on the inside, which by good fortune may have minimised the casualties; the northern curve has the stanchions on the outside. I haven't seen any photos of the OHLE being involved.
But not all of the train has derailed clear towards the canal, as this photo shows that part of the train has collided with the bridge parapet, with clear evidence of the immense forces at work. Obviously some evidence of fire as well, quite easy to see how anything combustible there could be ignited.

Yeah, that photo gives an easy explanation for the fire at one end of the bridge. With that level of impact and destruction, I'd say it's pretty likely that some debris (or even an entire vehicle) could have been deflected upwards to bring the OHLE down.

Possible theory for the fire at the far end of the bridge. After that impact, the badly damaged vehicle (presumably the leading power car) could have run along the top of the steel parapet and deposited some additional debris at the far side as it dismounted the bridge and ran down the embankment into the field. the remains of that bogie almost provide a ramp for the vehicle to ride up onto to parapet. I.e. lead power car derails prior to the bridge, then has a catastrophic impact with the bridge parapet, runs on top of the parapet across the bridge, but the shock from that initial impact causes the remainder of the train to depart outwards from the bend, producing the scarring in the field below.

With the talk of those 2 single lines being bidirectional, a question springs to mind. Is there a crossover between the lines to the immediate west of the curves? Could we have a Greyrigg style derailment at that hypothetical crossover?

 

[richw]

With the talk of those 2 single lines being bidirectional, a question springs to mind. Is there a crossover between the lines to the immediate west of the curves? Could we have a Greyrigg style derailment at that hypothetical crossover?

Post 90 has some photos and links of the scene before the crash and I think from the distance taken there is a crossover although can't be 100% sure due to the distance and angle of the shots

If the driving vehicle hit the bridge parapet at such speed, the driver has done well to survive and apparently talking to the police/investigator/reporter already

 

[MarkyT]

The westbound line (which the train was running eastbound on) appears to have notably sharper curves, in fact it looks like two curves separated by a straight, whereas the more northerly eastbound track appears to be a continuous curve.

I measure both curves to be the same at around 900m radius, with a short straight section part way along the inner one as you describe. The outer curve has to describe a longer continuous arc in order to present at the correct angle to pass under the other line again at the junction flyover.

I would guess the splay is to allow any future eastward extension of the high speed alignment, in which case the curve would become the connecting tracks to another flying junction.

A side benefit of the above I think

I'm also pretty sure the high speed lines are signalled for bi-directional operation

That is standard on French LGVs and also applies to our UK HS1, using the same signalling system (TVM 430).

The 176km/h speed limit seemed rather odd, but reading the Railway Gazette version of the accident, it would be the 10% overspeed test for the 160km/h limit at the end of the LGV line.

That would likely be the intervention speed at which the ATP system would cut in to automatically control the speed if the driver hadn't responded to overspeed warnings.

 

[Emblematic]

Possible theory for the fire at the far end of the bridge. After that impact, the badly damaged vehicle (presumably the leading power car) could have run along the top of the steel parapet and deposited some additional debris at the far side as it dismounted the bridge and ran down the embankment into the field. the remains of that bogie almost provide a ramp for the vehicle to ride up onto to parapet. I.e. lead power car derails prior to the bridge, then has a catastrophic impact with the bridge parapet, runs on top of the parapet across the bridge, but the shock from that initial impact causes the remainder of the train to depart outwards from the bend, producing the scarring in the field below.

There was an earlier picture posted on twitter which showed debris below the bridge which I couldn't put in context before. Clearer now that it's shredded vehicle debris that's been hurled across from the other side of the canal from the parapet collision, with very few recognisable fragments.

 

[33Hz]

There was an earlier picture posted on twitter which showed debris below the bridge which I couldn't put in context before. Clearer now that it's shredded vehicle debris that's been hurled across from the other side of the canal from the parapet collision, with very few recognisable fragments.

It's also interesting to see the bar car upside down and apparently burned out, yet others right next to it intact. What caused it to catch fire?

 

[Taunton]

I hadn't realised until researching this, that the Paris-Strasbourg railway was built for left hand running throughout, with the eastern section converted to right hand running by the Germans after 1870.

The railway was built by the French before 1870, but was still relatively simplistic and thus straightforward to convert from left to right hand running when Germany took the province over. Between then and 1918 the rail network, and industrial Alsace generally, developed very substantially, pretty much to what you see today, and thus it was a far greater task to convert back, so it was left as it was. Alsace went back to Germany in 1940, and returned again to France in 1945 (it's been a real political football over the centuries), but the running side has not been altered since 1870. The initial French decision to run on the left, as in several other European countries, was due to British engineers advising on the early construction. The Paris Metro when built took a deliberate decision to have opposite side running, along with a restrictive loading gauge, to thwart attempts by the national rail companies to get running powers.

 

[edwin_m]

I'm thinking that it almost certainly will have brought the knitting down as it derailed, then ignited miscellaneous debris (either something already there, or fragmented bits of the train). The supply will only trip out if the fault current is large enough and sustained. 25kV laying on the deck could easily dance around sparking for more than an instant if it doesn't sustain a current to earth and trip the circuit.

Really? I always understood that the fault current on AC was so much higher than the service current that it was pretty much guaranteed to trip out if it touched anything. Examples being the OLE falling on the platform at St Pancras (ignoring the fact it was then turned back on) and the trip when it touched a radio aerial protruding up from a Tunnel HGV train.

https://www.gov.uk/raib-reports/overhead-line-failure-at-st-pancras-international

https://www.gov.uk/government/news/fire-on-board-a-freight-shuttle-in-the-channel-tunnel-update

 

[Murph]

Really? I always understood that the fault current on AC was so much higher than the service current that it was pretty much guaranteed to trip out if it touched anything. Examples being the OLE falling on the platform at St Pancras (ignoring the fact it was then turned back on) and the trip when it touched a radio aerial protruding up from a Tunnel HGV train.

It certainly should trip out quite quickly, especially since it should land on or close to the rails which are used for the other half of the current path. It's not instant, however, so there's a brief moment where there's a huge amount of energy available to produce heat. Remember that normal service usage could be into the 10s of megawatts (e.g. a 91 or 92 is up to approx. 5MW). Then there's the issues of the trips getting reset, with another brief burst of very high energy each time.

 

[Gordon]

The railway was built by the French before 1870, but was still relatively simplistic and thus straightforward to convert from left to right hand running when Germany took the province over. Between then and 1918 the rail network, and industrial Alsace generally, developed very substantially, pretty much to what you see today, and thus it was a far greater task to convert back, so it was left as it was..

An excellent summary thanks, Taunton. It is a common misconception that the former Alsace Lorraine (A-L) railway network has always been right hand running but it was indeed the German authorities that changed it after 1871.

You are right.
The actual border was at Avricourt about 15km further west, where frontier formalities were made.

Probably due to travelling extensively across Europe as a youngster, I have always been fascinated by border crossings, so I know quite a bit about the A-L border geography.

When the border at Avricourt was created a new station was needed on the German side, which became Deutsch-Avricourt, and a railway settlement grew up around it.
After the return of A-L to France, Deutsch Avricourt became Nouvel-Avricourt. It was closed many years ago now. The French station was called Igney-Avricourt and is still open as such.

A web search for 'Bahnhof Deutsche Avricourt' brings up some excellent links to old postcards

Remarkably the building still (partially) exists to this day - look around on street view and you will see it.

With reference to all the railway borders between ex-AL and the rest of the French railway network, a good list of the position of the Flyovers with exact co-ordinates can be found on this site:
https://fr.wikipedia.org/wiki/Saut-de-mouton


.

 

[Senex]

I measure both curves to be the same at around 900m radius, with a short straight section part way along the inner one as you describe. The outer curve has to describe a longer continuous arc in order to present at the correct angle to pass under the other line again at the junction flyover.

I get the same figure of about 900 m radius for the curves, and some of the photographs that have appeared seem to shew considerable cant on the curves, as would be expected if the permissible speed round the corner and down on to the Haguenau-Strasbourg is indeed the 160 km/h that has been reported. Indeed, if the curve is only about 900 m radius then both cant and cant deficiency must be near the permissible limits.

In connection with the question of cant on the LGVs in France, what is allowed is described thus in the book Le grand livre du TGV which is the official SNCF publication about the LGVs (though now ten years old). On page 108 we read: "Les courbes ont un rayon minimal de 4000 m. Il y a toutefois trois exceptions à 3 250 m [on LGV1 - my note]. Le dévers est de 1 000 x 700 R [my bold] pour les rayons compris entre 4 000 et 20 000 m et de 180 mm pour les rayons compris entre 3 250 et 4 000 m. I'insufficance de dévers est au maximum de 85 mm."
This is a different formulation from what I know from this country, Austria, Germany, Sweden, and Switzerland, and it doesn't make sense to me. If the cant is 1000 x (700 x R), then for a radius of 1000 metres the calculation gives 1000 x 700,000 which is 700 million, which is crazy -- and I can't see a way to re-write the formula to get any sense. Am I making some very basic error, or does there have to be some sort of mis-print here?

 

[edwin_m]

That formula doesn't make sense. For the same speed the cant should be inversely proportional to the radius. If it's 1000/700R and the result is in mm not metres then this would be 175mm at 4000m radius. This lines up with the 180mm for smaller radii, for which presumably a speed restriction would apply. More recent TGV lines have had a higher maximum speed, which may have changed these figures.

 

[NicholasNCE]

SNCF just announced during a press conference that the train derailed at 243 kph when it should have been doing 176 kph, a human error as speed control was disabled for the tests. Braking began approximatively 11 seconds or 1km too late as shown on the speed graph:

The initial report can be found here: http://www.sncf.com/ressources/rapp...ordement_de_la_lgv_ee_le_14_novembre_2015.pdf

 

[notadriver]

I had a feeling he had braked too late.

 

[Senex]

I see they say it is a 945 m radius curve with 163 mm of cant.