Terrible accident at Santiago de Compostela, northern Spain - 24 July 2013

[whole diary]

[James]

Gosh, not good, my prayers go out to all the people who were injured and killed.
Hopefully the authorities get to the bottom of this serious crash, as spain is such a nice country. Get well soon folks .

[TonyK]

No mention yet of the presence of the second driver in the cab. He had apparently driven the first half of the route, and also survived the crash.His evidence will be invaluable in establishing the chain of events.

[DidcotPunter]

I'm not sure if this has already been referenced but this article from the IRJ confirms that (a) the ETCS^▸ (European Train Control System) controlled section of the line ends 4km from the accident site and (b) the Alvira type train involved in the accident are not (as yet) able to use ETCS on that section of the line. 

http://www.railjournal.com/index.php/europe/etcs-not-operable-on-santiago-crash-train.html

Therefore the train was under full manual control for the complete journey from Ourense to Santiago under the Spanish ASFA system which, like our AWS^▸ (Automatic Warning System), monitors acknowledgement of the lineside signals but (unlike TPWS^▸ (Train Protection and Warning System)) does not provide any form of speed supervision. So, other than the driver's vigilance, which appears to be lacking in this unfortunate accident, there was no other system in place to prevent trains entering the curve at excessive speed.

As the article states, even trains which are equipped with ETCS are only advised that automated supervision is disabled at the ETCS/ASFA transition point, there is nothing other than the driver's route knowledge to enforce braking prior to the curve.  As we found out at Morpeth on a number of occasions, this appears to be an accident waiting to happen.

[stuving]

That is rather surprising, if true - certainly it contradicts some other sources in its details. I assume this still counts as a journalistic source, even if it is of the more reliable technical kind. We will just have to wait for more definitive information, I guess.

On the other hand it does confirm what I was finding already, that saying a line or a train is equipped with system x does not tell you whether and how it is used.

[DidcotPunter]

Indeed.  AWS^▸ (Automatic Warning System) did not prevent the Southall accident (it was isolated on the HST^▸ (High Speed Train)) and ATP^▸ (Automatic Train Protection) didn't prevent the Ladbroke Grove one (fitted and operational on the HST, not fitted to the class 165 Thames Turbo unit involved).

[ChrisB]

No mention yet of the presence of the second driver in the cab.

The BBC^» (British Broadcasting Corporation - home page) reported he wasn't in the cab, but sat in one of the coaches.

[eightf48544]

Indeed.  AWS^▸ (Automatic Warning System) did not prevent the Southall accident (it was isolated on the HST^▸ (High Speed Train)) and ATP^▸ (Automatic Train Protection) didn't prevent the Ladbroke Grove one (fitted and operational on the HST, not fitted to the class 165 Thames Turbo unit involved).

Don't forget that at Southall the ATP was also out of use as well.

As stuving puts it.

On the other hand it does confirm what I was finding already, that saying a line or a train is equipped with system x does not tell you whether and how it is used.

This crash should make European Railway authorites more aware of the the problems of boundary conditions between old and new sytems and increasingly between software versions of similar systems. Just look at the problems the Belgians and Dutch have had around the boundary point on the High Speed line with different software versions of ERTMS^▸ (European Rail Traffic Management System.) 2 on either side of the border.

Thi will start to become a problem in this country as ERTMS evolves how will Cambrian ERTMS Fitted 158s interface with what will undoubtably be a more modern version when installed between Shrewsbury and Birmingham?

It's not the same as an AWS TPWS^▸ (Train Protection and Warning System) fitted unit working on a line without AWS or TPWS if a line is Level 2 and above with no linside signals then all stock using the line must be able to interface with the system whatever software version is in use.

Do you stick with the software version installed on the first line or do you take what are hopefully safety/performance advances in later software version for the next instalaltion? So do you upgrade the first system to new the version or do you upgrade the trains to read both (multiple?) versions (the SNCB/NS problem) or do you end up with increasing isolated train fleets confined to one route?   

You can liken it to having trains equipped with Windows 3, XP, Vista, 7 and 8 at the same time.

Not an easy problem.

[broadgage]

In the UK^▸ (United Kingdom) speed limits are strictly enforced with drivers at risk of dismissal for all but the most trivial breaches.
This is however fairly recent, I can remember trains routinely exceeding speed limits, was there not a "140 club" years ago of those HST^▸ (High Speed Train) drivers who had reached 140 MPH.

I'm not sure that's true Broadgage.  I was around when HST's came in on the Western region in1976, and I recall they all had speed limiters either from the outset of public service runs or very soon after which prevented them doing more than about 130 mph.  It maybe that early driver training runs etc pre-dated the speed limiters. 

I do remember doing 110 mph behind a Class 50 between Didcot and Reading: unlike Class 47's, which ran out of power above about 80-85 mph, Class 50's could really go!

AFAIK^▸ (as far as I know) the HSTs did not have speed limiters originaly, they were fitted later due to concerns about "over enthuisiasm"

[bobm]

BBC^» (British Broadcasting Corporation - home page) are reporting the train driver was on the phone to members of railway staff at the time of the derailment.

[stuving]

... and French TV that he was looking at a plan or map. Not that these are exclusive.

I wonder where these reports come from. Some are attributed to "tribunal", so does that mean this kind of evidence is in effect read into a public record as they go along? Otherwise, one would expect the judge-led investigation to be strictly confidential - in theory.

[onthecushions]

It struck me that the Eastern approach to Reading (down main) also had a 125mph limit into (old) platform 4, limit 50mph.

IIRC^▸ (if I recall/remember/read correctly) this was in recent years always approach controlled, usually with single yellow aspect and time in section. I believe that the criterion was line speed above 75mph and more than 2/3 speed reduction.

The Morpeth curve already mentioned seems to have had 5 smashes, three of which were down to over-speeding, (in 1969, 1984 and 1994). At least one had a distracted driver.

Sadness

OTC

[Chris from Nailsea]

More details, from the BBC^» (British Broadcasting Corporation - home page):

Spain train driver 'on phone' at time of deadly crash

The train driver in last week's crash in Spain was talking on the phone when it derailed, investigators say.

The train was travelling at 153km/h (95mph) at the time, investigators at the Court of Justice of Galicia said.

Francisco Jose Garzon Amo was speaking to members of staff at the state-owned railway company, Renfe, they added.

Crash investigators had opened the train's "black-box" data recorder to find the cause of the crash, which left 79 people dead.

Moments before the accident the train was travelling at a speed of 192km/h (119mph), the court said in a statement.

Investigators say the brakes were activated shortly before the crash. The speed limit on the sharp bend where the train derailed was set at 80km/h (49mph).

"Minutes before the train came off the tracks he received a call on his work phone to get indications on the route he had to take to get to Ferrol. From the content of the conversation and background noise it seems that the driver consulted a map or paper document," a court statement said.

Mr Garzon is suspected of reckless homicide, but he has not yet been formally charged. He was released from custody in Santiago de Compostela, where the crash occurred, on Sunday but remains under court supervision. He must appear before a court once a week and was not allowed to leave Spain without permission. His passport has been surrendered to the judge and his licence to drive a train has been suspended.

Under Spanish law, his legal status is that he is suspected of being involved in 79 counts of reckless homicide but has not been formally charged. But officials said he had admitted negligence by being careless when rounding a bend too fast.

All eight carriages of the train careered off the tracks into a concrete wall as they sped around the curve on the express route between Madrid and the port city of Ferrol on the Galician coast.

[stuving]

This English report from El Pais http://elpais.com/elpais/2013/07/31/inenglish/1375288407_411161.html refers to the phone call the driver was making when he missed his braking point:

Antonio Mart^n told EL PA^▸ (Public Address)^S that he called driver Francisco Jos^ Garz^n on the corporate phone to give him instructions on how to arrive at Pontedeume in A Coru^a. The call was made as Garz^n entered the sharp A Grandeira curve in Santiago's Angrois neighborhood at double the speed limit for the stretch of track.

Mart^n said the reason he made the call was to make it easier for a family with children who were traveling on the train to get off. The supervisor told police that he was in carriage number three and was accompanied by a rail security officer.

Renfe regulations prohibit anyone from calling the drivers of a train except for emergency purposes while they are operating a rail service.

Martin is described variously as a RENFE^▸ (Spanish National Railways (Red Nacional de Ferrocarriles Espanoles)) controller or supervisor, but where he fits into the organisation is not clear. The bit about "making it easier" is explained in the original as a choice between paths into the station; but there is a RENFE quote making the obvious point that neither driver nor supervisor had any control over this.

El Pais is also credited with releasing an audio tape of part of Garz^n's interview with the judge, which been played on TV news. Now I know he has not been formally charged, but I still find this degree of openness very hard to square with his rights if and when he is. 

[broadgage]

Both openness, and keeping things confidential have their merits.

In the UK^▸ (United Kingdom), criminal prosecutions are normally held in open court, anyone including the press, may make use of the public gallery and report as to whom said what.
Inquiries into serious accidents are sometimes held in view of the public via either the coroners court, or a public inquiry.

Other inquiries, especially if potentially leading to disciplinary action against an individual are normally private.

It would seem that the Spanish approach favours more openness than the UK, this may or may not be a good thing, but should not be criticised for being simply different to the UK approach.
There is very considerable interest in this awful accident and the causes thereof, and in my view openness as to the on-going inquiries and evidence revealed is a good thing, it will certainly reduce the amount of ill informed speculation and rumour.

[stuving]

There's some further details from El Pa^s in English: http://elpais.com/elpais/2013/08/02/inenglish/1375456623_554470.html
In part this says:

Garz^n, who was going at twice the speed limit, heard an automatic signal alerting the driver that he was approaching the A Grandeira curve, where the crash occurred, while he was talking on the phone, according to EFE, which obtained a copy of a police summary report on the black boxes.

The train, which was traveling from Madrid to Ferrol, derailed at 8.41pm while it was speeding at 179km/h. Some four second later the driver threw the emergency brake, which dropped the speed to 153km/h, but the 13 rail cars had already begun to jump the tracks.

There is also a much longer item in Spanish, which despite its technical content translates to make some sense (at least in Chrome): http://politica.elpais.com/politica/2013/08/02/actualidad/1375462076_167730.html.

If I have got this right, it says that the ASFA protection systems always had the capability to apply emergency braking (presumably to a halt) when a track speed limit is exceeded, on all equipped trains. For some reason this had never been seen as worth doing, and even now it takes a crash programme to implement it at specific high-risk locations. I presume this is because the ASFA balises and warning signs are not there, or not in the right places.

For the Angrois curve they are doing it belt and braces, with three successive speed traps ending at 30 km/hr. How this fits with the first report of a (speed**) alarm, and in particular its timing, is not clear from the rather garbled text. I found a longer report that gives two different times and positions for the brake application, so the confusion is not just in this translation.

(PS: ** I've re-read the second piece, and it refers to an audible announcement of the next signal's identifier and its being clear - i.e. a reminder of where he has reached, but nothing specifically about the speed limit.)